Sie sind auf Seite 1von 549

Bosch Professional Automotive


Konrad Reif Ed.

Automotive Networking · Driving
Stability Systems · Electronics
Bosch Professional Automotive Information
Bosch Professional Automotive Information is a definitive reference for
automotive engineers. The series is compiled by one of the world´s largest
automotive equipment suppliers. All topics are covered in a concise but
descriptive way backed up by diagrams, graphs, photographs and tables
enabling the reader to better comprehend the subject.
There is now greater detail on electronics and their application in the motor
vehicle, including electrical energy management (EEM) and discusses the
topic of intersystem networking within vehicle. The series will benefit
automotive engineers and design engineers, automotive technicians in
training and mechanics and technicians in garages.
Konrad Reif

Automotive Mechatronics
Automotive Networking, Driving Stability
Systems, Electronics
Prof. Dr.-Ing. Konrad Reif
Duale Hochschule Baden-Württemberg
Friedrichshafen, Germany

ISBN 978-3-658-03974-5 ISBN 978-3-658-03975-2(eBook)

DOI 10.1007/978-3-658-03975-2

Library of Congress Control Number: 2014946887

Springer Vieweg
© Springer Fachmedien Wiesbaden 2015
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,
reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication
or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965,
in its current version, and permission for use must always be obtained from Springer. Violations are liable
to prosecution under the German Copyright Law.

The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are exempt from the relevant protective laws
and regulations and therefore free for general use.

Printed on acid-free paper

Springer Vieweg is part of Springer Science+Business Media
Foreword V

▶ Foreword

As the complexity of automotive vehicles increases this book presents operational

and practical issues of automotive mechatronics. It is a comprehensive introduction
to controlled automotive systems and provides detailed information of sensors for
travel, angle, engine speed, vehicle speed, acceleration, pressure, temperature, flow,
gas concentration etc. The measurement principles of the different sensor groups are
explained and examples to show the measurement principles applied in different

Complex technology of modern motor vehicles and increasing functions need a

reliable source of information to understand the components or systems. The rapid
and secure access to these informations in the field of Automotive Electrics and Elec-
tronics provides the book in the series “Bosch Professional Automotive Information”
which contains necessary fundamentals, data and explanations clearly, systemati-
cally, currently and application-oriented. The series is intended for automotive pro-
fessionals in practice and study which need to understand issues in their area of work.
It provides simultaneously the theoretical tools for understanding as well as the
VI Contents

▶ Contents

2 Basics of mechatronics 165 Overview of the physical effects for sensors

2 Mechatronic systems and components 167 Overview and selection of sensor
4 Development methods technologies
6 Outlook
168 Sensor measuring principles
8 Architecture 168 Position sensors
8 Overview 195 Speed and rpm sensors
11 Vehicle system architecture 207 Acceleration sensors
212 Pressure sensors
18 Electronic control unit 215 Force and torque sensors
18 Operating conditions 224 Flowmeters
18 Design 230 Gas sensors and concentration sensors
18 Data processing 234 Temperature sensors
22 Digital modules in the control unit 244 Imaging sensors (video)
26 Control unit software
30 Software Development 246 Sensor types
246 Engine-speed sensors
44 Basic principles of networking 248 Hall phase sensors
44 Network topology 249 Speed sensors for transmission control
48 Network organization 252 Wheel-speed sensors
50 OSI reference model 256 Micromechanical yaw-rate sensors
52 Control mechanisms 259 Piezoelectric “tuning-fork” yaw-rate sensor
260 Micromechanical pressure sensors
56 Automotive networking 262 High-pressure sensors
56 Cross-system functions 263 Temperature sensors
57 Requirements for bus systems 264 Accelerator-pedal sensors
59 Classification of bus systems 266 Steering-angle sensors
59 Applications in the vehicle 268 Position sensors for transmission control
61 Coupling of networks 271 Axle sensors
61 Examples of networked vehicles 272 Hot-film air-mass meters
275 Piezoelectric knock sensors
70 Bus systems 276 SMM acceleration sensors
70 CAN bus 278 Micromechanical bulk silicon acceleration
84 LIN bus sensors
90 Bluetooth 279 Piezoelectric acceleration sensors
100 MOST bus 280 iBolt™ force sensor
111 TTP/C 282 Torque sensor
124 FlexRay 283 Rain/light sensor
136 Diagnosis interfaces 284 Two-step Lambda oxygen sensors
288 LSU4 planar wide-band lambda oxygen
144 Automotive sensors sensor
144 Basics and overview
147 Automotive applications 290 Electric Actuators
150 Details of the sensor market 290 Electromechanical actuators
151 Features of vehicle sensors 295 Fluid-mechanical actuators
152 Sensor classification 296 Electrical machines
154 Error types and tolerance requirements
155 Reliability 302 Electrohydraulic Actuators
158 Main requirements, trends 302 Application and Function
Contents VII

302 Requirements 404 Hydraulic modulator

303 Design and Operating Concept 404 Development history
304 Actuator Types 405 Design
313 Simulations in Development 408 Pressure modulation

316 Electronic Transmission Control 412 Sensotronic brake control (SBC)

316 Drivetrain Management 412 Purpose and function
317 Market Trends 414 Design
318 Control of Automated Shift Transmission 414 Method of operation
322 Control of Automatic Transmissions 416 Overview of common-rail systems
338 Control of Continuously Variable 416 Areas of application
Transmission 417 Design
340 ECUs for Electronic Transmission Control 418 Operating concept
347 Thermo-Management 422 Common-rail system for passenger cars
349 Processes and Tools Used in 427 Common-rail system for commercial
ECU Development vehicles

350 Modules for Transmission Control 430 High-pressure components of common-rail

350 Application system
351 Module Types 430 Overview
432 Injector
354 Antilock Braking System (ABS) 444 High-pressure pumps
354 System overview 450 Fuel rail (high-pressure accumulator)
356 Requirements placed on ABS 451 High-pressure sensors
357 Dynamics of a braked wheel 452 Pressure-control valve
358 ABS control loop 453 Pressure-relief valve
362 Typical control cycles
454 Electronic Diesel Control (EDC)
370 Traction Control System (TCS) 454 System overview
370 Tasks 456 Common-rail system for passenger cars
370 Function description 457 Common-rail system for commercial
372 Structure of traction control system (TCS) vehicles
373 Typical control situations 458 Data processing
374 Traction control system (TCS) for four 460 Fuel-injection control
wheel drive vehicles 468 Lambda closed-loop control for
passenger-car diesel engines
378 Electronic Stability Program (ESP) 473 Torque-controlled EDC systems
378 Requirements 476 Data exchange with other systems
379 Tasks and method of operation 477 Serial data transmission (CAN)
380 Maneuvers
388 Closed-loop control system and controlled 478 Active steering
variables 478 Purpose
478 Design
394 Automatic brake functions 480 Method of operation
394 Overview 481 Safety concept
396 Standard function 481 Benefits of active steering for the driver
398 Additional functions
VIII Contents

482 Drive and adjustment systems 496 Electromagnetic compatibility (EMC) and
482 Power windows interference suppression
483 Power sunroofs 496 EMC ranges
484 Seat and steering column adjustment 497 EMC between different systems in the
485 Heating, ventilation and air conditioning 504 EMC between the vehicle and its
485 Electronic heater control surroundings
485 Electronically controlled air conditioning 508 Guarantee of immunity and interference
system suppression

488 Vehicle security systems 510 Fault diagnostics

488 Acoustic signaling devices 510 Monitoring during vehicle operation
489 Central locking system (on-board diagnosis)
490 Locking systems 513 On-board diagnosis system for passenger
494 Biometric systems cars and light-duty trucks
520 On-board diagnosis system for heavy-duty
Authors IX

 Authors

Basics of mechatronics Dipl.-Ing. Christian Gerhardt,

Dipl.-Ing. Hans-Martin Heinkel,
 Dipl.-Ing. Klaus Miekley,

Dr.-Ing. Klaus-­Georg Bürger. Dipl.-Ing. Roger Frehoff,

Dipl.-Ing. Martin Mast,

Architecture Dipl.-Ing. (FH) Bernhard Bauer,

Dr. phil. nat. Dieter Kraft,
 Dr. Michael Harder,

Dipl.-Ing. Stefan Mischo. Dr.-Ing. Klaus Kasten,

Dipl.-Ing. Peter Brenner, ZF Lenksysteme GmbH,

Electronic control units Schwäbisch Gmünd,

Dipl.-Ing. Martin Kaiser, Dipl.-Ing. Frank Wolf,

Dr. rer. nat. Ulrich ­Schaefer, Dr.-Ing. Johann ­Riegel.

Dipl.-Ing. (FH) Gerhard Haaf.

Electric Actuators
Basic principles of networking
Dr.-Ing. Rudolf Heinz,
Automotive networking
Dr.-Ing. Robert Schenk.
Bus systems
Dipl.-Ing. Stefan Mischo,
Electrohydraulic Actuators
Dipl.-Ing. (FH) Stefan Powolny,
Electronic Transmission Control
Dipl.-Ing. Hanna Zündel,
Modules for Transmission Control
Dipl.-Ing. (FH) Norbert Löchel,
Dipl.-Ing. D. Fornoff,
Dipl.-Inform. Jörn Stuphorn,
D. Grauman,
Universität Bielefeld,
E. Hendriks,
Dr. Rainer Constapel, Daimler AG Sindelfingen,
Dipl.-Ing. T. Laux,
Dipl.-Ing. Peter Häussermann,
Dipl.-Ing. T. Müller,
Daimler AG Sindelfingen,

Dipl.-Ing. A. Schreiber,
Dr. rer. nat. Alexander Leonhardi,
Dipl.-Ing. S. Schumacher,
Daimler AG Sindelfingen,
Dipl.-Ing. W. Stroh.
Dipl.-Inform. Heiko Holtkamp,
Universität Bielefeld.
Antilock Braking System (ABS)
Traction Control System (TCS)
Automotive sensors
Electronic Stability Program (ESP)
Sensor measuring principles
Automatic brake functions
Sensor types
Hydraulic modulator
Dr.-Ing. Erich Zabler,
Dipl.-Ing. Friedrich Kost
Dr. rer. nat. Stefan Fink­beiner,
(Basic Principles of Vehicle Dynamics),
Dr. rer. nat. Wolfgang Welsch,
Dipl.-Ing. Heinz-Jürgen Koch-Dücker
Dr. rer. nat. Hartmut Kittel,
(Antilock Braking Systems, ABS),
Dr. rer. nat. Christian Bauer,
Dr.-Ing. Frank Niewels and
Dipl.-Ing. Günter Noetzel,
Dipl.-Ing. Jürgen Schuh
Dr.-Ing. Harald Emmerich,
(Traction Control Systems, TCS),
Dipl.-Ing. (FH) Gerald Hopf,
Dipl.-Ing. Thomas Ehret
Dr.-Ing. Uwe Konzelmann,
(Electronic Stability Program, ESP),
Dr. rer. nat. Thomas Wahl,
Dipl.-Ing. (FH) Jochen Wagner
Dr.-Ing. Reinhard Neul,
(Automatic Brake Functions),
Dr.-Ing. Wolfgang-Michael Müller,
Dipl.-Ing. (FH) Ulrich Papert
Dr.-Ing. Claus Bischoff,
(Wheel-Speed Sensors),
Dr. Christian Pfahler,
Dr.-Ing. Frank Heinen and
Dipl.-Ing. Peter Weiberle,
Peter Eberspächer
Dipl.-Ing. (FH) Ulrich Papert,
X Authors

Sensotronic brake control (SBC) Active steering

Dipl.-Ing. Bernhard Kant. Dipl.-Ing. (FH) Wolfgang Rieger,
ZF Lenksysteme, Schwäbisch Gmünd.
Overview of common-rail systems
High-pressure components of common-rail Drive and adjustment systems
system Dipl.-Ing. Rainer Kurzmann,
Electronic Diesel Control (EDC) Dr.-Ing. Günter Hartz.
Dipl.-Ing. Felix Landhäußer,
Dr.-Ing. Ulrich Projahn, Heating, ventilation and air conditioning
Dipl.-Inform. Michael Heinzelmann, Dipl.-Ing. Gebhard Schweizer,
Dr.-Ing. Ralf Wirth Behr GmbH & Co., Stuttgart.
(Common-rail system),
Ing. grad. Peter Schelhas, Vehicle security systems
Dipl.-Ing. Klaus Ortner Dipl.-Ing. (FH) Jürgen Bowe,
(Fuel-supply pumps), Andreas Walther,
Dipl.-Betriebsw. Meike Keller Dr.-Ing. B. Kordowski,
(Fuel filters), Dr.-Ing. Jan Lichtermann.
Dipl.-Ing. Sandro Soccol,
Dipl.-Ing. Werner Brühmann Electromagnetic compatibility
(High-pressure pumps), Dr.-Ing. Wolfgang Pfaff.
Ing. Herbert Strahberger,
Ing. Helmut Sattmann Fault diagnostics
(Fuel rail and add-on components), Dr.-Ing. Matthias Knirsch,
Dipl.-Ing. Thilo Klam, Dipl.-Ing. Bernd Kesch,
Dipl.-Ing. (FH) Andreas Rettich, Dr.-Ing. Matthias Tappe,
Dr. techn. David Holzer, Dr,-Ing. Günter Driedger,
Dipl.-Ing. (FH) Andreas Koch Dr. rer. nat. Walter Lehle.
(Solenoid-valve injectors),
Dr.-Ing. Patrick Mattes and the editorial team in cooperation with the
(Piezo-inline injectors), responsible in-house specialist departments of
Dipl.-Ing. Thomas Kügler Robert Bosch GmbH.
(Injection nozzles),
Dipl.-Ing. (FH) Mikel Lorente Susaeta, Unless otherwise stated, the authors are all
Dipl.-Ing. Martin Grosser, employees of Robert Bosch GmbH.
Dr.-Ing. Andreas Michalske
(Electronic diesel control),
Dr.-Ing. Günter Driedger,
Dr. rer. nat. Walter Lehle,
Dipl.-Ing. Wolfgang Schauer,
Rainer Heinzmann
2 Basics of mechatronics Mechatronic systems and components


The term “mechatronics” came about as Mechatronic systems

a made-up word from mechanics and and components
electronics, where electronics means
“hardware” and “software”, and mechan- Applications
ics is the generic term for the disciplines Mechatronic systems and components are
of “mechanical engineering” and “hy- now present throughout almost the entire
draulics”. It is not a question of replacing vehicle: starting with engine-management
mechanical engineering by “electronifi- systems and injection systems for gasoline
cation”, but of a synergistic approach and diesel engines to transmission control
and design methodology. The aim is to systems, electrical and thermal energy
achieve a synergistic optimization of me- management systems, through to a wide
chanical engineering, electronic hard- variety of brake and driving dynamics sys-
ware and software in order to project tems. It even includes communication and
more functions at lower cost, less weight information systems, with many different
and installation space, and better quality. requirements when it comes to operability.
The successful use of mechatronics in a Besides systems and components, mecha-
problem solution is dependent upon an tronics are also playing an increasingly
overall examination of disciplines that vital role in the field of micromechanics.
were previously kept separate.
Examples at system level
A general trend is emerging in the further
development of systems for fully automatic
vehicle handling and steering: more and
more mechanical systems will be replaced
by “X-by-wire” systems in future.

1 Mechatronic system


Forces, travel, etc. Forces, travel, etc.

Basic system
(mostly mechanical)


Actuator Sensor
engineering technology

Correcting Measured
variables variables
Feedback Processor variables

K. Reif (Ed.), Automotive Mechatronics, Bosch Professional Automotive Information,

DOI 10.1007/978-3-658-03975-2_1, © Springer Fachmedien Wiesbaden 2015
Basics of mechatronics Mechatronic systems and components 3

A system that was implemented long ago is Examples at component level

the “Drive-by-wire” system, i.e. electronic Fuel injectors are crucial components in
throttle control. determining the future potential of Diesel-
engine technology. Common-rail injectors
“Brake-by-wire” replaces the hydrome- are an excellent example of the fact that an
chanical connection between the brake extremely high degree of functionality
pedal and the wheel brake. Sensors record and, ultimately, customer utility can only
the driver’s braking request and transmit be achieved by controlling all the physical
this information to an electronic control domains (electrodynamics, mechanical en-
unit. The unit then generates the required gineering, fluid dynamics) to which these
braking effect at the wheels by means of components are subjected.
One implementation option for In-vehicle CD drives are exposed to partic-
“Brake-by-wire” is the electrohydraulic ularly tough conditions. Apart from wide
brake (SBC, Sensotronic Brake Control). temperature ranges, they must in particu-
When the brake is operated or in the event lar withstand vibrations that have a critical
of brake stabilization intervention by the impact on such precision-engineered sys-
electronic stability program (ESP), the SBC tems.
electronic control unit calculates the re- In order to keep vehicle vibration away
quired brake pressure setpoints at the in- from the actual player during mobile de-
dividual wheels. Since the unit calculates ployment, the drives normally have a
the required braking pressures separately spring damping system. Considerations
for each wheel and collects the actual val- about reducing the weight and installation
ues separately, it can also regulate the space of CD drives immediately raise ques-
brake pressure to each wheel via the tions concerning these spring-damper sys-
wheel-pressure modulators. The four tems. In CD drives without a damper sys-
pressure modulators each consist of an tem, the emphasis is on designing a me-
inlet and an outlet valve controlled by chanical system with zero clearances and
electronic output stages which together producing additional reinforcement for
produce a finely metered pressure reg- the focus and tracking controllers at high
ulation. frequencies.
Only by combining both measures
Pressure generation and injection are mechatronically is it possible to achieve
decoupled in the Common Rail System. good vibration resistance in driving
A high-pressure rail, i.e. the common rail, mode. As well as reducing the weight by
serves as a high-pressure accumulator, approx. 15 %, the overall height is also
constantly providing the fuel pressure re- reduced by approx. 20 %.
quired for each of the engine’s operating
states. A solenoid-controlled injector with The new mechatronic system for electri-
a built-in injection nozzle injects fuel di- cally operated refrigerant motors is based
rectly into the combustion chamber for on brushless, electronically commutated
each cylinder. The engine electronics re- DC motors (BLDC’s). Initially, they are
quest data on accelerator pedal position, more expensive (motor with electronics)
rotational speed, operating temperature, than previous DC motors equipped with
fresh-air intake flow, and rail pressure in brushes. However, the overall optimization
order to optimize the control of fuel me- approach brings benefits: BLDC motors
tering as a function of the operating condi- can be used as “wet rotors” with a much
tions. simpler design. The number of separate
parts is therefore reduced by approx. 60 %.
4 Basics of mechatronics Mechatronic systems and components

In terms of comparable cost, this more Development methods

robust design doubles the service life,
reduces the weight by almost half and Simulation
reduces the overall length by approx. 40 %. The special challenges that designers face
when developing mechatronic systems are
Examples in the field of micromechanics the ever shorter development times and
Another application for mechatronics is the increasing complexity of the systems.
the area of micromechanical sensor sys- At the same time, it is vital to ensure that
tems, with noteworthy examples such as the developments will result in useful
hot-film air-mass meters and yaw-rate products.
Because the subsystems are so closely Complex mechatronic systems consist of
coupled, microsystems design also re- a large number of components from differ-
quires an interdisciplinary procedure that ent physical domains: hydraulic compo-
takes the individual disciplines of mechan- nents, mechanical components and elec-
ical components, electrostatics and possi- tronic components. The interaction be-
bly fluid dynamics and electronics into tween these domains is a decisive factor
consideration. governing the function and performance
of the overall system. Simulation models
are required to review key design deci-
sions, especially in the early development
stages when there is no prototype avail-

2 Model library for a micromechanical yaw-rate sensor


Mechanical Electro-
components mechanical

Rigid Elastic Comb-like Detection

bodies bodies structures electrodes

From From Bending Segment Undivided Divided From From


segments segments beams of a circle stator comb stator comb segments segments
of a circle of a rectangle of a circle of a rectangle
Basics of mechatronics Development methods 5

Basic issues can often be clarified by pro- However, an analysis of the typical compo-
ducing relatively simple models of the nents in mechatronic systems shows that
components. If more detail is required, they can be composed of a few simple ele-
more refined component models are ments specific to the domains. These stan-
needed. The detailed models focus mainly dard elements are, for example:
on a specific physical domain: • In the hydraulic system: throttle, valve
• This means that detailed hydraulic mod- or electric line
els of common rail injectors are avail- • In the electronic system: resistor, capac-
able, for example. These can be simu- itor or transistor
lated using special programs with nu- • In the mechanical system: ground with
meric calculation methods that are friction, transmission or clutch (or the
exactly tailored to hydraulic systems. equivalent for micromechanics)
Cavitation phenomena have to be taken
into consideration, among other things. The preferable solution is that these ele-
• Detailed models are also needed to de- ments should be stored in a central stan-
sign the power electronics that trigger dard model library that is also decentrally
the injector. Again, this involves the use accessible to product development. The
of simulation tools which must be devel- essence of the standard model library is
oped specifically to design electronic a documentation of all the standard ele-
circuits. ments. For each element, this comprises:
The development and simulation of the Description of physical behavior in
software that controls the high-pressure words
pump and the power electronics in the The physical equations, parameters
control unit with the aid of the sensor (e.g. conductivity or permeability),
signals also takes place using tools that state variables (e.g. current, voltage,
are specially designed for this area of magnetic flux, pressure) and
the overall system. The description of the associated inter-
As the components in the overall system
interact with each other, it is not sufficient In addition, a major part of the environ-
to consider specific detailed models of the ment is a reference model written in a
components in isolation. The optimum so- modeling language that is independent
lution is also to take into account the mod- of the tool. Overall, the library includes
els of other system components. In most reference models from the mechanical,
cases, these components can be repre- hydraulic, electronic, electrodynamic
sented by simpler models. For example, and software areas.
the system simulation that is focussed on
the hydraulic components only requires
a simple model of the power electronics.

The application of various domain-specific

simulation tools during the development
of mechatronic systems is only efficient if
there is some sort of support for exchang-
ing models and parameters between the
simulation tools. The direct exchange of
models is highly problematic due to the
specific languages used for describing the
models of each of the tools.
6 Basics of mechatronics Development methods

V model depending on the technologies applied, for

The dependencies of the different product each of the associated domains (mechani-
development phases are illustrated in the cal engineering, hydraulics, fluid dynam-
“V model”: from requirement analysis to ics, electrics, electronics, and software).
development, implementation, testing and
system deployment. A project passes Recursions at each of the design levels
through three “top-down” levels during shorten the development stages signifi-
the development stage: cantly. Simulations, rapid prototyping, and
• Customer-specific functions simultaneous engineering are tools that al-
• Systems and low rapid verification, and they create the
• Components conditions for shortening product cycles.

A requirements specification (what) must Outlook

first be produced at each level in the form
of specifications. This results in the design The major driving force behind mecha-
specification, which is drawn up on the ba- tronics is continuous progress in the field
sis of design decisions (the actual creative of microelectronics. Mechatronics benefits
engineering work). The performance spec- from computer technology in the form of
ifications describe how a requirement can ever more powerful integrated computers
be met. The performance specs form the in standard applications. Accordingly,
basis for a model description which allows there is a huge potential for further in-
a review (i.e. validation) of the correctness creases in safety and convenience in
of each design stage together with previ- motor vehicles, accompanied by further
ously defined test cases. This procedure reductions in exhaust emissions and fuel
passes through each of three stages, and, consumption. On the other hand, new

3 Recursion method at one level


specification (what) Tool-supported
test-case creation

Specifications Validation,
ent p

Design decisions
(”creative engineering

Test cases



specification (how)
Basics of mechatronics Outlook 7

challenges are emerging with regard to The design approaches of mechatronic

the technical mastery of these systems. systems should strive toward continuity
However, future “X-by-wire” systems in several aspects:
without the mechanical/hydraulic fall- • Vertical:

back level must also provide the pre- “Top-down” from system simulation,
scribed functionality in the event of a with the objective of overall optimiza-
problem. The condition for their imple- tion, through to finite element simula-
mentation is a high-reliability and high- tion to achieve a detailed understanding,
availability mechatronic architecture and “bottom-up” design engineering
which requires a “simple” proof of safety. from component testing through to
This affects both single components as system testing
well as energy and signal transmissions. • Horizontal:

“Simultaneous engineering” across

As well as “X-by-wire” systems, driver-as- several disciplines in order to deal with
sistance systems and the associated man/ all product-related aspects at the same
machine interfaces represent another area time
in which the consistent implementation of • Beyond company boundaries:

mechatronic systems could achieve signifi- Step by step, the idea a “virtual sample”
cant progress for both users and vehicle is nearing our grasp
Another challenge is training in order to
further an interdisciplinary mindset and
develop suitable SE processes and forms
of organization and communication.

4 V-model general overview



Validation Test
analysis Acceptance test

Model Test


Validation Test

System design System test

Model, Test
System target prototype cases


Validation Test

Component design, Component test

development Model, Test

prototypes cases



Component manufacture
8 Architecture Overview


Over the last three decades, tremendous Overview

progress has been made in automotive
engineering. Modern injection and ex- History
haust-gas treatment systems drastically The on-board electrical network of a car
reduced pollutants in the exhaust gas, around the year 1950 comprised approx.
while occupant-protection and vehicle 40 lines. Essentially, cables were only re-
stabilization systems improved safety quired for the battery, starter, ignition and
on the road. Much of this success is due the lighting and signaling systems.
to the introduction of electronically-con- With the first electronic injection and
trolled systems. The proportion of these ignition systems, cabling complexity began
systems used in cars increased continu- to increase fast. Sensors fitted in the en-
ously. The requirements of safety and gine compartment (e.g. speed sensor,
environmental compatibility, but also engine-temperature sensor) had to deliver
the demand for comfort and convenience signals to the engine control unit, while
functions, will increase yet further and the fuel injectors required their triggering
this will in no small part be achieved signals from the electronic control unit.
through the use of electronics. Up to A further increase in cabling complexity
around 90 % of innovations in the motor resulted from the introduction and rapid
vehicle will be realized by electronics widespread adoption of the antilock brake
and microprocessor-controlled systems. system (ABS). Meanwhile, comfort and
The networking of these electronics cre- convenience systems, e.g. electrical power-
ates the prerequisite for having this wide window units, would also form part of the
variety of electronic systems integrated standard equipment. All these systems re-
within the complete vehicle system to quire additional connecting lines for the
form a whole. However, this results in connection of sensors, control elements
a complexity that can only be overcome and actuators to the control unit.
at considerable expense.

1 Proportion of electrics/electronics in the motor vehicle

Automobile 2000 Automobile 2010

Electronics Electronics Electronics Electronics Electronics Electronics
Hydraulics Pneumatics
Hydraulics Electronics Hydraulics Elec- Elec-
60% Pneumatics Hydraulics tronics tronics

Mechanics Mechanics Mechanics
Mechanics Mechanics Mechanics
20% Mecha- Mecha-
nics nics
40% 30% 22% 8% 30% 30% 25% 7% 8%
Driving Safety Con- Info- Driving Safety Con- Info- Commun-
and braking venience tainment and braking venience tainment ication/
Fig. 1

Proportion of electrics/electronics, Proportion of electronics,
approx. 22% approx. 35%
Mercer management

K. Reif (Ed.), Automotive Mechatronics, Bosch Professional Automotive Information,

DOI 10.1007/978-3-658-03975-2_2, © Springer Fachmedien Wiesbaden 2015
Architecture Overview 9

Technology of the present day shorten, the airbag and seat-belt preten-
In the 1990s the cabling work in a luxury sioners are set to emergency standby.
class vehicle amounted to around 3 km. The communication between the elec-
This figure clearly demonstrates how tronic control units cannot take more than
complex the vehicle system has become. fractions of a second. The more electronic
The growth of the proportion of electron- control units interact in the one complete
ics in the motor vehicle (Fig. 1) can mainly system, the more difficult it becomes for
be attributed to the growth in microelec- them to communicate undisturbed.
tronics and sensor technology. With the number of electronic control
At first, many of the new systems were units and the associated need for mutual
integrated into the vehicle by means of communication, the costs of developing
their own dedicated electronic control the systems rose as did the adaptation
unit. For the most part, the individual costs for making interfaces compatible.
electronic control units operated in mutual With the CAN bus (Controller Area Net-
independence. All the same, connecting work) developed by Bosch, a powerful and
lines became increasingly necessary be- widely used data bus system has become
tween electronic control units to enable commonplace in vehicles for the first time.
the exchange of data by means of PWM The data line of the CAN bus makes it pos-
signals, for example. Depending on the sible for the electronic control units to
vehicle class, there are between 20 and exchange specific and relevant items of
80 electronic control units fitted in today’s information with each other. At the start,
vehicles. They control such equipment as the network only comprised a few elec-
the engine, antilock brake system or the tronic control units, such as the engine-
airbags. The number of microcontrollers management system, the electronic stabil-
in the vehicle has therefore risen continu- ity program and the transmission control.
ously in recent years (Fig. 2). Gradually, further systems would expand
The components of the individual sys- this network, especially in the areas of
tems are optimally matched to each other. comfort and convenience and infotain-
The systems may originate from different ment. The CAN bus has gradually evolved
manufacturers that use previously agreed, into the standard for networking systems
albeit still their own, interfaces. The rain in the motor vehicle. Today it is the stan-
sensor, for example, “speaks” in a different dard for communication between elec-
way to the sensors for the engine manage-
ment. The following example demon- 2 Number of microcontrollers in the motor vehicle

strates just how networked the functions

in a modern vehicle are: the radar sensor 140
of the adaptive cruise control system 130
(ACC) measures the distance to the vehicle 120
Number of microcontrollers

traveling in front. If this distance is shorter
than a specified minimum distance, the 90
ACC electronic control unit sends this in- 80
formation to the engine management, the 70
ESP electronic control unit and the airbag
electronic control unit. The engine man- 40
agement reduces torque and thus driving 30
speed. If this is not sufficient, the elec- 20

tronic stability program (ESP) must also
generate brake pressure to decelerate 1980 1985 1990 1995 2000 2005
the vehicle. If the distance continues to
10 Architecture Overview

tronic control units within different areas objectives simultaneously, development

of the electronics (drivetrain, suspension, partners are more frequently tapping into
body electronics and infotainment) and resources that are already available in sub-
forms a powerful backbone for networking systems. These can be sensors or actuators
these areas with each other. Additional bus as well as realized functions that are avail-
systems (e.g. LIN bus, MOST bus) are used able to different systems over the commu-
as subbuses or for transmitting at high nications network. For new systems and
data rates with comparatively low realtime functions, manufacturers strive to get by
requirements in the motor vehicle. on a minimum of additional resources.
In the meantime, engineers are faced with
Development trends a new challenge in the form of "networked"
The proportion of electrics and electronics thinking and subsystem integration, espe-
in the motor vehicle will continue to in- cially when the assemblies for the subsys-
crease. In the drivetrain, the number of tems originate from different development
components in the exhaust line (e.g. ex- partners (suppliers).
haust-gas sensors) is increasing due to Complaints in the field (i.e. with series-
stricter exhaust-emissions legislation. production vehicles) due to electrical or
While the demands for reductions in fuel electronic failures could be the conse-
consumption can, for example, be fulfilled quence of not having taken the interac-
by means of new valve-gear concepts, even tions of the subsystems into consideration.
this requires additional electronic compo- The causes – unmanageable behavior of
nents. A further increase in the proportion functionality spread among networked
of electronics results mainly from the systems, and their integration – are avoid-
growth of electronic systems in the areas able through the logical application of cer-
of safety, comfort and convenience, and tified development processes as early as
infotainment. in the specification phase. Furthermore,
modeling and tools for authoring a formal
Objectives description of architectures are gaining
Drivers demand a high level of reliability ever more in importance.
from a car. The vehicle manufacturer and
the supplier of assemblies, meanwhile, Broadened requirements for a complete
are constrained by other requirements motor vehicle system in the future are
such as minimization of manufacturing leading to increased networking of vehicle
costs, space restrictions and the weight components and subsystems. In this re-
of components. An opportunity to fulfill gard, new functions are being developed
these requirements in the face of the in- that go beyond the frontiers of traditional
creasing complexity of the “vehicle” sys- applications – and this is without addi-
tem is seen in the shift of the traditional tional expenditure on hardware wherever
implementation technologies of mechan- possible.
ics, hydraulics and electrics towards New development methods and technol-
microprocessor-controlled, electronic ogies are required to make this achievable.
systems. For this reason, the development With a top-down approach, new functions
of software will continue to gain in impor- are viewed from the perspective of the
tance in future. complete vehicle. This means that, in ac-
The current situation in the electrical cordance with the method of systems engi-
and electronic architecture of motor vehi- neering, functional requirements and non-
cles is characterized by an increase in functional requirements (e.g. quality ob-
functionality and an increasingly strained jectives, safety, costs, etc.) are set for the
costs situation. To achieve both of these vehicle as a whole and derived as specifi-
Architecture Overview 11

cations for the subordinate subsystems. Vehicle system architecture

These requirements are formulated as a
model and can thus be used as a specifica- Architecture
tion for the subsystems and the creation The architecture of a system represents
of test cases. This is what is known as an its “construction plan”. It describes the
“executable specification”, which makes structural and dynamic system character-
it possible to prove the completeness and istics as a whole. The architecture is usu-
the traceability of the requirements, for ally specified in a description language.
example, or to identify the requirements Special draft mechanisms are used for
for interaction and communication be- specific requirements. With architecture
tween subsystems. In this way, it is possi- being a construction plan for different
ble to form an optimized architecture for realization technologies and a means of
the complete vehicle and its subsystems proving that functional and nonfunctional
and components. The functional relation- requirements have been fulfilled in the
ships between the complete motor vehicle system draft, different views of the system
system and the subordinate subsystems architecture are required. Examples of this
can be surveyed in different levels of detail include:
and suitable interfaces can be defined for • Hardware architecture
the functions. This approach supports an • Software architecture
expanding networking of functions. Syner- • Network architecture in the area of real-
gies are exploited between vehicle areas ization technologies
(domains such as the drivetrain, interior, • Cost and resource consumption in the
infotainment) that were hitherto consid- area of economical analysis and
ered in isolation and resources are spared. • For the area of social requirements,
As an element of the development pro- aspects such as safety, availability and
cess that works in the opposite direction, legal conformity
the generation of new functions from avail-
able resources and existing systems (bot- The problems that arise in the integration
tom up) should also be taken into consid- of differently structured subsystems can
eration to minimize innovation risks. be reduced by means of an architecture.
This is how new functions are integrated
into existing systems, for example. Exam- Functional structure
ples of this approach are measures to avert The domain of vehicle motion has the task
the consequences of an accident by “pre- of ensuring the controlled movement of
paring” subsystems for an imminent crash the vehicle as well as its directional stabil-
(closing windows, closing the sliding sun- ity. This task can be subdivided into vari-
roof, activating the airbag, etc.) or the as- ous levels (Fig. 3).
sistance of the driver in emergency brak- The navigation level is home to the plan-
ing situations in ESP in future. In this way, ning tools for the driving route. These are
it is possible to reduce the number of elec- merely informational in nature and have
tronic control units and counteract rising no interventional influence on vehicle
system costs. motion.
The development process described At vehicle guidance level, the decisions
characterizes the CARTRONIC® concept of the driver are implemented by means of
that Bosch developed in the 1980s. the steering wheel and accelerator pedal
The results of this concept are being but also various assistance systems for ve-
incorporated into the Autosar Initiative hicle handling (e.g. ACC, course stability
(see Autosar Initiative)
12 Architecture Vehicle system architecture

systems). At this level, the driver is able functional component represents the tasks
to overrule the assistance systems at any of the navigation level, which are to inform
time. the driver of the driving route determined
At the stability level, there are the sub- by means of a mapping system (Fig. 3).
systems that are able to correct the deci- Vehicle guidance represents the guidance
sions taken at handling level if these hap- level, and stability intervention the tasks
pen to be outside the range of safe refer- of the stabilization level. The vehicle mo-
ence variables (e.g. ABS, ESP). This may tion coordinator determines the correcting
be the case when cornering or on wet road variables for the actuators, e.g. of the
surfaces, for example. drive and electronic stability program
At stabilization level, correcting vari- (ESP), from the information input by
ables for implementation by the vehicle’s vehicle guidance and stability interven-
actuators are determined. Information tion.
about the environment (e.g. road condi- Figure 4 shows how the functional com-
tion, air temperature, rain sensor signal) ponents of guidance level, stabilization
is still required at the various levels for level and vehicle actuators are related in
the implementation of the relevant tasks. a hierarchical structure within vehicle
These tasks can be assigned to func- motion. Communication relationships
tional components, which are the architec- between the components and interactions
tural elements of the functional architec- with other domains, e.g. body and interior,
ture. In this way, the driver information are also featured in the model.

3 Levels in the vehicle motion domain

Environment Driving and assistance

Navigation level
Road network
Driver information

Calculated distance

Road layout Guidance level

Traffic regulations
Vehicle guidance

Vehicle guidance data Vehicle

Road conditions
Stability level Actuators
Vehicle condition
Vehicle motion $ive
View Stability
Weather conditions intervention $

Vehicle motion
Architecture Vehicle system architecture

In the same way as Vehicle motion is Systematic creation of EE system

refined, these functional components architectures
require further detailing until the refined The increasing amount of networking in
components represent manageable, traditional vehicle domains for the realiza-
clearly delimited tasks that make flexible, tion of new functions can be illustrated
modular implementation possible through using the ACC (Adaptive Cruise Control)
different realization technologies. Defined driver-assistance system as an example.
interfaces between the components enable Adaptive, same-lane driving is made possi-
communication and the exchange of data. ble by the networking of a combined
For example, the transmission control cruise and distance control system with
issues a request through the engine-man- the engine-management system, brake sys-
agement system for a specific reduction tem, transmission and cockpit. Here, sub-
in torque during a gearshift. This value is systems from the drivetrain, chassis and
exchanged as a physical variable via the infotainment (interaction with the driver)
interface. domains are used to realize the new func-
With its integration into a suitable pro- tion with minimal cost.
cedural model, the functional structure is The decision as to whether a function
the starting point for subsequent stages in (e.g. ACC) is realized in a dedicated logic
the development process. close to the sensor or in one of the exist-
ing, subscriber electronic control units
has no bearing on the function itself.

4 Example of a functional structure for the vehicle motion domain

Vehicle motion
Vehicle guidance


Steering angle

Vehicle motion coordinator

Brake torque
Drive torque Steering angle
Steering angle

Drive Brake Steering Chassis Stability intervention

Stop lamp

Body, interior
14 Architecture Vehicle system architecture

Rather, the decision is affected by non- ization stages. This required a decoupled
functional requirements such as safety, development process and the exploitation
availability, costs or resource availability. of synergies between subsystems. The de-
In addition to the functional requirements, velopment frameworks took into consider-
these requirements mainly determine how ation the dependencies and interface con-
the function is realized. The “how” is de- tents within the individual domains and
scribed by the architecture of the system. with the rest of the vehicle, as is the case
Different requirements result in different with a networked system such as ACC,
system architectures. for example.

CARTRONIC® concept Bosch introduced this concept to the Auto-

With the CARTRONIC® architecture con- sar Initiative (Workpackage 10.x).
cept, all closed and open-loop control
tasks in the vehicle have been structured Software architecture
in accordance with logical, functional The independence of the functional struc-
viewpoints and modeled in the form of ture, or architecture, from the later real-
a functional architecture. Delimited func- ization stage results in a decoupling of
tions (and their dependencies) that imple- functionality and technology and thus
ment specific functional requirements forms the first stage of a model-based
have been represented by defined archi- development process. The functional
tectural elements. The functional struc- structure can be used on several occasions
ture, i.e. the structural description, repre- and expanded as the foundation for draft-
sented a hierarchical decomposition of ing system architectures. This architecture
the subsystems down to manageable size. is characterized by architecture drivers
Interactions between elements of the func- (specific criteria of the architecture) that
tional structure have been described by are essentially the product of nonfunc-
communication relationships. Since the tional requirements (e.g. costs, quality,
use of the architecture concept could have reusability, relocatability).
led to different functional structures, Further precision of the development
it was essential to reach agreement on frameworks devised from the functional
the tasks and interfaces. It was necessary structure, and of their interfaces in partic-
to choose interfaces that were based on ular, is required if it is to be possible to
physical variables and thus supported evaluate an electronic control unit for the
aspects such as reusability and inter- relocatability of functions and the integra-
changeability. tion of software, which is a contribution
of various participants in the project.
The motor vehicle system with all its open While retaining the realization-indepen-
and closed-loop tasks was dismantled into dent information from the functional
subsystems that implement clearly defined structure – such as an agreed torque inter-
tasks. These subsystems include the en- face – the frameworks are supplemented
gine management, brake system, transmis- by realization-specific information such
sion control, ACC, lighting management, as data type, quantization, runtime prop-
etc. Different levels of functional structure erties or resource requirements.
detail can be assigned to the system and
subsystem levels (Fig. 5). It was therefore In the same way as hardware and disci-
possible to create development frame- plines such as mechanics or hydraulics,
works for selected functional components software can also be classified as a realiza-
and component groups on which to base tion technology. Product-line or platform
implementation in the form of partial real- approaches have long been a familiar
Architecture Vehicle system architecture

feature of mechanical development or pro- Network architecture

duction optimization and their application With the spreading of open standards
is virtually universal. The trend towards such as the CAN bus, the integration of
software-based system solutions and im- functions into application-specific elec-
provements, in conjunction with the vastly tronic control units, and satellites linked
expanding software scopes of electronic by subnetworks, network architecture has
systems, has given rise to the demand for become the synonym for the complexity
this strategy to also be transferred to soft- management of distributed systems.
ware-intensive systems. Extensions and “attachment solutions”
The decisive challenge faced today is are easily integrated until the limits of
less to do with technical feasibility but network capacity are reached. If these
more about how to develop methods fur- possibilities were to be exploited without
ther and apply already developed methods checking the system draft, this would
and processes in product developments result in unmanageable increases in com-
in a systematic and disciplined way, and plexities and integration conflicts. Biologi-
to anchor them within the organization. cal systems solve these unmanageable
The product-line approach in software de- complexities through specialization and
velopment was transferred to the motor the creation of subnetworks with new
vehicle domain with the participation of forms of organization. Their objectives
Bosch with methodological support from are stability and the ability to survive.
the software engineering institute (SEI). This model has, to a certain extent,
The method will be used systematically evolved on its own in motor vehicle sys-
in future Bosch product generations. tems through assignment to traditional

5 Functional architecture

System level


Hardware Software
architecture architecture

Network architecture
16 Architecture Vehicle system architecture

fields of application or domains and the reliability are fulfilled by the multiple use
comparatively slow growth of networking of proven standards. Autosar concerns
within these domains. itself with all vehicle domains.
Based on the uniform electronics plat-
Bus systems for the individual domains are form, which primarily consists of standard
becoming more specialized due to their software modules, each vehicle manufac-
plainly different requirements. With the turer is then free to build its own specific
CAN in the drivetrain as the point of ori- content. They enable integration into the
gin, new bus systems such as the LIN sub- electronics network. These software func-
bus have begun to infiltrate the area of tions permit differentiation between the
body electronics or FlexRay in the case of competition.
safety-relevant x-by-wire systems. In the Not only does software have to conform
multimedia field, where demands for high to the Autosar standard. The electronic
data rates but low safety requirements control units must be built in such a way
prevail, bus systems such as Bluetooth that the Autosar software is able to run on
have started to make an appearance. them. The Autosar members are hoping
Breaking through these traditional do- that the new development methods yield
mains with ever more applications leads such benefits as shorter development
to known consequences, e.g. dramatic in- times and lower development costs.
crease in complexity, high start-up costs,
increasing integration times and costs, and Until now, it was often the case that dedi-
more demanding work in customer service cated electronic control units would be
as a consequence of diagnostics no longer developed and fitted for new functions
being manageable. A solution for these (e.g. electronic transmission control,
multidimensional optimization tasks has in antilock brake system, air conditioning).
the past been sought in the software field. The number of electronic control units
In the case of technical systems in particu- fitted in the vehicle grew continuously;
lar, the paradigm is still king, especially in today’s generation of vehicles are
software realizations, because the absence equipped with between 20 and 80 elec-
of physical boundaries supports unlimited tronic control units. In future vehicle gen-
growth. erations, it is intended that all functions
be covered by a network of 10 to 20 elec-
Autosar Initiative tronic control units. Some of these will
The Autosar Initiative (AUTomotive Open function a little like main computers that
Systems ARchitecture) was founded in will bundle the important function groups
July 2003 by several vehicle manufacturers together. These include the drivetrain,
and suppliers – Bosch among them. Their suspension management system, body
global objective is the joint development and interior and the multimedia/telematics
of an open system architecture for future domain. On data buses, sensors with inte-
automotive applications. The aims of the grated electronics output processed and
partnership include the standardization verified signals, while the buses carry the
of fundamental system functions (basic relevant control commands to actuators
software) and function interfaces; they with integrated triggering electronics.
will replace the company-specific, individ- In future, new functions will often be
ual solutions used to date. Model-based able to use the existing computer architec-
concepts and methods ought to reduce ture up to its performance limit and will
complexity in spite of an expanding range be widely realized in the form of a soft-
of functions. The demands for quality and ware add-on. This would therefore render
Architecture Vehicle system architecture

unnecessary the additional electronic con- Outlook

trol unit that would have been required Increasingly greater demands for safety,
today. The system only needs to be supple- reliability and availability are being placed
mented by the sensors and actuators re- on the network architecture of modern
quired. vehicles. This is where energy network
architecture will play a key role. In the
Software will no longer be an inevitable face of vehicle functions increasingly be-
component of hardware, but will increas- ing realized electronically, this architec-
ingly become a stand-alone product. ture makes for a reliable supply of power
The first examples of business and col- to systems and thus forms the basis for
laborative models between supplier and the reliability and safety of future systems.
manufacturer have already been put into One possible future technology is the
practice at Bosch, e.g. in drivetrain man- transmission of power and information
agement. on the supply line. The following benefits
arise from the powerline communication
Examples (PLC) concept used in the public grid:
Individually-controlled drive components • Weight and cost reductions as well as
at each of the wheels with different wheel space savings from the discontinuation
positions and wheel loads permit optimum of data lines
use of tire force potential. This results in • Easier retrofitting for retrofit systems
increased driving dynamics and safety (spare parts trade)
while at the same time reducing consump- • Reduction in complexity of the wiring
tion, wear and emissions. For this to be harness in respect of manufacture and
possible, all active elements in the drive- installation
train, suspension and steering must be • Increase in system safety, especially in
networked. mechanically stressed zones (e.g. door,
One example of superordinate functions mirror) that are characterized by pre-
realized by networking is the ASIS (Active mature aging of lines and increased risk
Shift Strategy) drive strategy for automatic of failure
transmissions in passenger cars. Based on • Powerline as a redundancy path for sys-
the evaluation of various control elements tems relevant to safety
(e.g. accelerator pedal) and conclusions • Simultaneous, or parallel, implementa-
drawn from the information of other sys- tion of several bus systems or services,
tems (e.g. cornering detection from the e.g. diagnostics
wheel speeds), this strategy is able to con-
trol the gearshift in such a way as to meet
the driver’s real-time demand for agility
and power through selection of the appro-
priate gear. In future, the telematics will
be able to support additional, improved
driving strategies, e.g. through the use
of GPS signals for transmission control
in terms of predictive driving.
18 Electronic control unit Operating conditions, design, data processing


Digital technology furnishes an extensive Design

array of options for open and closed-loop
control of automotive electronic systems. The printed circuit board with the electri-
A large number of parameters can be in- cal components (Fig. 1) is installed in a
cluded in the process to support optimal housing of plastic or metal. A multiple
operation of various systems. The control plug connects the control unit to the sen-
unit receives the electrical signals from sors, actuators and electrical power sup-
the sensors, evaluates them, and then ply. The high-performance driver circuits
calculates the triggering signals for the that provide direct control of the actuators
actuators. The control program, the are specially integrated within the housing
“software”, is stored in a special memory to ensure effective heat transfer to the
and implemented by a microcontroller. housing and the surrounding air.
The control unit and its components are The majority of the electrical compo-
referred to as hardware. The Motronic nents are of the surface-mounted device
control unit contains all of the algo- technology type. This concept provides
rithms for open and closed-loop control extremely efficient use of space in low-
needed to govern the engine-manage- weight packages. Only a few power com-
ment processes (ignition, induction and ponents and the connectors use push-
mixture formation, etc.). through assembly technology.

