New methods to engineer and

seamlessly reconfigure time triggered

Ethernet based systems during runtime
based on the PROFINET IRT example
1st Edition Lukasz Wisniewski
Visit to download the full and correct content document:
https://textbookfull.com/product/new-methods-to-engineer-and-seamlessly-reconfigur
e-time-triggered-ethernet-based-systems-during-runtime-based-on-the-profinet-irt-exa
mple-1st-edition-lukasz-wisniewski/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Back Trouble A New Approach to Prevention and Recovery

based on the Alexander Technique Deborah Caplan

https://textbookfull.com/product/back-trouble-a-new-approach-to-
prevention-and-recovery-based-on-the-alexander-technique-deborah-
caplan/

Practical System Programming with C: Pragmatic Example

Applications in Linux and Unix-Based Operating Systems
1st Edition Sri Manikanta Palakollu

https://textbookfull.com/product/practical-system-programming-
with-c-pragmatic-example-applications-in-linux-and-unix-based-
operating-systems-1st-edition-sri-manikanta-palakollu/

Practical System Programming with C - Pragmatic Example

Applications in Linux and Unix-Based Operating Systems.
1st Edition Sri Manikanta Palakollu

https://textbookfull.com/product/practical-system-programming-
with-c-pragmatic-example-applications-in-linux-and-unix-based-
operating-systems-1st-edition-sri-manikanta-palakollu-2/

Multivariate Techniques An Example Based Approach 1st

Edition Saman Hanif Shahbaz Muhammad Hanif

https://textbookfull.com/product/multivariate-techniques-an-
example-based-approach-1st-edition-saman-hanif-shahbaz-muhammad-
hanif/
Biometric Identification Technologies Based on Modern
Data Mining Methods Stepan Bilan

https://textbookfull.com/product/biometric-identification-
technologies-based-on-modern-data-mining-methods-stepan-bilan/

Cases on Research Based Teaching Methods in Science

Education 1st Edition Eugene De Silva

https://textbookfull.com/product/cases-on-research-based-
teaching-methods-in-science-education-1st-edition-eugene-de-
silva/

Location-Based Social Media: Space, Time and Identity

1st Edition Leighton Evans

https://textbookfull.com/product/location-based-social-media-
space-time-and-identity-1st-edition-leighton-evans/

Mastering Kafka Streams and ksqlDB: Building Real-Time

Data Systems by Example 1st Edition Mitch Seymour

https://textbookfull.com/product/mastering-kafka-streams-and-
ksqldb-building-real-time-data-systems-by-example-1st-edition-
mitch-seymour/

Wearable Computing From Modeling to Implementation of

Wearable Systems based on Body Sensor Networks 1st
Edition Giancarlo Fortino

https://textbookfull.com/product/wearable-computing-from-
modeling-to-implementation-of-wearable-systems-based-on-body-
sensor-networks-1st-edition-giancarlo-fortino/
Technologien für die intelligente Automation
Technologies for Intelligent Automation

Lukasz Wisniewski

New methods to engineer

and seamlessly reconfigure
time triggered Ethernet based
systems during runtime based
on the PROFINET IRT example
Technologien für die
intelligente Automation
Technologies for Intelligent Automation
Band 6

Weitere Bände in dieser Reihe

http://www.springer.com/series/13886
Ziel der Buchreihe ist die Publikation neuer Ansätze in der Automation auf wissenschaft­
lichem Niveau, Themen, die heute und in Zukunft entscheidend sind, für die deutsche und
internationale Industrie und Forschung. Initiativen wie Industrie 4.0, Industrial Internet
oder Cyber-physical Systems machen dies deutlich. Die Anwendbarkeit und der indust­
rielle Nutzen als durchgehendes Leitmotiv der Veröffentlichungen stehen dabei im Vorder-
­grund. Durch diese Verankerung in der Praxis wird sowohl die Verständlichkeit als auch
die Relevanz der Beiträge für die Industrie und für die angewandte Forschung gesichert.
Diese Buchreihe möchte Lesern eine Orientierung für die neuen Technologien und deren
Anwendungen geben und so zur erfolgreichen Umsetzung der Initiativen beitragen.

Herausgegeben von
inIT – Institut für industrielle Informationstechnik
Hochschule Ostwestfalen-Lippe
Lemgo, Germany
Lukasz Wisniewski

New methods to engineer

and seamlessly reconfigure
time triggered Ethernet based
systems during runtime based
on the PROFINET IRT example
Lukasz Wisniewski
Lemgo, Germany

Dissertation, Otto-von-Guericke-Universität Magdeburg, 2016

Technologien für die intelligente Automation

ISBN 978-3-662-54649-9 ISBN 978-3-662-54650-5 (eBook)
DOI 10.1007/978-3-662-54650-5

Library of Congress Control Number: 2017935954

Springer Vieweg
© Springer-Verlag GmbH Germany 2017
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the
material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage
and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or
hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, 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.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are
believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors
give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions
that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.

Printed on acid-free paper

This Springer Vieweg imprint is published by Springer Nature

The registered company is Springer-Verlag GmbH Germany
The registered company address is: Heidelberger Platz 3, 14197 Berlin, Germany
Kurzfassung

Der Einsatz Ethernet-basierter Kommunikationstechnologien in industriellen Umge-

bungen ist heutzutage Realität. Sie ermöglichen den Aufbau vernetzter Steuerungssys-
teme (Networked Controls Systems, NCS), die sehr leistungsfähig in Bezug auf Echtzeit
und Zuverlässigkeit sind. Allerdings erfordern diese Systeme häufig einen hohen Pla-
nungs- und Inbetriebnahmeaufwand, der ihre Flexibilität bei notwendigen Änderun-
gen einschränkt. Der Markt für diese Systeme ist auf Grund einer hohen Produktsätti-
gung unter den Herstellern eng umkämpft. Um wettbewerbsfähig zu bleiben sind die
Hersteller bestrebt, kundenspezifische Produkte in guter Qualität und zu akzeptablen
Preisen anzubieten. Dies hat eine Reduzierung der Losgröße bis zur Losgröße eins zur
Folge. Darin besteht eine Herausforderung an Produktionssysteme und ebenso an NCS,
die einen bedeutenden Bestandteil der Produktionssysteme darstellen.
Gegenstand dieser Dissertation ist die Gestaltung eines Konzepts zur Erhöhung der
Flexibilität aktuell verfügbarer zeitgesteuerter Ethernet-basierter (time triggered eth-
ernet based, TTEB) Systeme, ohne deren Leistungsfähigkeit und Robustheit zu beein-
trächtigen. Die hauptsächlichen Herausforderungen beziehen sich dabei unter anderem
auf die Planung der zeitgesteuerten Kommunikation, die einen beträchtlichen Zeitraum
und die Ausführung auf einer leistungsstarken Plattfom erfordern kann. Zusätzlich ist
die Zuverlässigkeit zu betrachten und auf dem erforderlichen Niveau zu halten. Ebenso
muss die Rekonfiguration ausgeführt werden optimal ohne das laufende System zu
beeinträchtigen.
Um all diese Anforderungen zu erfüllen wird ein aus drei Komponenten bestehen-
der Lösungsansatz vorgeschlagen. Die erste Komponente beinhaltet den Planungsal-
gorithmus. Dieser Algorithmus erlaubt es, innerhalb einer kurzen und vorhersagbaren
Zeit einen allgemeingültigen und in Bezug auf Netzwerktopologien, welche zu Baum
Graphen gehören, optimalen Zeitplan zu berechnen. Im Fall von Ringtopologien ist
das berechnete Ergebnis nicht optimal, obgleich es kommerziell vefügbare Planungs-
werkzeuge übertrifft. Die zweite Komponente befasst sich mit der Zuverlässigkeit von
Netzwerken und mit Redundanzkonzepten. Einen Beitrag dazu liefert ein Ansatz, dy-
namische Redundanzkonzepte durch die Senkung der Zeit zur Erkennung von Netzw-
erkfehlern (time of network failure detection, TD ) auf bis zu 1ms zu verbessern. Da-
raus ergeben sich Potenziale für die Anwendung dynamischer Redundanzkonzepte für
TTEB Systeme. Ein weiterer Beitrag besteht in einem Algorithmus zur Suche möglichst
disjunkter Pfade, der in statischen Redundanzkonzepten zur Gestaltung redundanter
Kommunikation und zur Minimierung singulärer Fehlerquellen (single point of failure,
SPOF) anwendbar ist. Die dritte Komponente enthält ein Konzept zum Erfassen aller
durch einen Planungsalgorithmus benötigten Eingangsparameter auf der Feldebene,
VI Kurzfassung

ohne die Notwendigkeit der Einbeziehung eines Planungs- und Inbetriebnahmewerkzeugs.