Operating conditions Hybrid versions combining compact

dimensions with extreme resistance to
The control unit is subjected to very thermal attack are available for mounting
high demands with respect to directly on the engine.
• Extreme ambient temperatures
(in normal vehicle operation from Data processing
–40 to +60...+125 °C)
• Extreme temperature changes Input signals
• Indirect materials and supplies In their role as peripheral components,
(oil, fuel etc.) the actuators and the sensors represent
• The effects of moisture and the interface between the vehicle and the
• Mechanical stress such as vibration control unit in its role as the processing
from the engine unit. The electrical signals of the sensors
are routed to the control unit via a wiring
The control unit must operate reliably harness and the connector plug. These sig-
when the vehicle is started with a weak nals can be of the following type:
battery (e.g. cold start) and with high
charge voltages (vehicle electrical system Analog input signals
fluctuations). Within a given range, analog input signals
can assume practically any voltage value.
Other requirements arise from the need Examples of physical quantities which
for EMC (ElectroMagnetic Compatibility). are available as analog measured values
The requirements regarding immunity to are intake-air mass, battery voltage, in-
electromagnetic interference and limita- take-manifold and boost pressure, coolant
tion of high-frequency interference signal and intake-air temperature. They are con-
emission are extremely stringent. verted into digital values by an analog-
digital converter in the microcontroller

K. Reif (Ed.), Automotive Mechatronics, Bosch Professional Automotive Information,

DOI 10.1007/978-3-658-03975-2_3, © Springer Fachmedien Wiesbaden 2015
Electronic control unit Data processing 19

of the control unit and used for calcula- Signal conditioning

tions by the microcontroller CPU. The Protective circuits limit the voltages of in-
maximum resolution of these analog sig- put signals to levels suitable for process-
nals is 5 mV. This translates into roughly ing. Filters separate the useful signal from
1,000 incremental graduations based on most interference signals. When neces-
an overall measuring range of 0 to 5 V. sary, the signals are then amplified to the
input voltage required by the microcon-
Digital input signals troller (0 to 5 V).
Digital input signals only have two states. Signal conditioning can take place com-
They are either “high” or “low” (logical 1 pletely or partially in the sensor depend-
and logical 0 respectively). Examples of ing upon the sensor’s level of integration.
digital input signals are on/off switching
signals, or digital sensor signals such as Signal processing
the rotational-speed pulses from a Hall The control unit is the switching center
generator or a magnetoresistive sensor. governing all of the functions and se-
Such signals are processed directly by quences regulated by the engine-manage-
the microcontroller. ment system. The closed and open-loop
control functions are executed in the mi-
Pulse-type input signals crocontroller. The input signals that are
The pulse-shaped input signals from provided by the sensors and the interfaces
inductive-type sensors containing infor- to other systems (such as the CAN bus) are
mation on rotational speed and reference used as input variables. and are subjected
mark are conditioned in their own control to a further plausibility check in the com-
unit stage. Here, spurious pulses are sup- puter. The control unit program supports
pressed and the pulse-shaped signals generation of the output signals used to
converted into digital rectangular signals. control the actuators.

1 Design of a control unit using the example of an ME Motronic (sectional view through housing cover)
20 Electronic control unit Data processing

Output signals Control unit-internal communication

The microcontroller uses the output sig- In order to be able to support the micro-
nals to control output stages that usually controller in its work, the peripheral
provide enough power for connecting the components must communicate with it.
actuators directly. It is also possible to ac- This takes place using an address/data
tuate certain output stage relays for con- bus which, The microcontroller outputs
sumers that use up a great deal of power the RAM address whose contents are to
(e.g. motor fans). be read (for example) via the address bus.
The output stages are proof against The data bus is then used to transmit the
short circuits to ground or battery voltage, relevant data. For former automotive ap-
as well as against destruction due to elec- plications, an 8-bit bus topology sufficed.
trical or thermal overload. Such malfunc- This meant that the data bus comprised
tions, together with open-circuit lines or 8 lines which together could transmit
sensor faults are identified by the output- 256 values simultaneously. 65,536 ad-
stage IC as an error and reported to the dresses can be accessed using the 16-bit
microcontroller. address bus in this system. Presently,
more complex systems demand 16 bits,
Switching signals or even 32 bits, for the data bus. In order
Actuators can be switched on and off using to save on pins at the components, the data
the switching signals (e.g. motor fans). and address buses can be combined in a
multiplex system, i.e. addresses and data
PWM signals are dispatched through the same lines
Digital output signals can be in the form but offset from each other with respect
of PWM (Pulse-Width Modulated) signals. to time.
These are constant-frequency rectangular Serial interfaces with just a single data
signals with variable on-times (Fig. 2), line are used for data that does not have
Various actuators can be moved to various to transmitted extremely quickly (e.g. fault
operating positions using these signals memory data).
(e.g. exhaust-gas recirculation valve,
boost-pressure actuator). EOL programming
The extensive variety of vehicle variants
with differing control programs and data
records, makes it imperative to have a sys-
tem which reduces the number of control
2 PWM signals unit types needed by a given manufac-
turer. To this end, the Flash-EPROM’s com-
a plete memory area can be programmed
at the end of production with the program
and the variant-specific data record
(this is the so-called End-of-Line, or EoL,
Signal voltage

Another way of reducing the type

a diversity is to store several data variants
(e.g. transmission variants) in the memory,
which are then selected using coding at the
end of the production line. This coding is

Fig. 2
a Period duration stored in an EEPROM.
(fixed or variable)
b Variable on-time
| Performance of electronic control units 21
Control unit

▶ Performance of electronic control units

The performance of electronic control units I/O facilities for timer-controlled signals and
goes hand-in-hand with advances achieved in an integrated analog-digital converter at the
the field of microelectronics. The first gasoline end of the 1980’s. It was then possible to cre-
injection systems were still analog – with lim- ate relatively powerful systems. Figure 3 shows
ited flexibility in the implementation of control a comparison between the performance of a
functions. These functions were constrained fuel-injection system (LH3.2) and an ignition
by the hardware. system (EZ129K) – equipped with 80C515
Progress advanced in quantum leaps with controllers – and that of the succeeding
the arrival of digital technology and the micro- Motronic systems. The ME7 has approximately
controller. The entire engine management sys- 40 times the performance capability of the
tem was taken over by the universally applica- LH/EZ combination with a clock frequency of
ble semiconductor microchip. The actual con- 40 MHz. With the benefit of a new generation
trol logic in microcontroller-controlled systems of microcontrollers and a further increase in
is in a programmable semiconductor memory. clock frequency on the ME9, this figure will
From systems that initially simply con- increase to a factor of well over 50.
trolled fuel injection, complex engine-manage- In the foreseeable future microcontrollers
ment systems were then developed. They con- will process more than just digital control se-
trolled not only fuel injection but also the quences. Signal processors are integrated that
ignition system including knock control, ex- can also directly process the signals provided
haust-gas recirculation and a whole variety by knock sensors, for example.
of other systems. This continuous process of
development is bound to continue in a similar Advances in the development of semiconduc-
vein over the next decade as well. The integra- tor memory chips are also worthy of note.
tion of functions and, above all, their complex- Complex control programs require an enor-
ity are constantly increasing. This pattern of mous amount of memory space. The capacity
development is only possible because the of memory chips at the start of the 1980s
microcontrollers used are also undergoing was still only 8 kilobytes. The ME7 now uses
a similar process of improvement. 1-megabyte chips and soon memory capacities
Microcontrollers in the Intel 8051 family of 2 megabytes will be required. Figure 3
were used quite some time until they were re- shows this pattern of development and likely
placed with 80515 derivatives with additional future trends.
Fig. 3
Chart illustrating
▶ Performance
3 Development of electronic control units capability of

MPC555 56MHz systems




Flash: 2.5MB ME9.0

▶ Number of control
C167 40MHz
Connector pins (control unit)


unit connector pins


ME7.0.1 Flash: 1MB

1,024 kB

C167 24MHz Program memory



ME7.0 Flash: 1MB capacity


C167 24MHz
ME7.0 ▶ Data memory

Flash: 512 kB


80C517A 16MHz
capacity (RAM)
128 kB

Flash: 128 kB

80C517 15.8MHz
8 kB

M4.3 Flash: 64 kB
By way of comparison:
The performance
80535 12MHz

capability of a state-
32 kB

EPROM: 32 kB
0.5 kB

of-the-art engine-

80535 12MHz
LH3.2 + EZ129K

EPROM: 32 kB
management system
1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 far exceeds that of
Apollo 13.
22 Electronic control unit Digital modules in the control unit

Digital modules in the In a microcomputer, however, the above-

control unit mentioned functions are integrated on a
silicon wafer (system-on-a-chip). This is
Microcontroller not functional on its own (standalone) and
Structure is therefore referred to as a single chip
A microcontroller consists of the following microcomputer.
interacting components (Fig. 1): The microcontroller is used to control
• Central processing unit (CPU): this con- self-regulating systems such as an engine-
tains the control unit and the arithmetic management system. Depending on the
and logic unit. The control unit executes application, they may also have expansion
the instructions from the program mem- modules connected to them (e.g. addi-
ory, whereas the arithmetic and logic tional memory for data and program
unit performs arithmetical and logical code).
operations. The user program is fixed and is not re-
• Input and output devices (I/O, Input/Out- placed for different applications. This is
put), which handle the exchange of data the difference between a microcontroller
with peripheral devices. Peripheral de- system, for example, and a PC.
vices include input and output devices
and external data storage media. Programming
• Program memory, in which the operating The only command form capable of direct
program (user program) is permanently interpretation by a microprocessor is a
stored (ROM, PROM, EPROM or flash bit pattern, i.e. the binary representation
EPROM). of a number. Since, however, this form of
• Data memory, which is accessed for instruction is not easy to work with for a
reading and writing (RAM). This con- programmer, and is therefore susceptible
tains the data that is currently being to errors, easily memorable abbreviations
processed. Non-volatile memory (mnemonics) are used. These are automat-
(EEPROM) is used for data that must ically translated by an assembler program
not be deleted when the supply voltage into bit patterns (machine code) that can
is switched off. be understood by the microprocessor.
• The bus system connects the individual For more complex systems and pro-
elements of the microcontroller. grams, high-level programming languages
• A clock generator (oscillator) ensures such as C are needed, as otherwise it would
that all operations in the microcon- be impossible to keep extensive programs
troller take place within a defined timing manageable and free of errors. Such lan-
pattern. guages require sophisticated translation
• Logic circuits are modules with special- programs (compilers) which convert the
ized tasks such as program interrupts. text of the high-level language into a form
They are integrated in individual that can be processed by the microcon-
I/O units. troller.
The machine code is stored in the
The chief components of a microcomputer program memory, where it remains
are generally separate modules connected permanently. The CPU accesses these
to one another on a printed-circuit board. components via the bus system, reads
The microprocessor within such as system the numerically coded commands and
– the CPU – is not functional on its own: executes then.
it is always part of a microcomputer.

Microprocessor Data memory Program memory Data memory

Central processing unit (user memory) Read-only memory Non-volatile read/

(CPU) Volatile read/write (ROM, EPROM, write memory
memory (RAM) flash EPROM) (EEPROM)
for variable data For programs and
Arithmetic and logic unit permanent data records
(ALU) Memory capacity
4-, 8-, 16-, 32-bit Memory capacity Memory capacity 32 bytes to
64 bytes to 32 kbytes 2 kbytes to 512 kbytes 1 kbyte

Clock Bus 4-, 8-, 16-, 32-bit data circuit



Interrupt Event Signal acquisition Analog/ Digital Serial Bus
controller counter and output with digital inputs/ interface controller
time reference (A/D) outputs
converter (I/O) (UART,

(Timer, time SPI,
processing unit, CAN)
Electronic control unit

input capture,
Monitoring Resolution Resolution with external
circuit 50 ns 8 to 10 bit Data rate chips via
(watchdog) Counter Time range 4 to 32 8 to 32 200 bit/s to address/
8 to 64 bit 50 ns to 1s channels channels 1 Mbit/s data bus

… … … … … …
Digital modules in the control unit
24 Electronic control unit Digital modules in the control unit

Semiconductor memories Memory modules are organized on a bit or

Applications word basis, depending on the application.
Memories are used to store large volumes A “word” is a group of bits that can be pro-
of cessed as a single unit. The word length is
• Digital signals representing data equal to the number of bits processed as
(I/O data, statuses, intermediate results a single unit. Eight bits are referred to as
involving frequent and rapid reading a byte.
and writing) Memories can be organized on the basis
• Program code (usually permanently of different word lengths. An 8 M x 8-RAM,
stored) and for example, has a memory capacity of
• Constants (permanently stored) 8 million times 8 bits (64 Mbits). The data
is organized into bytes (8 bits), making the
Storage involves memory capacity 8 Mbytes.
• Recording (writing)
• Permanent retention (actual storage) Word lengths of 4, 8, 16 and 32 bits are
and common in microcontroller systems.
• Location and retrieval (reading) of The word length is one of the factors that
information determines the performance capability of
the system. The word length that is used
Memories utilize physical effects that depends on the performance capability
clearly and easily create and show two requirements of the system.
different states (e.g. conducting/non-con-
ducting or loaded/not loaded). The infor- The most important terms are explained
mation that is to be stored must therefore below according to their standardized def-
exist in binary form, i.e. encoded as a initions, where applicable, or their most
series of “yes or no” statuses (logical “1” common usage (see Figure 2 for overview).
or logical “0”). Such a “yes or no” unit of
information is called a bit (binary digit).

2 Overview of semiconductor memories

Semiconductor memories

Non-volatile memories Volatile memories

Factory programmed User programmed

on a program- Programmable Static Dynamic
in the circuit memories memories
ming device

UV Electrically
Read-only erasable erasable


Read- Program- Erasable EEPROM Electrically Static Dynamic
only mable read- PROM erasable RAM RAM
memory only memory PROM
Electronic control unit Digital modules in the control unit 25

Random-access memory (RAM) Erasable ROM

Random-access memory or RAM is a There are also ROMs whose contents can
short-term memory that allows direct ac- be erased and reprogrammed as outlined
cess to any storage location. Information below.
can be read/written from/to the memory
any number of times. EPROM (Erasable PROM)
This type of erasable read-only memory
Static RAM (SRAM) can have its contents completely wiped
Static RAMs use bistable switching ele- by irradiation with UV light and can then
ments as the data storage cells. Their func- be reprogrammed using a programming
tionality is similar to that of a flip-flop, device.
a simple circuit with two transistors, of
which either the one (logical “1”) or the EEPROM (Electrical EPROM)
other (logical “0”) conducts at any one The EEPROM (also known as E2PROM) can
time. In SRAM, the information remains be electrically erased and reprogrammed.
stored until the storage cell concerned Every storage cell of an EEPROM be indi-
is addressed and overwritten, or the vidually overwritten. For that reason,
operating voltage is switched off. this type of memory module can also
SRAM is therefore volatile memory. be used as nonvolatile data memory
(e.g. for learned information in engine
Dynamic RAM (DRAM) management systems).
Unlike SRAM, the information is stored
as an electrical charge in the gate capacity Flash EEPROM
of a CMOS transistor in dynamic RAM A more sophisticated variant of the
(DRAM). As such capacitors are suscepti- EPROM and EEPROM is flash EEPROM.
ble to leakage, the charge is gradually lost. In this case, electrical flash pulses are used
In order to retain the information, the to erase specific storage areas or the entire
charge has to be refreshed at regular contents of the memory. The erased areas
intervals (every few ms). can subsequently be reprogrammed.
The flash memory can be reprogrammed
Read-only memory on a programming station. However, the
Read-only memory (ROM) is permanent- advantage of flash EEPROM is that is can
storage memory that allows any memory also be reprogrammed while still inside
location to be accessed directly but – as the the sealed control unit.
name indicates – allows the information Flash EEPROM is used in cases where
only to be read and not modified. relatively large quantities of data need
A ROM is nonvolatile memory, i.e. the to be stored, but must also be modifiable
information it contains is retained even (e.g. program memory in vehicle control
when the operating voltage is switched off. units).
It is usually used to store program code
(control programs) and fixed data (func-
tion tables, encoding rules, engine charac-
teristic data maps) that need to be retriev-
able at any time. The information may
be indelibly entered in the memory by
the manufacturer or the user by means
of appropriate programming of specially
prepared memories (PROMs or program-
mable ROMs).
26 Electronic control unit Control unit software

Control unit software execute the command depends on the

microcontroller that is used and the clock
Real-time capability frequency. The microcontrollers that are
One of the requirements on electronic sys- currently used in vehicles can execute up
tems is real-time capability. This means to 1 million commands per second.
that control procedures must react to in- Because of the limited speed at which
put signals within an extremely short time. the program can be executed, a software
For example, a wheel that has a tendency structure with which time-critical func-
to lock must be detected so quickly that tions can be processed with high priority
the ABS control algorithm in the control is required.
unit can reduce the brake pressure via The engine-management program has
the hydraulic modulator quickly enough. to react extremely quickly to signals from
Brake slip is therefore reduced before the speed sensor, which records the en-
the wheel can lock. Engine management gine speed and the crankshaft position.
systems make considerable real-time ca- These signals arrive at short intervals that
pability demands so that crankshaft angles can be a matter of milliseconds, depending
can be adhered to with extreme accuracy on the engine speed. The control unit pro-
at fast engine speeds for injection and igni- gram has to evaluate these signals with
tion timing purposes. high priority. Other functions such as
The complexity of an electronic system reading in the engine temperature are not
therefore makes extremely high demands as urgent, since the physical variable only
of the software that is developed. The soft- changes extremely slowly in this case.
ware structure is explained below on the
basis of an example. Interrupt control
As soon as an event occurs that requires
Software structure an extremely rapid response (e.g. speed
The microcontroller in the control unit ex- sensor pulse), the program that is cur-
ecutes commands sequentially. The com- rently running must be interrupted. This
mand code is obtained from the program can be done using the microcontroller’s
memory. The time taken to read in and interrupt control facility. Events can trig-
ger a program execution interrupt, where-
upon the program jumps and executes
1 CPU power distribution principle the “interrupt routine”. When this routine
has been executed, the program resumes
Tooth at the point at which it was interrupted
(Fig. 1).
An interrupt can be triggered by an ex-
interrupt ternal signal, for example. Other interrupt
sources are timers integrated in the micro-
Ignition controller, with which timed output signals
interrupt can be generated (e.g. ignition signal:
microcontroller ignition output is switched
at a point in time that is calculated before-
Time frame
hand). However, the timer can also gener-
ate internal time frames.
Fig. 1 Background
Depiction of several program

program levels on the Time t

example of the software
from a Motronic system
Electronic control unit Control unit software 27

The control unit program reacts to several to the speed sensor signal. For this pur-
of these interrupts. An interrupt source pose, the engine-speed signal is connected
can therefore request an interrupt while to a microcontroller interrupt input. Every
another interrupt routine is currently falling signal edge at this input interrupts
being executed. Every interrupt source the current calculations that are in prog-
therefore has a fixed priority assigned to ress and forces a branch to the interrupt
it. The priority controller decides which routine. After executing the commands in
interrupt is allowed to interrupt another the interrupt routine, the program contin-
interrupt. ues execution at its point of origin.
In order to perform certain operations
Tooth interrupt the control unit program requires the time
The crankshaft is equipped with a pulse taken for the crankshaft to travel between
wheel (Fig. 2a) that has a certain number one tooth and the next. This calculation is
of teeth on its circumference. The teeth performed by an internal timer. This is a
are scanned by the speed sensor. This freewheeling 16-bit counter (Fig. 3) that
allows the crankshaft position to be re- increments at a certain rate, depending on
corded. The typical distance between the microcontroller oscillator clock cycle.
a pair of teeth on the crankshaft sensor This time frame amounts to about 0.5 µs.
wheel is 6°. In order to determine the When the falling tooth flank occurs,
crankshaft position, the control unit pro- the current counter status is recorded.
gram must execute certain routines as The difference (and therefore the tooth
each tooth is detected. At 6,000 rpm the interval) is calculated using the stored
detection time between two teeth is ap- counter from the previous tooth.
proximately 300 µs. Every command in
these routines must be executed within Example: crankshaft position calculation
this time. This requires a rapid response The engine-management system (Motronic
for gasoline engines, EDC for diesel en-
gines) must know the crankshaft position
2 Crankshaft sensor ring with speed sensor at any given point in time. This is a pre-
requisite for injecting into the right cylin-
der at the right time and ensuring that ig-
nition takes place at the calculated ignition
angle (Motronic systems). In order to

3 Time calculations via internal timer

Counter status


Time t
Speed sensor



Tn: Counter status at tooth n Fig. 2

Tn+1: Counter status at tooth n+1 a Design
b Speed sensor signal
28 Electronic control unit Control unit software

detect the engine position and the engine Since the synchronization program runs
speed, the control unit evaluates the speed over several teeth at fast engine speeds,
sensor signal (Fig. 2b). it has to be interrupted by the tooth inter-
There is gap in the crankshaft sensor rupt. The tooth interrupt is given higher
wheel in which two teeth are missing. priority than the synchronization pro-
The tooth space has a defined position gram.
in relation to the top dead center (TDC) of
cylinder no. 1. The control unit program Ignition interrupt
has to synchronize itself with this tooth The ignition output takes place within a
space. This is done by measuring the times certain crankshaft range, depending on
between two consecutive falling tooth the value from the ignition map. Since the
flanks. The time for the tooth space is specified ignition angle has to be adhered
considerably greater than the time before to exactly, the ignition output is controlled
and after the gap. Following a “short – by an interrupt. Like the synchronization
long – short” sequence the last thing to program, the ignition interrupt is also
be scanned was the falling flank of the called up once per combustion cycle.
second tooth after the space. The control unit program is aware of the
The crankshaft has rotated by 6° for crankshaft position in the 6° framework.
each falling tooth flank that has been de- However, this framework is not accurate
tected by the control unit program. This is enough for ignition angle output. For this
how the control unit program knows the reason, accurate ignition output between
crankshaft position within this time frame. two teeth must take place as well as this
Since cylinder no. 1 is in the tooth space approximate counting for the last 0 to
position in the vicinity of top dead center 6 crankshaft degrees. This is done using
(TDC) or bottom dead center (BDC), an ad- a timer (Fig. 5). Ignition angle output that
ditional signal is required to determine the was purely timer-controller would lead
position. The camshaft sensor provides to an ignition angle output error at high
a different voltage level in both cases. engine speed dynamics.
The control unit is therefore able to Firstly, the ignition coil must be enabled
uniquely assign the crankshaft and for a defined time (the so-called dwell pe-
camshaft positions. riod). In order to do this, the program cal-
culates the switch-on time by calculating
Combustion-synchronous interrupt
Some calculations have to be performed 4 Triggering the interrupt synchronously with
for every combustion cycle. For example, combustion
the ignition angle and the injection have
to be recalculated synchronously with
combustion for each cylinder. The pro-
gram does this by branching to the “syn-
Start synchronization
chronization program” after certain teeth
(Fig. 4). This interrupt takes place after
every 30 teeth (ignition interval) for a four-
Tooth counter elapsed
cylinder engine, and after every 20 teeth Synchronization program triggered
for a six-cylinder engine.
The synchronization program is fixed to Second tooth after the gap: reference mark
a certain tooth position and has to be exe- Preset tooth counter for triggering the

cuted with high priority. For this reason it synchronization program

is activated via an interrupt (triggered by a
command in the tooth interrupt routine).
Electronic control unit Control unit software 29

backwards from the ignition angle at which 5 Dwell and ignition time output
the ignition coil has to be switched off.
This makes it possible to calculate the
tooth after which the ignition coil has to
be switched off (approximate counting Ignition output
in 6° time frame). The remaining angle
(detailed counting 0 to 6°) is converted Switch on the ignition coil
Enter the time for ignition
into an output time using the current en-
gine speed. As soon as the specified tooth Enter time for detailed counting
position has been reached using approxi-
mate counting, a time is loaded with the Synchronization program:

Calculate the dwell period and ignition angle
output value from the detailed counting.
Preset the tooth counter for the dwell period
When this time period expires, the timer time (approximate counting)
triggers an interrupt. The commands that
switch on the ignition coil are programmed
in this interrupt routine. Then the timer Background program
is preset to the dwell period value, which All other activities that do not run in
causes an interrupt to be triggered when an interrupt routine or a time frame are
the timer elapses, switching off the igni- processed in the background program.
tion coil and therefore initiating ignition. At fast engine speeds, the synchronization
program and the tooth interrupt are called
Time frame frequently, leaving little CPU time for the
Many control algorithms have to run background program. The time taken for a
within a certain time frame. Lambda con- complete run-through of the background
trol, for example, has to be processed program therefore increases rapidly with
within a fixed time frame (e.g. 10 ms) so the increasing engine speed. The back-
that the correcting variables are calculated ground program must therefore only
quickly enough. contain low-priority functions.
30 Electronic Control Units (ECUs) Software Development

Software Development QA2F

When: After function implementation
A study of the current series projects together What: Review of each individual function
with the use of development capacities and checking of following docu-
demonstrate that approximately 60% of the ments:
time spent on ECU development has to be 쐌 specification,
devoted to creating the necessary software. 쐌 function description,
For this reason, it is absolutely essential that 쐌 source code,
modern tools and processes be used. 쐌 data definitions, and
쐌 test documentation.
Development Process
Definition of Development Process QA2
A depiction of the development steps in the When: Prior to software delivery
form of a V-model (Figure 1) serves as the ba- What: Review of all QA2F documents
sis for all software-development activities.
This model is used to detail the process steps QA3
which facilitate implementation within a When: Prior to start of series production
product-development department. What: Series-production review for hard-
ware and software
Quality Assessment
Quality assessments are scheduled at defined An essential part of the development process
points of the development process (Figure 2) is also the distinction between specification
for the purpose of process monitoring: and implementation. This separation allows
programming by contract, whereby project
QA1 teams use the software knowledge of ReUse
When: At the start of the project teams who implement the functions (e.g.
What: Resource check (capacity, develop- Keyword 2000 protocol) for various cus-
ment environment, responsibility) tomers. To this end, the project teams write
out function contracts which establish the
QA1F boundary conditions of implementation. The
When: Prior to function implementation scopes and levels of testing of the individual
What: Function specification check functions are determined in the project-spe-
cific quality plan (PQSP) with the customer
requirements in mind.

1 Simplified V-model 2 Development process in detail

System System
Specification Test initialization delivery
Function analysis QA2
QA1 integration/test
Function Function
initialization delivery
Function analysis QA2F
æ UTS0325E

æ UTS0326E

QA1F integration/test
Implementation development
Electronic Control Units (ECUs) Software Development 31

This also includes determining the QA x F Programming Guidelines

scopes. In any heterogeneous system of development
that is spread over countries and continents, a
The PQSP is a central element of project im- standardized procedure for creating software
plementation and should be fully discussed is a vital and integral part of the time to mar-
between vehicle manufacturer and supplier. It ket process. These guidelines address the fol-
lays down, among others things, the responsi- lowing points and are binding for all pro-
bilities, the customer relationships, the devel- grammers:
opment tools, the scopes of testing and docu-
mentation, etc. 쐌 general guidelines (terminology, vocabu-
lary, variation handling),
The process in the ReUse team is set out as 쐌 guidelines for software developments in C
follows: (templates, structure),
쐌 definitions and declarations (include,
defines, typedefs),
쐌 The person in charge of the project
쐌 check instructions (if, for, while, break,
formulates the task (if necessary,
return ...),
adoption of customer request).
쐌 coding specifications and instructions
쐌 The function contract is drawn up
(typecast, arithmetic, pointer),
(with details of task, project, desired
쐌 particular features when using variables
date, scope, and reference documents)
(alignment, address),
and given to the ReUse teams.
쐌 instructions on data consistency
(preemptive), and
쐌 instructions on resource relief.

쐌 The persons in charge of ReUse and These guidelines also serve as a source of
the project discuss the application and knowledge for effective code configuration in
establish the scopes and deadlines order to counteract the limitations in relation
together. to memory capacity and run time in the pro-
gramming of microcontrollers.

쐌 The person in charge of ReUse decides

on the variant and version handling
and implements the task.
쐌 All documents are handed over to the
person in charge of the project on

In order to ensure that the created software

can be reused to the greatest possible extent,
there are C-programming guidelines binding
an all programmers which are called up in the
relevant reviews (e.g. QA2F).
32 Electronic Control Units (ECUs) Software Development

Tools for Creating Software As the wide variety of tools demonstrates, the
As well as the formal aspects such as process process involved in creating the software for
and programming guidelines, it is crucially an ECU of the latest generation is highly
important to ensure that the tools are subject complex. Figure 4 provides a simplified
to constant support in the interests of prod- overview of the interplay between the indi-
uct quality. Figure 3 provides an overview of vidual tools from the specification through to
the tools currently used for the various devel- the finished ECU program.
opment phases. Significant features of this
tool chain are: By way of example, two component parts of
쐌 constant support throughout the entire de- the tool chain will now be explained in closer
velopment process and detail:
쐌 product-specific, optimized solutions with 쐌 Design with ASCET-SD and
tools partly developed in-house. 쐌 Vehicle simulation with TCM-Simutec.

3 Tools in the development process 4 Simplified process sequence

Organization: MS Project Design

ASCET-SD Function model
Proto- Implemen- Object model
Design Test
typing tation Prototyping
SD SD Innovator TCM-
StP Codewright LabCar Code generation
æ UTS0327E

Documentation: MS Word

ASCET: advanced simulation and control
æ UTS0328E

engineering tool
ASCET-SD: ASCET software developer
StP: software through pictures
(Aonix) for OO modeling
ESPRIT: engineering software-production
5 Function design with ASCET-SD
user interface for tools
Innovator: Software-development
environment (MID)
Codewright: Software-development
environment (Premia)
DAMOS: database for microcontroller-
oriented systems
INCA-PC: integrated car application system
TCM-Simutec: Vehicle simulator
æ UTS0329Y

ASCET-LabCar: Vehicle simulator for HiL simulation

ClearCase: Configuration-management tool
Electronic Control Units (ECUs) Software Development 33

Design with ASCET-SD Vehicle simulation with TCM-Simutec

ASCET-SD (Figure 5) offers the following To test the functions of a transmission-
functions for designing software: control system in the laboratory, there is a
쐌 interactive creation of function simulator for the vehicle and transmission
descriptions and function models, environment which provides the input signals
쐌 a graphical user interface, for ETC. One such simulator is shown in Fig-
쐌 support of object-oriented design, data- ure 7.
flow-oriented design and state machines. The front panel of the simulator is
equipped with assorted rotary potentio-
The ERCOSEK operating system is an integral meters, switches, and pushbuttons which en-
part of the development environment, and fa- able input variables such as output speed, se-
cilitates real-time simulation of the function lector-lever position, transmission
model. temperature, etc., to be specified.
The top of the simulator accommodates a
ASCET-SD offers the following support for breakbox which permits access to every ECU
the rapid prototyping of functions: pin. Measuring instruments can be easily
쐌 ASCET-SD operates as a bypass computer connected to these sockets to enable, for
for the series ECU, i.e., individual ECU example, a PWM signal of a pressure-regula-
functions run on the PC, while the other tor output to be viewed on an oscilloscope.
functions continue to be executed by ETC The laboratory car also contains computer
(Figure 6). cards which simulate the other ECUs in the
쐌 The connection is established via the CAN vehicle network (e.g. ECUs for engine man-
or the INCA probe. agement, for ABS, etc.) and also their signals.

The next step is the automatic C-code creation Process and Maturity Model
and the creation of the corresponding data A clear definition of the development process
files for the application from the models. and the corresponding implementation in the
projects are made possible by a software de-
For further information, log on to velopment which can be evaluated with a ma-
turity model such as CMM (capture maturity model).

6 Test setup for ASCET-SD in bypass 7 TCM-Simutec (laboratory car)

2 4

1 Fig. 6
1 ASCET-SD and
2 ASCET hardware
æ UTS0330Y

æ UTS0311Y

(ETAS ES 1000.2)
4 ETC-Simutec
(laboratory car)
34 Electronic Control Units (ECUs) Software Development

8 Software layer model The operating system with its services and the
hardware-compatible software are imple-
Transmission software from mented on this hardware:
vehicle manufacturer or Bosch
(OS) driver Input/Output driver
library Device Device Device
driver driver driver

driver The interface layer and program library for
the application software contain:
Operating system Diagnosis Diagnosis Security
handling monitoring software

æ UTS0331E
functions (SSK)
Hardware EEPROM KWP2000 Shift by wire
handling application functions

The application software

Software Structure (customized software) comprises:
The software structure described in closer de- e.g. ASIS (RB/ZF) Transmission software
tail in the following is implemented within AGS (BMW)
the transmission-control system. This layer Software sharing (interface)
model (Figure 8) comprises
쐌 the application software (transmission soft-
ware) with program library provided by Operating System
the vehicle manufacturer or supplier (in It is absolutely essential to use an OSEK-con-
this case by Bosch), forming operating system to fulfill the cur-
쐌 the operating system, rent real-time demands on an ECU. The ER-
쐌 the component driver, and COSEK operating system from ETAS is used in
쐌 the hardware. Bosch transmission control units (available
for all kinds of microcontroller).
The separation of hardware and application
software ensures that the software can be eas- An operating system is subdivided into
ily ported to new hardware platforms. Only processes and tasks (Figure 9):
the second layer, consisting of the operating A process is a function which has no call or
system and component driver (BIOS), has to return parameters.
be adapted. A task consists of different processes and is
The contents of the individual layers will characterized by
now be broken down in the following: 쐌 the sequential execution of processes,
쐌 the allocation of processes 씮 task,
The ECU hardware, the first layer of the soft- 쐌 each task being assigned a priority,
ware layer model, consists of the microcon- 쐌 tasks being assigned to a time base.
troller (here, by way of example, the
MPC555), the memory, the interfaces (SPI, For task changing, there is either cooperative
CAN and UART), and the peripheral chips scheduling or preemptive scheduling (task
(ASICs): management):

Cooperative Scheduling
CPU core SPI TPU MIOS CAN UART In the case of cooperative scheduling,
timer and (2) (2)
a task can only be interrupted between two
Memory, hardware, driver, etc.
processes by a higher-priority task (Figure 10).
Electronic Control Units (ECUs) Software Development 35

The advantages of this procedure are low and a response time that is not dependent on
memory requirement (register banks, stack), process implementation. The disadvantages
simple management, and data consistency. are increased memory requirement (stack,
The disadvantages are the limited response register banks) and data-consistency prob-
time (dependent on the process run time) lems.
and the jitter over the task period.
Mixed Scheduling
Preemptive Scheduling ERCOSEK offers the option of mixing both
Owing to the drawbacks of cooperative types of scheduling in one application.
scheduling, preemptive scheduling is used in A combination of hardware and software
operating systems which operate as real-time scheduling serves this purpose. Figure 12
systems. shows the distribution between cooperative
With this form of scheduling, a higher-pri- and preemptive using the priorities assigned
ority task can interrupt a lower-priority task to the tasks.
at any time (Figure 11). The advantages of A software call starts the operating system.
this procedure are the very short response It can support different application modes
times, the minimal jitter over the task period, (e.g. different task sets for initialization, oper-

9 Processes and task 11 Preemptive task change

Task Activation
and start
Task B

Task B
Process 1

p1B p2B p3B p4B

Process 2

Process 3

p1A p3A p4A

æ STS0332E

æ STS0334E

Prozess n Task A

Time t

10 Cooperative task change 12 Priority distribution

Activation Task B

Start Task B

Task B preemptive

p1B p2B p3B p4B


p1A p2A p3A p4A

æ STS0333E

æ STS0335E

Task A

Time t Distribution
36 Electronic Control Units (ECUs) Software Development

13 Application-mode change ation and ECU run-on, Figure 13). Each ap-
plication mode consists of an initialization
phase and an execution phase. Interrupts are
prohibited during initialization of an applica-
Mode n Mode n+1 tion mode.
Further documents on the subject of
ERCOSEK / OSEK can be found on the

Internet at:
Init Execution Init Execution

æ STS0336E
Acquisition of Input and
Output Variables
Zeit t
Access to the hardware is obtained within the
framework of the software layer model in ac-
14 Hardware access in the layer model cordance with three layers (Figure 14):
쐌 user layer,
쐌 configuration layer, and
User layer 쐌 hardware layer.

Access to hardware capsule: The first implementation example to be

featured is access (A) via global RAM cells.
via global as direct access Table 1 describes the name of the RAM cell,
RAM cells (function call)
the signal direction (input or output), the sig-
nal type (analog, digital, frequency, or PWM),
the description, and the physical conversion.
The second example featured is access via
(B) function interfaces. Table 2 is structured
Configuration layer along the same lines as Table 1, but here the
global RAM cell is replaced by a function call.
Configuration: Filtering:
Access to Elimination of The objective in the configuration layer is
hardware malfunctions to obtain independence from platform and
project in the conversion of the hardware ac-
Coherence: Scaling: cesses into real software. This is achieved on
Data timing Conversion
into physical the one hand by using tools which automati-
variables cally create the C-code for access to the hard-
ware, and on the other hand by using C-
macros which are then resolved on a proces-
Hardware layer
sor-specific basis.


æ STS0337E

Electronic Control Units (ECUs) Software Development 37

1 Hardware access via global RAM cells

RAM cell I/O Type Content Scaling

ugt_Batt In ANA 16 bit Battery voltage 0...25 000 mV
CGT In ANA 8 bit Oil temperature –40...+215°C
ccu_Chip In ANA 8 bit Substrate temperature –40...+215°C
fgt_Fet Out DIG 8 bit Status HSD-Fet 0.. 1 (On/Off)
fpo_L1 In DIG 8 bit Selector lever pos 1 0..1 (On/Off)
fpr_PinM In DIG 8 bit M button 0..1 (On/Off)
NAB In FREQ 16 bit Output speed 0...20 000 rpm
NAB32 In FREQ 8 bit Output speed/32 0...255 rpm/32
NTU In FREQ 16 bit Turbine speed 0...20 000 rpm
NTU32 In FREQ 8 bit Turbine speed/32 0...255 rpm/32
hmv1 Out PWM 16 bit Solenoid-valve output 0...1000 per mil
idr1s Out ANA 16 bit Nominal current 0..12 000 mA
pressure regulator
Table 1

2 Hardware access via functions

Software function I/O Return value Content Scaling

GetHWIO_U_IgnRunCrnk() In ANA Battery voltage 0...32 V
16 bit
GetHWIO_T_TransOil() In ANA Oil temperature –40...+215°C
16 bit
GetHWIO_b_HSD() In ANA Status HSD-Fet 0...1 (on/off)
8 bit
GetHWIO_e_TapUpDwnReq() In ENUM Tip (+/–) function 0 x 00...0 x 40
8 bit
TsHWIO_PRNDL In DIG Transmission control panel 0...1
GetHWIO_s_PRNDL(void) 8 bit
TsHWIO_FreqParams In FREQ Turbine speed Time stamp +
GetHWIO_s_NTU(void) struct counter value
TsHWIO_NAB_DualEdgeParams In FREQ Output speed, Time stamp +
GetHWIO_s_NAB_DualEdge() struct edge can be changed over counter value +
operating edge
SetHWIO_e_NAB_DualEdgeCptr Out – nout Rising, falling, both
Mode( BYTE ) Edge changeover Table 2

A further signal-acquisition component 15 ECU network

involves the exchange of signals via the a

communication interface. The CAN bus
(controller area network) has gained
acceptance in this field in the last few years. ECU 1 ECU 2 ECU 3 ECU 4

CAN replaces the conventional wiring

harness or the previously standard network of
ECUs (Figure 15). The bus system must b
satisfy the following requirements here:
Real-time updating for
safety functions: 10 ms ECU 1 ECU 2 ECU 3 ECU 4
Convenience functions: 10...100 ms
Maximum cable length: 40 m
æ STS0338E

CAN - Bus
Fig. 15
a Conventional
b With CAN
38 Electronic Control Units (ECUs) Software Development

3 Bit rates as a function of cable (bus) length The system must also be resistant to tempera-
Maximum bit rate Bus length
ture and moisture. The CAN bus has also
kbit/s m gained acceptance in the field of automation
1000 40 technology. Table 3 lists the maximum possi-
500 100 ble data rates for different cable lengths.
250 250 Figure 16 shows the circuit-engineering
125 500
implementation of the CAN interface in an
40 1000
Table 3

16 CAN interface In the microcontroller itself, message

handling is conducted via a dual-port RAM
Microcontroller (Figure 17). Since this RAM chip, as the name
suggests, can be described from two sides
CAN controller (CAN transceiver and microcontroller), the
CPU workload is substantially relieved for
signal transfer.
TxD RxD Ref Rs +6V
VCC Complete arbitration (message organization,
CAN transceiver 100nF who sends what when) on the CAN bus is
performed automatically by the CAN trans-
ceiver. It does not require any computing
power in the microcontroller (Figure 18).
A distinction is made within CAN messages
æ UTS0339E

RT CAN-BUS RT between standard and extended data frames

BUS CAN_L BUS (Figures 19 and 20).
termination termination

17 Use of dual-port RAM with the CAN bus 19 CAN standard data frame

Standard Data Frame Inter Frame Space

Message 1 low high
1 11 1 1 1 4 0…64 15 1 1 1 7 3
Message 2 dominant
Bus idle
Start of Frame

RTR Bit (D)

IDE Bit(D)
(reserved (D))
Data Length Code
Identifier Field

Data Field

CRC Sequence
CRC Delimiter
ACK Slot
ACK Delimiter
End of Frame

Message n
æ UTS0340E

æ UTS0342E

Acceptance Message Host CPU
filter management
Arbitration Control CRC Acknowledge
Field Field Field Field

18 CAN-bus arbitration 20 CAN extended data frame

Bus idle recessive

Extended Data Frame Inter Frame Space

NODE A recessive
dominant 1 11 1 1 18 1 2 4 0…64 15 1 1 1 7 3
Bus idle
Start of Frame
SRR Bit (R)
IDE Bit(R)
Extended Identifier
RTR Bit(D)
(2 reserved (D))

Data Field
CRC Sequence
CRC Delimiter
ACK Slot
ACK Delimiter
End of Frame
Bus idle
Data Length Code

æ UTS0343E
æ UTS0341E

NODE B loses the arbitration
switches to receive Arbitration Control CRC Acknowledge
mode Field Field Field Field
Electronic Control Units (ECUs) Software Development 39

The standard data frame is characterized by Object-Oriented Approach

the following data (Figure 19): Vehicle Control
Data capacity: 0...8 bytes As an introduction, Figure 21 shows the 1-2
Identifier length: 11 bits US and 2-1 DS shift curves for a driving pro-
Message length: max. 130 bits gram.
The shift curves shown in Figure 22 extend
In contrast, the extended data frame this system for different driving programs
(Figure 20) has the following from super economy (XE) to super sport
characteristic data: (XS). They clearly show that the upshift point
Data capacity: 0...8 bytes in the sporty driving program moves towards
Identifier length: 29 bits higher vehicle speed or higher engine speed
Message length: max. 150 bits and thereby achieves optimum utilization of
engine performance.
Gear Selection and Adaptive Functions
Transmission control has undergone various
phases or expansion stages within the frame-
work of development. 21 1-2 US shift curve

The basic functions and the adaptive

programs for shifting points and pressure
control are now standard in the field of elec-
Accelerator-pedal position

tronic transmission control (ETC). Where

2-1 RS

1-2 H

the various marques and vehicles differ is in

the different strategies employed in the auto-
matic adaptation of the transmission to the 50 1
driving style and the traffic situation. This
also represents an area of software which is
being increasingly taken up by the vehicle
manufacturer directly and which is no longer
in the hands of a supplier. The adaptive func-
æ STS0344E

tions for shifting-point control and pressure 0 30 km/h
control have already been discussed in the Vehicle speed υF
chapter sections entitled Shifting-Sequence
Control and Adaptive Pressure Control.
The following text will now deal with 22 1-2 curve with several driving programs

Bosch-specific implementations of automatic

adaptations (learn functions). The adaptive %
shifting strategy determines a gear cyclically 100

from the driver command, the vehicle status,

Accelerator-pedal position

2-1 RS XS
2-1 RS XE


and the driving situation. It is adaptive in re-

1-2 H
1-2 H

lation to the driver type (sportiness) and also

takes into account automatic or manual gear 50 1
preselections (tip/nudge operation, as is fa-
miliar, for example, from Porsche Tiptronic©
or BMW Steptronic©). The complete software
package has been modeled with an object-
oriented approach for optimum reuse.
æ STS0345E

0 50 km/h
Vehicle speed υF Fig. 21 and 22
1 Upshift
40 Electronic Control Units (ECUs) Software Development

To select a shift curve using the driver type On the other hand, there are the variables
and the total running resistance, it is necessary which are crucial to shifting-point selection:
for them first to be recorded and evaluated 쐌 driver type,
once. The overall structure of vehicle control 쐌 driving situation, and
shown in Figure 23 serves this purpose. 쐌 driving program.
On the one hand, there are the variables This is made clearer with the newly
which determine the vehicle and its status: arranged graphic overall structure of gear
쐌 transmission control panel (TCP), selection in Figure 24. Each of these variables
쐌 the transmission itself, is then further divided into different sub-
쐌 the engine, evaluations.
쐌 the accelerator-pedal position, and
쐌 the vehicle variables (e.g. vehicle speed, The driver-type evaluation specifies whether
wheel speed, etc.). the current driving style is economical or
sporty. The driver-type determination as
23 Overall structure of vehicle control featured in Figure 25 can be shown for this
Determine Cyclical ()
The result of the driver-type determination is
a driver-type counter (Figure 26) with an allo-
Determine status () Determine driver type () cated driving program (XE to XXS).
Transmission Driver type
control panel
Driver-type evaluation is followed by hill
Engine Determine drive situation () recognition (based on the vehicle running
Driving situation resistance), which distinguishes between
different types of uphill and downhill driving
Vehicle variables Set transmission () with the following allocation (Figure 27):
Drive program B0 Downhill 2
Accelerator pedal
B1 Downhill 1
æ STS0346E

B2 Level
Vehicle control
B3 Uphill 1
B4 Uphill 2

24 Overall structure of gear selection 25 Driver-type determination

Determine driver type ()

variable Determine driver type () Give driver type ()
Accelerator Administrator Determiner
Transmission Driving
control panel situation Startup evaluation
Gear selection
Engine Driver type
Gradient evaluation
Electronic Kickdown evaluation
stability program Linear-acceleration Counter
Display Special evaluation
æ STS0348E
æ STS0347E

Driver-type recognition
Electronic Control Units (ECUs) Software Development 41

This is followed by determination of the Summary

driving situation (Figure 28). The following features can be summarized for
the object-oriented adaptive driving strategy:
Ratio Criteria (RC) 쐌 cyclical determination of the gear,
Driver type and driving situation form one of 쐌 consideration of driver command,
three ratio criteria (RC), all of which demon- vehicle status, driving situation,
strate different functions: 쐌 adaptive in relation to driver type
RC Driver Type 쐌 division into static and dynamic driving
RC driver type makes a gear suggestion using programs,
a shift curve, depending on the relevant dri- 쐌 automatic and manual gear preselection
ver type. The driver thus serves as the (tip/nudge operation),
ratio supplier (RS). 쐌 new, flexible prioritization procedure, and
쐌 object-oriented structure.
RC FastOff
RC FastOff prevents upshifts if FastOff has
been detected, i.e., a
shift prevention (USP, DSP) takes place. 27 Hill recognition

RC TCP changes the order of priority of the
ratio supplier in accordance with operation of B2
the transmission control panel (TCP). An RC B3
selection adapted to the B4
shift situation (SS) B0
takes place.
æ STS0350Y

26 Driver-type counter 28 Driving situation

Determine drive situation ()

Startup evaluation
Gradient evaluation
Lateral-acceleration evaluation Coordinator
"Kick Fast" evaluation
Special evaluation
Determine status ()

Driver-type counter Hot mode Curve

Urban driving Winter

Variable Hill
shifting sequence
æ STS0349E

æ STS0351E

Fast Off Warm-up

Drive program (characteristics) Driving situation
42 Electronic Control Units (ECUs) Software Development

Diagnostic Functions 쐌 contains filters, time stamps, and assorted

As well as the actual transmission-control flags,
functions, monitoring functions are taking up 쐌 controls the internal program run via the
an ever-increasing percentage of the software. flags.
At present, approximately 30% of the soft-
ware stored in the ECU is assigned to Backup Fault Memory (BFM)
diagnostics. The backup fault memory (BFM) exhibits the
Full diagnostics serves to enhance driving following features:
safety and also with the aid of substitute 쐌 optional use,
functions to increase system availability. 쐌 ring memory, saved in the non-volatile
Diagnostic management must make the memory,
following functions available: 쐌 typically 5 to 10 memory locations,
쐌 management of the fault memory, 쐌 contains fault entries which have been
쐌 CARB fault management, applicable to deleted from the PFM.
each individual fault code (only specific
faults alter the vehicle’s exhaust characteris- Snapshot Memory
tics and are therefore relevant to CARB), The snapshot memory is optional and con-
쐌 filtering (time- and event-controlled), tains further environmental conditions for
쐌 requirement of measures (substitute the first PFM entry.
functions, limp-home),
쐌 provision of data for the application of The entries in this fault memory can be read
diagnostic management for each individual out in the workshop with the diagnosis tester,
fault type, and provide important information relating
쐌 monitoring functions call up diagnostic to vehicle repair.
management; this therefore only has a few
calls in the task lists. Monitoring Functions
The following text deals with the most
Fault Memory important monitoring functions in a
The fault memory is divided into different transmission-control system:
subsections, which are handled differently in
relation to the memory location and accord- Solenoid-Valve Monitoring
ing to the input of the “ignition off ” signal. The following conditions apply to monitor-
ing a solenoid valve in the transmission:
Primary Fault Memory (PFM) 쐌 activation in PWM mode,
The primary fault memory (PFM) exhibits 쐌 alternatively on/off function
the following features: (100%/0% PWM),
쐌 saved in the non-volatile memory 쐌 PWM with fundamental frequency 1 kHz,
(EEPROM), 쐌 pulse-to-no-current ratio dependent on
쐌 typically 10 memory locations, the supply voltage,
쐌 contains fault code/type, environmental 쐌 jump from approx. 95% to 100% in PWM
conditions, CARB/“warm-up” counter, and mode,
flags. 쐌 analog feedback for diagnosis, separate for
“on” and “off ” states,
Secondary Fault Memory (SFM) 쐌 general test condition:
The secondary fault memory (SFM) exhibits battery voltage ≥ threshold (7 V),
the following features: 쐌 fault detection as per Table 4.
쐌 one memory location for each fault code
(only in the volatile RAM),
Electronic Control Units (ECUs) Software Development 43

4 Solenoid-valve monitoring a PRC section must be followed by a “com-

plete check” which generates the relevant
0 – HW HW –
PRC part answer with the aid of correction
0...5 – SW SW – values. The part answers are XORed and form
5...95 – SW SW SW the complete part answer of the PRC to the
95...100 – – – SW question/answer communication. The cur-
100 HW – – HW rent question serves as the input variable into
Key: the PRC (or, more precisely, into each of the
PWM Pulse width modulation
SCGO Short circuit to ground or
PRC parts), i.e., the program run is moni-
open circuit tored on a question-dependent basis. A
SCG Short circuit to ground dummy must simulate the PRC-monitored
O Open circuit
SCP Short circuit to positive points during the initialization phase because
HW Diagnosis by hardware these points are not run through during ini-
SW Diagnosis by software
tialization. Table 4

A further method of solenoid-valve monitor- If faults occur in the PRC, i.e. are detected, an
ing is ISIG evaluation (inductive signature). incorrect complete part answer of the PRC is
This method serves to monitor the voltage formed, which for its part causes an incorrect
characteristic at the solenoid valve and evalu- complete answer to be sent to the monitoring
ates the drop (UISIG) which occurs when the module (watchdog in the external ASIC, see
spool is moved (Figure 29). The objective also section entitled “ASIC”). This increases
here is to monitor the function of the on/off its fault counter by 1 (a fault-free complete
valve. Since the voltage drop is only very answer results in a decrease to the minimum
small and very short (tISIG), a special evalua- fault-counter content 0). The monitoring
tor circuit must be used here. module shuts down the driver stages when
the fault-counter content 5 is reached and
Pressure-Regulator Monitoring initiates a reset at fault-counter content 7.
The pressure regulator must be permanently
monitored because its function is crucial to
the function of the transmission.

Program-Run Check (PRC)

The program-run check (PRC) ensures that
the following incidents are detected:
쐌 mix-up of components,
쐌 double execution of code parts, and 29 ISIG signal characteristic

쐌 skipping of code parts.

Each module or each safety-relevant code

part must have a check point at the beginning
UISIG ≈ 10 mV

and the end in order to ensure a correct run

with maximum high probability. The num-
Voltage U

bers represent the relevant module (function

or process) and range from 0 to 9. The check-
sum algorithm (MISR procedure) ensures the tISIG ≈ 10 ms
sequence relationship.
æ STS0352E

Fig. 29
Each reference (every 10, 20, and 30 ms ref- UISIG Voltage drop
erence) has its own enumerator and delivers tISIG Time interval for
Time t
its own part answer. The last checked point in voltage drop
44 Basic principles of networking | Network topology


With the tremendous speed at which Network topology

computer technology is advancing, the
number of electronic systems in use is A network is understood to be a system
increasing more and more. This growth in which a group of elements can exchange
is also continuing in automotive engi- information via a transportation medium.
neering. However, this also means that If the elements are visualized as nodes and
the complexity of an overall system (the the communication relationships as lines,
vehicle in this case) is on the increase. a picture is created of a network where
Individual systems such as engine man- many nodes are related to several other
agement have been improved over the nodes. The nodes in a communication net-
last few years. However, innovations are work are also often referred to as network
mainly achieved by means of interaction subscribers or stations.
between several individual systems. In motor vehicles, complex control units
The individual components need to be such as those for the engine management
networked so that the multitude of infor- system (Motronic or electronic diesel con-
mation that is managed by the individual trol, EDC), the electronic stability program
systems can also be used elsewhere (ESP), the transmission control system
throughout the system. Different com- or the door modules can be network sub-
munication systems are used depending scribers (Fig. 1). However, a sensor with a
on requirements (e.g. transmission reli- conditioning circuit that merely prepares
ability, fault tolerances, costs). and digitizes a measured value can also
act as a network subscriber and make the
measured signals available to other net-

1 Vehicle system networking


K. Reif (Ed.), Automotive Mechatronics, Bosch Professional Automotive Information,

DOI 10.1007/978-3-658-03975-2_4, © Springer Fachmedien Wiesbaden 2015
Basic principles of networking | Network topology 45

work subscribers. The transport medium Bus topology

via which the communication takes place This network topology is also referred
is referred to as a bus or a data bus. to as a linear bus. The core element of a
bus topology is a single cable to which all
A network topology is understood to be nodes are connected via short connecting
the structure consisting of network nodes cables (Fig. 2). This topology makes it ex-
and connections. This merely shows which tremely easy to add other subscribers to
nodes are interconnected, but does not de- the network. Information is transmitted by
pict underlying details such as the length the individual bus subscribers in the form
of the connection. Every network sub- of so-called messages and distributed over
scriber must have at least one connection the entire bus.
to another network subscriber in order Nodes transmit and receive messages.
to participate in network communication. If a node fails, the data that is expected
Different requirements are made of the from this node is no longer available to
network topology for a variety of commu- the other nodes on the network. However,
nication network applications, while the the remaining nodes can continue to ex-
topology determines some of the charac- change information. However, a network
teristics of the overall network. All net- with a bus topology fails completely if the
work topologies are based on the following main line is defective (due to a cable break,
four basic topologies for example).
Bus topology
Star topology Star topology
Ring topology and The star topology consists of a main node
Mesh topology (repeater, hub) to which all other nodes
are coupled via a single connection (Fig. 3).
Other structures (hybrid topologies) can A network with this topology is therefore
be created by combining these basic topol- easy to extend if free capacity is available
ogies. (connections, cables).

2 Linear bus topology 3 Star topology

46 Basic principles of networking | Network topology

In star topologies, data is exchanged be- Ring topology

tween the individual node connections In the ring topology, each node is con-
and the main node, whereby a distinction nected to its two neighbors. This creates
is made between active and passive star a closed ring (Fig. 4). A distinction can
topologies. In active star topologies, the be made between single rings and double
main node contains a computer that pro- rings.
cesses and relays information. The perfor- In a single ring, data transfers are uni-
mance capability of a network is essen- directional from one station to the next.
tially determined by the performance The data is checked when it is received.
capability of this computer. However, If the data is not intended for this station
the main node does not have to have spe- it is repeated (repeater function), boosted
cial control intelligence. In passive star and relayed to the next station. The data
systems, it merely connects the bus lines that is being transferred is therefore re-
of the network subscribers together. layed from one station to the next in the
The following applies to active and pas- ring until it has reached its destination or
sive stars: if a network subscriber fails or arrives back at its point of origin, where
a connecting line to the main node is de- it will then be discarded.
fective, the rest of the network continues As a soon as a station in a single ring
to operate. However, if the main node fails fails, the data transfer is interrupted and
the entire network is disabled. the network breaks down completely.
In the automotive area, star structures
are under discussion for safety and secu- Rings can also be set up in the form of a
rity systems such as brakes and steering. double ring (e.g. FTTI), in which the trans-
In this case, the risk of a complete network fer of data is bidirectional. In this topology,
failure is prevented by designing the main the failure of a station or a connection
node to be physically redundant. This between two stations can be overcome,
means that several main nodes are used since all data is still transferred to all
to which the nodes whose information is operational stations in the ring.
needed for safe operation of the vehicle However, if several stations or connec-
can be connected in parallel. tions fail, the possibility of a malfunction
cannot be ruled out.

4 Ring topology 5 Mesh topology


Basic principles of networking | Network topology 47

Mesh topology Hybrid topologies

In a mesh topology, each node is con- Hybrid topologies are a combination of
nected to one or more other nodes (Fig. 5). different network topologies. Examples of
In a fully meshed network, each node is such combination are:
connected to every other node. Star bus topology: the hubs of several
If a mode or connection fails it is possi- star networks are interconnected as a
ble for the data to be rerouted. This type linear bus (Fig. 6).
of network therefore has a high degree of Star ring topology: the hubs of several
system stability. However, the cost of net- star networks are connected to the main
working and transporting the message is hub (Fig. 7). The hubs of the star net-
high. work are connected in the form of a ring
Radio networks form a type of mesh to- in this main hub.
pology, since the transmissions from each
station are received by every other station
that is within range.
A mesh topology is bus-like as far as
exchanging messages is concerned, and
star-like regarding data transfers, since
every station receives all transmissions
from every other station, but connection
failures can be overcome.

6 Star bus topology 7 Star ring topology

48 Basic principles of networking | Network organization

Network organization Message-oriented method

In this method it is not the receiver node
Addressing that is addressed, but the message itself
In order to make it possible to transmit (Fig. 8b). Depending on the content of
messages via a network and evaluate the the message, it is identified by a message
contents thereof, the useful data (payload) identifier that has been predefined for
that is transmitted is also accompanied by this message type. In this method, the
data transfer information. This can be ex- transmitter does not need to know any-
plicitly contained within the transmission thing about the destination of the message,
or implicitly defined using preset values. since each individual receiver node de-
Addressing represents important informa- cides whether or not to process the mes-
tion for data transfer information. It is sage. Of course, the message can be re-
needed in order for a message to be sent to ceived and evaluated by several nodes.
the correct recipient. There are different
ways of doing this. Transmission-oriented method
Transmission characteristics can also be
Subscriber-oriented method used to identify a message. If a message is
The data is exchanged on the basis of node always transmitted within a defined time
addresses. The message sent by the trans- window, it can be identified on the basis
mitter contains the data to be transmitted of this position. By way of a safeguard,
and also the destination node address this addressing is often combined with
(Fig. 8a). All receivers compare the trans- message or subscriber-oriented address-
mitter receiver address to their own ing.
address, and only the receiver with the
correct address evaluates the message. Bus access method
The majority of conventional communi- A node must access the bus in order to
cation systems (such as Ethernet) operate transmit a message. In the bus access
using the subscriber addressing principle. method a distinction is made between
Predictable methods in which the bus
access is determined by certain time-
dependent network characteristics,
whereby only one node can transmit
at a time and
8 Addressing modes Random methods whereby any node
can attempt to transmit data if the bus
appears to be free

Adr 1 Adr 2 Adr 3 Adr 4 In the predictable method the bus access
right is determined before bus access.
Adr 3 It can thereby be ensured that only one
subscriber is using the bus at a time.
Access collisions because of simultaneous
bus usage will be prevented if all subscrib-
Id 1 Id 2 Id 2 ers use this method.
Id 4 Id 3
Id 3 Id 3 Id 5
Id 6 Id 5 Id 6
Fig. 8
Id 7

a Subscriber-oriented
Id 3
b Message-oriented
Basic principles of networking | Network organization 49

In the random method, the nodes can Multimaster

simultaneously attempt to use the bus as In a Multimaster network, several nodes
soon as it appears to be free. The timing can access the transport medium indepen-
of the bus access is therefore random. dently without the assistance of another
There is a risk of transmission collisions node. Bus access is uncontrolled. Every
using this method, which will require at- node can access the bus and transmit a
tention. This can be dealt with by repeat- message if the bus appears to be free.
ing transmissions after a collision has been This means that each node is its own mas-
detected (e.g. Ethernet), by giving the ter, and that any node can start a message
transmissions different codings (CDMA), transfer with equal status. However, this
controlling communication via a master or also means that collision detection and
prioritizing message types or transmitters. handling methods have to be in place.
For example, this may be in the form of a
Time division multiple access (TDMA) decision-making phase with prioritization
TDMA is a deterministic (predictive) ac- or delayed transmission repeats. The use
cess method. In this case each node is as- of priority control prevents a bus conflict
signed a time window in which it is allowed if several nodes wish to use the bus at the
to transmit (a priori). A fixed schedule same time, since the network node that
is therefore required for the network. has high priority or wishes to transmit
There is not usually a main communication a message with high priority forces its
subscriber controlling the communication way through in the event of a conflict and
procedure. However, concepts exist in transmits its message first. Normal mes-
which in which it is possible to switch be- sage transmission resumes when the line
tween different schedules if necessary. is free again.
The internal clocks of the different stations The Multimaster architecture has a posi-
must run extremely synchronously with tive effect on the availability of the system,
TDMA, since the transmit windows have since no individual node is in control of
to be adhered to with extreme precision. communication whose failure would lead
to total communication breakdown.
In the master-slave system, one node
on the network operates as the master.
This node determines the communication
frequency by interrogating its subordinate 9 Master-slave method

nodes (slaves). A slave only replies if it is

spoken to by the master (Fig. 9). However,
some master-slave protocols allow a slave Master
to contact a master in order to transmit a
message (e.g. transmit information about
the position of the power-window unit to
the door module).

Slave Slave Slave

1 2 3
50 Basic principles of networking | OSI reference model

OSI reference model ranged functional areas (layers). Not all of

the layers in the OSI model are needed in a
Network protocols are usually defined simple communication system. Layers can
in layers, which combine properties and also be combined for many applications.
tasks. The properties of the deeper-lying Network protocols in the automotive area
layers are assumed in the next level up. are often divided up into
This has the advantage that individual Physical layer
layers are exchangeable, provided that Communication layer and
the interfaces that are provided between Application layer (user layer)
the layers remain unchanged.
The ISO OSI reference model (Open Physical layer
Systems Interconnection) provides a basis The electrical and procedural parameters
for describing and comparing many com- of the physical connection between the
munication protocols. This was developed network subscribers is defined in the
by the ISO (International Standardization physical layer.
Organization) and led to the adoption of
international standards by ISO and IEEE Signal level
(Institute of Electrical and Electronic In digital technology, data is represented
Engineers). by sequences of the two binary statuses,
In the OSI model, data communication 0 and 1. In order to transmit the data on
systems are depicted in different layers a bus, these statuses must be represented
(Fig. 10). The complex task of data commu- on the transmission agent. It is particularly
nication is distributed among clearly ar- important to avoid short-circuits on the

10 OSI reference model 11 Creating dominant and recessive levels


Data Data
User layer line

Application layer E1 E2
T1 T2

Communication layer
Presentation layer
Control Status T Bus level
Session layer
0 conducts 0V
Transport layer
1 blocks 5V
Network layer
E1 E2 Bus level
Data link layer 0 0 0V dominant
0 1 0V dominant
Physical layer 1 0 0V dominant
1 1 5V recessive
Physical layer


Physical connection
Basic principles of networking | OSI reference model 51

bus when one node is transmitting a status If no data is being exchanged, the bus level
of 1 and another is transmitting a status is 5 V (microcontroller operating voltage,
of 0. Fig. 11). When the start bit is transmitted
The binary statuses can be depicted in (dominant level), the other station con-
many different ways. The serial interface nected to the bus (receiver) is notified
of the PC, for example, uses +12 V and that a data transfer is starting (Fig. 12).
-12 V, and CAN-B uses voltages of 0 V and The length of the start bit determines a
5 V. The voltages of the serial interface are bit time that represents the basis for the
unsuitable for a bus, since short-circuits entire data transfer. Every subsequent
can occur if several subscribers wish to data bit has the same length. The recipro-
transmit conflicting binary statuses simul- cal of this time corresponds to the data
taneously. transfer rate, i.e. the number of bits that
If the coding allows one level to over- can be transmitted in one second in a
write another, the overwriting level is re- continuous data stream. All participat-
ferred to as dominant, and the subordinate ing stations must be set to the same data
level as recessive. transfer rate.
It is also possible to depict dominant After the start bit has been received,
and recessive levels using visual media. the transmission of an 8-bit data word
A status of 1 (recessive) then corresponds commences (1 byte) with the lowest signif-
to e.g. dark, and a status of 0 (dominant) icant bit (LSB, Low Significant Bit). The re-
corresponds to light. In an optical fiber, ceiver that has synchronized itself to the
an individual node can override all of the start bit scans the data bus between each
others by feeding light into the conductor. data bit and therefore assembles the trans-
ferred data byte.
Bit stream The eighth data bit is followed by the
The application information cannot usu- parity bit. This bit indicates whether the
ally be transmitted directly. In order to number of transmitted ones is odd or even.
make transmission possible, the informa- It therefore allows the receiver to perform
tion is first incorporated as a payload in a simple check for possible transmission
the frame of a message that contains infor- errors. The sequence is completed with
mation to be transmitted. Since all proto- the stop bit, which is placed onto the bus
cols have been developed in accordance
with different requirements, the frame
format differs from protocol to protocol. 12 UART interface transmission frame

The frame needs to converted into a

bit stream to actually transmit the infor-
mation, i.e. a sequence of bits that can be
transmitted via the transport medium as
physical states.

Example of the UART interface

Start bit
The microcontrollers that are used in
8 data bits (1 byte)
control units have a simple interface
(UART, Universal Asynchronous Receiver/ Parity
Transmitter) on the chip, via which they Stop bit
can communicate with the outside world

(e.g. with a PC). The essential features of

a data transmission are read out via this
52 Basic principles of networking | Control mechanisms

with a dominant level. The next data trans- Control mechanisms

fer can then take place.
Event control
Communication layer In an event-driven bus system, messages
Control units can only interconnect and are transmitted as soon as an event that
exchange data if they speak the same triggers the transmission of a message
“language”. This language determines has occurred (Fig. 13a). Examples of such
the rules that are used to exchange data events are:
between the individual network sub- Pressing a button on the air conditioning
scribers. system control panel
The communication layer accepts data Operating the hazard warning flasher
from the application layer, prepares it for switch
transmission and forwards it to the physi- Incoming message that requires a reac-
cal layer. tion (e.g. information from rpm sensor
The essential features of this protocol to speedometer needle motor)
layer are: Expiration of a fixed time period (time
Message frame format frame, e.g. 100 ms), after which mes-
Bus access control sages are transmitted cyclically
Message addressing
Detection and handling of collisions Since the stations are not synchronized
Network node synchronization with each other, situations where several
Checksum calculation stations wish to access the bus simultane-
ously are unavoidable. In order to allow
Application layer
The application layer consists of the appli-
cation that processes and provides the in- 13 Event control

formation. The application layer is the only

protocol layer to be affected by user or
a Occurrence of events
sensor input. high



1 2 3


b Occurrence of events

3 7

2 6

1 3 4 2 5 6 7

Basic principles of networking | Control mechanisms 53

a message to be transmitted without falsifi- Timer control

cation, only one station at a time can trans- In the most recent developments in dy-
mit data on the bus. Collision avoidance namic driving systems such as brakes and
mechanisms are available for preventing steering, an increasing number of mechan-
or solving bus conflicts. ical and hydraulic components are being
If a node wishes to transmit a message replaced with electronic systems (x-by-
whilst the bus is occupied, the transmis- wire). Mechanical connections such as
sion is delayed (Fig. 13b). A station that is the steering column are becoming super-
ready to transmit must then wait until the fluous, and the functionality thereof is be-
transmission that is currently in progress ing taken over by sensors and actuators.
has been completed. The reliability, safety and failure tolerance
Since bus access is subsequently rene- requirements of these systems are ex-
gotiated, the transmission may be delayed tremely high. This means:
yet again. These delays become problem- Messages must be received on time
atic if the bus becomes overloaded by a The latency time of critical messages
large number of network subscribers that must be extremely small
wish to transmit messages. In this case The system must have a redundant
messages may be lost if the transmitter design
abandons the transmission due to exces- The failure of a node must affect the rest
sive delays. of the system as little as possible and
Event-driven bus systems are suitable It must be possible to achieve a safe op-
for reacting to asynchronous (unforeseen) erating status from any fault situation
events as quickly as possible. In an ideal
case, they reduce the delay between the X-by-wire systems require close network-
occurrence of the event and the message ing by the various components. The exter-
transmission (latency time) compared to nal increase in complexity places new de-
time-driven systems. However, the latency mands on the safety, failure tolerance and
time can vary considerably depending on availability of the communication system.
the network loading. The demands that are made of the elec-
tronic and network architecture therefore
Advantages also increase. A reliable, fault-tolerant net-
High level of flexibility and capability of work architecture is required so that data
retrofitting new nodes in the network is transmitted with guaranteed transmis-
Good response time to asynchronous sion characteristics, and electronic system
external events malfunctions are handled in the most effi-
Bus usage depending on event fre- cient way.
quency in line with requirements
No network loading by unused events, System architectures for real-time applica-
since only events that have actually tions meet these requirements because
occurred trigger a transmission their behavior is predictable and verifiable
because of the way in which they are con-
Disadvantages structed. In these protocols, time windows
Static bus occupancy, non-deterministic within which a node is permitted to trans-
(i.e. not possible to prove that a message mit are assigned to the control units in the
was transmitted at the right time) communication network (nodes) during
network planning (Fig. 14). In order to com-
ply with the time window, the nodes must
be synchronized as precisely as possible.
54 Basic principles of networking | Control mechanisms

All transmissions are processed sequen- Advantages

tially in accordance with the network plan- Deterministic system
ning (without collisions). Once each node Punctual data transmission
has transmitted its message, the cycle re- Reliable detection and isolation of
starts with the first transmitter. This makes defective network nodes
it possible to determine how chronologi-
cally up-to-date the data is at any time. Disadvantages
Since missing messages are detected Overall system must be planned for
immediately, time-triggered concepts are distributed developments
more reliable than event-driven systems. Capacity for expanding the communica-
If a fast data rate is required in a time- tion system must be planned in
triggered system, the time delay between Good response time to asynchronous
the occurrence of an event and the trans- external events
mission of the data can be so small that
the system complies with strict real-time Composability
requirements. If a communication system allows inde-
The bus can be protected from unau- pendently developed subsystems to be
thorized access by a bus guardian. The bus integrated in an overall system, it is said
guardian prevents a defective node from to support composability. An important
interfering with network communication criterion when doing this is that the prop-
by transmitting messages outside the rele- erties that have been assured for the func-
vant transmit window. tionality of a subsystem are not adversely
These characteristics make it possible affected by adding other subsystems.
to create redundant, fault tolerant systems If this has been ensured, the checking
in which transmission errors can be reme- of system functionality is restricted to
died and faults in the network can picked subsystem checking that can be carried
up by network nodes that can provide the out by the constructor of the subsystem.
functionality without errors. If a communication system supports
composability, changes can be made to a
control unit without affecting the function-
ality of other control units. It is therefore
not necessary to recheck the entire system
after integrating a modified control unit –
it is sufficient to check that the individual
subsystems are operating reliably. Com-
14 Timer control posability therefore reduces the time and
cost of integrating new subsystems. This is
the only way to increase the complexity of
Occurrence of events
the electronics in the vehicle.

Max. output
Time slots delay
Sensor measuring principles 55

왘 Overview of bus systems used in vehicles


high-speed CAN low-speed CAN
Definition Controller area Controller area Local interconnect Time-triggered protocol
network network network
Bus type Conventional bus Conventional bus Conventional bus Conventional and
optical bus
Domains Drivetrain Comfort/ Comfort/ Safety-related
convenience convenience networking
Applications Engine management, Body and comfort Low-cost expan- Networking in safety-
transmission control and convenience sion of CAN bus for related environments
and ABS/ESP net- electronics net- simple applications such as brakes, steering,
working working in the comfort and railway signal boxes or
convenience electron- aircraft landing gear
ics area
Most frequently Linear bus Linear bus Linear bus Star topology
used topology
Data transfer rate 10 kbit/s to 1Mbit/s Max. 125 kbit/s Max. 20 kbit/s Unspecified,
typ. 10 Mbit/s
Max. number of nodes 10 24 16 Unspecified
Control mechanism Event-driven Event-driven Time-driven Time-driven
Bus lines Copper conductors Copper conductors Copper conductor Copper conductors
(twisted pair) (twisted pair) (single wire) (twisted pair)
Deployment in all vehicles in all vehicles in all vehicles Premium class vehicles,
aircraft, rail control
Standard ISO 1198 ISO 11 519-2 LIN consortium TTA group
SAE classification Class C Class B Class A Drive-by-wire

MOST Bus Bluetooth Flexray

Definition Media oriented Proprietary name Proprietary name
systems transport (Danish king)
Bus type Optical bus Wireless Conventional and optical bus
Domains Multimedia and Infotainment Multimedia and Infotainment Deployment across all domains
Applications Transmission of control, Data transfers over short A network system for use
audio and video information distances, e.g. mobile phone in safety-related and simple
integration in the infotain- applications
ment system
Most frequently Ring topology Network topology (radio) Star topology
used topology
Data transfer rate Max. 22.5 Mbit/s Max. 3 Mbit/s (v2.0) Typ. 10 Mbit/s
Max. 723 kbit/s (v1.2) Max. 20 Mbit/s
Max. number of nodes 64 8 active (up to 256 passive) Theoretically up to 2,048
Max. 22 per passive bus/star
Control mechanism Time and event-driven Event-driven Time and event-driven
Bus lines Plastic or glass optical Electromagnetic radio waves Copper conductors
waveguides (twisted pair)
Deployment Premium class vehicles made All vehicles, connection be- Pilot application
by European manufacturers tween multimedia equipment
and infotainment system
Standard MOST cooperation Bluetooth SIG Flexray consortium
SAE classification Mobile Media Wireless Drive-by-wire

Table 1
56 Automotive networking | Cross-system functions


Electrical and electronic systems in Cross-system functions

motor vehicles are often not indepen-
dent of each other but influence and If you examine the signals that are pro-
complement each other. For this reason, cessed in the individual systems, it be-
signal lines were used in previous injec- comes evident that many signals are
tion and ignition systems in order to needed in several control units. For exam-
simplify communication between these ple, the driving speed is evaluated in the
two systems. However, the increasing electronic stability program (ESP) for the
number of electronic systems rapidly vehicle dynamics control, in the engine
increased the demand for and the scope management for the automatic speed
of the information that was being ex- control (cruise control) and in the car
changed. The number of signal lines sound system for the speed-dependent
and plug connections that is required volume control.
increased accordingly, meaning that the The preparation of these variables from
technology that has so far been used was sensor signals requires computing power
approaching the limit of its capability. and therefore hardware and software re-
sources. It is therefore advisable for these
The solution was provided by the develop- variables always to be calculated in a con-
ment of serial bus systems, with which trol unit and transmitted to other control
large volumes of data from different units via a communication network.
sources can be transferred. A serial bus However, in order to implement cross-
system was first used in a vehicle in 1991, system functions, electronic systems can
when the CAN bus was used in the also exchange information and therefore
Mercedes-Benz 500E. influence each other. Intelligent sensors
The demand for additional driving are also considered to be electronic sys-
safety, convenience, economy and stricter tems that prepare the sensor signal in an
legal requirements on the environmental evaluation circuit and put the information
compatibility of motor vehicles can only on the data bus via a bus interface. Pre-
be achieved with the aid of additional elec- crash sensors detect a pending collision,
tronics. The number of electronic systems for example; the airbag control unit then
in vehicles is therefore increasing all the sends the door modules and the overhead
time (Fig. 2). control panel a request to close the windows

1 Automotive networking

K. Reif (Ed.), Automotive Mechatronics, Bosch Professional Automotive Information,

DOI 10.1007/978-3-658-03975-2_5, © Springer Fachmedien Wiesbaden 2015
Automotive networking | Cross-system functions 57

and the sliding roof. This protects the Requirements for

occupants from penetrating objects. bus systems
Another example of a system-encom-
passing function is the adaptive cruise Both financial (e.g. cable costs, component
control (ACC), in which the radar sensor, costs) and technical constraints must be
the engine management, the electronic taken into consideration when a bus sys-
stability program (ESP) and the transmis- tem is being selected. The most important
sion control communicate with each other. technical selection criteria are explained
Distance control from the vehicle in front in the following.
that is adapted to the flow of traffic is
made possible in this system by means Data transfer rate
of engine torque adjustments, automatic This variable specifies the volume of data
brake system intervention and gear that is transmitted during a time unit. The
selection. smallest unit of data is the bit, and the data
Coordination between the individual transfer rate is usually specified in bits/sec-
systems is therefore required for cross- onds. Alternative names for this expression
system functions. Large volumes of data are transfer rate, data rate or bit rate.
must be exchanged to do this. As well as The required data rate is dependent on
powerful components, a powerful commu- the application. A slower transfer rate is
nication system is also required, with a required to switch the air-conditioning
low-cost network that is suitable for auto- compressor on and off than to transfer
motive vehicles. Special serial databus sys- audio signals, for example.
tems have been developed for this purpose.
Interference immunity
The use of bus systems has the following Ideally, the data should be transferred
advantages in comparison to a solution without interference. However, this cannot
that uses conventional wiring: be guaranteed in a motor vehicle because
Reduced costs with less weight and in- of electromagnetic effects. The interfer-
stallation space because of fewer cables ence immunity requirements that are
in the wiring harness made depend on the safety relevance of
Better reliability and functional reliabil- the electronic systems concerned. Lesser
ity due to fewer plug-in connections requirements are made of comfort and
Simplification of vehicle assembly
during production Number of control units in the Mercedes S-class
2 that are networked via CAN
Multiple use of sensor signals
Simple connection of system compo- 60
nents to a bus
Easier handling of equipment and spe- 50
cial equipment variants in a vehicle




W140 W220 W221
(1991) (1998) (2005)
58 Automotive networking | Requirements for bus systems

convenience systems than the antilock Hard real-time requirement: the time
brake system (ABS), for example. specification must be strictly adhered
In order to meet these requirements, to. If the specified response time was ex-
mechanisms that detect transmission er- ceeded, the calculated result would not
rors are incorporated in the network pro- be able to be used. This can lead to seri-
tocols. A simple check can be carried out ous problems in safety-critical systems.
using the parity bit, which is calculated in
the transmitter and is transmitted together For example, if time allowances were ex-
with the useful data. This specifies whether ceeded in the ABS system, the incipient
the number of 1's in the transferred byte is locking of the wheels would not be de-
even or off. This information is checked by tected soon enough and the pressure in
the receiver. Single errors can be detected the master cylinder would not be reduced
using this method. in time. This would result in locked wheels.
Another method is the checksum check. The time allowances must also be strictly
If several data bytes are being transmitted, adhered to for many engine-management
the transmitter calculates a checksum system functions. Delays in transmitting
from the individual data bytes using a pre- injection and ignition signals could lead
defined formula and transmits this value. to engine judder and even misfiring.
The receiver also calculates the checksum These reactions must be avoided, since
of the data bytes that have been received they represent a potential danger. Hard
and compares it with the checksum that has real-time requirements must therefore be
been received. If a data transmission error made of these systems.
is detected, the received data is not used However, this does not necessarily mean
and a repeat transmission is requested. that the transmission of data via a bus sys-
tem also has to be subject to these hard
Real-time capability real-time requirements. Adherence to
A real-time system guarantees that its re- soft real-time requirements is usually suf-
sults are calculated within a fixed time in- ficient. If signals from other control units
terval. The duration of the time interval are needed for functions (e.g. a torque re-
depends on the application. The antilock duction request during a shift operation),
brake system (ABS) must react to the the bus system must transmit the data at
incipient locking of a wheel within a few a faster data transfer speed and with a
milliseconds (wheel speed reduction), smaller time delay so that the overall sys-
whereas response times of 100 ms are tem complies with the specified real-time
adequate for actuating the power-window requirements.
motor. Human beings cannot perceive
delay periods of less than 100 ms. Number of network nodes
Different demands are made of real-time The maximum number of nodes to be inte-
behavior depending on the application: grated varies for different areas of vehicle
Soft real-time requirement: the system operation. The number of nodes for com-
generally adheres to the specified re- fort and convenience systems may be high
sponse time, and if these times are occa- due to servomotor networking (e.g. seat
sionally exceeded, it does not produce adjustment) and intelligent sensors (e.g.
any serious effects (e.g. image jerking rain sensors). Several identical busses can
during picture transmission). be used if necessary.
Automotive networking | Classification of bus systems 59

Classification of bus systems Applications in the vehicle

Because of differing requirements, The overall vehicle system can be divided
bus systems can be subdivided into into four domains or functional areas from
the following classes. the point of view or electrics/electronics:
Class A Chassis
Transfer rates Low data rates Interior and
(up to 10 kBit/s.)
Applications Actuator and sensor
Representative LIN In the drivetrain and chassis domains,
the emphasis is primarily on real-time
Class B
applications. In the interior domain, the
Transfer rates Average data rates
(up to 125 kBit/s.) main focus is on multiplex aspects in net-
Applications Complex mechanisms working. Mainly multimedia and infotain-
for error handling,
control unit networking
ment applications are networked in the
in the comfort functions telematics domain.
Representative Low speed CAN

Class C Real-time applications

Transfer rates High data rates The networking of these systems makes
(up to 1 MBit/s.) considerable demands of the performance
Applications Real-time requirements, capability of the communication system.
control unit networking
in the drive and running Crankshaft-synchronous processes or pro-
gear functions cesses within a fixed time frame with cycle
Representative High speed CAN times of a few milliseconds are typical.
Class C+ If the system response times are adequate
Transfer rates Extremely high data rates for the task in hand, it is described as hav-
(up to 10 MBit/s.)
ing real-time capability (e.g. rapid ignition-
Applications Real-time requirements,
control unit networking timing advance in the Motronic after a re-
in the drive and running quest from the traction-control system for
gear functions
reducing torque and therefore preventing
Representative FlexRay
the wheel from spinning).
Class D
Transfer rates Extremely high data rates
(> 10 MBit/s.)
3 Domains in the overall vehicle system

Applications Control unit networking

in the telematics and
multimedia functions
Representative MOST

Chassis Drivetrain

Interior Telematics
60 Automotive networking | Applications in the vehicle

The drivetrain and chassis systems are Multimedia networking

assigned to class C. These require fast Mobile communication applications com-
transfer rates in order to ensure the real- bine components such as
time behavior that is required for these Car sound system
applications. They also make considerable CD changer
fault tolerance demands. These require- Navigation system
ments are met by the event-driven CAN Driver-information systems
bus with a transfer rate of 500 kBaud Telephone
(high-speed CAN). Video system
Voice input
Examples: Internet, E-mail
Engine-management system (Motronic Back-up camera
or electronic diesel control, EDC)
Transmission control The networking of these components
Antilock brake systems, ABS makes it possible to have a centrally lo-
Vehicle dynamics control (e.g. electronic cated display and control unit for several
stability program, ESP) applications. Operating procedures can
Chassis control systems (e.g. active body be standardized in this way, and status
control, ABC) information can be summarized. Driver
Support systems (e.g. adaptive cruise distraction is therefore minimized.
control, ACC)
A distinction must be made between con-
Multiplex applications trol data and audio/video data in multime-
The multiplex application is suitable for dia networking. Transfer rates of up to
controlling and regulating components in 125 kBit/s are sufficient for control tasks
the body and comfort and convenience (such as CD changer control), meaning that
electronics area (class B), such as the low-speed CAN bus can be used, for
Displays example. The direct transmission of audio
Lighting or video data requires extremely high
Access authorization with anti-theft transfer rates of more than 10 MBit/s.
warning device The MOST bus is used for this purpose,
Air-conditioning for example.
Seat and mirror adjustment
Door module (power-window unit,
door-mirror adjustment)
Windshield wipers
Headlamp adjustment

The transfer rate requirements are not

as high for class B systems as they are for
class C systems. For this reason, low-speed
CAN with a transfer rate of 125 kBit/s or
single-wire CAN with 33 kBit/s. can be used.
If the transfer rate requirements drop
to less than 20 kBit/s, the low-cost LIN is
more frequently used. Applications are
mainly in the mechatronics area; examples
being the transfer of switch information or
the activation of actuators.
Automotive networking | Coupling of networks 61