Sie unterstützt auch das anschließende Ausrollen und Aktivieren eines Kommunika-
tionsplans ohne jegliche Beeinflussung einer laufenden Echtzeitkommunikation.
Abstract

Deployment of Ethernet based communication technologies in industrial environments

is a fact. They allow to build Networked Controls Systems (NCS) able to achieve high
performance in terms of real-time and reliability. However, often require time consum-
ing engineering process, which limits their flexibility whenever changes are required.
Due to the current market turbulences and high saturation of products, a fierce compe-
tition between manufacturers can be observed. In order to remain competitive, manu-
facturers are pushed to provide customers individual products, while keeping price and
quality at a reasonable level. This causes that the lot sizes have to be decreased and can
go even down to one. It is a big challenge for production systems and also NCS that are
an important part of it.
The objective of this dissertation is to design a concept that would allow to increase
the flexibility of currently available TTEB systems, however, without affecting their per-
formance and robustness. The main challenges are related to scheduling of time trig-
gered communication that may take significant amount of time and has to be per-
formed on a powerful platform. Additionally, the reliability has to be considered and
kept on the required high level. Finally, the reconfiguration has to be optimally done
without affecting the currently running system.
In order satisfy all these requirements an approach is proposed that consist of three
main components. The first component includes a scheduling algorithm, which al-
lows to calculate a valid and in case of network topologies that belong three graphs
an optimal schedule in a short and deterministic time. In case of ring topologies, the
scheduling result is not optimal, even though, it outperforms the commercially avail-
able scheduling framework. The second component deals with the network reliability
and redundancy methods. One contribution is an approach to improve dynamic re-
dundancy methods by decreasing the time of network failure detection (TD ) down to
1ms. This makes it worth to re-think, if dynamic redundancy methods can be applied
in TTEB systems. The second contribution is a possibly disjoint path search algorithm
that can be used by static redundancy methods to organise redundant communication
and minimise single points of failure (SPOF). The last component contains an approach
that allows to collect all required by scheduler input parameters at the field level, with-
out the need to use an engineering tool. Afterwards, distribute a new schedule and
activate it, without affecting the running real-time communication in any case.
Acknowledgement

I would like to thank all the people who in some way contributed and helped me to
finish this thesis.
First of all, I would like to thank Professor Christian Diedrich, for accepting me into
his group and giving me a freedom during my Ph.D research. I’m very thankful for his
guidance and critical feedback to my work. It was really an honour to work under his
supervision.
My special and heartily thanks to Professor Jürgen Jasperneite for giving me the op-
portunity to join the Institute Industrial IT and pursuit my work towards Ph.D. I would
like to thank him for his substantial as well as mental support during the whole du-
ration of my thesis. I’m grateful for the superb environment and the opportunity to
attend many conferences around the world, where I could share and discuss results of
my research.
I would like to also thank Professor Thilo Sauter who kindly agreed to serve as third
reviewer of my thesis.
I would like to especially thank to my colleagues and also ex-colleagues of the Insti-
tute Industrial IT and Fraunhofer IOSB-INA in Lemgo for their countless helpful discus-
sions and many inspirations that helped me in completing this work. Moreover, many
thanks to Henning Trsek, Shiraz Gilani, Tim Tack, Uwe Mönks, Verena Wendt, Arne Neu-
mann, Dominik Henneke, Lars Dürkop, Sergej Gamper, Carsten Pieper, Alexander Bien-
darra, Sebastian Schriegel and Holger Flatt for all discussions and the help in improving
the quality of this work. Special thanks goes to Markus Schumacher for his continuous
support and help in validation of many ideas. Your support was greatly appreciated.
Lastly, I would like to thank my family for all their support and encouragement during
this exciting but also sometimes difficult time. I want to thank my wife Gosia for taking
care of everything, especially our two sons Alan and Ian and allowing me to focus on my
Ph.D work. I greatly appreciate your faithful support and patience, especially during the
final phase of this thesis. Thank you!
Contents

1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Outline of the document . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 Fundamentals 7
2.1 Production Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.1 Automation Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.2 Networked Control System . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 NCS Communication Requirements . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.1 Timeliness in NCS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Data Collection and Data Distribution . . . . . . . . . . . . . . . . . 13
NCS System Reaction Time . . . . . . . . . . . . . . . . . . . . . . . . 15
Deterministic Communication . . . . . . . . . . . . . . . . . . . . . . 16
2.2.2 Communication Reliability in NCS . . . . . . . . . . . . . . . . . . . . 19
2.2.3 Industrial Communication Requirements - Summary . . . . . . . . 21
2.3 Industrial Network Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.1 Industrial Ethernet based Protocols - Real-Time . . . . . . . . . . . . 23
Class 3 – Polling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Class 3 – Token Passing . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Class 3 – TDMA Based . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3.2 Industrial Ethernet based Protocols - Reliability . . . . . . . . . . . . 25
Physical Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Data Link Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Application Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.4 Engineering of Industrial Network Systems . . . . . . . . . . . . . . . . . . . 28
2.5 Re-engineering of Industrial Network Systems . . . . . . . . . . . . . . . . . 30
2.6 Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.7 Solution Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

3 Timeliness in Ethernet Based Network Control Systems 37

3.1 Network Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.2 Ethernet Standard . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.3 Ethernet Standard Real-Time Capabilities . . . . . . . . . . . . . . . . . . . 42
XII Contents

3.4 Ethernet Based Time-Triggered Protocols . . . . . . . . . . . . . . . . . . . . 43

3.4.1 IEEE 802.1 TSN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Advanced Queueing, Forwarding and Resource Reservation Meth-
ods in IEEE 802.1 TSN . . . . . . . . . . . . . . . . . . . . . . 45
Time Triggered Communication Scheduling in IEEE 802.1 TSN . . . 48
Common Notion of Time in IEEE 802.1AS - TSN . . . . . . . . . . . . 49
Summary of TSN Real-Time Capabilities . . . . . . . . . . . . . . . . 51
3.4.2 Time Triggered Ethernet (TTEthernet) . . . . . . . . . . . . . . . . . 51
Advanced Queueing, Forwarding and Resource Reservation Meth-
ods in TTEthernet . . . . . . . . . . . . . . . . . . . . . . . . 51
Time Triggered Communication Scheduling in TTEthernet . . . . . 52
Common Notion of Time in TTEthernet . . . . . . . . . . . . . . . . 52
Summary of TTEthernet Real-Time Capabilities . . . . . . . . . . . . 53
3.4.3 PROFINET IRT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Advanced Queueing, Forwarding and Resource Reservation Meth-
ods in PROFINET IRT . . . . . . . . . . . . . . . . . . . . . . 54
Time Triggered Communication Scheduling in PROFINET IRT . . . 56
Common Notion of Time in PROFINET IRT . . . . . . . . . . . . . . 57
Summary of PROFINET IRT Real-Time Capabilities . . . . . . . . . . 59
3.4.4 EtherCAT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Advanced Queueing, Forwarding and Resource Reservation Meth-
ods in EtherCAT . . . . . . . . . . . . . . . . . . . . . . . . . 59
Time Triggered Communication Scheduling in EtherCAT . . . . . . 61
Common Notion of Time in EtherCAT . . . . . . . . . . . . . . . . . . 61
Summary of EtherCAT Real-Time Capabilities . . . . . . . . . . . . . 62
3.4.5 Summary of Ethernet Based Time Triggered Protocols . . . . . . . . 62
3.5 Communication Scheduling in TTEB Systems . . . . . . . . . . . . . . . . . 63
3.6 Complexity of the Scheduling Problem in TTEB-Systems . . . . . . . . . . . 65
3.7 Scheduling Problem Formalisation and Modelling . . . . . . . . . . . . . . 66
3.7.1 Release Time (r i ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.7.2 Start Time (s i ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.7.3 Processing Time (p i ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.7.4 End-to-End Time (e i ) . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.7.5 Task Deadline (d i ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.7.6 Completion Time C i . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.7.7 Scheduling Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . 72
No Preemption Constraint . . . . . . . . . . . . . . . . . . . . . . . . 74
Precedence Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
3.8 Scheduling Using Mixed Integer Linear Programming (MILP) . . . . . . . . 75
3.8.1 The Disjunctive TTEB Scheduling Model . . . . . . . . . . . . . . . . 75
3.8.2 MILP Scheduling Results . . . . . . . . . . . . . . . . . . . . . . . . . 77
Results for Line, Comb and Tree Topology . . . . . . . . . . . . . . . 77
Results for Ring Topology . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.9 Proposed Scheduling Approach . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.9.1 Slip-Stream Effect - Working Principle . . . . . . . . . . . . . . . . . . 80
Contents XIII

3.9.2 Theoretical Analysis of the Slip-Stream Scheduling - Line Topology 80

Downlink Scheduling . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Uplink Scheduling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3.9.3 Slip-Stream in Tree Topology . . . . . . . . . . . . . . . . . . . . . . . 83
3.9.4 Theoretical Analysis of the extended slip-stream effect - Tree Graphs 86
Downlink Scheduling . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
Uplink Scheduling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
3.9.5 Slip-stream in Ring topology . . . . . . . . . . . . . . . . . . . . . . . 89
Adaptation of the Slip-Stream Effect for Redundant Topologies . . . 90
Slip-Stream Effect - Solving Disjunctions . . . . . . . . . . . . . . . . 91
Linear Program - Calculating FSOs . . . . . . . . . . . . . . . . . . . . 94
3.9.6 Theoretical Analysis of the proposed approach - Graphs with Cy-
cles (Rings) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.9.7 Slip-stream in Topology, Where Multiple PLC Ports are Involved . . 97
3.10 Clock Synchronisation Accuracy and Scheduling . . . . . . . . . . . . . . . 97
3.10.1 Accuracy Determination Methods . . . . . . . . . . . . . . . . . . . . 98
Accuracy through Measurements . . . . . . . . . . . . . . . . . . . . 98
Accuracy through Network Modelling and Simulation . . . . . . . . 99
Accuracy through Analytical Calculation and Approximation . . . . 99
Accuracy through Tests, Certification and Classification . . . . . . . 99
Accuracy through Online Worst Case Distributed Calculation . . . . 100
Accuracy through System Integrated Measurement Capabilities . . 100
3.10.2 Benchmark of the Synchronisation Accuracy Determination Meth-
ods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3.10.3 Considering Synchronisation Accuracy in the Schedule . . . . . . . 102
3.11 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

4 Communication Reliability 105

4.1 Redundancy methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
4.2 Dynamic Redundancy Methods . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.2.1 An Approach to Improve Dynamic Redundancy Protocols . . . . . . 108
Decreasing the Link-Loss Propagation Delay . . . . . . . . . . . . . . 108
Decreasing the Link-Loss Detection Time . . . . . . . . . . . . . . . 109
Decreasing the Port State Change Delay . . . . . . . . . . . . . . . . . 110
4.2.2 Theoretical Analysis of the Failure Detection Time . . . . . . . . . . 110
4.2.3 Dynamic Redundancy and Scheduling . . . . . . . . . . . . . . . . . 111
4.3 Static Redundancy Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.3.1 Path Search Aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
4.3.2 2-Disjoint Paths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
4.3.3 Possibly Disjoint Path Search Method . . . . . . . . . . . . . . . . . . 117
4.3.4 Theoretical Analysis of the Possibly Disjoint Path Search Approach 122
4.3.5 Static Redundancy and Scheduling . . . . . . . . . . . . . . . . . . . 123
4.4 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
XIV Contents