Coupling of networks Examples of networked

The network topologies and protocols that
are most suitable for requirements are Topology
used for the different applications. How- The topologies of the communication net-
ever, the different network protocols are works can differ considerably depending
incompatible, meaning that data cannot be on the vehicle equipment. Figure 5 shows
simply exchanged between networks. examples of how the network can be struc-
In this case, help is provided by a gate- tured for different vehicle classes. In some
way. A gateway can be compared to an in- cases, different car manufacturers use dif-
terpreter that receives the “data” from a ferent bus systems for communication.
discussion partner, translates it and passes
it to another discussion partner. Techni- Signal transmission
cally speaking a gateway is a computer Signal types
that reads in the data that is transmitted A wide variety of information can be trans-
by the networks and converts it into an- mitted in a communication network in a
other format. The use of gateways there- vehicle. Some examples are:
fore makes it possible to exchange infor- Engine operating conditions (e.g. engine
mation between different networks. temperature, engine speed, engine load)
A central gateway (Fig. 4a) or several Physical measurements recorded by
distributed gateways can be used (Fig. 4b) sensors (e.g. outside temperature)
to interconnect the bus systems. All bus Control signals for activating servo-
lines are routed to the central gateway. motors (e.g. power-window units)
In the other case, one gateway connects Control-element switch positions
two or more busses. (e.g. for the windshield wiper)
Multimedia data (audio and video) for
transmitting music and speech (e.g. from
radio stations or when handsfree talk-
ing with a cellular phone) and moving
pictures (e.g. when playing a DVD or the
displays from a reversing camera)

4 Gateway structures

a b

LIN Diagnostics Gateway 2

FlexRay Central
CAN-C gateway Gateway 1 Gateway 3
CAN-B ECUn (telematics) (interior)
ECU1 ECU1 MOST Gateway 4
ECU1 ECU2 ECU2 (chassis)
ECUn ECUn ECUn Fig. 4
ECU4 ECU 4 ECU 1 ECU 1 ECU n a Network with
Chassis Drive- Interior Telematics

train ECUn ECU n ECU n ECU 2 ECU 3 central gateway

b Network with
several distributed
62 Automotive networking | Examples of networked vehicles

5 Network topologies

Sensor CAN


cluster CAN

Comfort CAN

Infotainment CAN

Comfort CAN


b Distance control

cluster CAN
Diagnostics CAN

Drive CAN

Fig. 5 LIN
Typical network
topologies for

vehicles Comfort CAN

a Compact class
b Luxury class
Automotive networking | Examples of networked vehicles 63

Resolution (e.g. 1 second). The engine speed, on

The signals must be available with a suit- the other hand, can change extremely
able resolution. Switch positions can rapidly. The times at which the engine
simply be shown as a 1-bit value (0 for management carries out measurements
switch open, 1 for switch closed). Other and calculations depends on the crank-
signals such as digitized analog voltages shaft position and is performed once per
from the engine-temperature sensor or combustion cycle. At fast engine speeds,
the calculated engine speed are shown as this corresponds to a time interval of a
1-byte or 2-byte values, for example, de- few milliseconds, i.e. approx. 3.3 ms at an
pending on resolution requirements. One engine speed of 6,000 rpm for a 6-cylin-
byte can be used to represent 256 values, der engine. However, not every system
and two bytes can be used to represent that needs the speed information for its
65, 536 values (= 256?256). The resolution control and regulation functions depends
for sensor signals with a voltage range on this availability. The engine control
of 0…5 V is approximately 20 mV with unit therefore does not need to output the
a 1 byte representation (= 5 V/256). A speed information on the data bus as soon
resolution of 5 mV requires a 10-bit data as it is calculated. In this case too, data
representation. transmission takes place according to a
The conversion between the binary cyclic time frame. A time frame of 10 ms
value and the physical value must be uni- is normal in the engine control area. This
form so that the transferred signals rep- means that the speed information is trans-
resent the same physical value in all sys- mitted on the bus 100 times per second.
tems. A resolution of 30 rpm is sufficient
for accessing the ignition map as far as Multimedia data
the engine speed n is concerned. A value A multimedia-compatible digital bus
range of 0 to 255?30 rpm (= 7, 650 rpm) system is frequently used in luxury-class
and therefore the entire speed range can vehicles to transfer audio data – as an
consequently be represented by one byte alternative to using analog cables. In ad-
(8 bits). An incrementation of 30 rpm is too dition to improved audio quality, a bus
little for idle-speed control, on the other system of this type offers the advantage
hand. More bits are required to represent that various audio streams and the associ-
the signal with a higher resolution, pro- ated check commands can be transferred
vided that the same measuring range is in parallel. In the automotive industry,
being recorded. the optical MOST bus (Media Oriented
Systems Transport) has proven successful
Output as a the multimedia bus system.
In event-driven systems such as the ones The transfer of audio signals requires
that are primarily used in the automotive on the one hand synchronized transfer
area, the signals can be transferred on between the transmitter and one or more
the data bus when an event occurs. Ex- receivers and on the other hand a high
amples of such events are the operation data rate. In view of the fact that data
of switches for switching on the air-condi- transfer on the MOST bus takes place syn-
tioning system or the windshield wiper. chronously with a fixed clock-pulse rate,
Signals that represent the operating state synchronization is already assured by its
of the engine, for example, are not nec- transfer mechanism.
essarily associated with an event. The The digital transfer of CD-quality audio
engine temperature, which only changes signals, i.e. with a resolution of 16 bits and a
slowly, is cyclically measured in a fixed clock-pulse rate of 44.1 kHz, requires a con-
time frame by the engine control unit stant data rate of 1.35 Mbit/s for one stereo
64 Automotive networking | Examples of networked vehicles

channel. Up to 15 stereo channels can be driving speed to be read out in the work-
transferred in parallel with the current ver- shop via the connected diagnostic tester
sion of the MOST bus (MOST25 with a total (example: the correct assignment of the
transfer rate of up to 24.8 Mbit/s). wheel-speed sensors must be checked for
the ABS functional test).
Data transfer: examples
The following examples show which sig- Engine speed
nals are measured and evaluated in which The injection and (with a gasoline engine)
systems. the ignition timing are output with a reso-
lution of less than 1° of the crankshaft an-
Driving speed gle. In order to ensure real-time behavior,
The ESP control unit calculates the driving the crankshaft position must be recorded
speed from the wheel-speed sensors. This in the engine control unit. The engine-
variable is transmitted on the CAN-C bus speed sensor scans the crankshaft trigger
(drive CAN). The engine-management sys- wheel and relays the signal to the control
tem needs this value for the cruise control, unit, which calculates both the crankshaft
among other things, and the transmission position and the engine speed. This vari-
control unit determines gear changes from able is used to calculate the injection time
the driving speed. The adaptive speed and the ignition angle, for example.
control (ACC, Adaptive Cruise Control) The engine speed is a variable that is
needs the current driving speed to calcu- needed in many other systems. The engine
late the necessary distance from the ve- control unit therefore outputs it on the
hicle in front and use it as a setpoint value. data bus. The shifting points are defined
A gateway transmits the speed informa- in the transmission control unit depending
tion via another CAN bus (instrument clus- on the speed. The engine speed is needed
ter CAN) to the instrument cluster, which for the ASR function (acceleration slip
displays the value via a needle instrument. control) in the Electronic Stability Program
The CAN-B bus (comfort CAN) is also (ESP) – ASR intervention (torque reduc-
connected to the network via the gateway. tion) must not make the engine stall.
Some luxury class vehicles are equipped As in the previous example, the engine
with dynamic seats. The padding of the speed is transmitted to the diagnosis inter-
seats is inflated depending on the speed face and the instrument cluster (display on
and the acceleration, counteracting the rev counter).
the centrifugal force of the driver. This
increases comfort considerably when Turn signaling
cornering. The driver operates the turn-signal lever
The speed information is sent to the (Fig. 6, Item 1). A signal is relayed to the
Infotainment CAN via the gateway and re- steering column control unit via a discrete
layed to the car sound system. This allows line (2) depending on whether the driver
the volume to be adapted to the driving is indicating a right or left turn. This may
speed. The navigation system needs the be a resistance-coded signal, for example.
speed to calculate the position if the GPS The control unit evaluates the signal and
signal is missing (e.g. in a tunnel). detects that the driver is indicating a left
The diagnosis interface is directly con- turn, for example.
nected to the engine and transmission The comfort CAN relays this informa-
control unit via the serial K-line. All other tion to the vehicle power supply control
control units are connected to the diag- unit (3). The indication direction is defined
nosis interface via a virtual K-line that is on the basis of the received information
simulated on the CAN bus. This allows the (normal flash frequency, increased flash
Automotive networking | Examples of networked vehicles 65

frequency in the event of bulb failure). The Load management

front left and rear left turn-signal lights At low revs (idle speed) and when a con-
are then actuated via discrete lines (4, 5). siderable amount of power is being used
The vehicle power supply control unit also up by the electrical consumers that are
transmits the “left turn signaling” informa- switched on, the battery or alternator volt-
tion on the comfort CAN. The gateway (6) age can drop to a low value. The vehicle
relays the information to the instrument power supply control unit calculates the
cluster CAN. The indicator lamp then current status of the vehicle power supply
flashes on the instrument cluster (9). from the current battery voltage, the DF
If the vehicle has a trailer hitch, the in- signal from the alternator (alternator uti-
formation goes to the trailer-recognition lization) and the information about heavy
control unit (7) via the comfort CAN. This current consumers that are switched on
actuates the turn-signal lamp on the trailer with a short switch-on duration. The vehi-
(8) via cables. cle power supply control unit requests an
idle speed increase via the CAN bus if the
Wiper stage 1 vehicle power supply is insufficient at idle
The wiper switch (Fig. 7, Item 1) transmits speed. The engine control unit implements
a signal via a discrete line to the steering the request. If this action does not solve
column control unit (2), which evaluates the problem, the vehicle power supply
the information (e.g. wiper stage 1). The control unit switches off specific consum-
control unit transmits this information on ers such as the heated rear windshield, the
the comfort CAN bus. The vehicle power seat heating or the heated outside mirror.
supply control unit (3) picks up the infor- These consumers are connected to the
mation and relays it to the wiper motor (4) vehicle power supply control unit via dis-
via the LIN bus. The vehicle power supply crete lines.
control unit acts as a gateway between the
comfort CAN and the wiper LIN.

Fig. 6
1 Turn-signal lever
2 Steering column
control unit
3 Electrical-system
6 Data transfer during turn signaling 7 Data transfer during windshield-wiper operation
control unit
4, 5 Turn-signal lights
6 Gateway
4 7 Trailer-recognition
3 5 LIN control unit
3 8 Turn-signal light
Comfort CAN

on trailer
9 Instrument cluster
Comfort CAN

Instrument cluster
Fig. 7
CAN 9 1
2 1 Windshield-wiper
2 Steering column
control unit


7 3 Electrical-system
control unit
4 Wiper motor
66 Automotive networking | Examples of networked vehicles

Multimedia application to its current reception status (the current

The signal sequences during the output of station name, for example) which are up-
a radio signal on the amplifier is described dated where necessary.
in the following as an example of the trans- At the amplifier, the head unit connects
fer of signals and multimedia data in a the stereo channel with a selected input
luxury-class vehicle infotainment system and uses corresponding check commands
with MOST bus (Fig. 5b). to set the properties for outputting the au-
The functions of the vehicle infotain- dio signal, e.g. the set volume. When all the
ment system are controlled via the cen- settings have been made, it instructs the
tral operating unit, the head unit (Fig. 8, amplifier to fade in the audio signal of the
Item 1). In current systems, this unit usu- radio signal, which is then output through
ally has a screen in the instrument panel the infotainment system’s speakers.
and a rotary/pushbutton controller in the Before this, the head unit has if neces-
center console, by means of which the user sary ensured that the output of a previ-
interacts with the system. ously output audio signal has been faded
To output a radio station, the head unit, out, the output of the associated device
which in most systems is responsible for stopped, and the associated audio channel
managing the audio channels of the info- removed.
tainment system, establishes on the MOST In parallel the CAN/MOST gateway (4)
bus a stereo channel of the required audio constantly transmits the driving-speed
quality between the radio tuner (2) and the data, which it receives via the comfort
amplifier (3). It now tunes in the requested CAN, via the MOST bus to the amplifier,
station at the radio tuner via correspond- which has requested a notification for this
ing check commands and if necessary information when the system was started.
makes further settings. Finally, the head The amplifier can use the speed together
unit connects the tuner output with the with further vehicle information to calcu-
previously created stereo channel. For the late additional settings for outputting the
information to be displayed in the user audio signal, e.g. adapting the volume de-
interface, the head unit receives from the pending on the current speed.
radio tuner corresponding data relating

8 Multimedia networking

4 2
Comfort CAN


Fig. 8 3
5 1
1 Head unit
2 Radio tuner
3 Amplifier 6 7

4 CAN/MOST gateway
5 Screen Check commands
6 Control element Multimedia data stream
7 Speakers
Automotive networking | Examples of networked vehicles 67

Control of an automatic steel folding roof four CAN buses and a further four LIN sub-
Vehicles with automatic steel folding roofs buses. In this example, a total of 13 control
are currently very much in vogue. These units are involved in the control of the au-
roofs are automatically opened and closed. tomatic steel folding roof. The text below
The technical realization of this seemingly describes which functions are performed
simple mechanical function poses a huge by the individual control units to move the
challenge for the networking of the elec- roof.
tronics and control units.
The networking depicted in Figure 9 Roof control unit
shows a typical present-day mid-size ve- Power activation of the motors for the steel
hicle with the additional function of an au- folding roof is performed by this control
tomatic steel folding roof. At its maximum unit. This control unit also assumes the
equipment specification, this vehicle has role of complete monitoring of the move-
over 35 electronic control units (ECUs), ment process. Proximity-type sensors
which communicated with each other via which monitor, record and evaluate the

9 Control of an automatic steel folding roof

Instrument cluster
Central OBD socket
processing unit Electrical-system Door control unit Dynamic headl.
(Radio, RadioNavi, control unit driver leveling contr.
(master) Door control unit AWD
passenger control unit
Booster AMP Wiper
(slave) Door control unit Transmission
rear left control unit
Voice control RS/LS
Door control unit Airbag
SMLS control unit
rear right
TV tuner (steering column)
Comfort control unit
MF st.w. Selector lever
Telephone (Kessy)
interface box
Roof control unit IR Electric
(CSC) power steering
NAR Sounder
PTC heating Steering-angle
NGS sensor

Parking aid Engine control unit

A/C control unit NOx

Tire-pressure Seat, driver
monitoring (memory) Brake
control unit
Trailer Seat, passenger Fig. 9
(ABS, ESP, …)
control unit (memory) Systems in blue boxes

cluster are involved in the
Auxiliary heating control of the steel
control unit
folding roof
68 Automotive networking | Examples of networked vehicles

movement of the roof are used for this ated. This also applies to the sliding sun-
purpose. In order to inform the driver of roof, which is activated by the roof control
the current status while the roof is being unit.
opened and closed and to prevent any
damage to the vehicle or the surroundings, Trailer control unit
the roof control unit receives numerous The trailer control unit informs the roof
further parameters from other control control unit whether a trailer is hitched. If
units in the vehicle. this is the case, opening and closing of the
roof is disabled.
Instrument cluster
The instrument cluster receives via the Door control units
gateway the current status from the roof The status of the windows and doors is
control unit and from other participating interrogated by all four door control units
control units, and informs the driver ac- and transmitted to the roof control unit.
cordingly. Example: “Obstacle behind the Opening or closing of the roof is only
vehicle! Roof movement stopped”. begun if all the doors are closed and the
windows are in the correct positions. This
Gateway setting is automatically corrected if the
All the communication requests which are windows are not in the correct positions.
transmitted from one bus system to an-
other are routed via the central gateway. Comfort control unit
The comfort control unit informs the roof
Vehicle power supply control unit control unit of, among other things, the
The vehicle power supply control unit key position. Movement of the roof is en-
checks whether the vehicle battery has abled only if the correct key is inserted.
sufficient charge to enable the roof to be
moved. If necessary, the driver is informed Brake control unit
and no roof movement is performed. The brake control unit uses sensors to
record the wheel speeds and thereby iden-
Parking-aid assistant tify whether the vehicle is moving. The
Before the roof is opened, the parking-aid roof control unit receives the speed infor-
assistant monitors the area behind the mation and enables movement of the roof
vehicle for obstacles to ascertain whether only if the vehicle is stationary.
there is sufficient space available to un-
fold the roof. If an obstacle is detected, Future areas of FlexRay application
the driver is informed and movement Up to now the high-speed CAN bus (CAN
is stopped. In this event, the driver can C) has been used to network control units
decide for him-/herself whether he/she in the drivetrain and in the chassis area. In
wishes to continue moving the roof. In future X-by-wire systems, the mechanical
certain cases the parking-aid assistant can connections, e.g. between steering wheel
detect something which poses no danger and front axle (steer-by-wire) or brake
to the vehicle. pedal and wheel brakes (brake-by-wire),
will be replaced by electrical communica-
A/C control unit tion systems in conjunction with driving-
All windows are automatically closed dynamics control systems.
when the air-recirculation switch is actu-
Automotive networking | Examples of networked vehicles 69

Furthermore, vehicle architectures have ity Program), AFS (Active Front Steering),
up to new often used several CAN buses and ARC (Active Roll Control). Higher-
which are linked to each other via gate- level driving-dynamics control controls
ways to distribute the high data volume. the individual systems without limiting
Future architectures will use, for a fast, their functionality and prevents negative
powerful connection between several mas- interactions. Figure 10 shows the intercon-
ter computers (which assume, for example, nection of these systems with bus system
central functions in the safety and driver- and sensors.
assistance areas), a backbone bus with a The yaw sensor transmits the vehicle’s
high data rate to which in each case sub- acceleration values and yaw rate to ESP.
buses (e.g. drivetrain CAN and sensor CAN If a critical driving situation is detected
in the chassis area) are connected. by ESP using these data (e.g. vehicle over-
The new FlexRay bus system satisfies steer), situation-conditioned brake pres-
the demands which will be placed in future sures and engine-management interven-
on the vehicle architecture, such as, for tions are calculated. Higher-level global
example driving-dynamics control also evaluates
high data rates and guaranteed real- the driver-command steering angle from
time capabilities in the drive and chassis the steering-angle sensor and calculates
areas, a supplementary steering angle, which is
large date volume in the backbone, and converted by AFS. More effective and com-
high failure safety of safety-relevant fort-enhancing control can be achieved by
applications (e.g. X-by-wire) this interaction of the individual systems.
In this way, the vehicle can, for example,
through properties such as be stabilized by a corrective steering
high availability and redundancy by movement already at a very early stage
means of two physically independent such that braking interventions can be
channels, partially or even completely avoided.
high data rates with up to 10 Mbit/s per
data transfer with guaranteed latency,
and 10 FlexRay
synchronicity of all communication
users by means of a global time base. Global driving-dynamics control
Driving ESP
CAN dynamics FlexRay (active
FlexRay is currently still in the develop-
ment stage, although some initial systems
are already in volume production. The fol-
lowing example is a current approach to a Sensor AFS
technology (Active
FlexRay topology.
Excerpt from chassis domain
Concepts for global driving-dynamics Further ARC
compo- (Active
control are being developed to facilitate nents Roll

control which is as cooperative and thus Control)

as effective as possible. Components of
these concepts are ESP (Electronic Stabil-
70 Bus systems | CAN bus


CAN bus to damage. As a result of these different

requirements, buses with different data
In 1991 the CAN bus (Controller Area rates are used that offer an optimum cost-
Network) was the first bus system to be benefit ratio for the field of application
introduced to a motor vehicle in mass concerned. A distinction is made between
production. It has since established itself high-speed and low-speed CAN buses.
as the standard system in the automotive
sector, but the CAN bus is also commonly High-speed CAN (CAN-C)
used as a field bus in automation engineer- CAN-C is defined in ISO Standard 11898-2
ing in general. In imitation of other net- and operates at bit rates of 125 kBit/s to
work types, such as the local area network 1 MBit/s. The data transfer is therefore
(LAN), wide area network (WAN) or per- able to meet the real-time requirements
sonal area network (PAN), this bus system of the drivetrain.
was given the name, CAN. CAN-C buses are used for networking
the following systems:
Applications • Engine-management system (Motronic

The CAN bus is used in various domains in • for gasoline engines or EDC for diesel
the motor vehicle. These domains differ in engines)
the requirements they demand of the net- • Electronic transmission control

work. Due to the fast processes involved in • Vehicle stabilization systems (e.g. ESP)

the area of engine management, informa- Instrument cluster

tion is required much faster here than in
the area of comfort/convenience where Low-speed CAN (CAN-B)
the controlled systems are located further CAN-B is defined in ISO Standard 11898-3
apart and as such lines are more prone and operates at a bit rate of 5 to 125 kBit/s.
For many applications in the comfort/con-
venience and body area, this speed is suffi-
1 Networking of electronic control units by CAN cient to meet the real-time requirements
demanded in this area. Examples of such
applications are:
ECU 1 ECU 2 ECU 3 ECU 4 • Control of the air-conditioning system

• Seat adjustment
CAN-C • Power-window unit

• Sliding-sunroof control

• Mirror adjuster

• Lighting system
way • Control of the navigation system

The CAN bus is finding ever more use in

Diagnosis vehicle diagnostics. Here, the electronic
interface control unit is connected directly to the
CAN bus and thus receives the information
Fig. 1 ECU 5 ECU 6 ECU 7 ECU 8 it needs for diagnostics immediately. Pre-
ECU Control unit
vious diagnosis interfaces (e.g. KWP2000)
ECU 1 to ECU 4
CAN-B are becoming less important.
On the CAN-C

high-speed bus
ECU 5 to ECU 11 ECU 9 ECU 10 ECU 11
On the CAN-B
low-speed bus

K. Reif (Ed.), Automotive Mechatronics, Bosch Professional Automotive Information,

DOI 10.1007/978-3-658-03975-2_6, © Springer Fachmedien Wiesbaden 2015
Bus systems | CAN bus 71

Topology Incoming messages are processed by the

Bus topology transceiver and sent to the CAN controller
In the development of the CAN, particular on the RxD line.
focus was placed on eliminating the need The microcontroller, which runs the ap-
for a central control element for communi- plication program (e.g. Motronic), controls
cation. This approach is most effectively the CAN controller, prepares the data to be
supported by a bus topology in which all sent and evaluates the data received.
network nodes are connected to a bus and
each node is able to receive all information Logic bus states and coding
sent on the bus. The bus topology is most CAN uses two states for communication,
commonly selected during the conceptual dominant and recessive, with which the
design of the communications system. information bits are transmitted. The dom-
In addition to offering favorable electri- inant state represents a binary “0”, the
cal properties, the linear bus topology recessive a binary “1”. NRZ (Non-Return to
has the advantage that the failure of one Zero) is used as the encoding method for
station would not affect the functionality the data transmission. With this method,
of the data transmission system. Further- there is no compulsory return to zero be-
more, additional stations can be connected tween two transmission states of the same
to the system with little extra effort. value.
When it receives messages, the
Star topology CAN transceiver converts the signal level
The use of a central coupler makes it pos- back to logical states. In the process, a
sible to build star topologies. Active as well differential amplifier subtracts the CAN_L
as passive couplers can be used. The use level from the CAN_H level (Fig. 3). If lines
of a star topology achieves a high level of become twisted, disturbance pulses (e.g.
flexibility in adapting to the networking from the ignition system) have the same
task. effect on both lines. Differential data trans-
The coupler is used to build a star topol- fer therefore makes it possible to filter out
ogy and simply forwards the messages to interference on the line.
the individual segments. Since the signal
transit times remain unchanged, the cir-
cuit length of the star topology is the same
as that of a bus topology.
2 Network nodes in the CAN
Data transmission system
Network nodes
Network nodes
A network node (Fig. 2) comprises the mi-
crocontroller for the application software, Actuators Microcontroller Sensors
the CAN controller and the CAN trans-
ceiver (bus driver). The CAN controller is
responsible for the transmit and receive CAN controller
modes. It generates the bit stream for CAN TxD
data communication from the binary data
CAN transceiver
to be transmitted and forwards it to the
transceiver on the TxD line. This amplifies CAN_H CAN_L
the signals, generates the voltage level re-

quired for differential data transfer and CAN bus

transmits the processed bit stream serially
on the bus line (CAN_H and CAN_L).
72 Bus systems | CAN bus

Some transceivers also evaluate the volt- Single-wire line

age level on the CAN_H and CAN_L line The single-wire line is a means of reducing
separately. It would then be possible for manufacturing costs by dispensing with
operation to continue in single-line mode the second line. For this to be possible,
if one of the two bus lines were to fail as however, all bus subscribers must share
a consequence of a short-circuit or cable a common ground that would assume the
break. However, the bus subscribers would function of the second line. The single-
have to share a common ground that would wire version of the CAN bus is therefore
assume the function of the failed line. only an option for a communications sys-
tem of limited spatial dimensions.
Transmission agent and bus coupling The data transfer on the single-wire line
Two-wire line is more prone to interference radiation be-
For the CAN, any transmission agent on cause it is not possible to filter out distur-
which dominant and recessive states can bance pulses as it is with the two-wire line.
be transmitted is worthy of consideration. For this reason, a higher level increase is
An untwisted or twisted pair of wires, de- required on the bus line to improve the
pending on the ambient conditions, is usu- signal-to-noise ratio. This in turn has a
ally used, and the wires are either galvani- negative effect on interference radiation.
cally coupled or decoupled. The two bus The flank steepness of the bus signals
lines are designated CAN_H and CAN_L. must therefore be reduced in comparison
The two-wire line supports symmetrical to the two-wire line. This is accompanied
data transfer whereby the bits are sent on by a lower data transfer rate. Conse-
both bus lines and represented by differ- quently, the single-wire line is only used
ent voltages. This reduces sensitivity to for the low-speed CAN in the body elec-
in-phase interference because the interfer- tronics and comfort/convenience electron-
ence affects both lines equally and can be ics. Thanks to this feature, a low-speed
filtered out. Additional shielding of the CAN with two-wire line would still remain
lines reduces their own radiation emis- functional in the event of a line failure
sions, especially at high baud rates. (CAN_H or CAN_L).

3 Filtering out interference on the CAN bus

a V

3.5 CAN high level

CAN low level

a V

2.0 Differential


Bus systems | CAN bus 73

Voltage level Emission limits

The CAN transceiver converts the logical To permit correct evaluation of bit trans-
states 0 and 1 received by the CAN control- mission, the signal must arrive at each
ler into voltage levels that are fed to the node within the bit time without interfer-
CAN_H and CAN_L bus lines. ence, and yet still at the relevant sampling
The high-speed and low-speed CANs use point. Delays arise from the signal transit
different voltage levels for the transmis- time on the data bus. The maximum per-
sion of dominant and recessive states. missible bit rate thus depends on the total
The voltage levels of the low-speed CAN length of the bus. ISO 11898 specifies
are shown in Figure 4a, those of the high- the bit rate for a defined circuit length.
speed CAN in Figure 4b. The following recommendations exist for
In the recessive state, the high-speed longer lines.
CAN uses a voltage of 2.5 V on both lines. • 1 MBits/s for 40 m (specified)

In the dominant state, a voltage of 3.5 V is • 500 kBits/s up to 100 m

present on CAN_H and a voltage of 1.5 V is (recommendation)

present on CAN_L. • 250 kBits/s up to 250 m

On the low-speed CAN, a voltage of 0 V • 125 kBits/s up to 500 m

is present on CAN_H in the recessive state, • 40 kBits/s up to 1,000 m

and 5 V on CAN_L. In the dominant state,

voltages of 3.6 V and 1.4 V are present on It is possible to connect at least 30 network
the CAN_H and CAN_L respectively. nodes to the bus without the need for ad-
ditional measures.
Reflection-free termination
Reflections of the electrical signals at open
ends of lines would interfere with commu-
nication. To dampen these reflections, the
bus lines are terminated at each end with
a resistor of 120 V.
Alternatively, the terminating resistors
may be integrated into the electronic con-
trol units themselves.

4 Voltage level 5 Data transmission system

a V
Node 1 Node n
1.4 (max. 30)
recessive dominant recessive
Time t CAN_H
b V
CAN_H 120V CAN bus line 120V Fig. 4
a Voltage level of
2.5 CAN_L
CAN_L the low-speed CAN


b Voltage level of the
recessive dominant recessive
high-speed CAN
Time t
74 Bus systems | CAN bus

CAN protocol information. To fulfill this purpose, the

Protocol layers transport layer is responsible for such
For communications protocols, it is functions as arbitration or fault detection
standard practice to arrange thematically and signaling.
coherent tasks as a set of layers, which af-
fords a high level of flexibility in the imple- Physical layer
mentation of a bus system. With the CAN, The physical layer is the lowest level in the
both the CAN hardware and the CAN soft- transport stack. It consists of the physical
ware are subdivided into several layers components of the network, such as the
(Fig. 6). wiring and the voltages used to send the
Application layer
The application layer represents the infor- Multimaster principle
mation in the form of data structures used The CAN protocol supports communica-
by the application. These sets of data tion between network nodes without the
which are to be transmitted are forwarded need for a central control unit. Each node
to the object layer for this purpose. may attempt to send messages at any time.
Whether this attempt is successful or not
Object layer essentially depends on two factors:
The task of the object layer is to manage • Is the bus free before the start of trans-

the messages. The functions of this layer mission?

are used to decide which message should • Has the arbitration phase been passed
be sent at which time. For incoming mes- successfully?
sages, this layer is responsible for message
filtering. This design ensures that, even if any num-
ber of nodes were to fail, it can still be de-
Transport layer termined whether a node is authorized to
The transport layer furnishes the object send.
layer with received messages and pro-
cesses the messages prepared by the ob- Content-based addressing
ject layer for sending in such a way that Unlike other networks, the CAN does not
the physical layer is able to transmit this address the individual network nodes but
the messages that have been sent. Each
6 Protocol layers message has a unique marker, or identifier.
The identifier classifies the content of the
message (e.g. engine speed or power-win-
Application layer dow position). A station is therefore able
to broadcast a message to all other stations
(multicast or broadcast method). These
stations read only those messages whose
Object layer
identifiers are stored in their acceptance
list (message filtering, Fig. 7). In this way,
Transport layer
each station decides for itself whether or
not it needs a message sent on the bus.
The identifier has 11 bits (standard format,
CAN 2.0 A) or 29 bits (extended format,

Physical layer
CAN 2.0 B). With 11 bits in the standard
format, it is possible to distinguish be-
tween 2,048 different CAN messages;
Bus systems | CAN bus 75

in the extended format, this number rises ority (lowest binary value of the identifier)
to over 536 million. is assigned first access, without any data
The advantage of this addressing loss or delay (non-destructive protocol).
method is that the network nodes do not The arbitration principle permits the
require any information about system dominant bits transmitted by a given sta-
configuration and are thus free to operate tion to overwrite the recessive bits of the
fully independently of each other. This re- other stations (Fig. 8). Each station outputs
sults in a highly flexible complete system, the identifier of its message onto the bus
which makes it easier to manage equip- bit by bit, with the most significant bit first.
ment variants. If one of the ECUs requires During this arbitration phase, each station
new information which is already on the wishing to send data compares the level
bus, all it needs to do is call it up from the present on the bus with the level it actually
bus. It is possible to integrate additional possesses. Each station that attempts to
stations into the system (provided they are send a recessive bit but encounters a dom-
receivers) without having to modify the inant bit loses the arbitration process.
existing stations. The station with the lowest identifier, i.e.
the highest priority, makes its way onto
Controlling bus access the bus without having to repeat the mes-
Arbitration phase sage (non-destructive access control).
If the bus is unoccupied (recessive state) The transmitters of lower-priority mes-
and messages are available for sending, sages automatically become recipients of
each station is free to initiate the sending the message just sent by another station.
of its message. The message begins with a They repeat their attempt to send as soon
dominant bit (start-of-frame bit), followed as the bus is free again.
by the identifier. When several stations Without this access control, bus colli-
start to transmit simultaneously, the sys- sions would result in faults. To guarantee
tem responds by employing “wired-and” unequivocal bus arbitration, therefore,
arbitration (arbiter = logical AND operator) it is not permissible for more than one
to resolve the resulting conflicts over bus node to send a message with the same
access. The message with the highest pri- identifier.

7 Addressing and message filtering 8 Bit arbitration

(acceptance check)


Station 1 Station 2 Station 3 Station 4 Bus line
Fig. 7
1 Station 2 transmits
Accept Provision Accept Station 1
0 Stations 1 and 4 accept
the data
Selection Send Selection Selection Station 2
message 0 Fig. 8
1 Station 2 gains first
Station 3 access (signal on the
Reception Reception Reception 0
bus = signal from

Station 1 Station 3
loses the loses the station 2)
arbitration arbitration
0 Dominant level
1 Recessive level
76 Bus systems | CAN bus

With this access method, maximum-pri- • Error frame:

ority messages only have to wait for the If a station detects a fault or error, it
transfer of the message currently being communicates this to the other stations
sent and, with bit times of 130 (CAN 2.0 A) using an error frame.
or 150 (CAN 2.0 B), they have the lowest • Overload frame:
latency. At a data transfer rate of This can be used to create a delay be-
500 kBits/s, this equates to 260 µs or tween a preceding and subsequent data
300 µs. The higher the load on the bus, frame or remote frame. The transmitting
the greater the data transfer's temporal node reports that it cannot currently
offset becomes for messages of lower pri- process another frame.
ority, and thus the greater the uncertainty
over when a message to be sent will arrive For data transfer on the data bus, a mes-
at the recipient. sage frame is created. It contains the infor-
In order that all messages have a chance mation to be transmitted arranged in a
of accessing the bus, the data transfer rate defined sequence. CAN supports two dif-
must be matched to the number of bus ferent formats of frame, which are speci-
subscribers. fied in CAN 2.0 A and CAN 2.0 B. The most
important difference between these two
Priority assignments frame formats is the length of the identi-
The direct consequence of the arbitration fier. A CAN 2.0 A frame has an 11-bit iden-
process is that the identifier also has the tifier, while a CAN 2.0 B frame has a 29-bit
role of prioritizing the frame during trans- identifier divided into two parts (11 bit
mission in addition to identifying the and 18 bit).
frame content. An identifier correspond- Both formats are compatible with
ing to a low binary number has high prior- each other and can be used together in
ity and vice versa. Message priorities are a network. The frames compliant with
derived from the speed at which the mes- CAN 2.0 A and B are shown in Figure 9.
sage content changes or from the impor- They have a maximum length of 130 bits
tance of the message to safety consider- (standard format) or 150 bits (extended
ations, for example. It must not be possible format).
for messages to have the same priority.

Message format
The message transfer on the CAN bus is 9 CAN message format

based on four different frame formats:

• Data frame: Start of frame
The transmitted message contains data Arbitration field
(e.g. current engine speed) that is pro- Control field
Data field
vided by the transmitting station (data CRC field
source). ACK field
• Remote frame: End of
Stations can request the data they
need from the data source (example: frame
the windshield wiper requests how wet space
the windshield is from the rain sensor). IDLE 1* 12* 6* 0...64* 16* 2* 7* 3* IDLE
The data source responds by sending 0

the relevant data frame. Data frame

Fig. 9
0 Dominant level Message frame
1 Recessive level
Bus systems | CAN bus 77

In CAN 2.0 A and CAN 2.0 B data frames, Control field

the sequence of transferred information In a CAN 2.0 A frame, the control field
is arranged in the same way. The start-of- comprises the IDE bit (Identifier Extension
frame bit is followed by the arbitration, Bit), which is always sent as dominant
control, data and CRC fields. A frame is here, followed by a reserved bit for future
completed by the ACK field and the end- extensions, which is sent recessively. The
of-frame. remaining four bits in this field define the
number of data bytes in the next data field.
Start of frame This enables the receiver to determine
The bus is recessive in idle state. The start whether all data has been received.
of frame, represented by a dominant bit, The structure of a CAN 2.0 B frame is
indicates the start of a transmission and, practically identical. However, since the
together with the process of bit stuffing, IDE bit already belongs to the arbitration
serves to synchronize all stations. field, a further reserved bit for future ex-
tensions takes its place and this reserved
Arbitration field bit is sent recessively.
With a CAN 2.0 A frame, the arbitration
field comprises the 11-bit identifier and Data field
a control bit, the RTR bit (Remote Trans- The data field contains the actual mes-
mission Request). sage information comprised of between
With a CAN 2.0 B frame, the arbitration 0 and 8 bytes. A data field in which the
field comprises the 11-bit identifier, fol- length of the data is expressed in 0 results
lowed by the SRR bit (Substitute Remote in the shortest possible data frame with
Request) and the IDE (Identifier Extension a length of 44 or 64 bits. A frame like this
Bit). Both bits are sent recessively and can be used to synchronize distributed
thereby ensure that a CAN 2.0 A frame processes. A number of signals can be
always takes priority over a CAN 2.0 B transmitted in a single message (e.g. en-
frame if it has the same 11-bit identifier. gine temperature and engine speed).
The second 18-bit identifier follows. The
RTR bit completes the arbitration field. CRC field
The RTR bit indicates whether the trans- The CRC field (Cyclic Redundancy Check-
mitted frame is a data or remote frame. sum) contains a 15-bit checksum (frame
The RTR bit is dominant in the data frame check word) across the preceding frame
and recessive in the remote frame. If sta- from the start bit to the final bit of the data
tion A, for example, happens to send a field. The 16th bit (CRC delimiter) in this
message by data frame and station B re- field is recessive and closes the checksum.
quests this message by remote frame at The checksum is a means of detecting
the same time, the arbitration conflict can- possible transmission interference.
not be resolved in this situation by means
of the message identifier: the RTR bit is the ACK field
decisive factor for access authorization. Unlike all preceding fields, the ACK field
First of all, station A wins arbitration (acknowledgment) is not set by the sender
with the sending of the dominant RTR bit of the frame but by a different node that is
and continues to transmit the message. able to acknowledge receipt of the frame
Station B, which has requested precisely directly after the data field. This field
this message, prepares to receive and comprises the ACK slot and the recessive
is able to read the other data sent by ACK delimiter. The ACK slot is also trans-
station A. mitted recessively by the sender and over-
78 Bus systems | CAN bus

written as dominant by a receiver upon the Cyclic redundancy check

message being correctly received. Here, it For every transmitted message, the trans-
is irrelevant whether the message is of any mitter calculates a check sequence from
significance for the particular receiver in the start-of-frame, arbitration, control and
the sense of an acceptance check (message data field. A 15-bit checksum is derived
filtering). Only correct receipt is con- from the bit sequence by means of a gener-
firmed. This signals to the sender that no ator polynomial. The checksum is then
malfunction occurred during data transfer. used to detect errors in the data transfer.
With CRC generation, a defined genera-
End of frame tor polynomial is used to carry out poly-
The end of frame marks the end of the nomial division across a given frame
message and comprises seven recessive range. The remainder forms the checksum.
bits. With seven bits of the same value in Once the CRC field has been received,
succession, the stuffing rule is deliberately the receiver is able to check that the frame
broken (see Error detection). has been transferred correctly by carrying
out its own polynomial division of the re-
Interframe space ceived frame range using the generator
The interframe space comprises a succes- polynomial and checking whether the re-
sion of three recessive bits which serve to ceived checksum matches the calculated
separate successive messages. After a total remainder.
of 10 of these recessive bits, the stations
are permitted to begin transmitting within Frame check
the network again. Until then, the bus re- The frame check involves all bus subscrib-
mains in idle state. ers, senders as well as receivers, checking
Only data and remote frames need to the sent/received data frame for compli-
include the interframe space. Error and ance with the predefined frame structure.
overload frames can be sent immediately The CAN protocol contains a number of
after the last frame. This enables immedi- fixed-format fields (start and end-of-
ate signaling of errors and problems. frame, delimiter) which are checked
by all stations.
Transmitter initiative
The transmitter will usually initiate a data ACK check
transmission by sending a data frame. It is With the ACK check, a receiver acknow-
also possible for a receiving station to call ledges that the frame has been received
in data from a sending station by deposit- correctly by sending a dominant bit in the
ing a remote frame. The network node ACK slot. The sender of the frame can then
that is able to deliver the information re- verify whether a message was transferred
quested makes the information available correctly. An absence of this bit indicates
in response to this query. that a transmission error has been de-
Error detection
The bus line may suffer interference, e.g. Monitoring
electromagnetic interference. To eliminate The sender of a frame continuously moni-
the risk of faulty behavior, the transmitted tors the bus level. It is able to detect a
data must be checked for correctness. transmission error by comparing the
A number of control mechanisms for sent bit and the sampled bit.
detecting errors are integrated in the
CAN protocol.
Bus systems | CAN bus 79

Bit stuffing Fault confinement in the event of failures

Compliance with bit stuffing is checked by Defective stations could weigh heavily on
means of the code check. The stuffing con- bus traffic if they were to frequently send
vention stipulates that in every data frame erroneous messages or interrupt the trans-
or remote frame a maximum of five suc- mission of correct messages by repeatedly
cessive bits of equal state may be sent be- sending an error frame. The CAN protocol
tween the start of frame and the end of localizes station failures by means of
the CRC field. As soon as five identical bit statististical error analysis. A station rec-
states have been transmitted in succes- ognizes the probability of its own malfunc-
sion, the sender includes a bit with the op- tion by how often it aborts messages be-
posite state. The receiving station clears fore other stations send an error frame.
all of these inserted bits after the message The protocol's first measure in this case
has been received (destuffing). is to prevent a station such as this from
This measure permits detection of line continuing to abort transmissions. In an
faults, e.g. short-circuit or burst interfer- emergency, the station shuts down auto-
ence. Rare signal changes would affect matically.
the possibility for synchronization in the

Error handling
If a CAN controller detects a fault or for-
mat error, it interrupts the current trans-
mission by sending an error frame com-
prising six successive dominant bits. This
breaks the stuffing rule that prohibits this
type of bit sequence. If the sender detects
that its message has been interrupted by
an error frame, it stops transmitting and
makes another attempt at a later time.
This effect prevents other stations from
accepting the erroneous message and
thereby ensures consistency of data
across the entire system.
80 Bus systems | CAN bus

Hardware If messages on the bus are managed in

CAN controller of a subscriber extended format (29-bit identifier), the
The CAN controller of a subscriber gener- CAN controllers must be compatible with
ates a frame from the data to be sent con- the CAN 2.0 B specification. There are
taining all the fields required for compli- controllers that only support CAN 2.0 A
ance with the CAN protocol. It then con- and send and receive messages in the
verts the frame into a bit stream. standard format, but that generate errors
CAN only specifies the physical, trans- with messages in the extended format.
portation and object layers. The interfaces Other controllers are able to tolerate the
that are provided in the application layer extended format without generating an
can be arranged differently to suit the field error. These modules can be used in a CAN
of application. In order to provide proper together with controllers that administer
CPU support (electronic control unit mi- the sending and receiving of messages in
crocontroller for application software) for the extended format.
a wide range of different requirements,
semiconductor manufacturers have Basic CAN
brought onto the market products that In modules with basic-CAN implementa-
provide a variety of capabilities. They dif- tion, only the basic function of the CAN
fer neither in the frame format they pro- protocol for the generation of the bit
duce, nor in their error-handling methods, stream is implemented in the hardware.
but solely in the type of CPU support re- For the management of the messages to be
quired for message administration. sent and received, there is an intermediate

10 Basic-CAN module 11 Full-CAN module

Microcontroller Microcontroller
Output Input
buffer buffer
System bus interface
Transmission Test for: Control
control unit acceptance, unit
Transmit Receive
buffer buffer
Basic CAN
controller Transmission Test for:
Transmit Receive control unit acceptance,
buffer buffer faults



CAN transceiver CAN transceiver



CAN bus CAN bus

Bus systems | CAN bus 81

buffer to which the local computer (appli- Modules without local computer
cation software microcontroller) has ac- A further category of CAN module is one
cess (Fig. 10). Since the buffer capacity that is supplied without a local computer.
is limited, the computer must read the These SLIOs (Serial Linked Input/Output)
received data before new messages are are able to input and output data via ports.
received. Message filtering also takes They are therefore suitable for making
place in this computer. A part of the sensors and actuators bus-compatible at
computer's capacity is therefore used for low cost, but they do need a master that
CAN management. Since the computers controls them.
do not usually have sufficient processing
capacity, modules with basic CAN are pri- Transceiver
marily suitable for low bit rates, or for the The bit stream generated by the CAN con-
transmission of fewer messages but at troller is made up of binary signals. They
higher bit rates. do not yet correspond to the required volt-
The advantage of these modules, in com- age levels of the CAN bus. The CAN-bus in-
parison to modules with full CAN, is the terface module, or transceiver, generates
smaller chip surface and the lower manu- the differential signals CAN_H and CAN_L
facturing costs. and the reference voltage Uref from the
binary data stream.
Full CAN
Full CAN implementation is the protocol of Sleep mode
preference in cases where a station has to The CAN comfort bus must remain ready
manage several messages at high bit rates for operation even with the ignition
and the local computer has no free capac- switched off so that functions such as the
ity for communication tasks. They contain radio, power windows or parking lamp
several “communication objects”, each of may continue to operate. The bus sub-
which contains the identifier and the data scribers must therefore be supplied by
of a particular message. During the initial- terminal 30 (permanent positive). After
ization of the CAN module by the local terminal 15 has been switched off (ignition
computer, it is decided which messages off), a CAN node may enter sleep mode
the CAN controller should send and which (standby) to relieve the vehicle electrical
received messages it should process fur- system of as much load as possible. The
ther. Received messages are only accepted transmitter part of the transceiver module
(message filtering) if the identifier matches is switched off in this condition to mini-
one of the communication objects. mize the power consumption in this mode
CAN controllers with full-CAN imple- of operation. However, the receiver part
mentation relieve the burden on the local remains active and checks whether mes-
computer by performing all of the commu- sages are being sent on the bus. In this
nication including message filtering in the way, the CAN controller, which also enters
controller (Fig. 11). standby mode, is able to react to a wake-up
The CAN controller can be coupled to message and fully activate the CAN node.
the microcontroller in the electronic con-
trol unit as a stand-alone module by the
address/data bus. Powerful microcon-
trollers have the CAN controller integrated
on-chip. This type of bus coupling is the
more cost-effective and thus the more
common solution.
82 Bus systems | CAN bus

Data transfer sequence Once the engine-speed information has

Using the transfer of engine speed as an been stored in the transmit buffer, the
example, it shall be demonstrated how CAN controller begins to compile the CAN
data is transferred on a CAN with full-CAN frame. Over the RxD line, the CAN con-
module. troller now monitors the bus. If the bus is
free, it immediately starts to generate the
Transmission bit stream and directs it to the transceiver
The engine management's application soft- on the TxD line; if the bus is occupied, it
ware calculates the engine speed from the waits. From the bit stream, the transceiver
signal from the engine-speed sensor. This creates a signal with the requisite voltage
value is calculated once for every combus- level.
tion cycle. The measured value enters both
the receive buffer and the transmit buffer Reception
of the microcontroller (Fig. 12a). The message originally sent by the engine
The engine-management microcon- management and forwarded on by the
troller is coupled to the CAN controller transceiver is received by all stations
by a parallel interface. The contents of connected to the bus (Fig. 12b). The mes-
the transmit buffer are cyclically passed sage reaches the CAN controller on the
(e.g. every 10 ms) to the transmit buffer RxD lines in the form of a bit stream.
of the CAN controller. A flag notifies the At the first stage (monitoring layer),
CAN controller that a message is ready the CAN controller checks the incoming
to be sent. With this transmit instruction, message for errors using the CRC check-
the engine management has completed its sum. If the message is free of errors, each
part of the task.