5 Seamless Reconfiguration of the Industrial Network System 125

5.1 Reconfiguration of the TTEB Industrial Network System . . . . . . . . . . . 125
5.2 Identification of a INS Change - Scheduler Input . . . . . . . . . . . . . . . 127
5.2.1 Topology Discovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
5.2.2 Device Characteristics Discovery . . . . . . . . . . . . . . . . . . . . . 130
5.2.3 Traffic Characteristics Discovery . . . . . . . . . . . . . . . . . . . . . 132
5.2.4 Discovery Mechanisms Summary . . . . . . . . . . . . . . . . . . . . 134
5.3 Application of a New Schedule . . . . . . . . . . . . . . . . . . . . . . . . . . 135
5.3.1 Distribution of a New Schedule . . . . . . . . . . . . . . . . . . . . . . 135
5.3.2 Synchronisation of a Schedule . . . . . . . . . . . . . . . . . . . . . . 136
5.3.3 Synchronous Activation of a Schedule . . . . . . . . . . . . . . . . . . 139
Static Real-Time Channel Reservation . . . . . . . . . . . . . . . . . . 139
Real-Time Payload Piggy-backing . . . . . . . . . . . . . . . . . . . . 140
Synchronisation Frame Piggy-backing . . . . . . . . . . . . . . . . . 140
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
5.3.4 Appliance of a New Schedule - Summary . . . . . . . . . . . . . . . . 142
5.4 Proposed Seamless Reconfiguration Approach . . . . . . . . . . . . . . . . . 142
5.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

6 Proposed Approach Evaluation 149

6.1 Scheduling Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
6.1.1 Scheduling Approach Implementation . . . . . . . . . . . . . . . . . 150
XML Input Parser and Output Generator . . . . . . . . . . . . . . . . 150
Path Search Module . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
Visualisation Module . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
6.1.2 Scheduling Approach - Definition of Use-cases and Evaluation . . . 151
Use-case - Line Topology . . . . . . . . . . . . . . . . . . . . . . . . . 152
Use Case - Comb Topology . . . . . . . . . . . . . . . . . . . . . . . . 152
Use Case - Tree Topology . . . . . . . . . . . . . . . . . . . . . . . . . 153
Use Case - Ring Topology . . . . . . . . . . . . . . . . . . . . . . . . . 154
6.1.3 Scheduling Approach Evaluation - Summary . . . . . . . . . . . . . . 155
6.2 Communication Reliability Methods . . . . . . . . . . . . . . . . . . . . . . 156
6.2.1 Dynamic Redundancy Methods . . . . . . . . . . . . . . . . . . . . . 157
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
6.2.2 Static Redundancy Methods . . . . . . . . . . . . . . . . . . . . . . . 159
Disjoint Path Search - definition of Use Cases . . . . . . . . . . . . . 160
Disjoint Path Method - Results . . . . . . . . . . . . . . . . . . . . . . 161
Disjoint Path Search - Summary . . . . . . . . . . . . . . . . . . . . . 161
6.2.3 Communication Reliability Methods - Summary . . . . . . . . . . . 162
6.3 Seamless Reconfiguration Approach . . . . . . . . . . . . . . . . . . . . . . . 162
6.3.1 Distribution of a New Schedule . . . . . . . . . . . . . . . . . . . . . . 163
Simulation Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
Scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
Contents XV

6.3.2 Synchronisation and Activation of a New Schedule . . . . . . . . . . 166

Measurement Setup - Schedule Switching Time . . . . . . . . . . . . 166
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
Measurement Setup - Schedule Switching vs. Running RT Commu-
nication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
6.3.3 Seamless Reconfiguration Approach - Summary . . . . . . . . . . . 169
6.4 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169

7 Conclusions and Outlook 171

7.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
7.2 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173

Bibliography 175

List of Figures 195

List of Tables 199

XVI

Abbreviations
ADAG Almost Directed Acyclic Graph

AP Automation Pyramid

APDU Application Protocol Data Unit

ARP Address Resolution Protocol

AS Automation System

AVB Audio Video Bridging

BE Best Effort

BFS Breadth First Search

BMCA Best Master Clock Algorithm

CAD Computer Aided Design

CBS Credit Based Shaping

CC Communication Cycle

CIM Computer-Integrated Manufacturing

CM Compression Master

CN Controlled Node

CPS Cyber Physical System

CPU Central Processing Unit

CS Control System

DAG Directed Acyclic Graph

DC Distributed Clock

DCP Discovery and Configuration Protocol

DDF Device Description File

DFP Dynamic Frame Packing

DFS Depth First Search

DHCP Dynamic Host Configuration Protocol

DPWS Device Profile for Web Services

 XVII

FCS Frame Check Sequence

FSO Frame Send Offset

GM Grand Master

HMI Human Machine Interface

HSR High-Availability Seamless Redundancy

ICMP Internet Control Message Protocol

IFG Inter Frame Gap

INS Industrial Network System

IOD IO Device

IoT Internet of Things

IRT Isochronous Real Time

IT Information Technology

JSP Job Shop Scheduling

LAN Local Area Network

LLDP Link Layer Discovery Protocol

LOC Lines of Code

LP Linear Program

MAC Medium Access Control

MES Manufacturing Execution System

MII Media Independent Interface

MILP Mixed Integer Linear Program

MN Managing Node

MRM Media Redundancy Master

MRP Multiple Registration Protocol

MRPD Media Redundancy for Planned Duplication

MSRP Multiple Stream Reservation Protocol

NCS Networked Control System

NMS Network Management Station

XVIII

NRP Network Redundancy Protocol

OUI Organisation Unique Identifier

P2P Peer-to-Peer

PLC Programmable Logic Controller

PM Production Module

PRP Parallel Redundancy Protocol

PTCP Precision Transparent Clock Protocol

PTP Precision Time Protocol

QoS Quality of Service

RC Rate-Constraint

RSTP Rapid Spanning Tree Protocol

RT Real-Time

RTP Real Time Phase

SC Synchronisation Client

SCADA Supervisory Control and Acquisition

SDN Software Defined Networking

SFD Start Frame Delimiter

SII Slave Information Interface

SM Synchronisation Master

SNMP Simple Network Management Protocol

SOA Service Oriented Architecture

SPOF Single Point of Failure

SR Stream Reservation

SRP Stream Reservation Protocol

SST Schedule Switching Trigger

STP Spanning Tree Protocol

TC Transparent Clock

TDMA Time Division Multiple Access

 XIX

TLV Type Length Value

TP Technical Process

TRILL Transient Interconnection of Lots of Links

TSN Times Sensitive Networking

TT Time Triggered

TTEB Time Triggered Ethernet Based

URL Uniform Resource Locator

VLAN Virtual Local Area Network

VoIP Voice over IP

VPN Virtual Private Network

WLAN Wireless LAN

WSN Wireless Sensor Network

1 Introduction

The current technical processes in the industry are controlled by the Programmable
Logic Controllers (PLCs), which connected together with Sensors and Actuators via net-
work system form a Networked Control System (NCS)[1]. Depending on the type of
technical process, requirements in terms of sampling rates of controlled processes and
corresponding reaction times may differ [2]. They may require reaction times in the
range of few milliseconds, in case of motion control applications, up to hundreds of
milliseconds or more in case of process automation applications, such as control of
biochemical processes. Although, some specific requirements for the network system
responsible to transit data between PLC and Sensor and Actuator may vary, all of them
require to exchange process data in real-time. Real-time communication is achieved
through introduction of special prioritisation rules and dedicated hardware. In order to
achieve determinism and cycle times that are equal or less than 1ms, the time critical
communication has to be specially organised and isolated from other kinds of traffic.
Such highly deterministic network systems have one additional thing in common that
is limited flexibility. When it comes to changes of such systems, typically an offline
engineering process in a special tool has to be performed. Afterwards, the controlled
process has to be stopped and the communication system is restarted.
In this dissertation a concept is proposed that allows to implement selected engi-
neering functionalities into a field level devices, such as PLC. Thus allowing to perform
reconfiguration of industrial networks systems, which are part of NCSs in a seamless
way.

1.1 Motivation
Several industrial network systems (INSs) supporting exchange of information in real-
time had been developed in the past years. They can be categorised into three main
real-time performance classes, as defined in [3]. The first class supports low real-time
performance, i.e. cycle times around 100ms, which can be achieved with the standard-
ised network equipment. The second class supports medium performance, i.e. cycle
times around 10ms. This performance is typically realised by prioritised real-time traf-
fic delivery, using slightly modified network hardware. In order to achieve the high-
est real-time performance with the cycle times lower than 1ms, independent on the
amount of traffic present in the network system, special communication rules have
to be created (e.g. scheduled communication) and sophisticated hardware is needed.
This performance is achieved by Time Triggered Ethernet based (TTEB) protocols, rep-
resented by technologies, such as PROFINET IRT[4][5], EtherCAT [6][7][8], TTEthernet

© Springer-Verlag GmbH Germany 2017

L. Wisniewski, New methods to engineer and seamlessly reconfigure
time triggered Ethernet based systems during runtime based on the
PROFINET IRT example, Technologien für die intelligente Automation 6,
DOI 10.1007/978-3-662-54650-5_1
2 1.1. Motivation

[9], Powerlink[10], SERCOS III [11] or the newest IEEE 802.1 standard, which is still un-
der development, Times Sensitive Networking (TSN) [12][13][14][15][16]. All of them
are capable of providing high real-time performance, with cycle times going down to
12,5μ as in case of EtherCAT [17]. Apart from the real-time performance, these systems
also provide high level of reliability, which is required in case of critical applications.
This is typically realised by diverse redundancy approaches [18].

The current trends, such as Internet of Things (IoT) [19], vertical integration of the au-
tomation system using technologies such as OPC UA [20], camera based surveillance
of technical processes or camera based localisation [21] will be a big challenge for the
INSs. This is due to the increased amount of traffic produced by the additional net-
work members and their potential influence on the real-time communication. Here,
an important role can be played by TTEB systems. With the help of precisely defined
communication rules, such as Time Division Multiple Access (TDMA) with commu-
nication scheduling, token passing or polling based approaches [22], they allow the
assignment of critical data to dedicated, exclusive time slots, providing deterministic
data exchange. Such communication systems can be used not only to service the most
demanding application classes as the motion control or safety critical applications, but
also give the possibility to deal with large systems, such as automotive production sys-
tems, consisting of thousands of network entities. The possibility to assign a fraction of
bandwidth to selected traffic for providing defined, constant performance at any given
time, independent to the number of users and the current traffic load in the system,
may become a key feature in a crowded network systems.