12 Data transfer sequence

a b
Engine control unit ABS control unit Instrument cluster
Microcontroller Microcontroller Microcontroller
n n

area n n

CAN controller CAN controller CAN controller

Transmit Transmit Transmit
buffer buffer buffer


CAN bus
Bus systems | CAN bus 83

station responds with an acknowledge CAN is also widely used in industrial auto-
(ACK check). mation. These applications are supported
At the second stage, or acceptance layer, by an alliance of companies in the “CAN in
the message undergoes message filtering. Automation” users group (CiA).
Each station checks whether the received Bosch has concluded contracts with its
identifier is addressed to the particular licensees that guarantee that any CAN im-
station and whether the message is re- plementations will be able to communicate
quired in the application software. If not, with each other. Users will be able to rely
the message is rejected. Otherwise, it on the interaction of any CAN modules.
makes its way to the receive buffer. A flag
notifies the application software that a Characteristics
new message is ready for processing. • Standardized in accordance with
The instrument cluster, for example, ISO 11898
calls up the available message, processes • Prioritized communication

the engine-speed information and calcu- • Data transfer rates: up to 1 MBits/s

lates triggering signals for the actuator of • Data capacity: up to 8 bytes per message
the rev counter. • Real-time response: the data protocol is
sufficient for the real-time requirements
Standardization in the motor vehicle
The International Organization for Stan- • Non-destructive bus-access method
dardization (ISO) and SAE (Society of Au- • Low power consumption
tomotive Engineers) have issued CAN stan- • Flexibility of configuration

dards for data exchange in automotive ap- • Simple and economical design with

plications: twisted line pairs

• For low-speed applications up to • Very high reliability of data transfer
125 kBit/s: ISO 11 519-2 and 11 898-3 • Fault detection and signaling
• For high-speed applications faster than • Localizing of failed stations
125 kBit/s: ISO 11 898-2 and SAEJ 22 584 • Handling of intermittent and permanent
(passenger cars) or SAEJ 1 939 (commer- faults
cial vehicles) • Short-circuit resistance
• An ISO Standard for diagnosis via CAN • The number of nodes is theoretically un-
has also been published as ISO 15 765 limited. However, a limit arises in prac-
tice from the capacitive load of the bus
Standardization makes it possible for and the increasing latencies of messages
components of different manufacturers when a high number need to be sent.
to function together. No adaptations are
84 Bus systems | LIN bus

LIN bus The LIN bus is suitable for low data rates
of up to 20 kBit/s and is typically limited to
Overview a maximum of 16 bus subscribers.
The increasing use of mechatronic systems The electrical interface can be created
in the motor vehicle gave rise to the idea easily and cost-effectively in the network
of designing a cost-effective bus system nodes. As far as the nodes are concerned,
as an alternative to the low-speed CAN. a distinction is made between the master,
In 1998, several automotive manufacturers which is generally an electronic control
founded a consortium with the aim of de- unit connected to a superordinate bus sys-
veloping a specification for a serial bus for tem, and the slaves. These are intelligent
the networking of sensors and actuators in actuators, intelligent sensors or, quite sim-
the body electronics area. ply, just switches with additional hardware
It was believed that a bus system with for the LIN-bus interface.
simple bus protocol and a simple sequence The bus subscribers are usually ar-
control would make it possible to use even ranged in a linear bus topology and con-
low-capability microcontrollers without nected to each other by a single-wire line.
additional hardware for the communica- This topology, however, is not explicitly
tion interface. specified.
The workgroup's resulting LIN bus spec- Communication on the LIN bus takes
ification was introduced into mass produc- place in a time-synchronous manner,
tion with the Mercedes-Benz SL as early as whereby the master defines the time grid.
2001. Consequently, there arises a strictly deter-
The name, LIN (Local Interconnect ministic LIN bus response.
Network), is derived from the fact that Figure 1 shows an example of a LIN net-
all electronic control units are located work as a subbus in the roof/wiper area of
within a demarcated installation space the motor vehicle. Here, the bus comprises
(e.g. in the door). The LIN, therefore, is a central electronic control unit, as the
a local subsystem for supporting the vehi- master, and the four slaves: mirror, garage-
cle network by means of superordinate door opener, rain/light sensor and wiper
CAN networks. actuator. The master also functions as a
gateway to the Chassis CAN, the Body CAN
and the Diagnostics CAN.

1 LIN bus with master and slave nodes

Slave 1 Slave 2 Slave 3 LIN slave 4

Mirror Garage-door Rain/light Wiper
opener sensor actuator
CAN Master
Central ECU
Roof/wiper LIN

Body CAN Chassis CAN

Bus systems | LIN bus 85

Applications Data transmission system

The LIN bus as a means of networking me- The LIN bus is designed as an unshielded
chatronic systems can be used for many single-wire line. The bus may adopt two
applications in the motor vehicle for which logical states:
the bit rates and variability of the CAN bus • The dominant level corresponds to the

are not essential. Examples of LIN applica- electrical voltage of approx. 0 V and
tions: represents logical 0
• Door module with door lock, power-win- • The recessive level corresponds to
dow drive and door-mirror adjustment battery voltage UBat and represents
• Control of the power-sunroof drive unit the logical 1 state
• Control of the wiper motor for the wind-
shield wiper The recessive level is characterized by a
• Sensor for rain and light detection resistance to battery voltage of 1kV in the
• Air-conditioning system (transmission master and 30kV in the slaves.
of signals from the control element, Due to circuitry variations, there may be
activation of the fresh-air blower) differences in the voltage levels. A stable
• Headlight electronics data transfer is ensured by the defining of
• Control of motors for seat adjustment tolerances for sending and receiving in the
• Theft deterrence areas of the recessive and dominant levels.
• Garage-door opener The tolerance zones on the reception side
are wider (Fig. 2) to make it possible for
valid signals to be received despite inter-
ference radiation.
The data rate of the LIN bus is limited
to a maximum of 20 kBit/s. This is the out-
2 Tolerance ranges for transmitters and receivers come of a compromise between the de-
mand for high edge steepness for easy syn-
Transmitter chronization of the slaves on the one hand,
and the demand for lower edge steepness
80% for improved EMC characteristics on the
other. The standard bit rates of 2,400 Bit/s,
Voltage level

9,600 Bit/s and 19,200 Bit/s are recom-

mended. The minimum permissible value
20% for the bit rate is 1 kBit/s to prevent time-
dominant out conflicts. The edge steepness itself
Rise time Release time is defined in the LIN specification as
Time t 1 to 3 V/µs.
The maximum number of nodes is not
specified in the LIN specification. It is
UBat theoretically limited by the number of
available identifiers. In practice, the num-
ber of subscribers is restricted to 16 due
Voltage level

60% to the maximum permissible total capacity

40% of the bus system.


Time t
86 Bus systems | LIN bus

Bus access LIN protocol

Access to the LIN bus is determined by the Frame
master-slave principle. Each message is The information transferred on the
initiated by the master. The slave has the LIN bus is embedded in a defined frame
possibility to respond. The messages are (Fig. 3). A message initiated by the master
exchanged between the master and one, always begins with a header. The response
several or all of the slaves (point-to-point, (message field) contains different informa-
multicast, broadcast). tion depending on the type of message.
The following relationships are possible If the master wishes to transmit control
in the communication between master and instructions for a slave, it will populate the
slave: response with data to be used by the slave.
• Message with slave response: the master If the master is transmitting a data request,
sends a message to one or more slaves the slave that it is addressing will populate
and asks for data (e.g. switch states or the response with the data requested by
measured values) the master.
• Message with master instruction: the
master gives a slave a control instruction Header
(e.g. switch on a servomotor) The head is composed of the following
• The master initiates communication parts:
between two slaves • The synchronization break
• The synchronization field
No arbitration or conflict management is • The identifier field
required because, with the master-slave
access control, the master alone controls Synchronization
access to the data line. To guarantee consistent data transfer
between master and slaves, a synchroniza-
tion takes place at the beginning of each
frame. First of all, the beginning of a
frame is unambiguously marked by the
synchronization break (SynchBreak).

3 LIN message format

LIN frame

Header Response
Synch Synch Ident Data Data Data Data Data Data Data Data Check-
Break Field Field field 0 field 1 field 2 field 3 field 4 field 5 field 6 field 7 sum
Bus systems | LIN bus 87

The SynchBreak comprises at least 13 con- Of the 64 possible messages, 32 may only
secutive dominant levels and one recessive contain two data bytes, 16 may only con-
level. tain four data bytes, and the remaining 16
At the end of the synchronization break, eight data bytes each.
the master sends the synchronization field The last bits in the identifier field con-
(SynchField) consisting of the bit sequence tain two checksums, which are used to
01010101. The slaves are then able to ad- check the identifier for transmission er-
just themselves to the time basis of the rors and any resulting incorrect message
master and thus synchronize. assignments.
The synchronization method described
permits a loose specification of the timing Data field
of the bus subscribers. The clocking of Once the header sent by the master node
the master should not deviate more than has been transmitted, it is time for the
±0.5 % from the nominal value. The clock- transfer of the actual data to begin. The
ing of the slaves is permitted to deviate by slaves know from the transmitted identi-
up to 15 % before synchronization as long fier whether or not they are being ad-
as the synchronization reduces the devia- dressed and, if they are, they reply with
tion to a maximum of 2 % before the end their response in the data field.
of the message. Several signals can be packed into one
In this way, the slaves can be built with- frame. In this case, each signal has pre-
out an expensive quartz oscillator, e.g. cisely one generator, i.e. it is always writ-
using a simple RC circuit. ten by the same node of the network.
During the data transfer of the bytes,
Identifier it is always the least significant bit (LSB)
The third byte in the header is used as the that is output first. Each byte (8 bits) is
LIN identifier. As with the CAN bus, a con- preceded by a start bit and followed by
tent-based addressing method is used – a stop bit, which means that each byte
the identifier therefore provides informa- involves the transmission of ten bits.
tion about the content of a message (e.g. The purpose of the start and stop bits is
engine speed). Based on this information, to resynchronize the nodes and thereby
all nodes connected to the bus decide prevent transmission errors.
whether they would like to receive and The data response of the slaves is veri-
process the message further or simply fied by means of a checksum.
ignore it. This process is known as accep-
tance filtering. LIN description file
Six of the eight bits of the identifier field The configuration of the LIN bus, in other
determine the identifier itself. Their per- words the specification of network sub-
mutations give rise to a possible 64 differ- scribers, signals and frames is managed
ent identifiers (ID). They have the follow- in the LIN description file, or ldf. For this
ing meanings: purpose, the LIN specification provides for
• ID = 0 to 59: transmission of signals. an appropriate configuration language.
ID = 60: master request for the com- From the ldf, a set of C codes or header
mands and diagnosis files is automatically generated using suit-
• ID = 61: slave response to ID 60 able tools. These are used as the basis for
• ID = 62: reserved for manufacturer- implementing the master and slave func-
specific communication tions in the electronic control units con-
• ID = 63: reserved for future extensions nected to the bus.
to the protocol
88 Bus systems | LIN bus

The ldf is therefore a means of configuring Example: air-conditioning control

the entire LIN network. It represents a The LIN bus is typically used in the control
common interface between the vehicle of the air-conditioning system. The operat-
manufacturer and the supplier of the ing and display unit functions as the bus
master or slave modules. master. This is where the software for the
open and control algorithms are stored.
Message scheduling One of its tasks is to adjust the speed of
The scheduling table in the ldf defines the the fresh-air blower. The controlled vari-
sequence and time grid in which messages able for this is the actual temperature in
are sent. Frequently needed information is the passenger cell and the desired temper-
sent more often. Once the table has been ature set by the driver. The electronic con-
worked through, the master begins with trol unit receives the interior temperature
the first message again. The sequence in from a temperature sensor located in an
which the table is worked through may appropriate place in the interior. The de-
change depending on the operating status sired temperature is set at the control unit,
(e.g. diagnostics active/inactive, ignition e.g. using a sensor ring.
on/off). From the input variables, the electronic
The transfer grid of each message is control unit calculates that the blower
therefore known. This deterministic be- speed needs to be increased to a value of
havior is guaranteed by the fact that, with 200 rpm, for example. The electronic con-
the master-slave access control principle, trol unit transmits a message containing
all messages are initiated by the master. the master instruction to the LIN bus at a
defined time interval. The identifier in this
Network management example would be “set the blower speed”.
The nodes of an LIN network can be forced In this subbus, this instruction corre-
into sleep mode to minimize the no-load sponds to identifier 25, for example.
current of the entire electronics and elec- Following the header containing this iden-
trical system in the vehicle. Sleep mode tifier, the master transmits a numerical
can be achieved in two ways: value in the data field that equates to the
• The master sends the go-to-sleep com- physical value of 200 rpm. Each slave pos-
mand with the reserved identifier 60 sesses a list of identifiers that are relevant
• The slaves enter sleep mode of their own for this node. The fresh-air blower is the
accord if no data transfer has taken one and only slave that responds to the
place on the bus for a relatively long “set blower speed” identifier and executes
period (4 seconds) the master's request.
At speeds below idle, the fresh-air
Both the master and the slaves are able to blower must be switched off. At this low
wake up the network. To do so, they must speed, the heavy load could cause the en-
send the wake-up signal. This comprises gine to stall. The LIN bus is coupled to the
data byte 128. After a break of 4 to 64 bit CAN bus via the gateway and thus receives
times (wake-up delimiter), all nodes must the current engine speed on a continuous
have been initialized and be ready to re- basis. If the speed falls below the specified
spond to the master. engine-speed threshold, the LIN master
sends the message containing the “set
blower speed” identifier where the data
field contains the value 0. The fresh-air
blower switches off in response.
Bus systems | LIN bus 89

4 LIN networking (example)

LIN slave 1 LIN slave 2 LIN slave 3 LIN slave 4

Fresh-air Windshield PTC auxiliary PTC auxiliary
blower heating heater, right heater, left

LIN master 1
ECU for air-con-
ditioning system

LIN slave 1
Sliding sunroof

LIN master 2

Roof module

Intelligent actuators can send up-to-date Summary

operating-status information to the actuat- The essential characteristics of the LIN bus
ing unit. The fresh-air blower records the are:
speed by means of a sensor and sends it • Single-master/multiple-slave concept

back on the LIN bus as a numerical value. • Master/slave access control

Due to the master-slave access method, • Independent synchronization of the

the value is only contained in a message slave possible even without quartz
with slave response initiated by the mas- • Deterministic signal transmission

ter. The feedback signal with the current • Communication in the form of very

engine speed makes it possible to control, short messages

and therefore accurately maintain, the • Character-based transmission (UART)
desired value. • Bit rate max. 20 kBit/s
• Data transfer over an unshielded single-
wire line
• The reference potentials of the data line
are battery voltage and ground
• Maximum bus length 40 m
• Maximum number of nodes 16, typically
fewer than 12
90 Bus systems | Bluetooth

Bluetooth for transmitting data for multimedia

Overview In the meantime, the Bluetooth SIG has
Mobile communication is gaining ever expanded to involve around 2,000 com-
more importance in all domains. For con- panies from the telecommunications, data
venient communication between various processing and automotive engineering
devices, a wireless link is indispensable. fields.
Infrared connections, as often used in
the past, tended to be manufacturer- Applications
specific, they required a direct line of Bluetooth is an industry standard for
sight and imposed constraints on the the networking of mobile multimedia
area of movement. Only with a standard- devices, such as a car sound system, cell
ized, wireless link could mobile devices phone, headset, PDA (Personal Digital
of different manufacturers communicate Assistant), PC and peripheral equipment
with each other without problems. (Fig. 1). Bluetooth stands for the simple
The development of the Bluetooth exchange of data between portable ter-
standard began in 1994 at the telecom- minals and the wireless transmission of
munications company, Ericsson. As part audio and video signals for entertain-
of an initial study, Ericsson investigated ment and information.
the possibilities for completely replacing The short-range wireless connectivity
the cable connections between the cell of Bluetooth eliminates the need for con-
phone and additional devices. In cooper- nection cables between the cell phone
ation with other industry partners, the and the hands-free system: no more
Bluetooth SIG (Special Interest Group) tangling of cables when making a call.
was formed in 1998 to create a uniform, It is in this area that Bluetooth has been
internationally-accepted standard. The the most well received.
SIG set itself the aim of specifying a wire- Meanwhile, a large number of con-
less technology with low manufacturing sumer electronics devices, be they cell
costs, low energy consumption and ro- phones, PDAs, notebooks or car phones,
bust resistance to interference. The now offer an address book function.
wireless interface had to be suitable

1 Bluetooth networking of mobile multimedia devices (example)

1 2 3

Fig. 1
1 Cellular phone
2 Headset 5 4
3 Personal digital
assistant (PDA) FM2 EUROPE 1

4 Laptop computer
5 Car sound system
with navigation unit
Bus systems | Bluetooth 91

With Bluetooth, it is possible to conve- the specification, great emphasis was

niently synchronize the entries stored placed on backward compatibility with
in the various address books. The car the previous versions.
driver is spared the laborious task of
typing the phone number stored in the Transmission technology
cell phone or PDA or inputting an ad- Frequency band
dress into the navigation system. Bluetooth operates in the worldwide-
Bluetooth also makes it possible to license-free 2.4 GHz ISM band (Industrial
listen to the radio through a headset in Scientific Medicine). This is available
the rear of the vehicle and to control practically worldwide without approval
the car sound system using a PDA. – apart from certain national limitations –
and is reserved for applications in indus-
Bluetooth is predominantly used for try, science and medicine. For this rea-
the transfer of data in the multimedia son, the possibility of interference from
domain. However, this technology can garage-door openers, microwave ovens
quite easily be used for diagnostic and or other appliances which operate using
service purposes. Workshops or break- the same frequency band has to be ac-
down services would be able to conve- cepted. To minimize interference, the
niently read vehicle information during best possible use should be made of the
fault diagnosis and record the condition frequency spectrum. Devices that use
of the vehicle. the ISM band therefore implement
spread-spectrum techniques.
Bluetooth versions
The first specification (Bluetooth 1.0) of Frequency-hopping method
the Bluetooth standard was adopted in Bluetooth divides the 2.4 GHz band
July 1999. In December 1999 this specifi- into 79 channels at intervals of 1 MHz
cation was replaced by Bluetooth 1.0b, (f = 2,402 + k, k = 0…78). The band is
which contained improvements and clar- terminated at both the upper and lower
ifications for interoperability. February end by two guard bands with a width of
2001 saw the release of the Bluetooth 1.1 3.5 MHz (upper guard band) and 2 MHz
specification, which fixed fundamental (lower guard band). The transmission
interoperability problems between the
first devices and chips and is used in
most Bluetooth devices today. Three 2 Frequency-hopping method

years later, in November 2003, version

1.2 of the specification was published.
This used the IEEE (Institute of Electrical
and Electronics Engineers) language def-
inition for technical documentation and
introduced adaptive frequency hopping
Frequency f

(AFH) as a technical advancement that

improves resistance to interference in
the frequency band. The most significant
modification was introduced with ver-
sion 2.0 of the specification in November
2004. With the enhanced data rate (EDR), Time slots Time t

it was possible to triple the bit rate from

1 MBit/s to 3 MBit/s and reduce power
consumption. With each new version of
92 Bus systems | Bluetooth

uses a combined frequency-hopping/ 8 differential phase-shift keying (8DPSK),

time division duplex (FH/TDD) method: which modulate the phase of the carrier.
the frequency-hopping spread spectrum p/4DQPSK represents one of four com-
(FHSS, Fig. 2). The channels are switched binations of a bit pair (00, 01, 10 and 11)
1,600 times per second (i.e. in 625 µs and so transfers a maximum of two bits;
slots). The slots are assigned in accor- 8DPSK can represent any combination
dance with the TDD method, i.e. the of three bits. This is used to increase the
transmitter and receiver are authorized bit rate from 1 MBit/s (net 727 kBit/s) to
to send alternately. The frequency-hop- 2 MBit/s (net 1,446 kBit/s) with GFSK,
ping method guarantees optimal and or 3 MBit/s (net 2,169 kBit/s) with EDR.
uniform use of the ISM band and makes
Bluetooth resistant to interference from Power classes
other transmitters in the same frequency Bluetooth devices are categorized into
band (e.g. WLAN). In addition, it also different power classes. Manufacturers
offers some security because a hacker are free to decide which power class they
would not know the destination of a fre- implement. The present specifications
quency switch or indeed the time frame define three power classes:
for which a particular frequency would • Class 1 with 100 mW (20 dBm) trans-

be valid. This makes interception of a mission output for a range of 100 to

connection much more difficult. 150 m
The introduction of adaptive fre- • Class 2 with 2.5 mW (4 dBm) transmis-

quency hopping (AFH) in the Bluetooth sion output for a range of 10 to 25 m

1.2 specification further increased • Class 3 with 1 mW (0 dBm) transmis-

Bluetooth's resistance to interference sion output for a range of 10 m

and reduced the effect of Bluetooth itself
as a source of interference on other de- Topology
vices in the 2.4 GHz ISM band. The num- Bluetooth supports the automatic config-
ber of effective channels is dynamically uration of ad-hoc networks. This means
adaptable to the conditions in the fre- that two or more devices are able to
quency band. Frequencies used by other spontaneously form a network with no
transmitters are then removed from the prior knowledge of each other. Bluetooth
list of 79 possible channels. is based on the master-slave principle.
In principle, any Bluetooth device may
Modulation method be a master or a slave. The master is a
Bluetooth uses the Gaussian frequency- special device that coordinates commu-
shift keying (GFSK) method of modula- nication between the devices.
tion. This is a special variant of fre-
quency-shift keying (FSK) supplemented
with a Gaussian low-pass filter. GFSK
varies the frequency and delivers 1 bit
per keying interval. Bluetooth has a sym-
bol rate of 1 MBit/s. With the enhanced
data rate (EDR) introduced with the
Bluetooth 2.0 specification, it has been
possible to increase this data rate to
3 MBit/s. This is made possible by the
use of two variants of phase-shift keying
(PSK): the p/4 differential quaternary
phase-shift keying (p/4DQPSK) and the
Bus systems | Bluetooth 93

Piconet The piconet is defined by the frequency-

A network of Bluetooth devices is known hopping sequence, i.e. the sequence of
as a piconet (Fig. 3). For the devices switching between the 79 possible chan-
within a piconet, there are two roles: nels. The hopping sequence is specified
master or slave. Which device assumes by the master and is calculated from the
which role is not decided until communi- master's device ID (a worldwide unique
cation is established. A device sets up a 48-bit identifier) and its internal clock.
piconet and transmits its device ID and The slaves synchronize with the master's
the value of its internal clock. This de- clock and its hopping sequence.
vice becomes the master. A master is able to manage up to seven
Any device that is not participating in active slaves. Each active slave receives
a piconet but is not switched off remains a 3-bit active member address (AMA).
in standby, which is marked by a small Active devices can send data or even
current draw. simply remain connected. By deregister-
The first stage in the creation of a ing, a device can switch back to standby.
piconet involves the potential master A Bluetooth device may also switch to
searching for other devices in its range any of three energy saving modes: hold
(inquiry scan). Special frequency-hop- mode, sniff mode and park mode. The
ping sequences (inquiry sequences) are devices continue to be synchronized with
available for this purpose. Devices that the hopping sequence of the piconet. The
wish to enter into communication with lowest energy saving mode is the park
other devices respond with an inquiry mode, where the device relinquishes its
response message. By the end of the in- AMA and receives in its place an 8-bit
quiry sequence, the information about all parked member address (PMA). With the
devices willing to communicate is avail- PMA, the device is still a subscriber in
able. The potential master is now able to the piconet but makes way for another,
address a specific device by means of a active device. Parked devices remain
special paging sequence. Paging is a pro- synchronized with the hopping sequence
cess whereby a fixed connection is es- of the piconet and can be addressed and
tablished between a master and a slave. reactivated by the master using the PMA.
The slave contains all the information it
needs to synchronize with the master. Scatternet
In a piconet, only one slave is ever able
3 Piconet to communicate with the master simulta-
neously. When more slaves enter the net-
work, the data rate per device falls very
quickly. This is where scatternets pro-
vide a solution. A scatternet consists of
S up to ten piconets in which the individ-
P P ual piconets partly overlap. In this way, it
is possible to set up even larger networks
in which all network subscribers may
S S connect to one another. The scatternet
in Figure 4 comprises two piconets. One
of the Bluetooth devices is a member of
Fig. 3
both piconets. Through this device, the M Master

two networks are able to exchange data. S Slave

However, this device, which is assigned P Park mode
SB Standby
94 Bus systems | Bluetooth

to more than one piconet, can only func- It is conventional for the transmissions
tion as master in one of these networks. of the master to begin in the even time
A device that is assigned to two picon- slots and those of the slaves to begin
ets in a scatternet must synchronize with in the odd time slots. This characteris-
the network with which it wishes to com- tic conforms to the regular time-divi-
municate. It cannot remain connected to sion multiplexing method (TDD here)
both networks at the same time. Before whereby the master uses one half of the
the device leaves the old piconet, it must time slots and the slaves the other half.
inform its master that it will be unavail- In addition to packets that may occupy
able for a certain period of time. The one time slot, Bluetooth provides for
remaining network subscribers can packets that may occupy three or five
continue to communicate. time slots (multi-slot packets). As soon
A master, too, may leave its own as a master or slave sends a packet three
piconet and become a slave in another. or five time slots long, this transmitter
All communication in its old network, will remain on the same frequency.
however, is broken until it returns and There is no frequency change within
assumes its master function again. a packet (Fig. 6).
A master cannot become master of a Once the packet has been transferred,
second piconet. If this were to happen, the frequency hops as determined by the
both piconets would behave in an identi- hopping sequence (independently of the
cal manner and form a single network. transmission process). In the illustration,
for example, the frequency hops from
Physical data channel fk to fk+3 after the transmission of a 3-slot
A data channel is represented by a hop- packet). The reason for this behavior is
ping sequence between the 79 possible that not every station may have received
frequencies in the ISM band. Any device the transmission and cannot therefore
that is actively participating in a piconet respond specifically to the transfer of
must also hop at the same carrier fre- data in several time slots. For this rea-
quency (frequency fi). As soon as a mas- son, all stations that are not involved in
ter has sent data at frequency fk, a slave the transmission always continue to hop
can respond at frequency fk+1. This pro- in the sequence specified by the master.
cedure is illustrated in Figure 5.

4 Scatternet 5 Timing of data transfer

Piconet 1 Piconet 2

S SB f(k) f(k+1) f(k+2)



S S P Slave

P S 625 ms
Fig. 4
M Master


S Slave
P Park mode
SB Standby
Bus systems | Bluetooth 95

Physical connections which transfer data from the user level –

Bluetooth supports both circuit-switched the applications.
and packet-switched data channels. For
circuit-switched, synchronous communi- Data packets
cation, Bluetooth offers a synchronous A Bluetooth data packet generally
connection-oriented link (SCO), and an comprises three fields:
ACL (Asynchronous Connectionless Link) • Access code

for packet-switched, asynchronous • Header and

communication. Between two Bluetooth • Payload
devices, there may only ever be one
ACL link but there can be up to three A packet may simply comprise the access
SCO links. The ACL links are the founda- code, or the access code and a header,
tion for the data transfer. The SCO links or the access code, header and payload
are purely voice channels with a data (Fig. 7).
rate of 64 kBit/s. The voice is transmitted A packet begins with an access code.
by continuously variable slope delta If a header follows, the access code has
modulation (CVSD) or pulse-code modu- a length of 72 bits, otherwise it is 68 bits
lation (PCM). A Bluetooth master can long. The access code is used for the
maintain several different data channels synchronization and identification of a
between its slaves even at the same time. piconet, for device polling and for device
calls. There are three different types of
Logical data channels access code:
Logical channels refer to different types • Channel access code (CAC), the purpose

of channels streamed over a physical of which is to synchronize and identify

connection. The data transmitted over a piconet. The CAC is sent with each
a physical channel has different logical packet transmitted over the piconet.
meanings. Bluetooth distinguishes be- • Device access code (DAC), which is used
tween two categories of channel: link to transfer specific identifiers during
channels, which are used for the ex- the device call (e.g. when paging).
change of control information between • Inquiry access code (IAC), which is

master and slave; and user channels, used for device polling.

6 Frequency hopping with multi-slot packets 7 Structure of a packet

Access code Header Payload

f(k) f(k+1) f(k+2) f(k+3) f(k+4) f(k+5) f(k+6)
72 (68) bits 54 bits 0 to 2,745 bits

625 ms 4 preamble 3 AMA

f(k) f(k+3) f(k+4) f(k+5) f(k+6) 64 synchron- 4 type

1 flow
(4) appendix
f(k) f(k+5) f(k+6) 1 ARQN


8 checksum
96 Bus systems | Bluetooth

The header of a Bluetooth packet is Bluetooth architecture

54 bits long and contains information The Bluetooth architecture is complex.
about the link (link control information). This is inherently associated with the
It is made up of the address, packet type, idea behind general cable emulation.
flow control, error monitoring and The architecture is designed to enable
checksum fields. The actual length of new protocols and applications to be
the header (i.e. the sum of the bits in adapted for the use of Bluetooth. Accord-
the header) is 18 bits. The header is pro- ingly, the Bluetooth protocol stack is
tected by forward error correction (FEC). very extensive (Fig. 8 shows a simplified
The bits of the header are sent three version of the protocol stack).
times (1/3 FEC), giving a header length of The Bluetooth technology can be sub-
54 bits. A receiver can then simply take divided into three logical sections:
the majority decision; each bit triplet is • Hardware

resolved to the value with majority in • Host stack

the triplet. • Applications

The payload of a Bluetooth packet can

be up to 341 bytes in size (1,023 bytes The hardware layer and the host stack
with EDR). The data field of the payload are defined in the Bluetooth Core Specifi-
consists of up to three segments: the cation; the applications are part of the
payload header, the payload itself and, Bluetooth Profile Specification.
in certain circumstances, a checksum.
Bluetooth uses 16 different packet Hardware-side protocols
types, the common feature of all being The hardware is represented by the
the access code and the header. The Bluetooth chip. This is where the radio
structure of the payload depends on the layer, baseband and link management
packet type concerned. Along with pack- are contained. The radio layer deals with
ets for ACL and SCO links, there are also the radio transmission, used frequencies,
packet types for polling the slaves, syn- modulations and transmission output.
chronizing the hopping sequence and The baseband represents the link-estab-
acknowledging data transfers. lishment mechanisms, the packet struc-
ture and the timing. The link management
Bluetooth device addresses establishes and manages the link between
A Bluetooth device is assigned a world-
wide unique Bluetooth device address. 8 Protocol stack

The address is derived from the IEEE-

802 standard and implemented in the de- Applications
vice by the manufacturer. The 48-bit ad-
dresses are subdivided into three parts
• LAP (Lower Address Part, containing SDP TCS BIN RFCOMM
24 bits)
Audio L2CAP
• UAP (Upper Address Part, containing
8 bits) HCI (Host Controller Interface)
• NAP (Non-significant Address Part,
Link Manager Protocol LMP
containing 16 bits) (link management)

The LAP and UAP fields form the signifi-

cant part of a Bluetooth device address. Radio (radio layer)

Bus systems | Bluetooth 97

two devices and also implements security In addition to L2CAP, there are also au-
and authentication functions. dio and control protocols that govern the
The host controller interface (HCI) is handling of audio data and control data.
the interface between the hardware of a Audio applications, for example, can
Bluetooth module and the host-side pro- make direct use of the baseband layer
tocols. The HCI is a hardware abstraction once the audio signals have been coded
and realizes various interfaces for con- accordingly.
trolling the Bluetooth hardware and
transferring data (e.g. via USB or UART). The service discovery protocol (SDP) is
another important host-side protocol.
Host stack It serves to identify and scan for services
The host stack uses the HCI to control with specific properties and to describe
the Bluetooth hardware and transfer services within range of a Bluetooth de-
data. The format of this data is deter- vice. Bluetooth devices are meant to be
mined by the upper protocol layers. The able to interact on an ad hoc basis with
logical link control and adaptation proto- other devices in different environments.
col (L2CAP) implements the abstraction It is therefore necessary to know which
of the hardware's properties, adapts the services are made available by which de-
upper layers of the protocol stack to the vices within range. All devices wishing to
capabilities of the baseband and hides provide services must use an SDP server;
transmission details such as the connec- for all other devices, an SDP client is suf-
tionless or connection-oriented trans- ficient. The SDP prepares the services
mission type. The L2CAP layer essen- available on a device in a service data-
tially has three main functions: base. The service information in the pos-
• It can receive packets with a length of session of the SDP server is stored in a
up to 64 kB from the upper layers and service record. The service record con-
decomposes them into smaller data sists of a list with service attributes that
packets (segments) for processing in describe the properties of the service
the lower layers if necessary. At the more precisely and is identified by a
other end, the segments are recom- 32-bit service record handle.
posed back into packets.
• It manages the multiplexing and de- Two further protocols in the protocol
multiplexing of several packet sources. stack form the fundamental basis for the
If a packet is being recomposed, the interoperability of Bluetooth devices.
L2CAP layer determines to which pro- The RFCOMM (Radio Frequency Com-
tocol of the upper layers the packet is munication) cable emulation protocol
forwarded. above the L2CAP layer simulates up to
• L2CAP offers functions for negotiating 60 virtual serial interfaces derived in ac-
quality of service and configuration pa- cordance with the ETSI 07.10 standard.
rameters. This means, for example, that As a result, almost any software that pre-
the maximum size of the payload can be viously expected a serial interface can
negotiated so that a device with limited work with Bluetooth.
resources would not be overwhelmed TCS BIN (Telephony Control Protocol
by overly large packets. Using the con- Specification – Binary) is employed as
figuration parameter for the quality of the means of controlling telephone and
service, it is possible to define the telephony functions. This is a bit-ori-
properties of the data transfer: best ef- ented protocol for establishing voice
fort (best attempt but with no guarantee and data connections between Bluetooth
of data transfer) or guaranteed. devices. The use of appropriate, indus-
98 Bus systems | Bluetooth

try-wide standards has made it possible, The Bluetooth profiles form the hierar-
here too, to ensure wide-ranging com- chy (Fig. 10). The GAP forms the basis
patibility with legacy applications. and describes all the essential functions
that a Bluetooth device must fulfill at the
Applications lowermost level. These include, for ex-
Many more protocols have been adapted ample, the functions for establishing and
for the Bluetooth standard (adapted pro- managing the link, the supported operat-
tocols) and can be found in the protocol ing modes and the security of a link.
stack. Internet applications can, for ex- The SDAP defines the access interface
ample, continue to use TCP/IP through for the service discovery protocol (SDP)
the point-to-point protocol (PPP) or the with which devices can discover or poll
Bluetooth network encapsulation proto- the services offered by other devices.
col (BNEP). For the exchange of vCalen- SDP builds on the GAP. All Bluetooth de-
dars and vCards, it is possible to use the vices must implement these two profiles.
object exchange protocol (OBEX) cov- The SPP is used by most other
ered by the IrDA standard. Bluetooth profiles. One exception to
this is the telephony control profile.
Profiles This is always used when Bluetooth is
The uppermost layer contains the appli- being used for cable emulation or if a
cations and profiles. Profiles represent serial data connection is to be used.
“standard solutions” for a particular us- The SPP builds on the GAP and uses
age scenario. The Bluetooth specification the RFCOMM protocol.
currently brings together 13 different The GOEP defines the fundamental
applications described as profiles. Each functions necessary for the exchange
profile uses a certain choice of protocol; of complex objects. It defines a client-
in principle, application profiles make server relationship for the exchange of
a different protocol stack available for data. Like the SPP, the GOEP provides
each application. Profiles describe the the foundation for further profiles.
vertical slice through the Bluetooth pro-
tocol stack; the protocols represent the
horizontal layers (Fig. 9). Within the pro-
files, the required and optional functions
of the layers are defined. These stan-
dardized profiles make it possible to en- 9 Profiles and protocols

sure interoperability between different

The four fundamental profiles of the
Bluetooth specification are:
• Generic access profile (GAP)
• Service discovery application profile

• Serial port profile (SPP) and

• Generic object exchange profile

Bus systems | Bluetooth 99

10 Profile hierarchy

Generic access profile (GAP)

Service discovery Telephony control profile

application (SDAP) Intercom (IntP) Cordless telephone (CTP)

Serial port profile (SPP)

Dial up network (DUNP) Generic object exchange profile (GOEP)

Fax (FaxP) File transfer (FP)

Headset (HSP) Object push (OPP)

LAN access (LAP) Synchronisation (SP)

▶ Origin of the name, Bluetooth

The name, Bluetooth, originates from

King Harald Gormsen, a Danish Viking.
Harald had the nickname, Blatand.
In the Middle Ages, Bla actually meant
dark, while Blatand made reference
to the dark figure of Harald Gormsen.
Bluetooth, therefore, is not a direct
translation of Blatand.

In the 10th Century, Harald Gorm-

sen united large parts of Denmark
and Norway. The Bluetooth wireless
system developed 1,000 years later
unified a wide range of different infor-
mation, data processing and wireless
mobile devices. This is why it was
named after King Harald.
100 Bus systems | MOST bus

MOST bus The MOST Cooperation creates and ad-

ministers the specifications on which the
Introduction MOST bus is based. Furthermore, it de-
The MOST bus (Media Oriented Systems fines the requirements for implementation
Transport) was specifically developed for of MOST devices and provides appropriate
the networking of infotainment systems in compliance tests through accredited test
motor vehicles. In addition to traditional houses.
entertainment functions, such as radio
tuners or CD players, these infotainment Type of use
systems (which are becoming an increas- At present, the MOST bus is almost exclu-
ingly common feature of modern vehicles, sively used for the networking of infotain-
and premium class vehicles in particular) ment systems in motor vehicles. It can cur-
also offer video functions (DVD and TV), rently be found in over 35 vehicle models
route guidance capabilities and access to of various manufacturers, especially in
mobile communications and information. the premium and mid-range classes (as at
The functions of an infotainment system Features of the MOST bus
place high demands on a bus system. The MOST bus supports the logical net-
The transmission of multimedia data, working of up to 64 devices, at which point
both audio and video, requires a high constraints associated with the chipsets
data rate and a synchronization of the may be encountered.
data transfer between source and sink In its current version, the MOST bus
as well as between sinks themselves. offers a data rate of 24.8 MBit/s (MOST 25).
For the transmission of informational Versions with higher data rates of 50 MBit/s
data, e.g. detailed information about the (MOST 50) and 150 MBit/s (MOST 150) are
music tracks on an mp3 player, and the already available as a development model.
sending of software updates, it is neces-
sary for the data transfer to be flexible Data transmission channels
with varying and, at certain times, For the transmission of data, the MOST bus
equally very fast data rates. supports the following channels and their
Likewise, the requirements for use in a different attributes:
motor vehicle, and the associated require- • The control channel is used for the sim-

ments of electromagnetic compatibility ple transmission of control commands,

(EMC), must also be fulfilled. for the signaling of device statuses and
for the sending of messages necessary
MOST Cooperation to system management. With MOST 25,
It was with these requirements in mind the control channel has a gross band-
that the MOST bus was developed by the width of 705.6 kBit/s.
MOST Cooperation. This was founded in • For the transmission of multimedia data,

1998 by BMW, DaimlerChrysler, Harman/ the MOST bus has a flexible number
Becker and Oasis SiliconSystems (now part of synchronous channels that are able
of SMSC). By 2006, the MOST Cooperation to carry both audio and video data.
had 16 automotive manufacturers and In MOST 25, there are a maximum of
67 suppliers and tool makers among its 15 audio channels available in stereo
membership. quality.
Bus systems | MOST bus 101

• On the asynchronous channel, data is Each device has a bypass. When this is
sent in packets. It is therefore suitable closed, the device forwards the signal
for transferring information that does directly and is therefore invisible to the
not require a fixed data rate but does MOST system. This service is useful,
need a high data rate at certain times. for example, when the system is starting
Examples could be the transfer of track and a device requires more time for initial-
information of an mp3 player or a soft- ization, or in the event of a temperature-
ware update. With MOST 25, the asyn- dependent shutoff.
chronous channel has a gross bandwidth There are also other possible structures
of up to 12.7 MBit/s. for a MOST system, e.g. a star topology.
Internally, however, these must always
The available bandwidth can be flexibly be arranged to form a (logical) ring.
distributed between the asynchronous With a star structure, this can be achieved
channel and the synchronous channels through use of a central hub, for example.
by means of the “boundary descriptor”,
which can also be shifted to the transit Device model
period if relevant preconditions are met. The MOST standard defines the model
shown in Figure 2 for a MOST device for
Topology which the following elements are required:
A MOST system is arranged in a ring struc- At the lowermost level, the physical
ture whereby a device is connected to its layer provides access to the transmission
predecessor or successor in the ring by agent. Departing from the original defini-
an input or output respectively (Fig. 1). tion that provided for optical transmission
One of the devices acts as the “timing by POF cable (Plastic Optical Fiber), there
master” and generates the data frames are today various physical layers available
for data transfer with which the other with optical and electrical transmission as
devices synchronize. well as various speeds of transfer.

1 Ring structure of the MOST system 2 Model for a MOST device

Timing master MOST device

position 0
NetBlock Appl.
Rx Tx

Network service

MOST network

Rx tion 1 ve
Tim sition 2

po ng s

interface controller


MOST physical



102 Bus systems | MOST bus

The MOST network interface controller Transmission agent

(NIC) is a hardware controller that is re- The MOST bus was originally defined as
sponsible for controlling the physical layer an optical transmission agent that used a
and implements the basic transfer ser- plastic optical fiber (POF). The optical sig-
vices. While in the original version of the nal used here with a wavelength of 650 nm
NIC many control tasks had to be carried (in the red wavelength range) is generated
out by the main processor of the device, by an LED on the transmit side, which is
the current version (Intelligent Network designated Tx-FOT (FOT = Fiber Optic
Interface Controller, INIC) is, for the most Transceiver). On the receive side, Rx-FOT,
part, able to handle tasks autonomously the optical signal is converted back into an
and already implements substantial parts electrical signal by a PIN photodiode.
of the network service. Typical controllers The POF cables used for automotive
for the MOST are the OS8104 (NIC) and the applications (Fig. 3) consist of a 980 μm
OS81050 (INIC), manufactured by SMSC. thick optical core insulated by 20 μm thick
The MOST network service is the driver optical cladding with a low refractive in-
layer through which applications and dex. In total, the optical conductor there-
system services have access to the NIC. fore has a diameter of 1 mm. The optical
The lower layer of the network service fiber is insulated with a black buffer, which
(layer 1) offers basic communication and is in turn surrounded by a protective cable
management functions, while the layer sheathing. This gives the cable a total dia-
above (layer 2) offers functions for sup- meter of 2.3 mm.
porting the development of applications, POF lines offer the following benefits
such as a translator for the application when used in the motor vehicle:
protocol or support for the notification • No electromagnetic interference radia-

service. tion
The applications of the device are imple- • Insensitivity to interference irradiation

mented on top of the network service. • Lower weight than equivalent shielded

In a MOST system, the interface of an ap- electrical lines

plication is realized as a function block • More flexible routing in comparison to

(FBlock). Each device must at least imple- equivalent shielded electrical lines
ment a special FBlock, the NetBlock,
which is required for management func-
tions within the MOST system. In addition,
a device will usually implement one or 3 POF cable

more applications that can be used as

FBlocks of applications on other devices.
A device of a vehicle infotainment system
is often integrated with several functions,
Cable sheathing
e.g. those of a radio tuner and those of an
amplifier, which can be represented in the Buffer
MOST system as independent FBlocks.