In order to remain competitive, the future production systems have to react promptly
on the market turbulences, to satisfy diverse customer needs. Currently, flexibility of
production systems gives some angle of possible changes but for a hefty price. The
price is typically related to the retrofitting of production line, which is time consuming
and expensive due to necessity of manual engineering and configuration of machines,
robots, network devices, etc., and results in production stops. As a consequence, there
must always be a minimum amount of workpieces (the minimum number typically de-
pends on the product complexity) that have to be produced to reach the cost efficient
level. The current trend of the production paradigm is the mass customisation, which
suppose to allow the customer to define features of the ordered product (even at the
level of the computer-aided design (CAD)). For the production system, it would mean
that there might be a need for a production with the lot size 1. Therefore, the pro-
duction system should be capable of handling such requirements. The paradigm that
allows customisation of products, while keeping production costs at a reasonable level,
is the agile manufacturing [23], [24], [25]. Figure 1.1 illustrates the relationship between
a customer and an agile manufacturing system as a use case diagram. The possible
interactions can be defined as:

• Customise product – allows to exactly specify the product that the customer is
ordering. The customer is even able to change the shape of the product, e.g. in
order to easily integrate it into some machine.
1.1. Motivation 3

   
    

  

   

    

Figure 1.1: Agile production system - use case diagram

• Order product – allows to order even one piece. Modular structure of the pro-
duction lines together with production scheduling software allow to achieve cost
efficient level even in case of lot size equal to 1.

• Deliver product – there are several ways to perform delivery, but it is typically
done by shipment.

In order to allow realisation of such systems, there are several aspects that should be
tackled at different levels of the production system [26]. Starting from the highest levels,
such as:

• production planning and material ordering at the company level,

• collection and visualisation of plant data of Supervisory Control and Data Acqui-
sition (SCADA) tools at the cell level,

• issues related to engineering of PLCs, sensors and actuators at the field level.

The main focus of this dissertation is the lowest level, where the NCSs are located.

As mentioned before, the biggest advantage of NCSs, based on TTEB technology, is

the fixed high performance in terms of real-time and reliability, which allow to satisfy
the most demanding applications. Due to special rules for handling the time sensi-
tive data, they are also able to deal with real-time performance in large scale systems,
including many network members and high amount of traffic. The disadvantage of
these systems is the low flexibility, caused by the necessity of time intensive system en-
gineering, whenever the network system or requirements of the data to be transported
changes [26], e.g. exchange of device, new device in the setup or even swapping the
ports at the network switch. The engineering process is typically performed offline in
an engineering tool, and in order to apply changes, the whole system has to be restarted
and the application has to be stopped.
Furthermore, the engineering tool itself is also highly complex. Engineering tools, such
as Simatic Step 7 from Siemens [27], does not just consist of a framework for writing the
control logic in languages belonging to the IEC 61131-3 family [28] , and a software for
configuring network with all communication parameters. It includes also a framework,
for communication scheduling, which according to the existing literature, typically falls
4 1.2. Objective

into the family of hard mathematical problems [29][30]. This might be also the reasons
why, currently, only a very few companies offer a complete NCSs based on TTEB, such
as PROFINET IRT by Siemens, including PLCs, IODs and an engineering tool. Addi-
tionally, it has to be mentioned that none of the currently available solutions supports
flexibility features required by agile manufacturing systems, for integrating a new net-
work device that was not foreseen in advance at run time.
Apart from the vendor specific TTEB solutions, TSN standardised by the IEEE may play
an important role in the future industrial network systems [31]. It suppose to offer
high performance by utilisation of precise clock synchronisation protocol (802.1AS)
and support for scheduled traffic (802.1Qbv). However, no engineering or a scheduling
framework is specified in the current set of TSN amendments where such communica-
tion schedule could be calculated.
There are few scientific works dealing with the scheduling problem of TTEB systems,
which consider existing industrial requirements in terms of network topologies, com-
munication patterns or reliability. However, according to the author’s best knowledge,
no contributions exists that try to achieve the highest performance in terms of real-
time and reliability, while keeping the possibility to perform system reconfiguration at
runtime. In the dissertation of Dopatka [32], a framework for scheduling of communi-
cation based on graph colouring theory was developed. The assumption, however, was
that the system remains static, and the time necessary to calculate the communication
schedule is not of major importance. In his work, he did not considered redundancy to
improve the communication reliability. Hanzalek et al. in [30] also developed a sche-
duling framework that supports scheduling of PROFINET IRT protocol. The flexibility
and communication reliability were also not considered in their contribution. There are
also several contributions of Kopetz et al. about designing of Time Triggered communi-
cation systems that achieve high-time performance and reliability [33][9][34]. However
according to Poledna et al., the current deterministic systems based on TTEB are still
lacking flexibility features for performing online reconfiguration [35].

1.2 Objective
The main aim of this dissertation is to develop a scheduling framework for TDMA based
TTEB systems that considers typical industrial topologies (including redundancy) and
traffic characteristics, and is able to calculate a valid and possibly good schedule in a
short and deterministic time. Additionally, the objective is to develop a concept that
allows to automatically reconfigure TDMA based TTEB system at run-time, without the
necessity of stopping the running real-time communication and hindering its function-
ality.

Due to fast and deterministic scheduling approach the overall engineering time of
TTEB systems can be decreased. Furthermore, the scheduling algorithm can be also
executed on a resource limited hardware such as PLC. This will give the possibility that
in case of some changes, such as swapping of ports at a switch or exchanging network
device, the ”light“ version of the engineering (including only scheduling) can be per-
1.3. Outline of the document 5

formed directly at the PLC, without the necessity of performing manual efforts in an
engineering tool. This requirement is also induced by the TSN standard, that does not
have any additional engineering tool, where a communication schedule could be cal-
culated. Therefore, in this work it is assumed that the scheduling task will be overtaken
by one of the ordinary network members in a defined network segment.

The theoretical analysis of the proposed algorithms and methods will provide the ev-
idence about their efficiency as well as low complexity. The practical implementation
will help to asses theirs performance and allow to validate proposed methods. Further-
more, practical tests provide preliminary results of the proposed mechanisms and allow
to prove defined theories. Additional simulative experiments will complement the in-
vestigation with auxiliary results that were not possible or were difficult to obtain with
the existing hardware.

1.3 Outline of the document

The reminder of the thesis is structured as follows. In chapter 2 relevant background for
this work is provided and industrial environment with all important communication
requirements is described. Afterwards, the problem tackled in this work is precisely
formulated. Chapter 3 deals with timeliness aspects of currently available TTEB sys-
tems, highlighting the way how they achieve high real-time performance. This chapter
discusses also the complexity of the scheduling problem related to TTEB systems. Fur-
thermore, in this chapter one of the three main contributions is described that is the
simple scheduling approach. Chapter 4 deals with the communication reliability as-
pects, which have to be considered while designing robust and high performance TTEB
systems. It includes the second main contribution that is a special bypass for dynamic
redundancy methods, and also the disjoint path search algorithm for static redundancy
methods. In chapter 5 an approach is described that allows seamless reconfiguration
of TTEB systems that is demonstrated based on the PROFINET IRT example. All pro-
posed methods and algorithms are evaluated in chapter 6, and finally a conclusion and
outlook is provided in chapter 7.
2 Fundamentals

This chapter provides a general state of the art analysis that is relevant for this work.
Section 2.1 gives an overview about the production systems and provides deep insights
into NCSs build upon TTEB systems. Furthermore, in section 2.2 requirements in terms
of real-time and reliability that have to be fulfilled by INS are discussed and summarised.
Afterwards, in section 2.3, existing concepts to achieve deterministic and reliable real-
time communication are described. Section 2.4 elaborates on the engineering process
and all steps that have to be performed in order to bring an INS into work. Finally in
section 2.6 the problem statement of this dissertation is formulated and in section 2.7
the solution approach is briefly described.

2.1 Production Systems

Current production systems are complex structures composed of entities coming from
different technological areas such as: mechanical engineering, electrical engineering
and information technology. They are continuously evolving to fit current market cir-
cumstances to allow production companies to remain competitive. The short evolution
of the industry, including four major evolutionary steps is illustrated in figure 2.1. The
first industrial revolution (Industry 1.0) was started in 18th century with the invention
of mechanical machines and steam engine, which was was supplying machines with
the power. This significantly improved efficiency of production processes, allowed to
produce goods that were not possible before and had a positive influence on work-
ing conditions of human beings. Next, the second industrial revolution (Industry 2.0)
was introduction of the paradigm of mass production. It dramatically changed the pro-
duction systems by massively increasing production output at the same time lowering
the price. Further evolution of mechanical and electrical engineering allowed building
much more sophisticated production lines, at the same time increasing wideness of the
production assortment and shortening the production cycle time. The next significant
step in improving production systems and its third revolution (Industry 3.0) was ap-
pliance of the information technology (IT). It allowed further evolution of production
systems with the possibility for the system components to communicate among each
other and with the operators (e.g. through Human Machine Interfaces (HMI)). The cur-
rent trend in production systems is increasing connectivity of the “things” (robots, ma-
chines, etc. and also human beings) participating in the production processes. The
main idea is to increase accessibility and reduce complexity of the network configu-
ration and integration of new things into the existing environment. This is currently
an important research topic, called Internet of Things (IoT), which has been addressed

© Springer-Verlag GmbH Germany 2017

L. Wisniewski, New methods to engineer and seamlessly reconfigure
time triggered Ethernet based systems during runtime based on the
PROFINET IRT example, Technologien für die intelligente Automation 6,
DOI 10.1007/978-3-662-54650-5_2
8 2.1. Production Systems

        !"    !   

 

          
   

        
 

 

 
 

 

Figure 2.1: Industrial evolution (4 significant revolutions)

in several EU projects, such as IoT-a [36], IoT@Work [37] and many others that can be
found in the European Research Cluster of the Internet of Things [38]. The other im-
portant aspect is to make “things” smart or “intelligent”, e.g. through self diagnosis,
self optimisation, etc. This can be achieved by extending the “physical world” (devices,
machines, tools, etc.) with algorithms and methods that run in a “cyber world” and in-
teract with the “physical world”, thus forming a new concept of Cyber Physical System
(CPS) [39][40]. All these concepts are part of the so called fourth industrial revolution
(Industry 4.0) [41][42][43].

2.1.1 Automation Systems

In today’s factories, there are machines, robots, controllers, communication devices,
cameras, etc. that, if properly configured (engineered), can operate 24 hours per day,
sometimes without or only minimal intervention of a human being. Typically, when
rearrangements are necessary, human (engineer/expert) interaction is needed. Nowa-
days, such system spans the whole factory, including different departments that are
locally or even globally distributed. This has been mainly achieved through the use of
appropriate communication technologies. An automation system (AS) can be illus-
trated using well known automation pyramid (AP) illustrated in figure 2.2. According
to Sauter [44], through implementation of internet technologies into the AS, the pyra-
mid shrunk from six defined at the time of Computer-Integrated Manufacturing (CIM)
down to only three levels. Starting on top from the Company level including Enterprise
Resource Planning (ERP) systems [45], going down through the cell level that includes
Supervisory Control and Acquisition (SCADA) system [46] and Manufacturing Execu-
tion System (MES) [47] ending on the field level, where sensors and actuators coupled
with the controller directly interacts with the technical process (TP) [48]. The TP, as de-
fined in the VDI guideline 2206 – Design methodology for mechatronic systems [49], is
a physical system, composed of mechanical, electric, hydraulic, pneumatic or any com-
bination of these components. As shown in figure 2.2 there is a need for an information
exchange in the automation system among different layers defined as horizontal com-
munication as well as across the levels defined as vertical communication. However,
the important fact is that these data flows have different requirements in terms of time-
2.1. Production Systems 9

Data size

Flexibility
ERP

MES/
SCADA

Timeliness

Reliability
Co
Networke ntrol System /
d Contro
l System
(NCS)

Technic
al Proc
ess (TP
)

Figure 2.2: Automation pyramid

liness, amount of data sent in one data packet or communication reliability. Level of
available flexibility is also different, which in this work is measured as the effort related
to integration of a new network component with the existing network infrastructure. In
case of vertical exchange, the most crucial aspect is the interoperability between differ-
ent systems located at different levels.
The focus in this dissertation is the lowest level of the automation pyramid that di-
rectly interacts with the TP. It has the most demanding requirements in terms of short
latency and jitter (defined as timeliness and described in section 2.2.1) and communi-
cation reliability (described in section 2.2.2). However the level of flexibility at this level
is rather limited.

2.1.2 Networked Control System

At the bottom of the automation pyramid, we can find the following units: sensors,
actuators, Input/Output (IO) devices as well as controllers that are directly interacting
with the TP [48]. A brief description about each of these components is given below:

• Sensors are units responsible for monitoring of variables of the TP in time, con-
verting the physical sizes, such as: speed, torque, temperature, pressure, etc. into
a electrical signal, thus creating a “digital picture” of the TP.

• Controllers are sophisticated computer systems, designed to operate in a harsh

industrial environment to control the TPs. The control is performed by collecting
data about the current status of the TP from multiple digital and analogue sen-
sors, processing it with the internal control logic software and producing output
signals for actuators. This process is typically performed periodically, where the
cycle time used to collect sensor data and update actuators with the new data, de-
10 2.1. Production Systems

pends on the dynamics of the TP. For instance printing machines, having a high
process dynamics (motion control application), require short cycle times, i.e. be-
low 1ms with the jitter less than 1μs [50], while processes such as beer brewing
(process control application), require cycle times in the range of 10ms [50]. In
the industrial environment, the role of controller is typically performed by a spe-
cially designed Programmable Logic Controllers (PLCs) and therefore from now
on, this terms will be used to described such control units.

• Actuators are responsible to interact with the TP by changing digital signal into
energy or force using hydraulics, pneumatics, electric current, magnetics or me-
chanics in order to control it.

• IO devices (IOD) also called remote IO, are currently widely used to communi-
cate with multiple sensors and actuators, which are distributed in the field. In
comparison to older centralised systems, in which sensors and actuators are di-
rectly connected with a PLC, they allow to save a lot of wiring efforts, simplify the
system, thus save time and money. Communication between PLC and sensors
and actuators is performed using one physical interface (e.g. Ethernet) and one
communication protocol (e.g. EtherCAT).

These components, typically exchange small amount of data by using defined cycles
times [51] forming a Control System (CS). Two major classes of control systems can be
distinguished. These are open loop control and closed loop control systems, as shown
in figure 2.3.

Input Output
PLC actuator TP
Input Output +
PLC actuator TP -

sensor
a) b)

Figure 2.3: Control System a) Open Loop System, b) Closed Loop System with Feedback

Open loop system is controlling TP by setting selected parameters, without consider-

ing its current status, see figure 2.3a. Therefore, it can be used to control simple non-
critical (rather not industrial) processes, such as drying clothes in washing machine.
Here, after setting a particular program (defined by drying time), the drying process
starts. Since drying time has been defined based on a simple model, such as 1 kg of
clothes should be dried in 0.5h, there is no guarantee that all clothes will be dried af-
ter this time elapses. Another important characteristic of this control systems is that
whenever there are some disturbances of the process, such as air leakage, the quality of
this control loop may be significantly decreased. Closed Loop Systems with feedback
(figure 2.3b) consider the current status of the TP (e.g. using sensors) to much more
precisely control its behaviour. Since the status of the TP system is constantly moni-
tored, closed loop control systems can be used to control diverse industrial processes,
such as chemical reactions, motion control applications, etc. As opposed to the open
loop, closed loop control system can handle external disturbances that may influence
2.2. NCS Communication Requirements 11

the TP.
Today, components such as PLCs, IODs, sensors and actuators are highly distributed
among the production systems, thus there is a need to span them together with the
communication network. Such control systems are called Networked Control System
(NCS), where network becomes a part of the control loop, see figure 2.4. It is important

Input Output
PLC actuator TP
+
- Network IO Device

sensor

Figure 2.4: Networked Control System

to mention that the network used in the control loop has to fulfil certain requirements
in terms of timelines and reliability. Several proprietary and standardised network tech-
nologies and protocols have been developed [52] that can be used to build NCSs. They
were explicitly designed to fulfil requirements of a large variety of industrial applica-
tions. A detailed description about the requirements is given in the following section.
Discussion about the way how protocols were designed in order to fulfil these require-
ments is given in section 2.3.

2.2 NCS Communication Requirements

There is a large variety of different processes (applications) that have to be controlled by
a NCS. A simple example could be a boiling process (e.g. in a brewery), where temper-
ature of a liquid must be kept in a predefined temperature range. Such system can be
composed of the following components (see figure 2.5): a boiler that holds the liquid (it
is the TP), a heater (an actuator) and a temperature sensor (thermometer) responsible
for measurements. Sensor and actuator are connected together via IODs responsible
for exchanging information with the PLC that controls the boiling process. The HMI

Actuator TP
Set point Output
HMI +


PLC Heater Boiler
Network IO Device
Feedback
Temp. sensor

Figure 2.5: Boiling process controlled by a control loop with feedback

enables to set a reference temperature in the boiler. It is a typical example of a con-

trol loop that uses feedback to control the process (in this case heater turns on/off).
Looking at the process from a higher level, task of the control system is to keep tem-
perature of the boiler at some predefined temperature range. From the control system
point of view, the control process can be divided into several smaller tasks that have to
12 2.2. NCS Communication Requirements

be performed cyclically. In order to be able to control the boiler in real-time, all tasks
performed by the NCS are time constrained. It means that tasks, such as data collection
from sensors, data processing by the PLC and data distribution to the actuators have
to be performed in a defined time window, see figure 2.6. This time window defines its

task release time (ri) task deadline (di)

time window
time

Figure 2.6: Task with the real–time requirements

release time (r i ) for each task, e.g. in case of data collection, defines the earliest time
when the sensor data is ready to be sent. Each task have also its deadline (d i ), which is
defined as the latest time, where a task has to be accomplished. In case of data collec-
tion, it would describe the latest time a sensor data has to be delivered to the PLC. Only
in such a way, it is possible to achieve appropriate reaction time of NCS over the TP. The
following subchapters will discuss the most important characteristic parameters that
has to be considered in designing of NCS. The focus is put on the INS that has to allow
to fulfil certain real-time as well as communication reliability requirements as a part of
the NCS. The next chapter discusses the timing requirements of the NCS in more detail.

2.2.1 Timeliness in NCS

In order to be able control the industrial process, it is important that the communi-
cation is performed in real-time. Figure 2.7 illustrates a sample of communication ex-
change between an IOD (sender) and a PLC (receiver), and transmission latency Td (can
be td i or td o ) caused by the network system. All messages are delayed by the network