2.3 mm

1 mm 1.5 mm Optical fiber

Bus systems | MOST bus 103

To date, modern vehicles are equipped • An area for the control channel (with
exclusively with MOST 25, with POF data MOST 25, two bytes per data frame)
transmission via POF connections. In addi- • An area for the synchronous channels,
tion to this, however, further transmission whereby one or more bytes are always
techniques have been defined for the assigned to a specific synchronous
MOST: channel
• The optical transmission by glass fiber • An area for the asynchronous channel
cable (PCS) with laser diodes (VCSEL),
which offers a greater damping reserve, In the case of MOST 25, the synchronous
supports higher speeds and is less sensi- and asynchronous area demand a 60-byte
tive to high temperatures. share of the data frame. The distribution
• An electrical transmission by copper between the synchronous channels and
cable, which is also less sensitive to the asynchronous channel is determined
temperature and is more economical by by the value of the boundary descriptor
comparison, requires additional shield- with a resolution of 4 bytes. The synchro-
ing measures at higher bandwidths. nous area must have at least 24 bytes
This has consequences for costs and (6 stereo channels). This means that
cabling. 24 to 60 bytes are permitted for the
synchronous area and 0 to 36 bytes are
Data transfer permitted for the asynchronous area.
Data transfer on the MOST bus is orga-
nized into data frames, which are gener-
ated by the timing master with a fixed data
rate and passed on by subsequent devices
in the ring.

Data frames
The timing master normally generates
data frames with a cycle of 44.1 kHz,
and in rarer cases 48 kHz. The cycle is
defined by the system manufacturer
(i.e. the vehicle manufacturer in conjunc-
tion with its device suppliers) to suit the
predominant media formats in the system. 4 Data frame

The size of the data frames, by extension,

determines the bus speed of a MOST sys- Administrative area: 1 byte
(preambel, boundary descriptor)
tem. With MOST 25, the size of a data
frame is 512 bits.
Administrative area: 1 byte
A data frame is made up of the following (frame control, status and parity)
ranges (Fig. 4):
• Areas for administrative information
at the start and end of the data frame,
the tasks of which include the synchro- Syn- Asyn- Control
chronous chronous channel:
nization of the data transfer, determina- channels channel 2 bytes
tion of the devices' ring positions and
Total of 60 bytes: distribution
the transmission of the boundary by boundary descriptor

1 data frame for MOST 25: 64 bytes
104 Bus systems | MOST bus

System events • Group addresses: each device may also

The fundamental system events described be assigned a group address in the
below indicate the current status of the 0x300 to 0x3FF range, making it possi-
data transfer in the MOST system. These ble to address all devices in a given
are monitored by the NIC and forwarded group at the same time.
by the network service to the applications, • Broadcast address: the MOST addition-
which must then respond accordingly. ally supports a special broadcast ad-
• Lock, stable lock: a lock occurs as soon dress, hexadecimal 0x3C8, which makes
as communication is established in the it possible to address all devices in the
ring, i.e. the correct device receives the MOST system.
correct data frames. A stable lock is set
when the system has been in lock status Transmission of control commands
for a specific period. The control channel permits the transmis-
• Unlock, critical unlock: an unlock de- sion of control messages with a maximum
notes the loss of the lock, i.e. that no length of 32 bytes. These have to be shared
correct data frames are being received. among consecutive data frames. With
If the system is in the unlock state for MOST 25, the 16 data frames required
a certain period, the status switches to here are also known as a block. The format
critical unlock. of the control messages is defined by the
• Network change event: if the number of protocol described in the next section.
devices in a MOST system changes be- The sending of messages on the control
cause one device has opened or closed channel is supported by the NIC. It pro-
its bypass, this is signaled by a network vides the necessary services for data pro-
change event. tection and channel arbitration, i.e. for de-
termining the next opportunity for a mes-
Addressing sage to be sent. The reliable transmission
Devices on the MOST bus are addressed by of control messages is safeguarded by the
means of a 16-bit address. The following use of a cyclic redundancy code (CRC) and
different addressing types are supported: an automatic repeat request.
• Logical addresses: each device has a log- The control channel's gross bandwidth
ical address that is unique in the system of 705.6 kBit/s is reduced to a net data rate
and is set and distributed when the sys- of 406 kBit/s following the deduction of ap-
tem starts. The logical addresses can proximately 3 % for system administration
be set dynamically based on the ring overheads (two of 64 control messages)
position, for which, beginning with the and less the extra overhead to account
timing master, the hexadecimal range for arbitration and data protection. Since
0x100 to 0x13F is reserved. Or they arbitration only allows a particular device
can be defined statically by the system to populate every third available control
manufacturer in a range reserved spe- message, an individual device has a maxi-
cifically for this purpose. mum data rate of 135 kBit/s at its disposal
• Physical addresses: for system adminis- for sending.
tration services, e.g. determination of To enable the sending of messages of
logical addresses, each device also has a more than 32 bytes, the network service
physical address that is derived from its supports the segmentation and deseg-
current ring position. The physical ad- mentation of application messages up to
dresses are assigned the range 0x400 to 65,556 bytes in size on the control chan-
0x43F. nel. This service, also known as the appli-
Bus systems | MOST bus 105

cation message service (AMS), differs from to occupy a synchronous channel

the control message service (CMS), which sends an appropriate system message
only allows 32-byte control messages to request (ResourceAllocate) to the tim-
be sent. ing master. A corresponding message
(ResourceDeallocate) is used to free the
Transmission of multimedia data channel.
In the MOST system, the transmission of
multimedia data takes place on the syn- Transmission of packet data
chronous channels. The data transfer is On the asynchronous channel, data is sent
controlled by relevant control commands in the form of packets. The asynchronous
on the control channel. A synchronous channel currently supports two modes:
channel can be assigned a specific band- a slower 48-byte mode, where 48 bytes
width, which is achieved with a resolution are available in each packet for the net
of one byte of a data frame. A stereo audio data transfer but which places less heavy
channel with a resolution of 16 bits re- demands on device implementation, and
quires four bytes, for example. In the a 1,014-byte mode that is more complex
case of MOST 25, a maximum of 60 bytes to implement.
(depending on the value of the boundary With MOST 25, between 0 and a maxi-
descriptor) are available for the synchro- mum of 36 bytes of a data frame can be
nous channels, which equates to 15 stereo assigned to the asynchronous channel,
audio channels. which corresponds to a maximum gross
The sinks and sources for multimedia bandwidth of 12.7 MBit/s. For the 1,014-
data are each assigned to an FBlock, which byte mode, this produces a maximum net
provides the necessary functions for their bandwidth of almost 11 MBit/s, and a data
management. An FBlock may therefore rate of almost 3 MBit/s for the 48-byte
contain several sources and sinks, which mode. In practice, however, substantially
are numbered consecutively with a source lower data rates are achieved mainly due
and sink number respectively. to restrictions in device implementations.
An FBlock has functions that supply in- The transmission of data packets on the
formation about the quantity and type of asynchronous channel is also directly sup-
sources and sinks that it makes available ported by the NIC. Likewise, a packet on
(SyncDataInfo, SourceInfo and SinkInfo). the asynchronous channel contains a CRC
Furthermore, each FBlock with a source to permit detection of simple transmission
has an “Allocate” function with which it errors. Unlike the control channel, how-
requests a synchronous channel and asso- ever, there is no issuing of automatic re-
ciates its source to it. Its “Deallocate” peat requests on the data link layer.
function, on the other hand, is used to To ensure reliable transfer and flow con-
free up the channel again. trol in the large data volumes typical of the
Similarly, an FBlock with a sink has a asynchronous channel, it is conventional
“Connect” function for connecting the sink to use yet another transport protocol that
to a particular synchronous channel, and is implemented in a higher driver layer.
a “DisConnect” function for its disconnec- This may either be the MOST high protocol
tion. Several sinks may also be connected (MHP), developed specifically for the
to a particular channel at the same time. MOST, or the popular TCP/IP protocol,
With MOST 25, the channel assignment which is set up on an appropriate adapta-
for synchronous channels is managed tion layer known as the MOST asynchro-
by the timing master. A source wishing nous medium access control (MAMAC).
106 Bus systems | MOST bus

Administrative functions not necessarily have to be the case. The

The MOST standard defines the following other devices in the MOST system in this
administrative services that are required relationship are known as network slaves.
for the operation of a MOST system. The network master manages an image
of the current configuration of a MOST
Configuration status system in the central registry. In addition
A valid configuration status in the MOST to the addresses of all devices, this also
system is the prerequisite for communica- contains all FBlocks implemented by them
tion in the application layer. The configu- (down to the system FBlocks such as the
ration of the MOST system is managed by NetBlock). A network slave searching for
a single device, the network master. An ap- the device address for a particular FBlock
plicative communication is only permitted may request it from the network master.
once this device has signaled a conflict- If a network slave frequently accesses
free configuration status. other devices, it may, for faster access, also
• OK configuration status: the network store a local copy of the central registry –
master transmits the OK configuration a decentral registry – which it must update
status as soon as a conflict-free configu- accordingly in the event of any change in
ration is achieved. the configuration.
• NotOK configuration status: if a conflict To build the central registry, the net-
arises, e.g. because two devices have work master carries out a network scan
the same address, the network master when the system starts and as soon as
transmits its NotOK configuration status. a network change event has occurred.
In response to this, all devices initialize During this scan, it queries the NetBlocks
their addresses and communication set- of all devices for the FBlocks implemented
tings. by this device. If it detects a conflict while
• New and Invalid configuration statuses: doing so, e.g. duplicate device address or
these configuration statuses indicate duplicate instance of an FBlock, it takes
that a new application (FBlock) has measures to resolve the conflict. To do
been registered in the system or that a this, the network master may initiate a
previously registered application has recalculation of addresses, convert the
dropped out. InstId of an FBlock or, if necessary,
ignore a device.
It is mandatory for each device to imple- Connection master
ment an FBlock called the NetBlock, which The connection master manages the syn-
covers a range of different administrative chronous connections existent in the MOST
functions (e.g. for address initialization) system at a given time. It makes available
and supplies information about the device a table containing information about all
and the FBlocks that it implements. the connections currently present in the
MOST system (SyncConnectionTable).
Network master It also makes it possible to query the band-
The network master is implemented by width remaining for further synchronous
a special device in a MOST system and connections (AvailableChannels).
is responsible for system configuration. On request, the connection master
In present-day systems, the network mas- builds connections between a specific
ter is usually realized by the headunit source and sink (BuildSyncConnection) or
(i.e. the control element) of the infotain- removes them (RemoveSyncConnection)
ment system. This device often assumes by sending the necessary commands to
the role of timing master too, but this does the associated FBlocks.
Bus systems | MOST bus 107

It is even possible to dispense with the

1 Elements of a control message
connection master completely if a central
Field Size Description
entity in a MOST system, e.g. the headunit,
DeviceID 16 bits Device address
is responsible for the administration of the
FBlockID 8 bits FBlock identifier
synchronous channels.
InstID 8 bits Instance of the FBlock
FktID 12 bits Function identifier
MOST application layer OPType 4 bits Type of operation
The MOST standard defines a suitable pro- Length 16 bits Length of the data
tocol in the application layer for the trans- field
mission of control commands, status infor- Data 0 to Data field
mation and events. This protocol makes it 65,535 bytes Table 1
possible to address a specific function of
an application interface (i.e. of an FBlock) Function block (FBlock)
that is provided by any device within the A function block (FBlock) defines the in-
MOST system. For example, it is possible terface of a specific application or system
to start playback of a CD on a separate service. An FBlock is assigned an address
CD player or to query the number of the comprising an 8-bit FBlockID, which spec-
track currently playing in it. While the ifies the type of FBlock, and an additional
MOST protocol is mainly applied to the 8-bit InstID. The latter is a means of distin-
control channel, it can be transferred guishing between several instances of
across the asynchronous channel if FBlocks of the same type in a MOST sys-
necessary. tem.
An FBlock that is controlled by an asso-
Overview ciated application, and the device that im-
The protocol for MOST control messages plements the FBlock, are known as the
provides for the following elements of a slave in this relationship (e.g. an applica-
control message (Table 1): tion that addresses an external CD player
• The address of a device in the MOST sys- through the associated FBlock). The appli-
tem (DeviceID) cation that controls the FBlock is called
• An identifier for an FBlock implemented the controller.
by this device (FBlockID) and its in- The functions that an FBlock possesses
stance in the MOST system (InstID) are defined by the function catalog. In the
• The identifier for the requested function case of system services, the function cata-
within the FBlock (FktID) logs are defined by the MOST standard.
• The type of operation (OpType) that The MOST standard even defines the inter-
should be applied to this function, e.g. faces for common applications from the
the setting or querying of a property of area of vehicle infotainment systems (e.g.
the FBlock for an amplifier or a CD player/changer).
• A data range containing the parameters However, these interfaces tend to be ex-
of the function call (Data) and a corre- tended by the system manufacturer. Other
sponding length value (Length) proprietary FBlocks are defined in their
entirety by the system or device manufac-
108 Bus systems | MOST bus

Functions and operations Data field

A function block is made up of several The data field of the control message con-
functions that are addressed with a 12-bit tains the parameter values for a function
FktID. Depending on the type of function, call or its results. The data field is inter-
the function may be assigned differently preted as specified in the definition of the
predefined operations, which are de- function concerned and may contain one
scribed by a 4-bit OpType identifier. In the or more parameter values.
case of functions, a further distinction is
made between methods and properties. The MOST standard defines the following
A method describes an action that an parameter types:
FBlock is able to execute, e.g. starting the • Boolean and BitField for individual

station scan of a radio tuner. A method boolean values (1 byte) or sequence

can be initiated by the controller using a of individual bits (1, 2 and 4 bytes).
Start or StartResult operation. In response, • Enum for enumerations.

the slave returns a corresponding result • Unsigned and signed byte, word and

message if applicable (for StartResult). long for integer values with or without
If this result is not made available within sign and a size of 1, 2 or 4 bytes.
a specific period, the controller is notified • String for character strings. These

that the method is still being processed by are zero terminated and contain a
means of a processing message. In the end, description of the character coding
a method may even be terminated by the in the first byte.
sending of an abort message. • Stream, ClassifiedStream and

A property describes a particular attri- ShortStream for various types of

bute of the FBlock, such as the number of byte sequences of any length.
the track currently playing. A property
such as this can be requested (Get opera- A parameter type for real numbers does
tions) or set (Set or SetGet operations) not exist. The MOST standard provides for
by the controller. As the result of such a real numbers represented as fixed-point
query, the slave responds with a status values by the integer parameter types.
message, if applicable (for Get and SetGet), More complex data types are not defined
containing the current contents of the in the standard either. These have to be
property. realized by function classes instead.
Furthermore, the MOST standard pro-
vides for a notification service with which
a controller may register with a slave for
certain properties. If the value of one of
these properties changes, all controllers
that have registered for notification are in-
formed of the change by a status message.
Bus systems | MOST bus 109

Function classes Applications

To standardize the way in which functions As well as defining the lower layers neces-
are defined, the MOST standard specifies sary for data transfer, the MOST standard
a series of function classes for proper- defines the interfaces for typical applica-
ties. These determine which properties tions from the area of vehicle infotainment
the function has and which operations are systems, e.g. a CD changer, amplifier or
permissible. radio tuner.
In addition to simple function classes The FBlocks defined by the MOST
(with a single parameter for numerical Cooperation are summarized in a func-
values or texts), there also exist complex, tion catalog and listed with their current
composite function classes. These are versions in the next section. The function
used to represent composite data struc- catalog defines in detail all functions for
tures such as records and arrays. They can the FBlocks contained in them as well
also be nested at a specific nesting depth, as the permissible operations and their
which makes it possible to create two- parameters.
dimensional arrays or arrays of records, For the function catalog, there is a
for example. These nested data structures machine-readable description available in
are, for example, used to represent the XML format that can be used to import the
phone book of a mobile phone. catalog into various MOST tools.
With multidimensional function classes, In addition to the function catalog’s
the two parameters PosX and PosY are description of all functions of an FBlock,
used to address a particular position in the developments are underway to specify
first or second dimension, where 0 always the dynamic behavior for the use of an
stands for all elements of this dimension. FBlock. This involves the use of message
This makes it possible to access not only a sequence charts (MSCs) compliant with the
specific element but also an entire line of a MSC 2000 standard but with minor, MOST-
twodimensional array. specific enhancements.
Functions required by various applica-
tions, e.g. the aforementioned functions
for managing sources for multimedia data,
are defined in the GeneralFBlock. This is
used as a function compilation, i.e. when a
new FBlock is defined, all necessary func-
tions are copied from the GeneralFBlock to
the new FBlock.
110 Bus systems | MOST bus

Standardization The following application interfaces are

The MOST standard is maintained by the standardized by the MOST Cooperation.
MOST Cooperation, which also publishes The following versions are currently valid:
the corresponding specifications. The • AudioAmplifier (FBlockID: 0x22),

specifications are available through the Version 2.4.2: a simple amplifier.

home page of the MOST Cooperation • AuxIn (FBlockID: 0x24), Version 2.4: an

( interface for connecting MP3 players or

The following versions of the MOST data carriers with pieces of music.
specifications are currently valid (as at • MicrophoneInput (FBlockID: 0x26),

2006): Version 2.3.1: a microphone input.

• MOST Specification, Version 2.4. • AudioTapePlayer (FBlockID: 0x30),
• MOST Dynamic Specification, Version 2.3.1: a cassette player.
Version 1.2. • AudioDiskPlayer (FBlockID: 0x31),
• MOST Specification of Optical Physical Version 2.4: a CD player or CD changer.
Layer, Version 1.1. • DVDVideoPlayer (FBlockID: 0x34),
• MOST Specification of Advanced Optical Version 2.4.1: a DVD player or DVD
Physical Layer, Version 1.0. changer.
• MOST Specification of Electrical • AmFmTuner (FBlockID: 0x40), Version

Physical Layer, Version 1V1. 2.4.2: a radio receiver for FM/AM.

• MOST High Protocol Specification, • TMCTuner (FBlockID: 0x41), Version

Version 2.2. 2.3.1: a special receiver for traffic-

• MAMAC Specification, Version 1.1. message signals (TMC).
• TVTuner (FBlockID: 0x42), Version

The MOST Cooperation defines within 2.3.2: a television receiver.

the framework of the compliance process • DABTuner (FBlockID: 0x42), Version

requirements placed on MOST devices 4.0: a receiver for digital radio (DAB).
which must be satisfied by all devices in • SDARS (FBlockID: 0x44), Version 2.4: a

order to be able to carry the MOST logo. receiver for satellite radio.
MOST compliance is tested and awarded • Telephone (FBlockID: 0x50), Version

by test institutes which have received ap- 2.3.2: a telephone module or a connec-
propriate accreditation from the MOST tion to a cellular phone.
Cooperation. The compliance require- • GeneralPhonebook (FBlockID: 0x51),

ments are governed by the following speci- Version 2.3.1: access to a phonebook,
fications, which describe the compliance e.g. that of a cellular phone.
process itself and the compliance tests • NavigationSystem (FBlockID: 0x52),

for the physical layer, for system mecha- Version 1.11: an interface for a navi-
nisms (Core) and for application interfaces gation system; but only unofficially
(Profile): approved by the MOST Cooperation.
• MOST Compliance Requirements, • GraphicDisplay (FBlockID: 0x60),

Version 2.0. Version 2.3: an independent display

• MOST Compliance Test of Physical Layer screen.
Specification, Version 1.0.
• MOST Core Compliance Test Specifica-

tion, Version 1.1.1.

• MOST Profile Compliance Test Specifica-

tion, Version 1.0.

Bus systems | TTP/C 111

TTP/C Areas of application

TTP/C was originally developed for use
Overview in safety-relevant systems in the automo-
The Time-Triggered Protocol for SAE tive field. Because the core functions of
Class C networks (TTP/C) is a time-con- TTP/C were formally verified and a TTP/C
trolled protocol in which the network us- network can be put together economically
ers transfer data in special time intervals compared with other protocols with simi-
determined in advance. In other words, all lar reliability properties, TTP/C has also in
network users have a global time defini- the meantime come to be used in aircraft,
tion which is determined by means of a a field which is subject to very stringent
time-synchronization protocol. requirements with regard to safeguarding
TTP/C was developed specially to satisfy functionality. TTP/C is also used in train
real-time requirements in distributed, control systems.
fault-tolerant systems. It was developed
specifically with a view to its use in the Fault-tolerance strategy
automotive field, but has in the meantime The fault-tolerance strategy is used to
also come to be used in other fields, such describe which faults can be detected
aviation and railroads. and handled. TTP/C is able to detect and
According to the requirements of the handle each individually occurring fault
Society of Automotive Engineers (SAE) for (Single Fault Hypothesis). In addition, a se-
Class C protocols, TTP/C is a protocol for ries of multiple fault scenarios is handled.
high-speed networks (>125 kbit/s bit rate) However, some important multiple fault
which can be used for real-time commu- scenarios, such as a complete temporary
nication in check systems. For this reason, breakdown of communication for example,
particular emphasis was placed during cannot be detected on the node level. Ad-
development on the fields of fault detec- ditional methods of fault detection and
tion and tolerance, robustness against mal- handling are required on the application
functions, composability, and guaranteed level for this purpose.
latencies. The TTP/C network must consist of at
TTP/C is based on the principle of least four real member nodes to ensure
Time-Triggered Architecture (TTA), a that all single faults are detected.
framework for developing highly reliable, The network ensures on the architecture
distributed real-time systems and the ap- level that a faulty network node cannot
plications that run on them. prevent any properly functioning network
The development of TTP/C dates back node from transmitting its message. This is
to the MARS project (Maintainable Real- done by bus guardians, which permit bus
Time System), which was conducted in access only at defined points in time.
1979 at the Technical University of Berlin.
Further development followed within the Time-Triggered Architecture
framework of further projects, such as Time-Triggered Architecture (TTA) forms
TTA (Time-Triggered Architecture) or the the basis on which the TTP/C protocol was
Brite-EuRam III project “Safety Related specified (Fig. 1). The core element of TTA
Fault Tolerant Systems in Vehicles (X-by- is the network node, which comprises a
Wire)”, in which systems for activating host processor and a TTP/C controller,
safety-relevant automotive components, referred to in the following as the control-
such as brakes or steering systems, via ler. The network node is the “Smallest Re-
electronic networks without a mechanical placeable Unit” (SRU) in a TTP/C system.
fallback level were examined. The host processor processes inputs from
112 Bus systems | TTP/C

sensors and activates actuators. The Com- Communication Network Interface (CNI)
munication Network Interface (CNI) forms The CNI functions as a temporary firewall
the interface between the host processor because it makes available a memory area
and the controller and provides memory via which data received and to be sent are
areas via which the host processor and the exchanged between the host processor
controller can exchange information. and the controller. This interface repre-
The controller is in the end connected to sents the host processor’s sole possibility
the TTP/C bus, by way of which it connects of transferring information via the net-
the node with other nodes. The group of work. It is therefore not possible for the
all the nodes connected to a TTP/C bus and host processor to influence the communi-
the bus forms a cluster. cation sequence. In particular, the point in
time at which information is provided by
Host Processor the host processor has no influence on the
The host processor executes the applica- network’s communication sequence.
tion, the actual function of the network
node. In order to support the application, TTP/C Controller
an operating system, which also for the The controller (Fig. 2) operates entirely
most part provides the FT-COM layer, runs independently of the host processor. Its
on the host processor as well as the ap- most important components are the proto-
plication. The FT-COM layer introduces a col processor, a local bus guardian, and the
further abstraction level of the technical Message Descriptor List (MEDL).
realization of communication. The applica- The function of the protocol proces-
tion can provide data for transfer via the sor is to prepare information provided
interfaces of the FT-COM layer, thanks to by the host processor in the CNI in such a
which the application designer does not way that they can be transferred as TTP/C
have to deal with deeper concepts, such as frames.
the memory areas in the CNI. The controller is connected via two in-
terfaces with the remaining components
of the network node. The connections to
the host processor and to the transceivers
(drivers) are established via the CNI and
the Logical Line Interface respectively.

1 TTA network

Node Node Node Node

actuators Host Host Host Host
processor processor processor processor
communication Controller Controller Controller Controller

Channel 0

TTP/C bus
Channel 1
Bus systems | TTP/C 113

Message Descriptor List (MEDL) Strictly speaking, the designation Message

The MEDL contains all the check informa- Descriptor List is incorrect. Information
tion required by the controller to send and blocks within a frame are designated as
receive data. It is a list adapted for each messages in TTP/C (Fig. 3). However, this
controller containing, for example, infor- structure exists only from the perspec-
mation on the position and size of transmit tive of the host processor, not from the
and receive slots. perspective of the controller, which only
The information stored in the MEDL processes frames. The designations has
includes the points in time at which the persisted, however, because it was intro-
controller frame may be transmitted and duced in earlier specifications.
at which frames of other controllers are
expected which are processed by the host Local and Central Bus Guardians
processor. The primary function of the bus guardian
CNI memory positions in which informa- is to limit transmissions of a node to the
tion from the host processor is to be pro- defined time windows. For this purpose,
vided for transfer and in which received it accesses the information of the MEDL
information from the controller is to be in which the transmit slots of the node are
stored are also defined in the MEDL. stored.
The MEDL also contains further in- There are two different types of bus
formation needed for the operation. For guardians. Local bus guardians are con-
example, the identification of frames used nected between the controller and the
to synchronize the internal clocks and the TTP/C bus and prevent bus access out-
synchronization times. side defined transmit slots. Central bus
guardians operate in a central coupler.
If exclusively central bus guardians are
2 TTP/C node used, however, only faults are excluded for
the nodes which are directly connected to
the couplers. Connected bus topologies,
on the other hand, cannot be completely
I/O Interface


3 Message components

Communication Network
Interface (CNI) Application data:
Supplied/processed by the host processor

Protokoll- Messages:
MEDL Data areas interpreted
by the host processor

Bus Guardian Controller Message Message Message CRC

1 2 3
Logical Line
Interface Protocol overhead Protocol overhead
Frame type, explicit C-state Frame CRC
Driver Driver


Bit stream transmitted via the channel

114 Bus systems | TTP/C

In addition to monitoring bus access, the Network

central bus guardian can also assume the Network Size
function of detecting and containing SOS TTP/C is designed for networks of up to
faults (Slightly-Off Specification). These 64 nodes. A TTP/C network can consist of
are faults in which a node continually vio- a minimum of two nodes, but at least four
lates slightly one or more communication nodes are required to satisfy the Single
parameters, such as the voltages used or Fault Hypothesis. A problem in the data
the timing of the transmit slots. The upshot transfer lies in the danger that a transfer is
of this is that some network nodes which not received uniformly by all the network
were designed with greater tolerances nodes. If there are at least four real mem-
process these messages, but other reject ber nodes (nodes with sole access to one
them as faulty. As a result of SOS faults, it transmit slot) in a TTP/C network, TTP/C
becomes problematic to establish whether can identify reliably whether the message
data have been correctly transmitted or transmission was successful or whether
incorrectly received. faults occurred during transmission or
Logical Line Interface
The Logical Line Interface (LLI) is the Topology
interface between the controller and the All TTP/C networks have two independent
transceiver (driver). The LLI establishes physical channels via which a network
the structure of the logical information in node simultaneously transmits information
which the driver must accept information redundantly. This configuration is required
to be transmitted and in which it must pro- in order that the Single Fault Hypothesis,
vide received information. This makes it i.e. the handling of the failure of any com-
easier to adapt controllers to a new trans- ponent, can be satisfied. However, it is also
mission medium since only the drivers possible for nodes which do not fulfill any
have to be modified; however, the format safety-relevant functions to be connected
of the interface to the rest of the system only to one of the two channels.
remains unchanged. Since a controller always forwards
messages via the LLI to its drivers, it is
Driver possible to adapt TTP/C to very different
The driver is the transceiver for the trans- topologies. Bus, star and hybrid topologies
mission medium. It converts the physical from any combinations of bus and star to-
states of the TTP/C bus into logic states, pologies are described (Fig. 4) in the TTP/C
which can be processed by the controller. Specification (Version 1.1).
Bus systems | TTP/C 115

Bus topology Star topology

In a bus, all the network nodes are con- If a star topology is used, the network
nected in series to central cables. If a bus nodes are connected to each other via cen-
topology is used, the structure dictates tral couplers. Here, each TTP/C channel is
that no central bus guardians can be used. brought together in its own coupler.
However, the overall length of the cabling Star topologies facilitate the use of cen-
can be reduced to a minimum for this pur- tral bus guardians, which are able to pro-
pose. vide additional protection against faults.

4 Network topologies Hybrid topologies of bus and star

Even hybrid technologies composed of bus
and star topologies can be built. For ex-
a ample, it is possible to cascade several star
Node Node Node Node topologies into a multi-star or to connect a
sub-bus to a star topology.

Transmission Media
Within a TTP/C network, communication
takes place via two channels. The channels
are designated Channel 0 (Ch0) and Chan-
Node Node nel 1 (Ch1), or Channel A (ChA) and Chan-
SC nel B (ChB).
Because the TTP/C Specification is
SC formulated very openly with regard to
Node Node the transmission medium, it is possible
to build TTP/C networks using very dif-
ferent media. Both electrical and optical
media can be used here. In particular, it is
c also possible to design one channel to be
Node Node optical and the other to be electrical. Dif-
ferences in the latencies of the different
media can be compensated for by way of
Node Node
SC SC the configuration in the MEDL.
The achievable bit rate also varies to-
Node Node gether with this flexibility; however, the bit
rate itself is fixed during operation within
a network and cannot be altered.
It is possible with current implemen-
d tations to achieve bit rates of 5 Mbit/s
Node Node Node for asynchronous data transfer during
system starting and of up to 25 Mbit/s for
SC synchronous data transfer during normal
operation. Prototypes in the laboratory Fig. 4
SC a Bus
environment has achieved bit rates of up
b Star
to 1 Gbit/s.
Node Node Node c Multi-star
d Star/bus

SC Coupler
116 Bus systems | TTP/C

Communication cycle FTUs are used so that safety-relevant

TTP/C controls bus access by means of a information can also be provided in the
TDMA process (Time Division Multiple event of the failure or one or more net-
Access, Fig. 5). Here, the available band- work nodes. Here, all the nodes combined
width is divided into time windows (slots) in a FTU are able to process a task and use
in which in each case an established net- in a cycle alternately the transmit slot of a
work node can utilize the network’s full round allocated to the FTU.
TTP/C distinguishes between rounds Passive Nodes
and cycles in the sequence of slots. All A node transfers data only when current
rounds contain the same sequence of data are available for transfer, otherwise
slots whose size can be independently it remains silent. A node which does not
established. A cycle is a defined sequence transfer any data in its transmit slot is
of rounds in which the individual rounds a passive node. This is the standard be-
differ only in the content of the individual havior when the controller of the node
slots, but not in the sequence of size of the establishes that its host has not updated
slots. the “Host Life Sign”, by means of which
In round, each member node of the an application signals that it is active. The
cluster transmits one frame precisely in controller then assumes that the applica-
one slot. A distinction is made here be- tion cannot generate any valid data and
tween two types of member node: real does not transmit.
and virtual. Real member nodes are nodes This mechanism is also used for monitor
which have their own transmit slot which nodes, which merely monitor the commu-
only this node can access. Virtual member nication of the other nodes, but themselves
nodes consist of a group of nodes which do not influence the communication.
use a common transmit slot. This structure
is used to build up redundant structures
(Fault Tolerant Unit, FTU) which make the Time control
failure of nodes more tolerable. The time window of a slot can be subdi-
vided using the performed activities into
different phases (Fig. 6). Before a node
can transmit information, it needs time in
order to read in the transmission param-
5 Schematic bus-access diagram eters or to prepare the transceivers. This
phase is called the Pre-Send Phase (PSP).
This is followed by the time interval in
A Node C Frame Ch0 Frame
which information is actually transmitted
B Round slot D Frame Ch1
via the network, the Transmission Phase
(TP). After a message has been received,
A 0 1 2 3 0 1 2 3 0
a node needs time in order to prepare the
information. This takes pace in the Post-
B 0 1 2 3 4 5 6 7 0
Receive Phase (PRP). Because the time
needed for PSP and PRP is not fixed, the
PRP is followed by the Idle Phase, a phase
Node slot Time
through which the length of a node slots
TDMA round TDMA round is stretched to the time interval defined in

the MEDL.
Cluster cycle
Bus systems | TTP/C 117

A TDMA slot is fixed at points in time COM layer, also an application, which is
which can be observed using the transmis- processed on the host processor.
sion. But because this is not applicable to The lower layers are implemented in
all the activities of a controller during a the controller, where data communication
slot, a distinction is made between node takes place between the host processor
slot and TDMA slot. and the controller via the CNI interface.
For a node, the slot begins with the PSP. In the controller, the top layer is the Pro-
But because no signal is transmitted in this tocol Service Layer; under this is the Data
phase, external observers lack a precise Link Layer followed at the lowest level by
point in time at which the node slot begins. the Physical Layer.
The first precise point in time which can
be observed from an external source is the Physical Layer
start of the TP. This point is also called the The Physical Layer comprises the require-
Action Time (AT). A TDMA slot therefore ments which are placed on the physical
comprises the time interval which passes transmission of information. Often the
between two ATs. Therefore, there exists connections to be used, the cable types or
in a TDMA slot only the TP and the Inter- the voltage levels are established for this
frame Gap (IFG), in which the PRP, Idle purpose in a protocol specification.
Phase and PSP are combined. Efforts are indeed underway to formu-
A node slot, on the other hand, begins late a specification for a TTP/C physical
with the PSP, which is followed by the TP, layer, but the TTP/C protocol specification
the PRP and the Idle Phase. merely establishes requirements which
must be fulfilled by the physical layer
TTP/C Protocol used. This has the advantage that the pro-
TTP/C systems use the TTA framework tocol is very flexible with regard to the
(Time-Triggered Architecture) as the basis networking and bit rate. It is thus possible,
for which – with the exception of the appli- for example, for networking to use both
cation – finished solutions can be used. optical and electrical media, which can
The TTA protocol stack (Fig. 7) can be also be combined with each other.
divided hierarchically into three areas. However, the system designer is faced
The top layer is the host layer, the ex- with the problem of initially having to
ecuted application. Under this is the FT- identify and evaluate possible solutions.

6 Slot timing 7 Protocol stack

Executed application
Node Dslot Node
slot i-1 Node slot i slot i+1 FT-COM CNI
Redundancy management FT-COM
DTP DPRP DIdle DPSP Message permanence Layer
PRP Idle PSP TP PRP Idle PSP TP Communication services Protocol
Safety services Service
Action Time

Action Time

IFG Higher-layer services Layer



Frame structure Data Link

Endianess of bus Layer
TDMA slot i-1 TDMA slot i TDMA
slot i+1 Bit synchronization Physical
Bit coding/decoding Layer

Layers processed by the host processor

Layers processed by the controller
118 Bus systems | TTP/C

Essentially, three requirement are placed FT-COM Layer

on the physical layer: The FT-COM Layer offers functions for
• Two independent physical transmission realizing replicated, distributed applica-
channels must exist. tions. These include redundancy control
• The transmission channel must permit and the decision as to whether replicated
distributed broadcast transmissions, i.e. messages of other nodes are correct.
transmissions which are received by all In order to allow applications standard-
the network users. ized access to communication data, the
• The delay which occurs during the FT-COM Layer offers an interface similar
transmission of information via a me- to CNI in which the data for the Host Layer
dium (Propagation Delay) must be applications are provided.
Host Layer
Data Link Layer The Host Layer comprises the applications
The Data Link Layer contains the functions which run on the host processor. In addi-
which are needed to transmit frames be- tion to the actual applications, these in-
tween nodes. These include the structures clude the operating system and the check
which are required for access to the trans- structures (control loops).
mission channels and for the transmission In order to give the communication sys-
of frames. In particular, the Data Link tem as modular a structure as possible, the
Layer establishes the format of the mes- applications of this layer should access the
sage frames. FT-COM interface and not the CNI. This
makes it possible, for example, for an ap-
Protocol Service Layer plication which previously ran on a real
The Protocol Service Layer is responsible member node to be switched with minimal
for implementing the protocol services of- complexity to a series of virtual member
fered. Here, a distinction is made between nodes.
communication services, safety services
and higher-level services.
The functions of the communication ser-
vices are the safeguarding of reliable data
transfer, startup of a cluster, reintegration
of member nodes in the cluster, transmis-
sion acknowledgment, and fault-tolerant
clock synchronization.
The safety services include the manage-
ment of node membership, avoidance of
cliques, and the independent bus guardian.
The higher-level services include func-
tions for delayed mode changes, distrib-
uted alarm, external clock synchroniza-
tion, or reconfiguration of a node.
Bus systems | TTP/C 119

Protocol services The internal clocks of the network nodes 1)

Jennifer Lundelius-
Cluster Startup of a cluster can run with different levels of
All the nodes of a cluster act without syn- precision. Only nodes which have precise
Nancy A. Lynch
chronization during cluster startup; bus clocks should be used for time synchro-
access is therefore not yet controlled via nization. These nodes receive the flag as 2)

TDMA. Because there is always a trans- “Master Clocks” in the MEDL, whereby Hermann Kopetz,
mission delay caused by the medium the SYF (Synchronization Frame Flag) Wilhelm Ochsenreiter:

used when messages are distributed in a is set for their frames. Nodes with more “Clock Synchroniza-
tion in Distributed
network, two unsynchronized nodes of a imprecise internal clocks act as “Slave
Real-Time Systems”,
cluster can start simultaneously with the Clocks” and use the clock information of
IEEE Transactions on
message transmission if the transmission the master clocks. There must be at least Computers, 36(8):
delay is greater than the duration of the four master clocks in a cluster to ensure 933-940, August 1987.
startframe. This can occur during system that clock synchronization is Byzantine-
startup if several nodes are authorized to fault-tolerant. Byzantine faults are faults in
initiate synchronization (coldstart nodes). which different nodes interpret the same
The situation can arise where different transmission differently.
cliques arise within one cluster of which The nodes of a cluster are synchronized
only the nodes of one clique are synchro- by measuring the Action Time (AT) in
nized, but not all the nodes of the cluster. each transmitted slot. If the SYF flag of a
In order to avoid this, all the nodes re- frame is set and the received frame was
ject the first correctly received coldstart faultlessly received, the AT of the frame is
frame. The coldstart nodes wait for a time stored in a stack four deep. In other words,
interval (startup timeout) before trying the last four correction values determined
again to transmit a coldstart frame. Be- are stored in the stack.
cause the size of the startup timeout must The actual clock synchronization takes
be different for all the coldstart nodes in place in frames, for which the ClkSyn
the cluster and differ at least by the size flag (Clock Synchronization) is set in the
of the transmission delay, collisions can MEDL. If this is the case, the mean value
no longer occur during the second cluster is formed from the stack values, where
startup. This process is also known as the highest and lowest values are ignored.
“Big Bang”. The internal clock is now corrected by the
calculated mean value.
Synchronization TTP/C also supports the inclusion of
In a communication system which controls external reference clocks (e.g. GPS) in the
bus access by means of TDMA, it is impor- system. Time Gateway Hosts are nodes of
tant for the individual users to have as pre- this type which offer an external time ref-
cise an idea as possible of a global time. erence. These nodes transmit the external
Synchronization in a TTP/C system uses correction value as application data in a
for the operation time only the offset cor- frame. The remaining nodes of the cluster
rection by means of which the onset of a store this value in the CNI and include it in
point in time is synchronized to all the in- the calculation of the AT correction values.
ternal clocks of the nodes of a cluster. For
this purpose, TTP/C uses a variant of the
Welch-Lynch algorithm 1) which was de-
scribed by Kopetz and Ochsenreiter 2).
120 Bus systems | TTP/C

Membership Service Transmission acknowledgment

TTP/C uses Membership Service to ensure TTP/C offers an implicit transmission ac-
that all the nodes of a cluster at each point knowledgment. Because the member vec-
in time have the same consistent view as tor of the next sender in the round is also
to which nodes are present in the network transmitted as part of the frame, the origi-
and which are not. For this purpose, each nal sender can check whether its transmis-
node carries a member vector in which it sion was correctly received.
notes whether the last received transmis- But because the successor may be faulty,
sion of a node has run correctly. If a node the sender also observes the transmission
receives a message incorrectly, it changes of the second successor. If the latter iden-
the entry of the message sender to “incor- tifies in its member vector that it has also
rect”. received the transmission incorrectly, this
By comparing the member vectors of means for the sender that a transmit fault
the following frames, a calibration is made has occurred at its end. If the second suc-
with the remaining nodes of the cluster. cessor has received the message correctly,
the first successor detects that a receive
Clique Avoidance fault has occurred.
In order to achieve correspondence within Measures can now be taken to handle
the cluster with regard to the current state both fault types.
of the cluster, it is necessary to prevent the
formation within the cluster of subgroups Controller State (C-State)
or cliques whose view of the current state In order for the applications executed on
differs from the real state of the cluster. the host processors to be able to deliver
To prevent the formation of cliques, consistent results, it is necessary that
each node logs how many transmissions all the nodes of a cluster share common
were successfully and how many were in- knowledge of the state of the cluster. This
correct. knowledge is summarized in the C-state.
Before it can transmit a frame itself, Each node may indeed determine its C-
the Clique Avoidance Algorithm is used. state independently, but the cluster agrees
The controller checks whether more suc- by comparison with the other nodes to the
cessful transmissions were received than view of the majority of the users.
incorrect transmissions. Only if this is the The controller state is influenced by the
case does the node set both counters to values for global time, current position in
zero and transmit its frame. If this is not the TDMA round (MEDL position), current
the case, the controller switches to the cluster mode, requested mode change
“Freeze” state, in which the controller is (DMC), and the member vector.
deactivated and no data are transmitted.
From this state, it can be placed by the
host processor in the initialization mode
Bus systems | TTP/C 121

Cluster Mode the case, the controller assumes that there

In many real-time systems, operation has been a communication fault and begins
and check functions can be divided into with a cluster startup.
phases which are mutually exclusive. For
example, the operation of a motor vehicle Frame format
could be divided into the phases “Driv- Because TTP/C regulates bus access via
ing” and “Stationary”. In order to support TDMA, all messages are transmitted in
the different requirements placed on the broadcast mode and a priori knowledge
exchange of information in the different of the communication sequence and its
phases, it is possible to define in a TTP/C content is present in the MEDL, it is pos-
network modes in which the transmitted sible to keep very low the overhead of the
information in a frame can vary. However, protocol, i.e. the volume of information
the order of the slot in a TDMA round is needed to process the frame.
the same in all modes. Two different frame types are used in
For the purpose of correct interpreta- the operation of a cluster (Fig. 8). Cold-
tion, it is therefore necessary when modes start Frames, also known as Initialization
are used for all the nodes of a cluster to be Frames or I-Frames, occur exclusively
in the same mode. during system startup. Normal Frames,
also known as N-Frames, are used for
Deferred Pending Mode Change (DMC) communication during normal operation.
A change of cluster mode can be initi-
ated by applications if the controller is
permitted in the MEDL to change to the
requested new mode.
Because the change of cluster mode
comes into force only at the beginning of 8 Frame format

the following cycle, the information for

the deferred pending mode change is re-
corded in the C-state.
I/N Mode Change
If another change of cluster mode is re- Frame Flags
1 bit 3 bits
quested by another application later in the
cluster cycle, this request overwrites the
previous request. A special case is CPM
(Clear Pending Mode change), in which an I-frame
application prevents the requested change Header C-state CRC
of cluster mode.

Reintegration of a node
A node must synchronize itself with the N-frame (explicit C-state)
cluster if it has failed due to an incorrect Application data
Header C-state CRC
transmission from the member vector or up to 240 bytes

if it has been re-initialized. To this end, it

waits for a frame in which an explicit
C-state was transmitted, e.g. an initializa- N-frame (implicit C-state)
tion frame. The controller adopts this Application data
C-state and uses it for its transmission. Header CRC
up to 240 bytes

In order for this to be able to function,

a frame with an explicit C-state must be
transmitted every two rounds. If this is not
122 Bus systems | TTP/C

I-Frame C-state
An I-frame begins with the TTP/C header. The C-state consists of the global time,
This is followed by the current controller information on the MEDL and the member
C-state and finally the CRC value of the vector of the cluster. The MEDL position
frame. corresponding to the Round Slot Position
in the current cluster mode, the current
N-Frame cluster mode and the requested cluster
To ensure that Membership Service can mode change (DMC) is transmitted as in-
function, it is necessary for the C-state to formation on the MEDL.
be included in the frame structure. This An explicit C-state has a size of 96 bits
can occur explicitly or implicitly in the (six words of two bytes each).
case of N-frames.
If the C-State is explicitly transmitted, CRC checksum
the header is followed by the controller TTP/C uses the CRC checksum only to
C-state. This is followed by application identify transmission faults. Transmission
data and then the CRC value of the frame. faults are not corrected. Transmission
If the C-State is implicitly included in the faults are detected by a process whereby
frame, the frame consists of the header, each node determines a CRC checksum for
application data and finally the CRC value the received message and compares this
of the frame, the determination of which with the transmitted checksum.
was initialized with the current C-state. A polynomial is used to determine the
Download frames are a special form checksum. The TTP/C Specification in this
of the N-frame; these are accepted by all respect does not stipulate a specific poly-
the receiving nodes, regardless of cluster nomial; instead, it requires the polynomial
association and C-state. To facilitate this to facilitate a Hamming distance of at
behavior, a constant value which is known least 6. In other words, it must be possible
to all the nodes is used for these frames to within one transmission for at least five in-
determine the CRC. correctly transmitted bits to be detected.
The CRC calculation is initialized by the
Frame Fields Schedule ID, where the ID is split into two
Header parts. One part is used for initialization
The header of all the TTP/C frames is for Channel 0 and the other for Channel 1.
identical. Firstly, one bit signals the frame This prevents a node which does not have
type, whether an I-frame or an N-frame is the correct MEDL or which was connected
transmitted. crosswise to the channels from transmit-
This is followed by three bits, by means ting or receiving successfully.
of which a change of current cluster mode The C-state for initializing the CRC cal-
can be requested. These bits are set to culation is also used in the transmission of
zero if no change has been requested. frames with an implicit C-state. Thus, only
those frames whose CRC checksum was
Application data determined with the same C-state are ac-
The information of the application is trans- cepted as valid.
mitted in this area. The size of this field The calculated CRC value is transmitted
can be determined for each slot of a round at the end of a frame in a 3-byte field.
individually in the MEDL and must not
exceed 240 bytes per slot.
Bus systems | TTP/C 123

Composability Standardization
Networking technology has changed The development of TTA stretches back
greatly over the years. Where initially over a period of 20 years. Today, develop-
stand-alone systems (e.g. activation of a ment of the Specification is coordinated
turn-signal lamp via switches and relays) by the TTAgroup, a cross-industry consor-
were used, the changeover was quickly tium which was founded by the companies
made to cooperative systems in which Airbus, Audi, Delphi, Honeywell, PSA Peu-
individual components exchange informa- geot Citroën, Renault, and TTTech.
tion with each other. The Specification of TTP/C has been in-
However, the interaction of components fluenced in particular by Hermann Kopetz,
is increasingly making it difficult to ensure who between 1979 and 1982 was Profes-
the correct fulfillment of a function. For sor of Computer Process Control at the
this reason, the aspect of composability, Technical University of Berlin and since
i.e. the possibility of combining separately 1982 has run the Institute for Technical
developed subsystems into an overall sys- Information Technology at the Technical
tem without having to verify the function University of Vienna. Professor Kopetz
of the overall system again, is becoming also directed the MARS project and was
increasingly important. substantially involved in each further de-
In order to ensure that this functions, velopment of the protocol.
it is necessary to ensure that a subsystem
autonomously fulfills its functions, but also Characteristics
that this is the case when the subsystem is • Support of communication for safety-
integrated in an overall system. relevant functions.
TTP/C facilitates the implementation of • Guaranteed transmission properties.

composability through the communication • Guaranteed detection of all single faults.

parameters which are stored in the MEDL. • Fault-tolerant communication units

(FTU) for fault-tolerant provision of
Verification of the Specification data.
The functionality of TTP/C has been thor- • Local and central bus guardians
oughly analyzed. The core algorithms possible.
of the protocol specification have been • Low protocol overhead.

formally verified with regard to certain as- • Data efficiency of 85 % for transmission.

pects of consistency, stability and safety. • Supports composability with regard to

Furthermore, the function of TTP/C sys- time response and ranges of values.
tems has been tested using failure-injec- • Functions for the most part formally

tion experiments. This involved the use of verified.

both physical and software faults to check • Consistent view by all nodes of the clus-

the fault tolerance and fault-detection ter state thanks to Membership Service
properties. and Clique Avoidance Algorithm.
• Clear structuring of the protocol inter-

• No restriction of the bit rate from the

Specification. TTP/C components were

tested with a bit rate of up to 1 Gbit/s.
• Experience from product use.
124 Bus systems | FlexRay

FlexRay bus access is controlled by the use of mini-

slots. Minislots are small time windows
Overview in which a defined message can be trans-
FlexRay is a field that was designed to ferred. This access method is also known
support open and closed-loop control as FTDMA (Flexible Time Division Multi-
technologies in the automotive sector. ple Access).
Development focused on the suitability With messages transmitted in the static
for use in active safety systems in particu- component of the cycle, the transfer prop-
lar. FlexRay therefore offers transfers with erties can be assured in accordance with
guaranteed compliance with transfer known methods. In the dynamic compo-
properties, high bit rates and a fault- nent, messages are sent as required. In this
tolerant design. case, messages are prioritized based on
The development of FlexRay, managed the message ID. The division between the
by the FlexRay Consortium, goes back to static and dynamic component is freely
cooperation between German automotive configurable but it cannot be modified
manufacturers, BMW and DaimlerChrysler. later during system operation. The same
In 1999 these two companies began to applies to the slot lengths in the static
compile requirements for a new communi- range, which, while configurable, must re-
cation system. In the process, develop- main constant once operation is underway.
ment would incorporate experience FlexRay systems can be equipped with
gained from the byteflight system of two transmission channels, with each
BMW and the prototype development channel having separate lines. This makes
of DaimlerChrysler. it possible to create a redundant data
The aim of FlexRay is to provide a sys- transfer or increase the available band-
tem with high transfer rates that will work width for specific applications. The latter
in a deterministic and fault-tolerant man- can be used both for the parallel transmis-
ner while being as flexible as possible to sion of information of two network nodes
use and expand. The main fields of appli- and for the faster transfer of information
cation of the FlexRay are drivetrain sys- of one node.
tems (drive) and active safety systems with FlexRay operates at a maximum bit rate
no mechanical fall-back level (x-by-wire). of 10 MBit/s, which is achievable in opti-
However, the areas of passive safety sys- mal environmental conditions. A rate of
tems and comfort/convenience and body up to 20 MBit/s is achievable if two chan-
electronics are also supported. nels are used without redundant access.
To support these very different domains, In addition to the bit rate of 10 MBit/s,
FlexRay uses two different types of bus ac- which is the only bit rate defined in Speci-
cess. For deterministic transfer properties, fication 2.1 dated 15th December 2005,
it offers time-controlled bus access. In the there are endeavors to support lower bit
case of applications whose transfers place rates of 2.5 or 5 MBit/s.
less demanding requirements on transfer
properties, it is desirable to make as effec-
tive use of the available transfer capacity
as possible. To combine these two ap-
proaches, communication takes place in
cycles. In each cycle, there is firstly a static
transmission component in which bus ac-
cess is controlled by TDMA (Time Division
Multiple Access). There then follows a dy-
namic transmission component in which
Bus systems | FlexRay 125

Areas of application Active star topology

The areas of application of the FlexRay are An active star topology can be viewed as
very diverse. By offering redundant, time- a point-to-point link between a number
controlled and fault-tolerant transfers, of nodes and an active coupler. In this to-
not only is FlexRay suitable for use in pology, therefore, only the distance from
active safety systems with no mechanical a node to the coupler is of importance.
fall-back level and in the area of the drive- This cannot be more than 24 m.
train, but its optional, dynamic transmis- Apart from the connection lengths and
sion range means that it is able to support the number of connectable nodes, the
application in the areas of passive safety most important difference between an
systems and networking in the body and active and a passive star topology is in
comfort/convenience domains. the achievable bit rates. With the use of
With the high bandwidth of up to passive couplers, the achievable data
20 MBit/s for non-redundant transfers, rate (approx. 1 MBit/s) is below the rate
it is even conceivable, in theory, that that can be achieved with active couplers
FlexRay could be used for the transfer (up to 10 MBit/s).
of audio or highly compressed video.
Cascaded star topologies
Topology Cascaded star topologies are the product
Point-to-point of several active star topologies linked
The simplest configuration of a FlexRay together. In a FlexRay system, a topology
system is the direct link between two like this is limited to three cascaded star
network nodes. The maximum distance topologies.
between the two nodes is 24 m in this
configuration. Hybrid topologies
Hybrid topologies are a blend of topolo-
Bus topology gies from bus and star topologies, e.g. the
Between 4 and 22 nodes can be connected connection of several bus topologies to a
in a bus topology. The maximum distance star topology instead of individual nodes.
between any two nodes in a bus topology
may be 24 m. Two-channel topologies
Since it is possible to implement both
Star topology channels of a FlexRay system indepen-
A star topology is another architectural dently, different topologies can be used
variant of the FlexRay network. Both ac- for each channel. For example, one chan-
tive and passive couplers are supported. nel could be realized as an active star
topology, the other as a bus topology.
Passive star topology
3 to 22 nodes can be connected in a pas-
sive star. Here, too, the distance between
any two nodes cannot be greater than
24 m.
126 Bus systems | FlexRay

Hardware mission. The information received by the

FlexRay controller of a subscriber communication controller is forwarded to
A FlexRay node (Fig. 1) comprises the the actuators responsible for processing.
host processor, a communication control-
ler (CC) and one bus driver (BD) for each Communication controller
channel to which the node is connected. The communication controller (CC) real-
Optionally, each node may contain a izes all aspects of the FlexRay system
bus guardian (BG) for monitoring each relating to the protocol. Its tasks include
bus driver and the bus guardian may re- scheduling, synchronizing with other net-
ceive information through an additional work nodes, creating a macrotick signal,
host processor. creating a bit stream from the information
It is optional whether a network node of the host or controlling bus access.
is connected to one or both channels;
here, at the very most, the requirements Bus driver
from the field of application of the network The bus driver (BD) assumes the role of a
node take effect. transceiver. It converts logical information
into physical voltages that are carried on
Transmission agents copper wires, and vice versa. The role of
The transmission agent for a FlexRay sys- the BD also includes a protective function
tem is of a twisted-pair cabling design, against electrostatic discharge (ESD).
where shielded (STP) or unshielded (UTP) Furthermore, the BD monitors the states
cabling may be used. of the BP and BM and is thereby able to de-
Each of the two FlexRay channels con- tect whether the bus is affected by physical
sists of two wires designated bus plus (BP) faults.
and bus minus (BM). The BD also processes wakeup signals
and, optionally, is able to control the
Host processor power management of the network node
The host processor gathers information because functions of the node can be
from the sensors, which is forwarded to switched off.
the communication controller for trans-

1 Network node architecture

Communication data
Host Configuration data & Communication controller
status information
Communication data

Communication data

Configuration data & Bus guardian
Control signal

Control signal

status information (optional)

Control Status
signals signals
Control data &
status information Bus driver

Control signal (optional)

Power supply
Bus systems | FlexRay 127

Bus guardian Local bus guardian

The bus guardian (BG) is a device that will The local bus guardian (node-local bus
only permit transmissions if the connected guardian) is implemented as a stand-alone
network node is authorized to send a mes- CC inside the network node that monitors
sage. With this system, it is theoretically operation of a CC (Fig. 2). Its synchroniza-
possible for faulty network nodes to inter- tion process is independent of that of the
fere at the very most with their own trans- CC's; in particular, the CC and BG have
mission but not with the transmissions of separate oscillators.
other network nodes. In addition, it is also possible for the BG
The FlexRay protocol only provides the and CC to be controlled by dedicated host
framework for use of a BG: its use is not a processors that communicate with each
mandatory specification. Specification 2.0 other. This is, however, not mandatory.
only described a local bus guardian that As a consequence of direct control lines
would operate inside the network node (bus-guardian enable, BGE) between the
and monitor access to one of the channels. BG, CC and BD, information can only be
A critical aspect of this concept was that transferred from the CC to the BD if the
correct operation of the bus guardian in a BG permits this transfer to take place.
faulty network node could not be guaran-
teed as a consequence of the integration Central bus guardian
of the bus guardian in the node and the The central bus guardian (CBG, Fig. 3) was
permission for the bus guardian to access a new feature introduced with Specifica-
the oscillator of the network node. tion 2.1. It operates in the coupler of a
Specification 2.1 contains a refined star topology and monitors the operation
concept for the local bus guardian and of all connected nodes or networks. Faults,
introduces the concept of a central bus such as short-circuits on buses or network
guardian. nodes, that transmit information like a
“babbling idiot” with no consideration of
the transmission window, and thus inter-
fere with orderly communication, can be
detected by a CBG and suitable counter-

2 Local bus guardian 3 Central bus guardian

ECU Node 1 Node 2 Node 3 Node 4 Node 5

Host1 e.g. SPI Host2






A B Subbus



Channel A Optional Optional Optional

Node 6 Node 7 node 1 node 2 node 3
Channel B
128 Bus systems | FlexRay

measures may then be taken. As a result, the bus driver looks out for the transmis-
the effects of these faults can be confined sion of a wakeup signal, in response to
to small areas of the network. which the node is set to wakeup mode.

Operating modes Sending and receiving

In addition to normal operation, a FlexRay A bus driver connects a FlexRay network
system also supports the complete initial- node to the channel that contains a re-
ization of the network from sleep to ceiver and transmitter. If a controller is
standby (Fig. 4). In order for this system connected to both channels, it follows that
start to function, the BDs of all network there must be two receivers and transmit-
nodes have the capability to detect wakeup ters available in the network node.
signals on the bus and to start the remain-
ing other components of the dedicated Protocol
network node. Protocol layers
Special network nodes that may be con- The FlexRay protocol is built on five core
nected to the starter motor of a vehicle, mechanisms:
for example, generate the wakeup signal. • Coding and decoding

The system start is split into the wakeup • Control of bus access (media access

and startup phases. The wakeup phase in- control, MAC)

volves the activation of the network nodes • Processing of frames and symbols

by the wakeup signal. In the startup phase, • Clock synchronization and

the network nodes are initialized and syn- • Schedule monitoring by the bus

chronized. guardian
If the function of a node is not required
for the time being, the node can be set to Unlike in other network protocols, with
a standby mode in which all operations FlexRay there are interfaces between all
of the coding and decoding process are core mechanisms. This requires a process
stopped. This reduces the energy con- that coordinates and synchronizes the
sumption of the node. In sleep mode, the changes in the core mechanisms. This is
power consumption of a network node is the task of the protocol operation control
at its minimum. In this mode, all functions (POC).
of the network node are deactivated: only

4 Wakeup sequence

local wake POC state leave coldstart

wakeup event channel A inhibit mode
Node A
wakeup/coldstart node power config ready wakeup wakeup ready integration listen coldstart listen
channel A, B off/reset listen send
wake channel B
Node B
wakeup/coldstart node power off/reset config ready wakeup wakeup ready coldstart listen
channel A, B listen send

Node C
non coldstart node power off/reset config ready integration
channel B listen

Channel A wakeup

B pattern
Bus systems | FlexRay 129

FlexRay gives the host system the capabil- Coding on the physical layer
ity of influencing all five core mechanisms For coding, FlexRay uses a NRZ method
directly. To arrange this, communication (Non-Return to Zero) in which the two
takes place through an interface – the con- identical transmission states are not di-
troller host interface (CHI) – between the vided by a range in which the voltage falls
network node and host. back to a zero value. This type of coding
Concentrating on the main interfaces, makes it necessary to have mechanisms
the task areas can be arranged as a proto- with which a network node can divide the
col stack (Fig. 5). At the top level is the ap- transmitted states. To this end, FlexRay
plication, which forwards its commands to adds a sequence behind each transmitted
the CHI. Under the CHI is the POC, which byte: the byte start sequence (BSS). From
in turn has access to the MAC, clock syn- this sequence, each node can detect when
chronization and frame & symbol process- a byte has been transmitted and, with this
ing. In the next lower layer are the pro- information, encode the individual bits
cesses for coding and decoding. The low- of the transmission.
ermost level is represented by the transfer By applying different voltages to the
characteristics of the physical layer. two wires of one channel, it is possible
The bottom three layers are in nodes to create four bus states, which are desig-
that are connected to both channels: nated Idle_LP, Idle, Data_0 and Data_1
double presence because these functions (Fig. 6).
have to be fulfilled for each channel sepa- A bus state is identified by measuring
rately. More precisely speaking, this the differential voltage. Here, the bus volt-
means that for one node connected to age (uBus) is made equal to the difference
both FlexRay channels there is double between uBM and uBP (voltages at BP and
availability of synchronization, MAC, BM).
frame & symbol processing, coding uBus = uBP - uBM
and decoding, and physical layer.

5 Protocol stack 6 Bus states and voltages

Application V

Controller host interface Idle_LP Idle Data_1 Data_0

Protocol operation control

uBP uBus
Media Clock Frame & 2.5
access synchro- symbol
control nization processing

Coding and decoding uBM



Physical bus
130 Bus systems | FlexRay

With this method, the data transfer is pro- Generation of a frame bit stream
tected against external electromagnetic Before a node can transmit a frame con-
interference because these act equally on taining the data of the host, the frame is
both wires and are canceled out in the dif- converted into a bit stream. To this end,
ference. the frame is first decomposed into individ-
Idle_LP (LP = low power) is the state in ual bytes. The start of a frame is populated
which a very low voltage of -200 mV and with a transmission start sequence (TSS)
200 mV is present at BP and BM. This state, followed by a frame start sequence (FSS)
for example, is used to identify the start of (Fig. 7). From the bytes of the frames, an
a transmission. extended byte sequence (EBS) is then gen-
In Idle state, a voltage of 2.5 V with a tol- erated whereby each frame byte is pre-
erance of 500 mV is present at BP and BM. ceded with a byte start sequence (BSS).
To set the channel to Data_0 state, at The 24-bit checksum (CRC) for this
least one transmitting node must apply bit sequence (TSS+FSS+EBS) is now calcu-
a negative differential voltage of -600 mV lated and appended to the bit sequence.
to the channel. To finish the bit stream, a frame end
For the channel to be set to Data_1 state, sequence (FES) is appended.
at least one transmitting node must apply If the frame belongs to the dynamic seg-
a positive differential voltage of 600 mV ment, a dynamic trailing sequence (DTS)
to the channel. can be additionally appended to prevent
If the transmission of an information another node from beginning its transmis-
signal is neither blocked by the bus guard- sion on the channel prematurely.
ian nor the communication controller, a
HIGH bit is signaled by the Data_1 channel Communication cycle
state and a LOW bit by Data_0. Each FlexRay cycle (Fig. 8) contains a
static segment that is transmitted as the
first part of the cycle. The static segment
contains a fixed number of transmission
ranges, the “static slots”.
Optionally, there may be a dynamic seg-
ment in the FlexRay cycle and this is trans-
ferred in second place. Each dynamic seg-
ment contains a fixed, freely configurable
7 Bit stream (dynamic segment) number of “minislots”.
Also optional is the symbol window,
MAC MAC which is transmitted as the third element
of the cycle. It can be used for the trans-
mission of an individual symbol and has
CODEC a b cd the same size as a static slot.
To terminate the cycle, there is a phase –
network idle time (NIT) – in which the bus
is set to idle state. In general, the length
Low of the NIT corresponds to the remaining
TSS 1st byte last byte macroticks not used by the static and dy-
sequence sequence namic segment or symbol window. This is
1* gdBit
High not the case if it was detected during syn-

TxEN chronization that an offset correction is
Low required through which the length of the
NIT can be increased or reduced.
Bus systems | FlexRay 131

Static segment To identify the end of a transmission,

The static segment comprises a fixed num- a phase in which the channel is idle (dy-
ber of equally sized transmission windows namic slot idle phase) is added to the end
(static slots), which are sent simultane- of a dynamic slot. The phase in which the
ously on both channels. Exactly one frame transmission takes place is called the dy-
can be sent in each static slot. Both the size namic slot transmission phase.
and quantity of the static slots in a static Another difference between the dy-
segment are configured during integra- namic and static segment is that the coun-
tion. ters of slots transferred are incremented
Bus access in the static segment is synchronously on both channels in the
controlled by a TDMA method whereby, dynamic segment. In the dynamic seg-
in each slot, the frame with the corre- ment, the counter of both channels is in-
sponding frame ID is sent. Frame ID 1 cremented independently in line with
is therefore sent in slot 1, frame ID 2 in the current transmission status (Fig. 9).
slot 2 and thus frame ID n in slot n.

Dynamic segment 9 Slot counter

Unlike in the static segment, the transmis-

Slot counter channel A minislot
sions in the dynamic segment are not of
fixed length. To control bus access, “mini- m minislot 1 m+1 m+2 m+3 m+4 m+5 OfMinislots
Channel B Channel A

slots” are used that hold the same fixed frame ID m frame ID m+3 frame ID m+5
number of macroticks for all nodes con- t
nected to the network. Macroticks are
frame ID m+3 frame ID m+7
ranges in which exactly one frame can
m m+1 m+2 m+3 m+4 m+5 m+6 m+7 m+8
be transmitted.
“Dynamic slots” (transmission windows Slot counter Dynamic slot Dynamic slot
channel B with without
of variable size) are now created based on transmission transmission
the minislots. The maximum size of a dy-
Dynamic segment containing
namic slot is limited by a fixed number of

gNumberOfMinislots minislots
minislots specified during configuration. Transmission may only start within the first
pLatestTx minislots of the dynamic segment

8 Communication cycle

cycle level
static segment dynamic segment symbol window network
idle time
grid level
static slot static slot minislot minislot
action point action point action point

132 Bus systems | FlexRay

Symbol window For this purpose, selected messages from

The symbol window makes it possible to the static part of the cycle are used (sync
send a collision avoidance symbol (CAS) frames) and are transmitted by the net-
or a media access test symbol (MTS). work nodes that are connected to both
These two symbols are identical and channels.
help to prevent collisions during the In a distributed communication system,
system-startup phase. each network node generally needs its
The third symbol defined in the FlexRay own internal clock for setting transmission
protocol – the wakeup symbol (WUS) – is timing correctly. As a consequence of tem-
not permitted to be sent in the symbol win- perature and voltage fluctuations and pro-
dow. It serves only to generate a wakeup duction-related precision tolerances, these
pattern (WUP), which is used during the internal clocks can soon deviate from an
system startup. abstract, system-wide global time. In the
case of a system such as FlexRay, which
Network idle time (NIT) controls bus access by means of time slots,
During the NIT, a network node corrects the synchronization of all network nodes
the ascertained time deviations of its inter- plays an especially important role because,
nal clock. It also gives the network node without a standardized, global time preset
the opportunity in this phase to make im- for all nodes, the specified transmission
plementation-dependent adjustments and windows cannot be maintained and may
changes to settings relating to the commu- break down.
nication cycle.
Clock control hierarchy
Clock control In a FlexRay system, clocking is controlled
In both the static and the dynamic part of on three levels (Fig. 10). The bottom level
the communication cycle, FlexRay reverts is represented by the microticks, which
to the identifiers of the transmission win- are derived directly from the oscillator
dow in which a node may transmit. If this clock of the network node. A microtick
method is to be successful, it is necessary is therefore not a system-wide variable,
for all nodes connected to the network but only relevant to the node concerned.
to have synchronous time information. At the next level up, the clock is con-
trolled by means of macroticks – ranges
in which exactly one frame can be sent.
10 Timing Within fixed tolerances, macroticks form
the lowermost system-wide time unit.
Communication Each network node determines the dura-
cycle level 0 1 2 3 4 5 … 62 63 tion of a macrotick in microticks. The
number of microticks per macrotick does
vCycleCounter cCycleCountMax
not have to be the same for all macroticks.
gdCycle The top level is characterized by a clock
Macrotick control based on the communication cycle,
0 1 2 3 4 5 … n-1 n
level which always contains the same, fixed
gMacroPerCycle -1
vMacrotick number of macroticks.

Microtick 0 1 2 3 4 5 …

vMicrotick pdMicrotick
Bus systems | FlexRay 133

Time synchronization Gradient correction

A synchronization method is required in With gradient correction, deviations in
order to guarantee a standardized time the transmission frequency of a node are
across the entire system. The basic para- continuously corrected during the com-
meters for timing are the number of munication cycle. The node here checks
microticks per communication cycle the transmission frequencies of all other
(pMicroPerCycle), the duration of a com- nodes. By means of the FTAA, these
munication cycle (gdCycle) and the dura- values are used to adjust the node's
tion of a microtick (pdMicrotick), where own transmission frequency.
the relationship between each is deter-
mined by the following equation: Offset correction
Offset correction is a way to correct devia-
pMicroPerCycle = round( _____________ ) tions in the communication-cycle phase
by including an offset in the NIT. With this
During operation of a FlexRay network, method, the node similarly employs the
the clocks of the oscillators of various net- FTAA to determine its phase difference
work nodes will not be equal or even sim- in relation to the other nodes of the net-
ply remain constant. They will fluctuate work and uses it to determine the offset
in response to outside influences. This required to shorten or extend the NIT
deviation is balanced by two corrective phase. Since all nodes use this approach,
measures: the gradient correction and the all nodes can initiate the transmission of
inclusion of an offset in the communica- a cycle in a synchronous manner.
tion cycle.
Fault-tolerant average algorithm Transmissions in the FlexRay system
Since an individual node can only observe are timed by action points, which are
the differences between the expected time specially-marked macrotick boundaries.
at which a transmission should have been A transmission in a static slot begins once
actioned and the actual time at which the a static-slot action point has been reached.
transmission took place, a fault-tolerant The position of this action point in a mac-
average algorithm (FTM), as described by rotick is defined system-wide.
J. L. Welch and N. A. Lynch, is used in both In a dynamic slot, a transmission begins
corrective measures. The measured time once a minislot action point has been
differences of all transmitted sync frames reached and ends at a subsequent minislot
are sorted by their value. If more than action point. This method ensures that
seven sync frames were received, the two there is always an idle period between
highest and two lowest values are removed individual transmissions.
from the list. If between three and seven
sync frames were received, the highest and
the lowest value are removed. For fewer
than three sync frames, the average is
derived from all measured differences.

Welch, Lynch,
“A New Fault-Tolerant
Algorithm for Clock
Information and
Computation, vol.77,
No.1, April 1988
134 Bus systems | FlexRay

Frame format The startup frame indicator reveals that

FlexRay uses the same format in both this frame is a startup frame. Startup
the static and dynamic part of the frame. frames are used in the network's starting
The format can be broken down into three phase and may only be sent by special
parts: header, payload segment and trailer coldstart nodes.
segment (Fig. 11). The indicators are followed by the
frame ID, payload length, header CRC and
Header segment cycle count.
The header segment contains a total of The frame ID has a length of 11 bits and
five bytes, at the start of which is a series corresponds to the number of the slot in
of indicators. which the frame is sent. In one cycle, each
The reserved bit is intended for future frame ID exists only once on each channel.
modifications to the protocol and is sent as The payload length expresses the size
a logical “0”. of the information sent in the payload seg-
The payload preamble indicator indi- ment. Since the payload in a FlexRay frame
cates whether or not the payload segment is always sent in 2-byte words, a 62 in the
contains a network-management vector payload length field indicates, for exam-
(NMVector). An NMVector enables the host ple, that 124 bytes of payload are being
processor to transmit data directly without transmitted in the payload segment.
first being processed and prepared by the In the static segment, the payload length
CC. field always contains the same value
The null frame indicator indicates a null because the size of information here is
frame that contains no usable information. constant for all frames. Of course, this
The sync frame indicator indicates that property does not apply to frames in the
this frame is intended to be used for sys- dynamic segment.
tem synchronization. This indicator may In the header CRC, an 11-bit CRC value
only be set by network nodes that operate is transmitted. This is formed from the
as sync nodes. polynomial
x11 + x9 + x8 + x7 + x2 +1.

Since the information in the header seg-

ment is constant for all frames in the static
segment, the header CRC value can be cal-
11 Frame format culated offline and assigned to the network
node through configuration settings.
Reserved bit If a frame in the dynamic segment is
Payload preamble indicator
Null frame indicator always of the same size, it is possible here
Sync frame indicator also for the CRC value to be calculated off-
Stratup frame indicator
Header CRC
Covered Area The final field in the header segment
is the cycle count. This field contains the
Pay- count number of the cycle in which the
Frame load Header Cycle Data Data Data Data CRC CRC CRC
ID length CRC count 0 1 2 n
sending network node is currently active.
11 bits 7 bits 11 bits 6 bits 0…254 bytes 24 bits

Header Payload Trailer

Segment Segment Segment

FlexRay Frame 5 + (0 … 254) + 3 bytes

Bus systems | FlexRay 135

Payload segment development of FlexRay, depending on

The payload segment has a maximum the obligations that they have accepted.
length of 254 bytes, which are transmitted The uppermost group is represented by
in 2-byte words. the core partners, which first and foremost
The payload segment customarily trans- include the founding members. Alongside
fers the payload that is further processed the core partners are the premium associ-
by the host processors. Optionally, how- ates and associate memberships.
ever, an NMVector may also be sent in the By 2004 General Motors, Bosch and
payload segment, or a 16-bit message ID. Volkswagen had joined the FlexRay Group
With the assignment of message IDs, as core partners. In 2004 Freescale took
it is possible to send several data blocks in over Motorola's position in the group of
a frame. For this purpose, the message ID core partners.
is placed at the front of the other data by
the host processor as application data. Characteristics
• Deterministic transfers possible
Trailer segment • Optional bus guardian monitors network

The trailer segment contains a single access and protects against faults
field in which a 24-bit CRC checksum • Differential signal transmission

(frame CRC) is sent. The polynomial used • Collision-free transmission while opera-

to determine the frame CRC is: tion is in progress

x24 + x22 + x19 + x18 + x16 + x15 + x14 + • Redundant transmission of information
x13 + x11 + x10 + x8 + x7 + x6 + x3 + x+1 on two channels
• High transmission speed of up to
Standardization 10 MBit/s; up to 20 MBit/s with parallel
FlexRay Group transmission on two channels
FlexRay is a protocol specified by the • Startup and initialization from an inac-

FlexRay Group. The FlexRay Group was tive sleep state are directly supported
founded in 2000 by the companies, BMW, by the protocol
DaimlerChrysler, Motorola and Philips • Support for diverse fields of application

Semiconductors ( • Event and time-controlled transmission

The members of the FlexRay Group are of information possible

organized in a hierarchy. They also have • Published standardization by the

different levels of influence on the further FlexRay Group

136 Bus systems | Diagnosis interfaces

Diagnosis interfaces ple, individual actuators can be targeted

for activation to check them for correct
Diagnosis legislation (e.g. CARB, California operation. At the end of the line (EOL),
Air Resources Board) demands constant the diagnosis interface is used to test the
monitoring of emission-related compo- electronic control unit and make changes
nents in various electronic systems (OBD, to configurations (e.g. immobilizer, trans-
On-Board Diagnosis). Faults (e.g. electrical mission type). It is even possible to pro-
short-circuits of sensors, implausible op- gram the entire flash EPROM using end-
erating states) are stored in a fault memory of-line programming.
in the control unit. These faults can be
read using a scan tool (official testing sta- Diagnostics requires an interface for con-
tion) or workshop tester. For this purpose, necting the tester to the electronic control
there is a socket in the footwell, dashboard units. For communication, there are basi-
or center console of the vehicle to which cally two options available at the present
the tester is connected by means of a stan- time:
dard connector (ISO 15031-3) (Fig. 1). • K line and
Using a workshop tester, the after-sales • CAN bus
service is able to read the entire fault
memory: emission-related faults and The K-line network is not a bus in the true
vehicle-specific faults (e.g. from Motronic, sense, which means that collisions may
ABS). With the information stored in the occur. Since most electronic control units
fault memory, it is possible to diagnose a have a CAN interface for the exchange of
fault and repair it efficiently. The tester data between electronic systems, this op-
can also be used to clear the fault memory. tion for communication with the tester is
Furthermore, measured values and con- gaining ever more in importance.
trol-unit identification values can be evalu- Various protocols are implemented on
ated. Using the workshop tester, it is also the interfaces. These tend to be the CARB
possible to control special diagnostic func- or official protocol and the manufacturer-
tions. With actuator diagnosis, for exam- specific communication, which can be
closely related in their properties and
functions. It is even possible to use both
1 Diagnosis connector (diagnosis interface) interfaces: CARB-relevant functions run on
the CAN, manufacturer-specific functions
on the K line.
1 2 3 4 5 6 7 8
Figure 2 provides an overview of the
9 10 11 12 13 14 15 16
standards and the various layers of the
OSI reference model.
Pin 2: Bus line (high) from SAE J1850
Pin 10: Bus line (low) from SAE J1850
Pin 7: K line in accordance with ISO 9141-2
and ISO 14230-4
Pin 15: K line in accordance with ISO 9141-2
and ISO 14230-4
Pin 6: CAN_H from ISO 15765-4
Pin 14: CAN_L from ISO 15765-4
Pin 1, 3, 8, 9, 11, 12, 13
Not used by OBD

Pin 4: Vehicle ground

Pin 5: Signal ground
Pin 16: Battery positive
Bus systems | Diagnosis interfaces 137

Diagnostic protocol CARB

The communication method in diagnostics The Californian environmental authority
is characterized by the fact that the tester (CARB) introduced guidelines to promote
(client) addresses one or several electronic support for a standardized interface.
control units (servers) and requests data The CARB protocol is built on existing
output (e.g. from the fault memory) or an standards and specifies the data flow and
action (e.g. actuator diagnosis). These ser- timing (transport layer). This is established
vices and the transmission agent are de- in ISO 14230-4/ISO 9141-2 for the K line
fined in the various protocols. and in ISO 15765-4 for the CAN. The actual
communication messages (services) in the
KWP 71 application layer are precisely defined in
This protocol was developed in collabora- ISO 15031-5/SAE J1979 and permit no
tion with Bosch as the first standardized manufacturer-specific deviations.
serial interface and is based on a 5-baud
initialization and byte handshake. For ad- In California, every newly registered vehi-
dressing, the protocol provides for the use cle must support at least one of these pro-
of a unidirectional line (L line). The line tocols with all emission-related electronic
for data exchange (K line) is bidirectional. control units. The European counterpart
Alternatively, if the L line is not used at is the EOBD standard.
all in the system architecture, the K line
can also be used for addressing because
addressing and data exchange are two
chronologically sequential states. The way
in which communication is established is
predefined, the services are standardized.
2 Standards for diagnosis communication

KWP 2000 (ISO 14230: 1-3) a

This diagnostic protocol connects a tester

to the electronic control units using a K line CAN
K line. In the KWP 2000 standard, there 7 ISO 15031-5 ISO 15031-5 ISO 15031-5 ISO 15031-5

are various possibilities for the way in 6

which communication is established (fast 5 ISO 15765-4
initialization and 5-baud initialization).
The communication services are specified Fig. 2
ISO 15765-2
but there is some freedom for customer- ISO 15765-4 a CARB
ISO 14230-2 ISO 11898
specific adaptations. 2 ISO 9141-2
ISO 14230-4
SAE J1850
ISO 15765-4 communication

1 ISO 9141-2
ISO 14230-1
SAE J1850
ISO 11898 b Customer-specific
ISO 14230-4 ISO 15765-4
ISO 15765: 1-3 (CAN) b
This standard reproduces the existing Manufacturer-specific

Layers of the
protocols, ISO 14230: 1-4, on the CAN. K line: KWP 2000 CAN / UDS
ISO 15765-3 OSI reference model
It defines a way to leave untouched the 7 ISO 14230-3
ISO 14229-1 7 Application
fundamental format of messages ex- 6 6 Presentation
changed between tester and electronic 5 ISO 15765-3
5 Session

control unit and to transmit them on the 4 Transport

4 3 Network
CAN bus.
3 ISO 15765-2 2 Data link

1 Physical
2 ISO 14230-2 ISO 11898-1

1 ISO 14230-1 ISO 11898 UDS Unified Diagnostic

138 Bus systems | Diagnosis interfaces

Application protocols Communication on the K line

The increasing complexity of control-unit Physical layout
functions places high demands on the The diagnostic tester connects to one or
tuning and optimization of these systems several electronic control units using the
to suit the types of engine and vehicle con- K-line system (Fig. 3). All electronic con-
cerned. This process is known as applica- trol units have equal rights. Communica-
tion engineering. For this purpose, an tion flows bidirectionally between the tes-
application system is connected to the ter and the electronic control units. This
electronic control unit. The application means that data can be received and sent
protocol for the K line is the McMess pro- by all subscribers. However, only one sub-
tocol, and the CCP protocol for the CAN. scriber is ever able to send at a given time.
The communication is an asynchronous
McMess process. There is no additional line for the
With the McMess protocol, parameters in the transmission of a clock signal. The baud
electronic control unit can be modified and rate on the K line is between 1,200 and
operands recorded. In a special application 10,400 baud in accordance with ISO stan-
mode, the protocol describes the method dards. In special applications (e.g. manu-
and contents of communication between ap- facturer-specific flash programming),
plication system and electronic control unit it could even be as high as 250 kBd.
for the K line. With McMess, for example, the Data transfer on the K line is realized as
measuring device is able to read the contents follows: the resistor in the tester increases
of the variable memory (e.g. sensor values, the potential of the line – provided no elec-
measured values) quickly. tronic control unit is sending on it – to bat-
tery-voltage level (Fig. 4). If an electronic
CCP (CAN Calibration Protocol) control unit is activated, it connects its
The CCP protocol describes the contents K-line connection to the shared ground
of communication between an application and therefore switches the potential of the
system and an electronic control unit if K line to ground. Each of the connected
these communicate on the CAN bus. electronic control units is able to detect
CCP allows electronic control units in that one of the bus subscribers is actively
the network to function at higher payload accessing the K line.
rates than on the K line.

3 System architecture of bus systems 4 Circuit diagram of an electronic-control-unit network

CAN bus


K line

Diagnosis interface


Bus systems | Diagnosis interfaces 139

Message format Checksum

The message format of the KWP 2000 The 1-byte checksum is always positioned
protocol comprises three parts (Fig. 5): at the end of a message. It contains the
• Header value of the modulo 256 sum across all
• Data bytes and bytes in the message – apart from the
• Checksum checksum itself.

Header Initialization
The header is made up of no more than To be able to send messages, each elec-
four bytes. tronic control unit must support a subset
• The format byte (Fmt) contains informa- of the possible message formats. At the
tion about the form and composition of beginning of communication, the format
the message. Two bits (A0 and A1) pro- supported by the electronic control unit
vide address information, while six bits is communicated to the tester by means
(L0 to L5) provide the quantity of data of key bytes as part of initialization.
bytes. With this coding, there can be up The ISO standards specify which of the
to 63 data bytes. If L0 to L5 are set to possible initialization procedures are cus-
zero, the number of payload bytes must tomary for ECU diagnostics. The following
be coded using the length byte (Len). initialization procedures are supported:
• The target address (Tgt) indicates the • 5-baud initialization and
communication partner for which the • Fast initialization
information is destined.
• The source address (Src) indicates the
communication partner that sent the
• The length byte (Len) specifies the quan-
tity of payload bytes (max. 255 bytes).
Up to a value of 63, the coding can take
place in Fmt. For more than 63, the 5 Message format of the KWP 2000 protocol

length must be specified using this byte.

Header Data bytes sum
The existence of a target and source ad-
dress as well as the length byte is depen- Fmt Tgt Src Len SId
dent on the parameters from the format
byte. max. 4 bytes max. 255 bytes 1 byte

Checksum calculation
Data bytes
Up to 255 bytes of payload can be sent per
message. The first byte of the payload is
always a service-identification byte (SId). A1 A0 L5 L4 L3 L2 L1 L0
The subsequent data bytes are data that
differ depending on the service con-
cerned. 0 0 No address information
0 1 Exception mode (CARB)
1 0 With address information,
physical addressing

1 1 With address information,

functional addressing
140 Bus systems | Diagnosis interfaces

These different initialization methods are recognition pattern. From the bit sequence
used either for CARB diagnostics or for 01010101 (binary format), the tester de-
after-sales diagnostics. Furthermore, a termines the baud rate of the electronic
distinction can be made between physical control unit, which may be between 1,200
initialization and functional initialization, and 10,400 baud. The electronic control
which introduces the topic of “point-to- unit then transmits the two key bytes,
point” or “point-to-multi” communication. which communicate the header format
With functional initialization, a group of and timing method that the electronic
electronic control units is addressed and control unit supports. As acknowledgment
initialized. A function initialization can that communication has been successfully
only be successful if all the electronic con- established, the tester sends the comple-
trol units of this group support the same ment of the second key byte back to the
baud rate, the same transmission timing electronic control unit and, in return, re-
and the same protocol. By contrast, only ceives from the electronic control unit the
one electronic control unit is initialized in complement of the address. Initialization
the case of physical initialization. is then complete and the regular data
transfer may begin.
5-baud initialization
Communication is initiated by means of Fast initialization
a 5-baud address sent by the tester. It is Fast initialization establishes communica-
thus possible for each electronic control tion in a shorter time than 5-baud initial-
unit to be addressed individually and ad- ization. With fast initialization, the tester
dressing is clearly separate from the data sends a wakeup pattern (WuP) comprising
transfer. a low and high phase, each lasting 25 ms
The tester outputs the address onto the (Fig. 7). At the end of the WuP, the tester
K line at a transfer rate of 5 baud (Fig. 6). sends the StartCommunication Service at a
10 bits (eight data bits, one start and one rate of 10,400 baud. The electronic control
stop bit) take approx. 2 seconds. unit returns a positive reply containing the
On receiving the address, the electronic key bytes. Initialization is then complete
control unit responds with the baud-rate and regular communication may begin.

6 5-baud initialization 7 Fast initialization

Tester Control unit Tester Control unit

5-baud initialization Fast initialization

Address (2s) TWuP
5 baud 0x55 KB1 KB2 KB2 Address
W5 W1 W2 W3 W4 W4 Tidle

1,200 to 10,400 baud TiniL 10,400 baud

W1 60 to 300ms Baud-rate recognition
W2 5 to 20ms TiniL 25±1ms


W3 0 to 20ms TWuP 50±1ms

W4 25 to 50ms Tidle 300ms
W5 300ms
Bus systems | Diagnosis interfaces 141

CARB initialization If a falling edge was detected, the elec-

CARB initialization is a special case of tronic control unit delays sending its
5-baud initialization, although CARB may own data until the bus has returned to
also be activated by fast initialization. idle state and the P2min time period has
A functional initialization of all connected again elapsed. Only then may it attempt
electronic control units is achieved with to send its data again.
the fixed address 0x33. The baud rate is
permanently set at 10,400 Bd. The require- Collision detection
ments are described in more detail in A collision cannot be prevented if two
ISO 14230-4 and ISO 9141-2. electronic control units send at the same
One of the outcomes of these standards time. In this event, the collision still has
is that all emission-related electronic con- to be detected. This is possible by monitor-
trol units fitted in a vehicle may only sup- ing the K line. To this end, the electronic
port one protocol: ISO 14230-4, ISO 9141-2 control units reread their own transmitted
or SAE J1850. A mixture of protocols in the byte through the serial interface.
same vehicle is not permitted. In Figure 8, two electronic control units
(ECU 1 and ECU 2) send a start bit at the
Key bytes same time. ECU 2 then sends a 0 (dominant
The electronic control unit uses key bytes bit) and ECU 1 a 1 (recessive bit). The 0
to inform the tester of which header bytes, appears on the K line. With bit 2, the exact
length bytes and timing method are sup- opposite is true. Here, ECU 1 persists with
ported. The key-byte decoding process the dominant level. When the complete
is defined in ISO 9141 and ISO 14230-2. byte has been transferred, the electronic
control units compare the byte sent on the
Arbitration K line with the byte they have read. In the
Arbitration is a means of preventing or de- example, a recessive level was overwritten
tecting collisions of data sent by different by a dominant level for both electronic
electronic control units at the same time. control units. From this, both electronic
Arbitration is necessary if a correct response control units deem the transfer to have
to the tester request is to be received in been faulty and retreat from sending.
the case of functional communication
(e.g. reading the fault memory of all emis-
sion-related electronic control units using
a scan tool). Collisions can be expected 8 Sequence of unsuccessful communication

because more than one electronic control

unit is permitted to send following initial- Recessive
ECU 1 level
ization by the tester.
Arbitration is only relevant to data Start 1 2 3 4 level
transfer on the K line. bit

Collision prevention ECU 2

To prevent collisions following a tester re-
Start 1 2 3 4
quest, each electronic control unit checks
the K line for a falling edge after a time of
P2min. A falling edge may be interference, Bus
or another electronic control unit has

already been sending data. Start 1 2 3 4

142 Bus systems | Diagnosis interfaces

To prevent cyclical collisions, the elec- Different addressing methods exist:

tronic control unit reports after a time of normal and extended addressing, whereby
P2min, which is calculated in accordance normal addressing is the regular address-
with a specific algorithm. ing method and is analyzed below. Longer
Let us assume that ECU 2 in the previous messages that do not fit into a CAN frame
example sent a 1 as its first data bit. ECU 1 (seven bytes with normal addressing, six
would then receive nothing from the data with extended addressing) are segmented
transfer of ECU 2 and would continue to into several CAN frames and recomposed
establish communication. ECU 2, however, by the recipient.
detects the collision in bit 2 and retreats.
The collision in this event is non-destruc- Unsegmented messages
tive. If the quantity of data bytes to be transmit-
ted is no greater than seven bytes (for nor-
Communication on the CAN mal addressing), a single frame is trans-
Physical layout ferred. The first byte to be sent is the PCI
A CAN interface containing the necessary (Protocol Control Information). It identi-
hardware already exists in many electronic fies the frame in the high nibble (top four
control units. This bus system can also be bits of a byte) as being a single frame and
used for diagnostics. It is possible to com- indicates the data length (DL) in the low
municate with the electronic control units nibble (Fig. 9).
using a diagnostic tester connected to the The data is sent by the transmitter to the
bus by a diagnosis connector. receiver in a single message (Fig. 10a).