IOD Network PLC

(sender) System (receiver)

tS1 message 1
message 1 Td1
tR1

tSn message N
tRn Tdn
message N

Figure 2.7: Transmission time parameters

system. The transmission latency Td can be calculated by taking two time stamps, one
on the sender side at the time of sending the message (tS i ) and one at the reception
time of the message (tRi ). By the subtraction of tRi and tS i we obtain the Td . It is not
possible to exclude the network latency completely from the data delivery, however it is
important to keep the latency bounded. The parameter that describes the time differ-
ence between the worst case (Tdmax ) and the best case transmission latency is defined
Another random document with
no related content on Scribd:
concerned with Mr. Linder, the excellent agricultural inspector of the
Lindi municipality, to whom I owe many valuable suggestions, and
who, on account of his thorough acquaintance with this very district,
had originally been selected as my companion. Linder rendered
splendid service in suppressing the rebellion: while any action on the
part of the Field Force was still entirely out of the question, he had
already, with a small detachment of police, repulsed numerous
attacks of the rebels, and ultimately sustained a serious wound. But
while decorations have been simply raining down on the Navy and
the Schutztruppe, Bwana Linda still walks among mortals without a
single order. He is, however, a philosopher as well as a hero.
The song runs as follows:—
Ulendo wa Linda (er); pa kwenda ku Masasi na gumiri chikuo: mkasálile
mbwana mkubwa ngondo jaiche nand autwiche lunga yangadye. Mkasálile akida
Matora: ngondo jaiche na gombel(r)e lilōmbe. Tukujir(l)a Masasi; Mwera kupita
mchikasa mpaka pe Lindi. Ne wapere rukhsa. Yendeye ku mangwenu; mkapānde
mapemba.
The translation is as follows:—
“The journey of Linder, when he went to Masasi, and I shouted
with a shouting.—‘Tell the Bwana Mkubwa, war has come, and I ran
away without looking back. Tell the akida Matora, (that) the war has
come, and I have beaten the great-drum.’ Then we went to Masasi,
the Wamwera are beaten and go as far as Lindi, and they get
permission. ‘Go to your homes, and plant Mapemba (sorghum).’”
This is delivered in very quick recitative, and relates in a few words
the history of the whole campaign, of course making the singer the
central point. Mr. Linder comes to Masasi in the course of one of his
official tours, his principal duty being to ascertain whether the local
headmen have cultivated the various crops prescribed by
government. There the loyal Likoswe of course hastens to him and
warns him of impending hostilities on the part of the Wamwera.
Linder in his turn sends word to the District Commissioner at Lindi,
and at the same time despatches Likoswe with an urgent message to
Matola’s. Likoswe, on arriving, beats the war-drum (lilombe),
Matola’s warriors immediately hasten to the spot, six hundred men
with guns and many more with spears, bows and arrows, and the
chief marches on Masasi, to take the Wamwera in the rear. It is
related as a fact that Seliman Mamba and his subordinates had each,
at the beginning of the rising when their hopes were highest and they
already saw the Germans driven into the sea, fixed on a house at
Lindi with all its contents as his own share of the spoil. Possibly, the
line about the enemy’s going back to Lindi refers to these unrealised
plans. Matola, I believe, lost about forty men in fighting the rebels,
but certainly did not drive them back to Lindi. The last sentence
relates to the conclusion of peace:—the vanquished are pardoned,
and directed to go home quietly and plant their gardens once more.

YAO DANCE AT CHINGULUNGULU

My cinematograph, too, has been several times in requisition

during my stay at Chingulungulu, as I have found opportunity to take
a whole series of dances of the Wayao and Makua. The latter, it is
well known, are the hunting-tribe par excellence of the east—indeed
professional hunters of any tribe are generally described as Makua.
They are, moreover, typical for all other tribes in their method of
hunting, and in all appliances and customs connected therewith. One
day, by Matola’s orders, a troop appeared at Chingulungulu to
perform, as they said, the makwaru—a dance entirely based on the
details of the hunter’s life. I had quickly got my apparatus arranged
in a suitable place, not an easy matter here in the loose alluvial soil,
as, if one presses too hard on the legs of the tripod, they are apt to
sink into the sand up to their whole height. Grown wise by
experience, I now take the precaution of driving a wooden wedge
obliquely from above under each leg before beginning operations. It
is more difficult to remedy the results of a mistaken economy. In
order to save the African Fund about twelve shillings and a quarter of
a carrier, I did not bring the heavy stand necessary for the Ernemann
cinematograph, thinking that I could use my ordinary camera-stand.
This, though excellent of its kind, is far too light to stand the
continual jerks of the cinematograph, and I have to balance matters
by hanging a heavy stone or one of my packing-cases under it. If
matters become very serious one of the carriers has to sacrifice
himself and do duty as a tripod-holder. Everything being now ready
for the makwaru, the same band which figured at Sulila’s and
Likoswe’s performances takes its place. It consists of six or seven
men and youths, squatting before their long white logs with their
drumsticks in hand. Suddenly, a fantastically decorated something
flashes into the circle, moving so rapidly that it is impossible to
distinguish whether it is a man or a woman. Being compelled to
pause for breath it is revealed as a middle-aged man in a kilt of long
green leaves resembling a ballet-dancer’s skirt. The man scarcely
stirs from the spot, but his skirt flies in the wind, and he works his
feet in quick, regular time, while at the same time his arms move in a
manner difficult to describe, as there is nothing in European dancing
which in the least degree corresponds to it; and both, arms and legs,
keep exact time with the band. Whether the rest of the body in its
incessant motion backwards and forwards also keeps time it is
impossible to decide, as the vibrations are too rapid to let the eye
make out the details. This stage lasts so long that I am tempted to
regret the waste of my precious film.
 “BUSH SCHOOL” IN THE PORI, NEAR CHINGULUNGULU

At last the hunter changes his tactics. The dancer is, in fact, a
hunter, and not only that, but a very successful elephant-hunter; and
having just killed a large elephant, he is celebrating this deed of
prowess before the assembled inhabitants of his native village, just as
he does after his return from the actual hunt. Here, too, the people
have collected from far and near to see this celebrity, and to admire
his skill in the dance. His performance becomes more and more
vivacious—he no longer remains on one spot but trips forward, first
in a straight line, then in a zig-zag. At last he revolves in a circle,
moving round with short, cautious jumps, and all the time keeping
up the movements of his arms and hips without a moment’s
intermission. After one more rapid trip round the circle and a frantic
vibration of the whole body, the dancer stands still, breathing deeply.
This kind of dance is too peculiar, too divergent from all European
standards for us to judge of it critically according to the rules of art. I
had expected a pantomimic representation of an elephant-hunt, or at
least of the stalking and killing of the game, and I must confess that I
can find nothing in the performance which seems to have any such
reference, and must confine myself to admiring the incredible
dexterity shown by this acrobat in setting all his muscles a-quiver. I
have no sooner got a fresh film ready, than a second dancer has
appeared on the scene, whose action is still more curious and
perplexing. At first one sees nothing but a confused mass of green
leaves rolling and writhing on the ground in convulsive motions.
After a while, this resolves itself into a man much like the previous
one, except that his costume is much more voluminous. He quivers
in a masterly manner and shows as much staying power as his
predecessor; but his chief strength lies in his legs, whose suppleness
and power of assuming the most grotesque attitudes are nothing
short of marvellous. When he has exhausted his repertoire and made
way for a third performer, we at last get the expected pantomime.
Stooping as if for a spring, the hunter creeps up, noiselessly, making
use of every bit of cover, to stalk the elephant, whose scent is
exceedingly keen. At last the goal is reached—swiftly, but as
noiselessly as the hunter, the quarry, represented by another man,
has slipped into the arena, and squatted down, and the hunter circles
round him in diminishing spirals. We expect the deadly shot, but it
does not come off, and the third dancer, quite regardless of the
elephant he is supposed to represent, begins to “triumph” in
precisely the same way as the two others, practising highly artistic
short steps, swaying his hips and flourishing his arms. “Bassi”—
(finished,) I exclaim, as the last of my three films whizzes off the reel.
Quite in contrast to these are the typical unyago dances of the
Wayao. There seems to be a great variety of these; but so far I have
only seen two at Chingulungulu, a masewe, so called from the rattles
worn, as already mentioned, on the legs and feet, and a luwanja.
Both are essentially the same in character. The primitive xylophone
of the Makua hunting-dance is here replaced by a complete band of
drums, of the most various shapes and sizes. A certain musical
faculty inherent in the race is evidenced by the fact that the
musicians take care to tune up before the dance begins. Each beats
his own drum, listening carefully to hear whether it is in tune with
the rest, and if not, hurries away to the nearest hut and comes back
with a brand from the hearth and a large bundle of dry grass. The
grass is heaped on the ground and set on fire, and then every drum is
held with the open end over it, for a longer or shorter time—some for
a few seconds only, some for half a minute or more—the pitch being
tested by striking from time to time. At last all the skins are
sufficiently tense and the drumming begins.
 A YAO DRESSED FOR
THE MASEWE DANCE

MASEWE DANCE OF THE YAOS AT MTUA

At the same moment a dense cloud of dust is seen approaching

with lightning speed, and discloses a seemingly endless procession of
men, youths and boys, all decked in bundles of masewe at the ankle
and above the knee, and a kilt of leaves and strips of skin round the
waist. They take their places in the arena in front of the band, and
immediately fall into position and trot along in Indian file, till the
line closes up into a circle and moves round to the left, then round to
the right, and so on. It is astonishing how uniformly and accurately
the movements are executed by every individual performer, even the
youngest boys. There is nothing very exciting about this dance; in
fact, I find all native dances monotonous, perhaps owing to the
prevailing character of the continent, which is very uninteresting,
except in a few favoured spots. Perhaps a native critic, however,
might object that there is no great variety in our waltzes or polkas.
Just as these reflections were passing through my mind, the scene
changed, somewhat to its advantage, and the circle broke up into
groups which vied with each other in the most remarkable leg-
movements. These, in fact, seem to be the strong point of all these
dancers. One group floated along on tip-toe, another imitated the
dignified gait of some kind of wading-bird, yet another swayed
merrily in and out between the rest, and a fourth stalked along with
legs held perfectly stiff. Long after my last film was finished the
company were still disporting themselves, unable to leave off, but at
last this “turn,” too, came to an end; the band produced only horrible
discords; I was tired out with standing; Knudsen complained of the
first symptoms of fever, and the function was over.
The performance of dances like the one just described, which is
connected with the circumcision rite, have naturally increased my
interest in this tribal festival, and my desire to see and study it as
closely as possible.
My curiosity was increased by the two following incidents. One
afternoon I was strolling through the bush in the neighbourhood of
Chingulungulu; we had already obtained some interesting
photographs of graves, had studied the exterior and interior of some
outlying homesteads, and were about to take some views of the pori
showing the character of the vegetation. After straggling in Indian
file through the high grass and the underwood, which was here
exceptionally dense, we came to a little circular clearing, perhaps
from fifteen to twenty yards in diameter, and studded with a few
scattered bushes. The unique feature of the place was two concentric
circles of stumps having another stump in the centre. These stumps
were about a foot high, cut off with a perfectly smooth horizontal
surface, and excellently well adapted for seats. I took a photograph of
this remarkable object without loss of time, and, on my return to
camp, made inquiries of Matola and others as to its meaning. I found
that the stumps were seats for the wari, as the boys under initiation
are called after a certain point in the ceremony, and the seat in the
middle was that reserved for the instructor who has charge of the
boys during the months which they have to spend in a hut built for
the purpose in the bush. My informants added that the hut had stood
close to the circle, but was no longer in existence, as the unyago for
which it had been built had taken place some years ago.
Some days later, Knudsen and I were sitting under our baraza in
the early part of the afternoon, pressing our hands to our temples. It
was no wonder that every day about this time we both suffered from
excruciating headaches, for the temperature had been steadily rising
during the last few weeks, and on this particular afternoon the
thermometer stood at 93·36°F. We had given vent to our disgust at
the Dark Continent in the strongest of language, and I was just about
to soothe our ruffled feelings with a cigar apiece, when we saw two
black figures approaching. These proved to be Akundonde, the wise
old Yao chief, and his councillor, Akumapanje. We had sent to ask
Akundonde to find us some men capable of giving accurate
information, and now he came himself, though far from well. He was
suffering from the usual neglected ulcer on the leg, and could only
limp along painfully with the help of his staff, so that his taking a
four hours’ walk to oblige us shows a degree of goodwill deserving
the amplest recognition.
Akundonde being established in Knudsen’s long chair, while his
companion took a seat on a packing-case, I made an effort to divert
the conversation from the trifles which at first threatened to engross
it to the subjects which chiefly interest me, and succeeded, more by
luck than good guidance. As usually happens, we were soon
discussing the most recondite matters, such as the attitude of the
natives towards eclipses, the fall of meteorites, and the moon.
Meteorites are considered by the Yaos as of evil omen. When they are
heard to explode, people say, “Either a great chief will die this year,
or a great multitude of the people will perish.” An eclipse of the
moon is thought, as among all primitive people, to be a personal
encounter between two foes. The enemy of the moon is, of course,
the sun; they seize each other fiercely and wrestle together. As both
are equally matched, the battle remains undecided, which forces
mankind to interfere. The Wayao run in haste to fetch hoes and axes,
and strike them against each other, looking up at the scene of strife
and calling out:—
“Mlekangane, mlekangane, mwesi na lyuwa, mkamulene,
Mlekangane, mlekangane sambano.”
“Go asunder, go asunder, sun and moon, you have seized one
another. Go asunder, go asunder now.”
The same custom is observed in eclipses of the sun, as is only
logical.
The full moon with her pale light exercises the same magical
influence on the native mind as on the feelings of every other mortal,
except that our black brother is not like us filled with emotional
enthusiasm, but, quite in conformity with his views on other matters,
makes use of this favourable opportunity for heightening the virtue
of his medicines and charms. When the moon is at the full, the native
goes to the nearest cross-roads, or to a place where two paths meet,
carrying with him a sufficient quantity of a certain gum called ubani.
In perfect silence he then kindles a fire by means of the primitive
appliance of the drill (to be described later on). The dust produced by
boring catches fire, but the glimmer is at first so faint that it is
scarcely perceptible even to the keen eyesight of the savage. Very
carefully he blows on the tiny spark—it grows, catches the bunch of
dry grass and then the sticks, and when the flame leaps up, he drops
his powder into it. The flame now burns dimly, a thick smoke rises,
and the man takes the amulets he is accustomed to wear round his
neck, arms and waist, and holding them in the smoke, says: “You
moon, a little while ago you were not there, and the sky was dark.
Now you are there and shine down brightly. All beasts and plants are
glad and have new strength, so let my medicine also have new
strength.” Then he prays thus: “Let the medicine protect my body
against lions and serpents, against witchcraft and everything that
may hurt me, and let my body have new strength.” Once more he
swings his charms through the smoke, as it becomes thinner and
more transparent; the fire dies down, and as noiselessly as he came
the man creeps back to his hut.

FRESCO ON THE WALL OF A HUT AT AKUNDONDE’S,

REPRESENTING TWO EUROPEANS WITH THEIR ESCORT:
THE WORK OF A YAO BOY

Being now on the subject of magic, the three ethnographic

specialists, Knudsen, Akundonde and Akumapanje, keep to it, and
speak of the tying of knots. Akundonde relates how a man in this
country, if he has designs on any particular girl, takes a strip of bark,
makes a knot in it, without drawing it tight, and says to it, “You tree,
your name is sangalasa (joy)—you are to fetch me that girl, and as a
sign that it shall come to pass, I shut my words up in you.” He then
holds the open knot in front of his mouth, puts his tongue through it
and draws it tight. He afterwards wears the knotted piece of bark-
string tied round his wrist. This proceeding, though simple enough,
is connected with a long and important chapter in racial psychology.
The tying of a knot in fact, in many strata of mankind, has an occult
meaning; the binding power of the knot is supposed to be transferred
to certain persons, and, so long as the knot itself cannot be untied,
those persons are indissolubly attached to him or her who has tied it
according to certain rules and with the proper ceremonies.
Interesting as these matters were, and glad as I should have been
to know more of them, I was just now still more eager to hear about
the much-discussed unyago. I brought up the subject, but both
natives cleverly evaded it. After a while, I noticed the old chief’s eye
roaming wistfully about our study, saw that he was tired and thirsty,
and remembered that Daudi, the native clergyman, had sent us a
large pot of pombe whose quality precluded our drinking it
ourselves. “I suppose it will be quite good enough for these two old
sinners,” I remarked to Knudsen, who must have been revolving
similar cogitations; for he at once seized the import of my words,
fetched a huge tin mug from his tent, filled it with the yellow,
fermenting liquor, and handed it to Akundonde. The latter took it,
but did not drink, handing it to his companion instead. “There’s a
polite chief for you!” I thought to myself—but, seeing how very
cautiously Akumapanje touched the beer with his lips, it became
clear to me that I was witnessing an ancient traditional custom,
arising from the innate suspiciousness of the negro, who scents—not
indeed poison, but certainly witchcraft—everywhere, and dreads it
accordingly. The precaution is intended to divert the risk from the
superior to the subordinate.
Akumapanje, after tasting, handed the cup back to Akundonde,
who thereupon emptied it at a draught. A few seconds later it was
again at the lips of the prime minister, who faithfully copied his
master. Drink and counter-drink succeeded each other at the same
rapid rate, and we Europeans looked on with mixed feelings of envy
and admiration. This did not prevent me from remembering our
ethnographical purpose, and I found that what had previously
seemed impossible was now child’s play. The two old men, by turns
completing each other’s statements, gave a fluent description of the
general features of the boys’ unyago: the arrangement for holding
the festival at different villages every year (which was not new to
me); the introductory ceremony, held in an open square surrounded
by the huts erected for the candidates; and the operation itself, which
takes place in a special hut in the depths of the forest. I had heard
something of all this from Knudsen, who, in the course of his many
years’ residence among the Wayao, has acquired a wonderful
knowledge of their life and customs, and whom I have been pumping
at every spare minute with such persistency that the good fellow has
no doubt often wished one of us elsewhere.
At last, however, our two visitors, becoming more loquacious as
the pombe diminished, reached a part of the subject of which
Knudsen knows very little, but which attracts me most of all. This is
the instruction given to the boys during the months spent in the bush
by their teachers (anamungwi). These instructors, of whom every
boy has one from the time of his initiation into manhood, are
indisputably one of the most sympathetic features in the life of the
people. They watch over their pupils through the painful weeks of the
unyago, teach them what is fitting and unfitting, and remain
responsible for their welfare even after they have left their boyhood
far behind. I was anxious, above all, to ascertain the gist of the moral
teaching given in the bush hut, and, though I only partly succeeded
in doing this, it is a great satisfaction to have taken down verbatim a
fragment of a speech delivered on such an occasion.
Some extra well-filled cups having removed the last scruples of our
two jovial informants, Akundonde, with a little more encouragement
from Knudsen, began in a didactic tone:—
“Mwe mari, sambano mumbēle. Atati na achikuluwēno
mnyōgopĕ́. Nyumba mkasayinjila tinyisimana chimtumbánăgá.
Wakongwe mkasayogopa; mkagononawo, mesi akayasináwo.
Imālagắ akamtikĭté; imālagắ akamila muchisiḗ; masakam.
Munyitikisie: marhaba. Mkuona mwesi sumyógopé, ngakawa
kuulala. Kusimana timchiŭá; Miasi jili kogoya. Chilwele winyi.”[34]
The translation is as follows:—“You, my pupil, now you are
initiated. Your father and your mother, fear (respect) them. See that
you do not enter the house (unannounced), lest you should find them
embracing. Do not be afraid of women, but sleep with them, bathe
with them, when you have finished let her rub (knead) you; when
you have finished she should salute you (saying) ‘Masakam,’ and you
must answer, ‘Marhaba.’ You must be afraid (= take care) when you
see the (new) moon, you might get hurt. Beware of women during
their courses, this is dangerous, (it causes) many diseases.”
My notes were scarcely as complete and connected as the above
when first written down. The native is incapable even when sober of
taking his sentences to pieces, as it were, and dictating them bit by
bit; but taking down the words of these two jovial old sinners was a
difficult task, which, however, we accomplished successfully up to
the point when the inevitable catastrophe set in.
The two had invariably paused for refreshment at the end of every
sentence till they reached the point above indicated, when they
suddenly found the pombe jar empty. They had drunk at least five
gallons at a sitting, but with the strange logic of the intoxicated, they
considered themselves entitled to a further supply, and, when none
proved to be forthcoming, they indignantly broke off their lecture
and left in a huff. This is the reward of being hospitable overmuch.
The address here reproduced, which I have translated with the
help of Knudsen, Daudi, Matola and some others, is said to be the
same, both as to matter and form, at all unyago ceremonies. No
doubt this is correct, for I know nothing which could more exactly
express the feelings of the native than just these precepts. They are a
strange mixture of hygienic rules and moral instruction, and at the
same time contain a good deal of primitive tradition which still forms
part of daily life. I mean by this the fact that the youth, once
recognised as a member of the adult community, is forbidden to
enter his mother’s house unannounced. Here, in East Africa, we are
still in the matriarchal stage, where the husband is nothing, so to
speak, but a connection by marriage. He is his children’s father, but
is not related to them, in fact he belongs to a different clan. This clan,
as so often happens among primitive peoples, is exogamous—that is
to say that there is no impediment to a young man marrying a girl of
any clan but his own. This prohibition goes so far that the young Yao
has, as far as possible, to avoid his nearest female relations who, of
course, are his mother and sisters, and hence the injunction at least
to give warning of his approach when entering his mother’s house.
The stress here as elsewhere laid on the reverence to be shown to
father and mother must strike all right-thinking Europeans as a very
pleasing trait. Respect for parents and for grown-up people in
general is, as I have been told over and over again, the principal and
fundamental feature in native education, and Knudsen testifies that
the young people in general observe it in a marked degree in their
intercourse with their elders. We Europeans might well learn from
the natives in this respect, thinks Nils, who is no doubt, well qualified
to form an opinion.
But, in spite of all pleasant impressions as to native educational
maxims, I have lost the end of the unyago address—a misfortune for
which the good Daudi’s big pombe-jar is to blame. If the mountain
will not come to Muhammad, Muhammad will have to go to the
mountain. In other words, Akundonde having declared that he must
go home to put fresh dawa on his leg and cannot possibly come
again, we shall have to look up the old gentleman at his own
residence.
HERD OF ELEPHANTS. FROM A DRAWING BY BARNABAS,
AN EDUCATED MWERA AT LINDI
 CHAPTER X
FURTHER RESULTS

Chingulungulu, August 31, 1906.