Baud rates Segmented messages

The typical baud rates associated with If more than seven data bytes (for normal
CAN are 500 kBd or 1 MBd. addressing) are to be delivered, the diag-
nosis tester (client) sends a first frame
Addressing and message types on the CAN data bus (Fig. 10b) first of all.
Tester communication on the CAN bus The electronic control unit (server) ac-
is defined by ISO 15765. The communica- knowledges with a flow-control frame.
tion services of ISO 15765-3 or 14 229-1 The other data is subsequently sent in
are defined in a similar way to those of consecutive frames.
ISO 14230-3 (Fig. 2).
The fundamental difference between
these protocols is in the format of mes- 9 Format of an unsegmented message

sages and how they are transmitted. While

up to 255 data bytes can be transmitted in
0 1 2 3 4 5 6 7
a message on the K line, only eight data
PCI Data
bytes are possible with the CAN. To repro-
duce the service messages on the CAN, the
contents (data bytes) of the header and the
checksum are separated and embedded in
a similar, new message format. The ad-
dress of the electronic control unit is now 7 6 5 4 3 2 1 0
the CAN identifier itself. With different 0 0 0 0 DL
Single frame
CAN identifiers, it is now possible to sup-

port functional or physical addressing.

Bus systems | Diagnosis interfaces 143

The first frame contains the PCI, an ad- (Separation Time) specifies the interval
ditional length byte (DL, Data Length) and to be maintained between consecutive
the first six data bytes (Fig. 11). In the frames.
high nibble, the PCI contains the informa- The flow-control frame is followed by
tion for identifying the frames as the first the consecutive frames. Again, the PCI
frame. The extended data length (XDL) in the high nibble contains the identifier,
is stored in the low nibble. Together with while the low nibble contains the sequence
the DL, a 12-bit data word is formed with number (SN). With the first consecutive
which it is possible to express values from frame to be sent, the SN is set to 1. The SN
0 to 4,095. is incremented with each subsequent con-
The transmitter sends the first frame secutive frame; after 15, the SN restarts
and waits for an acknowledgment of re- from 0. By evaluating the sequence num-
ceipt from the receiver in the form of a ber, the receiver is able to detect whether
flow-control frame. This message again or not all the frames have arrived.
contains the PCI, the high nibble of which
identifies the frame as the flow-control
frame. The low nibble contains the flow
status (FS), which can be used to authorize
or delay the sending of further frames
(consecutive frames).
11 Format of a segmented message

BlockSize (BS) indicates how many consec-

utive frames can be received without the
need for a further flow-control frame to a 0 1 2 3 4 5 6 7
be sent. The values expressed in STmin PCI DL Data

7 6 5 4 3 2 1 0
10 Diagram of message transfer
0 0 0 1 XDL
First frame
a Transmitter Receiver
3 2 1 0 7 6 5 4 3 2 1 0
Single frame
b 0 1 2 3 4 5 6 7
PCI BS STmin Data

b Transmitter Receiver
First frame 7 6 5 4 3 2 1 0
0 0 1 1 FS
Flow control Flow control
frame Clear to send
Fig. 10
STmin Consecutive 0 0 0 0 (CTS)
frame Block a Message transfer
0 0 0 1 Wait (WT),
size (BS) receiver in a single frame
frame not ready (unsegmented
c 0 1 2 3 4 5 6 7 message)
Flow control PCI Data b Segmented
message transfer
Last block 7 6 5 4 3 2 1 0

Consecutive Fig. 11
frame ( ≤BS) 0 0 1 0 SN a First frame
Consecutive frame
b Flow control
c Consecutive frame
144 Automotive sensors | Basics and overview

Automotive sensors

The term sensor has become common, Basics and overview

as in the past 20 to 40 years measuring
gages have also come into use in con- Sensor, term/definition
sumer applications (e.g. motor vehicle The electrical sensor outputs are not only
and domestic appliance technology). provided in the form of current and volt-
Sensors – another term for measuring age, but are also available as current or
detectors or measuring sensors – convert voltage amplitudes, frequency, phases,
a physical or chemical (generally non- pulse widths, and cycles or periods of an
electrical) variable F into an electrical electrical oscillation, or as the electrical
variable E; this process often also takes parameters, resistance, capacitance, and
place over further, non-electrical inter- inductance. A sensor can be defined using
mediate stages. the following equation:

Table 1 summarizes and compares the (1) E = f (F, Y1, Y2…)

various fields of application for sensors. Sensor output signal
Figure 1 provides an overview of the (2) F = g (E, Y1, Y2…)
abundance of electronic vehicle systems Required measured variable
already on the market. Undoubtedly,
this number will increase immensely If the functions f or g are known, then they
in the years to come. represent a sensor model with the help of
which the measured variable sought may
be mathematically calculated from output

1 Automotive systems and their sensors

Co onic) and p rake) ic

Ra l loc enso tionin sor ing)

lic ron
so n Ra

i n n
(M tio drau lect

Hi ic) c-pr otro nic) )
on ri (M tro nic


ntr e s nd se itio
At ck me ontro or )



thr eler gnos nso onic

Pr and per (air n)

y (e
Kn mas esel sen troni

otr he r o tro

(ce ssur ir-co ure ond

(tra ed s en s omm


) u
oh r

ng em so ati
(M sp nso r (M , Mo

ng so

ctr nso
su )

mo sor ol)
Ac d di e s Mo

a at -c
es nic

ati ty/t en vig

Ai nic sure l, M


ori tra

jec sor

-up en ontr
c s

ele se

(he idi ity s r (na

le or tics)
oa su rol,
ox tion

sh or ste

nit , ul

mb ct-in en

ol, al

a e

otr n

ec re nt

n-b es nt

ind ns sy
os ion r

k r
La dire re s

ntr d
Bo iss enso

(ba anc iper

Hu ua enso
(el t-p co

(O k-pr n co

co -pe
mo se te

ra rt

(w n se ing
Ta iss or

or ota

line su
Pr etr

m s
tro s

n io
ns e s


Di ield-
Ai te s

ck e s
ns en

ns f-r
so es

ott at
o s


m l
(tra ssur

se le-o
(ga h-pr






Til tion enso ag)

(ai g)
SP or

y ing

rs )
an sor or P)

te d an (saf ABS


ce rat or ba
eri or

ll-o stem
(E ens

en ens r (ES


ft s rd
r ( ety

ste ns

Ac aw- ens r (air


ns -the gua


ed se



ti e
n so



ist e

ss qu

tio sen




a s
r-a or


ler e

pa n


we T

cu tio



oc era






at- el


Se Acc






K. Reif (Ed.), Automotive Mechatronics, Bosch Professional Automotive Information,

DOI 10.1007/978-3-658-03975-2_7, © Springer Fachmedien Wiesbaden 2015
Automotive sensors | Basics and overview 145

1 Areas in which sensors are used

Typical features Primary Precision Industrial Consumer

standards measurement measurement technology
Accuracy 10–11 to 10–7 2 to 5 · 10–4 2 to 5 · 10–3 2 to 5 · 10–2
€ 100 k to
Costs € thousands € hundreds € 1 to 10
Units/year single figures approximately 10 100 to 1 k 10 k to 10 m
Use – Research – Calibration – Process instru- – Auto
– Testing of mentation electronics
secondary – In-process – Building
normals measurement services Table 1

signal E and the influencing variables Yi These model parameters are generally
practically without error (intelligent or stored in a programmable, nonvolatile
smart sensors). memory component (PROM) in the digital
processes for conditioning sensor signals,
Adjustment which have come to dominate. By contrast
In the real world, the sensor model always with the conventional analog compensa-
includes some free parameters, with which tion of influencing variables, it is not only
the model can be adapted to the actual roughly linearly acting influences that may
properties of the individual sensor type be corrected, but also strongly nonlinear
in a kind of calibration process (Fig. 4a). processes that can be corrected success-
fully. Another significant advantage of this
type of calibration, which takes place over
2 Sensor symbol a purely electrical connection, is that each
sensor may easily be maintained under
operating conditions during the calibra-
tion phase.

▶ Concept of the smart sensor
In a somewhat more general form,

intelligent sensors (smart sensors)

may be defined as follows:
Intelligent sensors, sometimes also
known as integrated sensors or sensors
with specific local electronics, allow the
3 Basic sensor function (static or dynamic) accuracy inherent
in a sensor to be exploited by means of
(generally digital) microelectronics to
a far greater degree than conventional
sensors. The sensor information, in par-
Physical/ Electrical ticularly the complex information from
Sensor multisensor structures, can be com-
chemical output pressed here by local further processing,
quantity º signal E
that is to say, it is brought to a higher
level (than the sensor alone is capable
Disturbances Yi (temperature, of providing), without a multiplicity of

power-supply fluctuations ...) external connections being required

for this.
146 Automotive sensors | Basics and overview

There is no clear determination as to Saving the correction parameters

whether sensors can already incorporate The host computer calculates the model
a part of the signal processing or not; it is, parameters for the specific sensor from
however, recommended, for example, that the data previously obtained, e.g. for a lin-
no distinction be made between elemen- ear characteristic curve course and saves
tary sensor, sensor cell, or the like and these to the smart sensor’s PROM. In a
an integrated sensor. check process, these parameters may also
first be emulated in the host computer’s
Calibration process RAM before they are finally burnt to the
A smart sensor is generally programmed nonvolatile memory of the smart sensor.
or calibrated, in a procedure correspond- If characteristic curves are approximated
ing to the calibration of conventional ana- with higher order polynomials, program
log sensors, in three steps with the help of maps (look-up tables) may also be stored
an external computer (host) (Fig. 4): in the smart sensor to avoid protracted
computing processes. The solution of sav-
Recording the actual value ing a coarse-grid program map in conjunc-
The host computer systematically varies tion with a simple linear interpolation
both the measured variable xe and the between the data points (example shown
influencing variable(s) y, thus setting a in Fig. 5) has also proved very useful.
specific number of relevant and represen-
tative operating points. The smart sensor Operating phase
here passes the, still uncorrected, raw sig- The smart sensor is now disconnected
nals xa* to it. The host, however, is simulta- from the host computer and is in a position
neously also receiving the “true” values xe itself to calculate the measured variable xe
and y from substantially more accurate with very low error thanks to the stored
reference sensors. The host computes the model data. It is able to pass the value to a
correction values necessary by comparing connected control unit in digital, bit-serial
these two values, and interpolates these or even analog form (e.g. pulse-width
corrections over the full measuring range. modulated), for instance. The measured
variable can also be distributed digitally
to further control units via a bus interface.

4 Trimming/calibration of a smart sensor

1.) Actual xe Smart

value Measured variable xa*
y sensor
recording Influencing variables External
Reference xe computer
sensors y
Systematic variation
of measurement and
influencing variables
2.) Saving Smart External
correction sensor calibration
parameters PROM computer

xe Smart

3.) Operating Operation xe Control

phase values unit
Automotive sensors | "VUPNPUJWFBQQMJDBUJPOT 147

This process can, unlike conventional laser Automotive applications

calibration, in principle also be repeated if
an erasable PROM is used. This is a partic- With increasing demands on all functions
ular advantage in the development phase in the vehicle, the control and regulation
of sensors. functions that were previously imple-
mented by mechanical means have, in the
last 40 years, successively been replaced
by electronic devices (ECU, Electronic
5 Interpolation of measured values using a
data point program map Control Unit). This has necessarily caused
high demand for sensors and actuators
Interpolation with which these electronic control units
Measuring travel s

m = 0, …, 32 value s (T,q)
q can, on the one hand determine the rele-
ure b vant states of the vehicle, and on the other
e ratq s2 s4
mp q m+1
hand, actually influence these states.
dq s1 Over this period, the vehicle industry has
a s3
become one of the, previously unprece-
Pe dented, drivers of the development of sen-
ri od
du sors that could be manufactured in large
ion DT numbers.

T dT
Tn Whereas at the outset they still were of
n = 0, …, 64 T electromechanical or macromechanical
form of some sort, the trend beginning
Example: two-dimensional data point- in the 80s was for miniaturization, with
program map s (Tn, qm) of a smart sen-
sor for the measurement of a travel s:
6 Milestones in the development of sensors for
automotive applications
For a highly precise evaluation of the
sensor, which operates using variable
1950 Lambda oxygen sensor
inductance, its natural characteristic
curve and its temperature sensitivity 1960 Electromechanical pressure sensor
are each approximated using 5th order Piezoelectric knock sensor
polynomials. As a frequency-determin-
ing element of a very simple oscillator 1970 First integrated Hall sensor
circuit, it outputs the period duration T Strain-gage acceleration sensor
as an uncorrected output signal. As a for airbag
sensor model for the measuring travel s, First pressure sensor on silicone base
an outline program map embracing only
32 x 64 = 2,048 example-specific values 1980 Hot-wire air mass meter
sn,m is stored (in the PROM) along with Thick-film air mass meter
a simple interpolation algorithm (in the Integrated pressure sensor
ROM), instead of a total of 36 poly-
nomial coefficents and a protracted 1990 Micromechanical acceleration
polynomial evaluation. Should a signal T sensor for airbag
occur between these data points Tn and Piezoelectric yaw-rate sensor
Tn+1 in conjunction with a temperature q for ESP
between data points qm and qm+1, Micromechanical air-mass meter
the sensor interpolates in two dimen-
Micromechanical yaw-rate sensor
sions in accordance with the mapping
between the benchmark figures stored 2000 Yaw-rate sensor for roll-over sensing

“error-free” s1,...s4 and thus the mea-

sured value s(T, q) sought is determined
as the result of interpolation.
148 Automotive sensors | Automotive applications

semiconductor methods (batch process- gallium arsenide (GaAs) have a substan-

ing) for the manufacture of high use sen- tially greater operating temperature range
sors. For a while, sensors originating in than silicon, which can be very advanta-
hybrid technology using thick-film tech- geous, particularly at some points in the
niques also played a not insubstantial role. vehicle. Thin metallic layers are very
These are still used in isolated applications well suited to the manufacture of precise
today, e.g. in the wafer-shaped Lambda strain-gage resistors, accurate tempera-
oxygen sensors and high temperature ture sensors and magnetic field-dependent
sensors for the exhaust line. resistors.
With silicon it is possible also to inte-
Where temperature and magnetic field grate the electronics monolithically with
sensors could initially still be designed in the sensor. This technique has lost its im-
circuit-like structures and be produced in portance – with a few exceptions (e.g. Hall-
batches, this trend increased as it became effect ICs) – because of the generally very
possible also to micromachine silicon in different number and type of process steps
many various ways in two or three dimen- and also because of the inflexibility associ-
sions, and also to connect them, even in ated with this. Hybrid integration technol-
multiple layers, very soundly by very ogies in the tightest of spaces as a rule lead
efficient methods. to substantially more cost effective, but
Where the technologies of the electronic functionally equivalent solutions (Fig. 7).
semiconductor circuits were practically
exclusively based on silicon as the base While the development of sensors was ini-
material, quite different materials and tially concentrated almost exclusively on
technologies played a not insignificant systems inside the detail in the drivetrain,
role in sensors. Thus, for instance, quartz the suspension and the body and driving
can also be micromachined using anisotro- safety, the direction of sensing of newer
pic etching techniques, and unlike silicon developments is increasingly named to the
also has very advantageous piezoelectric outside and to the area close to and further
properties. III-V semiconductors such as from the vehicle:

7 Hybrid integration of sensor and electronics: surface micromechanical acceleration sensor

on a microhybrid circuit

Bonding wire Sensor chip Suspension spring Seismic mass with

movable electrodes

200 m Fixed electrodes

Automotive sensors | Automotive applications 149

• Ultrasound sensors detect obstacles on 8 Sensors in the multilayered process that is the
parking and will even allow automatic motor vehicle

parking in the foreseeable future, per-

haps in combination with other sensors. SE/ECU SA
• Near-range radar scans the area around
the vehicle to detect objects that proba- ECU
Fig. 8
bly could cause a collision, to gain time 1 Measuring sensors
and to prime safety systems before a E
2 Adapter circuit
collision occurs (precrash sensors). 2 Yi AZ 3 3 Driver circuit
• Imaging sensors can not only detect 4 Actuators
Y1 4
traffic signs and send them to the driv- 1 AK Actuator
SE AK AZ Display
er’s display, but also detect the edge
SA Operating switch
of the carriageway, warn the driver of
Φ SE Sensors

any hazardous deviation and, where ECU Control unit
required, in the long term also permit F Physical variable
automatic driving. In combination with E Electrical variable
infrared beams and a screen in the driv- Yi Disturbances

er’s field of vision, IR-sensitive imaging

sensors could permit long distance These customer-specific adapter circuits
observation of the carriageway, even are tailor-made for specific sensors and
at night (night vision) or in foggy con- are available in integrated design and in
ditions. a wide variety of versions. They represent
• Long-range radar sensors observe the a quite substantial and very valuable addi-
carriageway for 150 m in front of the tion to the sensors described here, without
vehicle, even in poor visibility, to adapt which their use would not be possible, and
the driving speed to vehicles ahead and, the measuring accuracy of which is, prop-
in the longer term, also to support auto- erly said, only defined in conjunction with
matic driving. these.
The vehicle can be regarded as a highly
As part of the vehicle’s peripheral equip- complex process, or control loop, which
ment, the sensors and actuators form the can be influenced by the sensor informa-
vehicle’s interface to its complex drive, tion from other processing units (control
braking, chassis, and bodywork functions, units), as well as from the driver using his/
as well as to the vehicle guidance and navi- her controls. Display units keep the driver
gation functions and the (usually digital) informed about the status and the process
ECUs which operate as the processing as a whole.
units (Fig. 8). An adapter circuit is gener-
ally used to convert the sensor’s signals
into the standardized form (measuring
chain, measured-data registration system)
required by the control unit.
150 Automotive sensors | Details of the sensor market

Details of the sensor market Micromechanical processes for the pro-

duction of sensors are continually being
The proportion of value added in the vehi- further developed in tandem with the
cle by the electrical and electronics sys- semiconductor processes. These compa-
tems is currently around 26 %. Nowadays, nies do not, however, have any specific
practically every second sensor is fitted know-how in the area of specification,
into a vehicle, with annual growth rates testing and packaging appropriately
that are, in some cases, still in double for the vehicle.
digits. Since the start of the 90s, microme- • Special, generally medium-sized manu-
chanical and microsystem-technical sen- facturers of sensors, who do not produce
sors have taken a rapidly increasing share, semiconductor circuits, but have gener-
which amounted to about a third in 2005. ally chosen just a few sensor types as
Unlike in the general sensor market, their product to supply the sensor mar-
Europe has a current market share of ket overall or just preferred branches
41 % of the vehicle sensors sector and such as the vehicle market.
Bosch is the world market leader; this has • Large vehicle suppliers and system man-
clearly surpassed America with a share ufacturers (e.g. Bosch) or large subsid-
of just 34 %. The total market for sensors iaries of vehicle manufacturer who have
for vehicle applications is expected specialized in the needs and support of
to grow from US$ 8.88 billion in 2005 to their mother groups. Here too, since the
around US$ 11.35 billion in 2010, that is introduction of electronics in the vehi-
by a total of 28 % (Fig. 9). cle, experience has been gained with the
production of semiconductor and hybrid
There are typically three groups of compa- circuits in close collaboration with semi-
nies offering sensors for use in the vehicle: conductor manufacturers (process de-
• The semiconductor industry: sensors velopment, licensing). Thanks to system
from semiconductor manufacturing knowledge, it has been possible to build
originated here with the application up comprehensive know-how in the
of a few process steps. They serve the field of specification, testing and pack-
entire sensor market including the auto- aging techniques appropriate for the
motive industry and have an efficiently- vehicle.
operating sales and distribution system.

9 Sensor market for automotive applications

US $
12 11.0511.35
10 9.66
8 7.51


2001 2005 2010
Fig. 9
Source: Bosch
Automotive sensors | Features of vehicle sensors 151

Features of vehicle sensors If no sensor principles are known for the

requirements set, research will be started
While general sensors are being developed into new procedures and methods for the
for as broad a user group as possible and measurement of the variables required.
in multilevel measuring ranges, often with- This phase is one of fundamental research
out the manufacturer being aware of the which ultimately also supplies new initial
application, vehicle sensors are generally technology prototypes. This procedure
specified and optimized for a special ap- can repeat itself recursively until a prom-
plication. They are a part of a system and ising solution which can then be sent to
are often not freely available in the trade. product development is found. It is not
Their development generally takes longer unusual for this development loop to be
than that of commercial sensors, not just repeated again in full or for a different
because of the enhanced demands. It is measured variable to be selected.
more associated with the development
of the system and generally lasts as long Five distinct phases are generally identi-
as this does, as the sensor specification fied in the development of a completely
can still change until system development new sensor, just as with other electronic
is concluded. products (Table 2). While prototypes and
A-samples mostly still come from advance
The high thrust for innovation in systems development or research, B- and C-samples
for the automobile sector very often com-
pels the development of new sensor tech-
nologies or significant extensions of their 10 Genesis of a functional Bosch sensor

specifications. Figure 10 shows the typical

development phases that vehicle sensors
pass through at the supplier’s works. Implementation phase Design phase
The development process begins natu-
Technology issues
rally with the system idea originating with Application issues

the vehicle manufacturer or the supplier.

Technical Technical
The initial task is to make a selection of the assistance clarification
measured variables required, whether or Research
not this is feasible. At this stage, the sen- (Technology pool) Process engineering
sors’ function is simulated as part of the New methods
Key problems
systems engineers’ usual system simula-
Advance development
tion and a first specification is drawn up.
Prototype, A-sample Sensor principle
If it is possible to use a sensor technology Laboratory test/ Sensor technology
that has already been introduced, the measuring technique Packaging
wishes in respect of the sensors are passed Product development
directly to product development or the Vehicle test, trial System definition
B-sample/C-sample Measurement selection
producing area. If no technology is imme- Sensor target
Production model
diately available, sensor and technology specifications
experts are gradually increasingly in- Product specification Vehicle parameters
cluded in research and advanced develop-
Customer specification
ment. Here, the first laboratory prototypes Product:
can be created using known technologies Vehicle system/ Customer requirements:
Components Vehicle system
and often with the help of external part-

ners. These prototypes can then be passed

to product development for initial tests.
152 Automotive sensors | Features of vehicle sensors

are products of product development. Sensor classification

If recursions are necessary in difficult
developments, there can easily also be a Sensors can be classified and grouped
number of B- or C-sample phases (B1, B2, according to very different points of view.
C1, C2). They can be classified with regard to their
use in the vehicle as follows:

Assignment and application

• Functional sensors (pressure, air-mass
flow) mainly used for open and closed-
loop control assignments
2 Sensor prototyping phases leading to series • Sensors for safety (passenger protec-
tion: airbag, ESP) and security (theft-
Prototype Function/ Manufacture deterrence feature)
phase specifications
• Sensors for vehicle monitoring
Prototype restricted Prototype con- (on-board diagnosis (OBD), fuel-con-
struction without
sumption and wear parameters) and
for driver/passenger information
A restricted Prototype con-
struction without
Characteristic curve type
• Continuous linear curves (Fig. 11a) are
B (possibly full Prototype con-
B1, B2) struction without
used mainly for control assignments
tools (construc- covering a wide measuring range.
tion identical to C) Linear curves are also distinguished by
C (possibly full Prototype con- uncomplicated testing and calibration.
C1, C2) struction with • Continuous nonlinear curves (Fig. 11b)
series production are often used for the closed-loop
tools control of a measured variable across
D full Pilot series, par- a very restricted measuring range
tially manual (e.g. exhaust-gas control to l = 1, vehicle
Series produc- full Automated spring-deflection level). When, for in-
Table 2
tion stance, a constant permissible deviation
relative to the measured value is de-
11 Characteristic curve types manded throughout the complete mea-
suring range (e.g. HFM air-mass meter),
curves which feature both pronounced
a b non-linearity and a special shape
S S (e.g. logarithmic) are at an advantage.
Fig. 11
• Such two-step curves (possibly even
S Output signal
X Measured variable featuring hysteresis, Fig. 11d) are used
for limit-value monitoring in such cases
a Continuous, c d where remedial measures are easy to
linear apply when the limits are reached.
b Continuous,
If remedial measures are more difficult,
then multi-step curves (Fig. 11c) can be
c Discontinuous,
used for an earlier warning.

d Discontinuous,
two-step (with
Automotive sensors | Sensor classification 153

Type of output signal For instance, the signal is bound to be

Sensors may also be distinguished by the discontinuous if it is digital and issued
type of output signals (Fig. 13): in bit-serial form.

Analog signals
• Current/voltage, or a corresponding
• Frequency/period duration
• Pulse duration/pulse duty factor 13 Signal shapes

Discrete output signal

• Two-step (binary coded). a f
• Multi-step, with irregular steps (analog U
coded). U
• Multi-step equidistant (analog or digital
• A further distinction must be made, as is Fig. 13
t a Output signal U,
represented in Figure 12 in a systematic
overview of the determined, i.e. nonran- b Tp
U parameter:
dom (stochastic) signal – as to whether frequency f

the signal is permanently available at U b Output signal U,
the sensor output (continuously) or only Tp
at discrete intervals (discontinuously). t parameter:
pulse duration TP

12 Classification of the determined signals according to the information parameter (IP) with examples

Infinite Finite
Deterministic signal
range range

Multiple signals Digital signals

Two-point signals Three-point signals Remaining Constant

(binary signals) (ternary signals) multiple signals quantific-
ation level

conti- disconti- conti- disconti- conti- disconti- conti- disconti- disconti-

nuous nuous nuous nuous nuous nuous nuous nuous nuous
1 2 3 4 5 6 7 8

1 IP 2 IP 3 IP 4 IP

0 0
Time t Time t Time t Time t

5 IP 6 IP 7 IP 8 IP Clock1 Clock2 Clock3


Time t Time t Time t Time t
154 Automotive sensors | Error types and tolerance requirements

Error types and ing), related to the measuring-range final

tolerance requirements value it becomes the percentage error of
the range.
The deviation of the actual characteristic Assuming a generally desirable linear
curve of a sensor from its reference char- characteristic, the absolute deviation eabs
acteristic curve is designated its error e. can be classified into three categories
It is best stated in relation to the input (Fig. 15):
range y (measured variable) and not to • Zero offset (offset error) ezero

the output range x (output signal): • Gradient deviation (gain error) egain
(3) e = yindicated - ytrue • Linearity deviation elin
yindicated = indicated value for the mea-
sured variable The causes of these errors are predomi-
ytrue = ideal value, setpoint value for nantly found in
the measured variable (is determined • The production scatter of the character-
with a measuring sensor that is more istic curve
accurate than the sensor being exam-
ined by at least 1 class)
15 Subdivision of the total error

The amount of the deviation represents,

as shown in Figure 14, the absolute a y
error eabs (unit as measured variable).
Related to the true measured value ytrue,
yi ezero
this becomes the relative error (% of read-

14 Characteristic curves and error graph for a sensor ys
a y eabs x0 x
b y
yi egain (x)
Fig. 14 er

a Actual and yi
c ha

reference ys


ris Dys

b Error graph
a ys
y Measured variable nce
e Dx
x Output signal Re Dx
Dx Measuring range
x0 x
e Error (deviation) x0 x c y
elin (x)
b eabs
eabs = yi – ys
Fig. 15 yi
a Zero error eabs
b Gradient error
raph Dys
c Linearity error rg
E ys
y Measured variable


x Output signal
Dx Measuring range x0 x x0 x
e Error
Automotive sensors | 3FMJBCJMJUZ 155

• The temperature sensitivity of the char- Reliability

acteristic curve and
• The production scatter of the tempera- Failure rate
ture sensitivity The operational reliability of a sensor is
a purely statistical quantity and, as with
The deviations referred to are exclusively any component, is characterized by its
systematic or terministic errors which, by failure rate l, which is stated in 1/h, %/h
contrast with random (stochastic) errors, or ppm/h. Here l is determined with a
are well defined, foreseeable and, as a mat- large number of parts. If it is wished to
ter of principle, correctable, and are also determine approximately the failure rate
for the most part more or less precisely with a number N (< 40) of sensors that is
corrected. not too great, it is possible to observe the
The non-correctable, stochastic errors failure behavior of this random sample
include, for example, under operating conditions until, after a
• Drift (low and higher frequency back- finite time, all the parts have failed. If this
ground noise) and observation is begun at time t0 and if the
• The effects of aging inventory of parts remaining at a later
point ti is identified as B(ti), then a good
In the specification of a sensor, the total approximation for the failure rate l is the
error in the new state and after aging is failure quota q:
prescribed in the product specifications
B(ti) – B(ti+1)
by means of a tolerance graph (Fig. 16). (6) q(Dti, ti) = –––––––––––––––––– Failure quota
Dti · B(ti)
Sometimes, however the permissible indi-
vidual error shares such as the offset, where Dti = ti+1 - ti
lead and linear deviations are additionally
specified. ti are the times at which individual parts
or several parts fail (Fig. 17). The ratio of
Strict metrology teaching states that, the current inventory of parts to the initial
in the case of systematic errors, the sum inventory of parts is known as the relative
of the amounts of the individual errors inventory BR:
must be assumed as the total error (these
may add up in the worst case). In the case
of stochastic errors, statistical addition,
which calculates the total error as the root 16 Tolerance graph for a sensor

of the sum of squares of the individual er-

rors, is permitted. As statistical addition K
leads to a smaller total error, it is, however, 4
often also applied to the systematic errors 3
Class B
in a less strict configuration: 2
n 1
(4) etotal = o|e | i Sum of n systematic 0
Class A

1 errors –1


(5) etotal = i
2 Sum of n stochastic
1 errors
–4 Fig. 16
(statistical addition) Tolerance graph

–200 0 650 850 C illustrated on the
Temperature T example of a resistance
temperature sensor
156 Automotive sensors | Reliability

Furthermore, it is absolutely necessary to

(7) BR(ti, t0) = –––––––– specify under what operating conditions
the failure rate defined in this way is to
With a very large number of parts (N → ∞), be understood. Particularly in the case
the survival probability R(t) at the, here of electrical components such as sensors,
now continuously variable, time t corre- for example, it is necessary to distinguish
sponds to this. The failure rate is calcu- between true, active operating time
lated here for a large number (in practice (switched on state) and service life in the
N ≈ 2,000) of sensors as a percentage varia- sense of pure storage time. Any statement
tion on the survival probability R per unit of a failure rate without this additional in-
of time dt as: formation is worthless.
Failure rates are normally determined
1 dR
(8) l(t) = — –––––– · –––––– Failure rate using time-acceleration methods. The ac-
R(t) dt
celeration factors are achieved here by ex-
posing the sensors to enhanced operating
Reliability is the inverse value of the conditions. A high degree of experience is
failure rate: required in order to apply time accelera-
tion methods that truly reflect reality.
(9) z = –––––– Reliability
The concept of the mean service life TM is
also used to characterize the reliability of
The definition of the failure rate l a sensor. A good approximation of this can
demands a failure criterion: be calculated from a random sample from
• Complete failure the sum of the individual service lives Ti:
• Partial failure N
• Sudden failure (sudden change in a
(10) TM ≈ ––––– · oT1
i or

• Drift failure (gradual change in a ∞

feature) (11) TM = eR(t) dt
for a very large
number of parts

17 Inventory and relative inventory of a random 18 Distribution of the failure rate l(t) over time

20 DBi = Bi – Bi+1 Ai = 1– BR (ti) 100 Early Random Ageing

18 % failures failures and
Bi Cumulative
16 Bi+1 failure 80 wear
frequency effects
Relative inventory
BR (ti) = B(ti)/ B(t0)

12 60
Inventory B(t)

Failure rate

8 40
6 Increasing
20 reliability
Fig. 17
0 0
Observation for a
0 2,000 4,000 6,000 8,000 10,000 h
random sample of

Dti Service life

N = 20 sensors;
ti ti+1
mean service Time
life TM = 4,965 h.
Automotive sensors | Reliability 157

The failure rate of a product against time chemical and electrical ambient influences
shows a typical bathtub pattern (Fig. 18). must be provided (packaging, passivation).
At the start, the failure rate is somewhat On the other hand, it is costly to create op-
high because of early failures, it then erational reliability by testing alone, i.e. to
passes through a longer, relatively low, exclude early failures through premature
horizontal range, then to rise again drasti- aging.
cally towards the end of the service life It is sensible to provide monitoring and
(aging and wear effects). In the case of sen- diagnostic capabilities for sensors in com-
sors for which high operational reliability plex systems (e.g. signal range check, etc.),
is demanded, an attempt is made to avoid so that any failures occurring may be de-
the elevated failure rate at the start by tected in good time. In an emergency, a
sorting out the early failures by pre-aging, sensor’s function can here be temporarily
e.g. by conditioning at higher tempera- replaced by other measured variables or
tures (“burn in”). Early failures are basi- sensible fixed values (limp-home mode,
cally nothing other than manufacturing back up). A sensor-less, often purely me-
defects that have not been recognized. chanical, limp-home mode can be pro-
Table 3 gives some examples of permis- vided. For instance, should the accelera-
sible failure rates l applicable for the tor-pedal sensor failure in a diesel vehicle,
vehicle. The ppm values given relate to the vehicle could also be driven slowly
a period of 10 years, or alternatively to back for repair under regulated idle speed
150,000 km covered, if this is completed alone (limp home).
in a shorter time. Where the sensors are Where reliability must be guaranteed
given a flat rate of <10 ppm, it means that with a probability bordering on certainty
in 10 years, only fewer than 10 of 1 million (e.g. for sensors in electronic braking and
sensors may fail. This value is, however, steering systems), redundancy, i.e. provi-
substantially lower for sensors in passen- sion of multiple parts, is generally the
ger protection systems. means used. Dual redundancy of a sensor
of the same type here only permits failure
Measures for increasing reliability detection for the case of widely different
The best method for ensuring high quality indications, whereas triple redundancy
is to engineer and design in reliability. with 2 from 3 analysis additionally still
This means that as soon as the sensor is supplies a correct measured value. Here,
designed, correspondingly durable mate- however, care must be taken to provide not
rials must be selected and solid protective only redundancy of the sensors, but other
measures against anticipated mechanical, essential parts such as the power supply,
signal evaluation and transmission means
must also be correspondingly redundant,
3 Reliability requirements on vehicle systems as otherwise the probability of simultane-
Warranty target: 150,000 km/10 years ous failure increases. It is often also advis-
→ ECU failure rate (field) < 50 ppm able to provide sensor redundancy in dif-
→ ECU failure rate (0 km) < 15 ppm ferent technologies.
→ Failure rate of modules < 10 ppm
and sensors
→ ASIC failure rate < 3 ppm
→ IC failure rate << 1 ppm
→ Failure rate of discrete < 0.5 ppm
→ For comparison ~ 5,000 ppm
Mobile phone Table 3
158 Automotive sensors | Main requirements, trends

Main requirements, trends sensors simultaneously. For example,

semiconductor sensors are manufactured
The vehicle sensors customized for the re- using “batch processing” in which 100 to
quirements of special electronic systems 1,000 sensors are manufactured simulta-
in the vehicle, unlike the universal sensors neously on a single silicon wafer. On the
conventionally found in the market, are other hand, such manufacturing equip-
subject to five strict requirements (Fig. 19) ment is only an economic proposition
which must be fulfilled by development when correspondingly large numbers of
and to which the most important develop- sensors are produced. These quantities
ment trends also correspond. sometimes exceed an automotive-industry
supplier’s own in-house requirements,
Low manufacturing costs and can typically often be between 1 and
Electronic systems in modern vehicles eas- 10 million per year. Here, the high num-
ily contain up to 150 sensors. Compared bers of sensors needed by the automotive
to other sectors of sensor application, industry has played an unprecedented and
this abundance forces a radical reduction revolutionary role, and set completely new
in costs. The target costs – typically in the standards.
range of € 1 to 30 – are here often less than
one-hundredth of those for conventional High reliability
sensors of the same capacity. Naturally, In accordance with their assignments,
particularly with the introduction of a new vehicle sensors are subdivided into the
technique or technology, the costs are gen- following reliability classes, given in
erally on a falling learning curve, starting descending order of severity:
from a higher level. • Steering, brake, passenger protection
• Engine/drivetrain, chassis/tires

Development trends • Comfort and convenience, diagnosis,

Automated manufacturing processes information, and theft deterrence

(Fig. 20) are largely used, working to a high
yield. This means that each process step
is always executed for a large number of

19 Principal requirements on vehicle sensors 20 Relationship of costs, markets, technologies

Vehicle sensor Requirement

Rational mass
Low costs
Sensor costs per unit

production 100
Robust, proven High reliability 10
Appropriate pack- Extremely severe
aging technology operating conditions 0.1
Production 10 100 1,000 100,000 10 million
Appropriate (units) 10,000 1 million 100 million
Low space per year
requirement Manufacturing manual semi-auto- fully automated mono-
technologies work mated discrete hybrid lithic

On-site error

High-level accuracy Market Traditional markets New markets

compensation classification
Automotive sensors | Main requirements, trends 159

The requirements on the highest class Severe operating conditions

here easily match the high reliability val- Like practically no other breed, vehicle
ues familiar from aerospace applications. sensors are exposed to extreme stresses
To some extent, they demand similar mea- because of where they are fitted, and must
sures, such as the use of the best materials, withstand all kinds of aggression:
redundant componentry, self-monitoring, • Mechanical (vibration, shock)

(short-term) backup power supply, multi- • Climatic (temperature, moisture)

ple programming of critical decision-mak- • Chemical (e.g. spray, salt mist, fuel,

ing algorithms. engine oil, electrolyte)

• Electromagnetic (incident radiation,

Development trends line-conducted interference, excess

Appropriate design measures guarantee voltages, polarity reversal)
built-in reliability. For instance, this neces-
sitates the use of reliable, top-quality com- It is this trend of using sensors immedi-
ponents and materials, coupled with rug- ately at the site where the measurement
ged and well-proven techniques and engi- is taken in order to exploit the benefits
neering. And furthermore, an effort is associated with this which itself leads
made to achieve a consistent integration to the requirements being considerably
of the systems to avoid connection points cranked up.
which may be broken and are at risk of Figure 21 illustrates this problem on
failure. This is possible, for instance, a sensor which might be a temperature
with radio-scanned sensors based on the sensor, speed sensor, flow-rate sensor
antenna-coupled SAW (Surface Acoustic or concentration sensor, for instance.
Wave) elements which do without wiring The sensor cannot in the slightest always
completely. If necessary, redundant sensor be surrounded by a hermetic protective
systems are fitted. sleeve. Although this sleeve could offer
rough protection for the measurement of
temperature, flow rates and concentration,
it must permit more or less direct contact
between the sensor and the generally very
aggressive monitored medium (an excep-
tion to this is the inertia sensor). Some-
times, thin, but very resistant, passivation
21 Technologies accompanying sensors layers are permissible on the sensor.
The form of the plug-in sensor demands
Monitored External environment a permanently tight seat of the mount in
5 6 7 the associated wall (inner packaging, outer
packaging). The sensor may be connected
Fig. 21
to the electronic control units via a fixed
1 Sensor
housing plug or also, as illustrated, via a 2 Protective sleeve
1 8 plug flexibly fastened with a length of ca- (coating)
ble (e.g. ABS wheel-speed sensor). Here, 3 Seal
too, there are three critical electrical con- 4 Mount
5 Seal, mounting
nection points that absolutely must be pro-
6 Data point
tected against conductive fluids and corro-
7 Seal, strain relief
sion: the inner connection on the sensor 8 Insulation (flexible)

element, the connection of the cable tail 9 Contacts

2 3 4 9 10 11
and finally the connection of the partially 10 Plug housing
shielded cable to the external plug connec- 11 Plug contact
160 Automotive sensors | Main requirements, trends

tion. If the seal is inadequate, it is only a Fiber-optic sensors in which the light
question of time until corrosive liquids guided in the optical fibers (glass, plastic)
penetrate from the external connector can be modified as a function of the mea-
into the internal sensor terminal. sured variable are particularly immune to
The plug-in connection itself must be electromagnetic interference. This is true
sufficiently tight overall, that no shunts up to the point at which the optical signals
form at the sensor output. The cable itself are converted back into electrical signals.
must retain its flexibility and tightness If these are to be used in the future, there
after many years of operation under the will have to be some development work
most adverse conditions. the provision of low-priced measuring
Separable plug-in connections in the elements and the accompanying technolo-
vehicle unfortunately still represent one gies. There are, for example, some very
of the most frequent causes of failure. interesting applications for this in the
Wireless signal connections (e.g. infrared field of force measurement, e.g. in provid-
light or radio) could break down this prob- ing finger protection for electric power
lem in the longer view, especially if the windows and sliding sunroofs (Fig. 22).
sensors were even to have a wireless Sensors of this type have also been tested
energy supply (autonomous sensors). very successfully as very early responding,
The costs of these indispensable tech- distributed sensors in the door and frontal
nologies complementary to the sensors area of the vehicle for triggering passenger
often exceed those of the actual sensor and pedestrian protective systems.
element many times over. They constitute
the actual value of an vehicle sensor, not
only in terms of cost, but also in terms of
function. 22 Microbending effect

Development trends Microbending effect

Protective measures against the stresses F=0 F>0
Damping ≈ 0 Damping > 0
referred to demand a very high degree of
specific know-how in the field of sensor
• Passivation and connecting techniques
• Sealing and joining techniques
• EMC protective measures Door frame
• Low-vibration installation
• Service life test and simulation methods

• The use of resistant materials, etc.

Sealing rubber
Furthermore, detailed knowledge of the
loading to which the sensor will be sub- Example of development: fiber-optic fin-
jected at the particular installation point ger protection in power windows based
on the microbending effect:
is required. It is often forgotten that the
The light in an optical fiber is attenu-
quality of a sensor stands and falls with ated proportionally to the corrugated
total competence with these protective distortion on the application of a force F
measures. transverse to the fiber (large measur-
ing effect), independently of where the
force is applied or whether it is acting
at a point or is distributed (distributed
Automotive sensors | Main requirements, trends 161

Low-volume design Micromechanical manufacturing refers

On the one side, the number of electronic on the one hand to dimensions in the µm-
systems in the vehicle continues to climb range and tolerances in the sub-µ-range
steadily. On the other, today’s vehicles are which cannot be achieved with conven-
becoming more and more compact. These tional machining methods. On the other
facts, together with the need to retain the hand, sensors which may have dimensions
high level of passenger-compartment com- in the mm-range, but which are manufac-
fort forces development to concentrate on tured with micromechanical methods, are
an extreme space-saving designs. Further- also considered micromechanical (Fig. 24).
more, the increasing demand for further We will discuss only the most well-known
improvements in fuel economy mean that and important method here, anisotropic
minimization of the vehicle’s weight is of etching of silicon. This is the most impor-
prime importance. tant method, because silicon is manufac-
tured to high precision and in large num-
Development trends bers at low cost, and is the most thor-
Widespread use is made of the familiar oughly researched and best known
technologies applied in circuit engineering material. In addition, it offers the capabil-
for the m