I am still at Chingulungulu, cursing the infernal heat, horrible dust

and dirty natives with more fervour than ever, but unable to get away
from them. The reason for this is the fact that while at first my stay
here seemed utterly barren of scientific results, this state of things
gradually reversed itself, so that the difficulty now lay in dealing with
the mass of new impressions and observations. It is impossible to
relate in full detail the exact way in which I obtained an insight into
native customs and ideas—this would fill several volumes, and my
time is limited. I shall therefore content myself with a few personal
touches and a small selection from the various departments of the
material and mental life of the tribes inhabiting this vast plain.
The most important incident affecting my expedition was the
engagement of Nils Knudsen as a permanent member of its staff,
subject, of course, to the consent of the Agricultural Committee. I
fancy the arrangement is satisfactory to both parties. As I have
already remarked, Knudsen is in the service of the Lindi
Municipality, as master of the Industrial School. At the request of the
District Commissioner, he had been granted leave of absence to
make a tour through the plain west of the Makonde Plateau and
exercise a sort of supervision over the village headmen. For reasons
of which I am not called on to judge, the plan of appointing such
European inspectors has been given up again, and, as the Lindi
municipality naturally saw no occasion to let their industrial teacher
travel about the country for his own amusement, he was recalled. I
must honestly confess that I had long found Knudsen quite
indispensable, and therefore took the opportunity of applying to the
District Commissioner for permission to engage him, when the
latter, a few days ago, visited us on one of his official tours. He has
seemed ever since to enjoy an increased sense of his own importance
and, in fact, the task of initiating a German scholar into the deepest
secrets of alien life is no doubt a far pleasanter one than that of
teaching lazy native boys to plane, saw, forge and solder.
The second incident is a severe attack of fever, with which I have
been laid up during the last few days. I was just about to photograph
the old Sudanese sergeant who had come up with Ewerbeck, and who
was chiefly remarkable for a cough which kept everyone awake at
night. When I saw him going to muster his men for roll-call in the
middle of the afternoon, I went to take down my 9 x 12 cm. camera
which hung from a nail on one of the pillars of the baraza; but let it
fall in lifting it down, and found, on picking it up, that the sliding
front had got bent and the instantaneous shutter injured by the fall.
The first accident was remedied by energetic pressure, for the second
nothing could be done. I do not to this day understand why the loss
of this instrument should have thrown me into such a state of
excitement; but there are moments in life when we do, or omit to do,
things for which we afterwards vainly try to account. I suppose I
never even remembered at the time that I still possessed a 13 × 18
cm. apparatus of excellent quality. That I did not recall the fact later
on, is easier to understand, as by sunset I found that my temperature
was rapidly rising. I tried a remedy previously found effectual for
bringing on perspiration—huge quantities of tea with citric acid in it,
but in vain. After a terrible night with an average temperature of over
104°, the fever had so far abated that I could exert myself to make
the working drawings for additional slides to my 13 × 18 cm. camera,
which I wished to send to the Indian fundi at Lindi. Up to this
moment I had thought my photographic equipment perfect, but the
possibility of such an accident as befell my smaller camera and of
remedying it by the use of simple wooden frames had not occurred
either to me or the firm who supplied me. By exerting all my
energies, I was just able to finish the drawings and send them off by a
runner to Lindi, when my temperature again rose above 100° and I
was forced to go back to bed. The attack then ran its course and came
to an end, as fever always does. To-day I should almost feel inclined
to smoke, if we had any tobacco worthy of the name. However, I have
now had quite enough of Chingulungulu, and as the Rovuma with its
green banks and clear, cool water, its sand-banks and islands is only
a day’s march distant, we intend to go thither shortly for a rest and
change after all the discomforts, great and small, of our stay here.
 Before leaving, I feel that I ought to set down at least a few of the
observations made at this place.
Among many other diseases, such as malarial, black-water and
remittent fever, sleeping-sickness, guinea-worm, beriberi, and
whatever other ills, great or small, mankind may suffer from in these
otherwise favoured regions, leprosy is unfortunately endemic in our
colony on the Indian Ocean. On the coast of the southern district, the
Government is trying to prevent the further spread of this terrible
disease, by establishing an isolation hospital on an island in the
Lukuledi Estuary, where the patients, at present about forty in
number, are treated by the medical staff at Lindi. Here in the
interior, lepers are for the present entirely dependent on the care of
their fellow-tribesmen. Among the Yaos this care is a mixture of
human sympathy and the crudest barbarity. The patient is taken to a
hut built specially for him in a remote part of the bush, where his
friends or relations bring him food, till the end seems to be
approaching. If the wise men of the tribe come to the conclusion that
this diagnosis is correct, a last and very abundant meal is carried out
to the hut, which is then fastened up from the outside, so strongly
that, even had the patient the power and the will to make an effort,
he could not free himself. He is thus, should he still have any vitality
left by the time the last of the food and drink is consumed,
condemned to perish of starvation.
 VILLAGE OF THE NGONI CHIEF MAKACHU

GRAVE OF THE YAO CHIEF MALUCHIRO, AT MWITI

Another picture connected with death presents itself. We have

already seen the mysterious, legend-haunted site of Hatia’s grave on
Unguruwe mountain; those of other mortals are unpretending
enough and quite prosaic in character. In the country round
Chingulungulu I have found graves, both old and recent, at various
places in the bush, none of them outwardly distinguishable from
graves in our own country, except that the mounds over those of
children are round or oval, instead of long like those of adults. So far
I have seen nothing of the custom reported to me by several
informants, of building a hut over the grave, and decorating it with
calico. Only one grave at Masasi had such a hut, but I was told that it
was an Arab grave, and there was no cloth.[35] The grave of Nakaam’s
predecessor, Maluchiro, at Meviti, has unfortunately quite lost the
traditional character. Here the traveller finds a large oval hut, and,
stooping under the wide, overhanging eaves to enter, he sees, in the
solemn twilight within, massive clay pillars at the head and foot of
the grave, and a somewhat lower wall on either side of it. Such
monuments are shown with pride by the natives to the passing
European, but they are a proof how far Islamitic culture has
penetrated the old African life.
European influence also has a share in the disappearance of old
customs, though, in one point, at least, it is less far-reaching than I
had supposed. I imagined that a box of matches would be found in
every native hut, but I have seen nothing of the sort, and, moreover,
have observed no other way of procuring fire. Yet no hut is ever
without it. Here we have the startling solution of a question which
has long occupied the attention of ethnographers. Not so many
decades ago, inquirers of the standing of Tylor and Lubbock
seriously believed in the existence of fireless tribes—even our brown
fellow-subjects in the Marianne Islands being classed with such
unfortunates. The contrary of this hypothesis has now been
irrefutably demonstrated, and it is known that there is no tribe in the
world ignorant of the use of fire, or even of the mode of producing it
artificially. The problem has therefore assumed another aspect. Did
men first use fire, and then learn to produce it? that is to say, did
they begin by making use of its natural sources, such as volcanoes
and lava currents, burning naphtha-beds, trees kindled by lightning,
or heaps of vegetable matter ignited by spontaneous combustion?—
or did they first learn to bring out the divine spark by boring,
friction, or percussion, and then proceed to harness the kindly
element to household tasks? Both sequences of events are a priori
possible, though, of course, the first is much the more probable of
the two. To-day we may say that it is the only one recognised. This
knowledge we owe entirely to ethnography.
 At a time when hundreds of students are continually busy
investigating and describing the remotest and most forlorn of
primitive tribes at present accessible—when the existing
ethnographic museums are filled to overflowing with new collections,
and new museums are opened every year, it is strange to think of the
earlier and less favoured period which had to be content with mere
arm-chair theories. Two branches of a tree rub together in a storm.
As the wind grows stronger, the friction becomes more rapid, till the
surfaces are heated; at last a tiny spark appears, it becomes a larger
spark, and then a devouring flame which consumes the whole tree.
Primitive man, standing under the tree, has been watching the
process with amazement. “Oh!” says he, “is that how it’s done?” and
thereupon takes a couple of sticks and does likewise.
In this description we have a typical
specimen of the old-fashioned theory devoid
of any concrete basis of fact. It is the
hypothesis propounded by Kuhn, the
philologist, who, fifty years ago, was at least as
famous for his “Origin of Fire”[36] as for his
work in comparative linguistics. We of a
generation which knows no reverence have
grown accustomed to laugh at the venerable
scholar; but such is the way of the world.
It is always well to remember, in the case of
a widely-distributed art, like the production of
fire, that it may have originated in more ways
than one. When we see to-day that by far the
greater number of primitive tribes make use
of a boring implement, while a smaller section
KINDLING FIRE BY uses friction, and a third an instrument like a
FRICTION
saw, and the rest have already advanced to the
principles of the flint and steel, the concave
mirror and the pneumatic fire-producer—it follows of itself that such
must be the case. At the same time this variety of method shows us
that the production of fire is everywhere a secondary matter, an
accidental discovery, made while pursuing some entirely different
end. This is even found to be the case with the Malay fire-pump of
South-Eastern Asia. This is a tube, closed below, into which a tightly-