Sie sind auf Seite 1von 6

Design and experimentation of a low-power IoT

embedded system for wireless underwater sensing


Redouane Es-sadaoui1, Lahoucine Azergui2, Youssef Ghanam3, Jamal Khallaayoune4
1,4
National Institute of Posts and Telecommunications, Rabat, Morocco
2
National School of Applied Sciences, Cadi Ayyad University, Marrakech, Morocco
3
Sultan Moulay Slimane University, Beni Mellal, Morocco.
Emails: red.essadaoui@gmail.com, 2lahoucine.azergui@edu.uca.ma, 3youssefghanam@gmail.com, 4laayoune@inpt.ac.ma
1

Abstract—This paper aims at presenting an power IoT platform that integrates underwater sensors. The
experimented low power Internet of Things architecture, rest of the paper is structured as follows. Section II presents
that could be integrated to underwater monitoring the related works to low power IoT schemes. Section III
systems like Autonomous Underwater Vehicles for describes the proposed architecture. In section IV, the
wireless underwater sensing. The proposed hardware implemented low power communication protocols are
architecture of the sensor nodes is based on an ultra- described. The experimental results are presented in Section
low-power Flash Micro-controller that takes V. Finally, conclusions are given in Section VI.
measurements from water sensors, and synchronizes the II. RELATED WORKS
network by means of a GPS. The receiver board
incorporates a Digital Signal Processor and a Flash A. IoT Technologies
MCU that communicates data to a local server The number of IoT applications is growing in many
implemented on a RaspberryPI unit. Moreover, this areas including smart home, healthcare monitoring, smart
paper proposes mechanisms to minimize the power city, utilities, smart agriculture, security, smart water,
consumption of the realized system. Laboratory industrial control, environmental monitoring, and more.
experiments and sea trials have been conducted Recent advancements in wireless communication and
successfully with water depth and temperature sensors embedded systems make wireless sensor networks as a
and showed the capability of the developed system to prominent asset of Internet of Things [5].
measure the real temperature and depth of the water Many relevant works have been published in relation to
and gather the measured data to the local server with IoT technologies: The paper [6] gave a survey of IoT
the minimum power consumption. platforms, protocols and applications and proposed a
framework integrating different protocols. In the paper [7],
Keywords—IoT; power saving; underwater sensors; authors proposed a wireless temperature monitoring system
wireless communication based on ZigBee module that showed the capability to solve
wiring issues, limited adaptability, and other issues related
I. INTRODUCTION to distributed wireless temperature monitoring system.
Nowadays, there is an increasing demand of battery- Moreover, there are a number of hardware platforms
operated IoT applications in underwater environment. developed for IoT purpose, as listed by [8], [9] and [10].
Indeed, battery power saving and optimization become one
of the major concerns in the IoT sector view that IoT B. Power saving in IoT
devices mostly operate on batteries and need to reach Many researchers have been carried out for power
lifetimes of several years without battery replacement [1]. saving of IoT devices. Methodologies were presented in
Hence, there are extensive researches happening both in [11] [15] [16] to characterize the power consumption of
hardware and networking areas to optimize the power embedded applications and operating systems based on
consumption of IoT devices. high-level specifications and physical measurements in
order to build power models.
Various IoT embedded systems have been studied in
terrestrial, however, they cannot be applied directly to The work [12] proposed an execution driven simulation
underwater because of attenuation of Global Positioning testbed that measures the execution behavior and power
System (GPS) and Radio Frequency (RF) signals consumption of embedded applications and RTOS. A
underwater, making the communication between sensors comprehensive model for the power consumption of
underwater particularly challenging [2]. wireless sensor nodes was examined in [21].
Recently, we proposed novel architectures and The paper [13] conducted a general study on aspects of
realizations that allow to sensors to communicate Operating System design and implementation. They
underwater through acoustic waves [3] [4]. In the present investigated low power modes of embedded devices and
paper, we follow up by proposing an experimented low proposed energy efficient techniques to use operating

978-1-5386-2123-3/17/$31.00 ©2017 IEEE


system functionalities. Furthermore, the energy
consumption analysis of the "C/OS RTOS when running
several embedded applications can be found in [14].
In [17], the influence of cycle-accurate energy
consumption models was presented to explore compiler and
source code optimizations for reducing energy consumption.
The paper [18] proposed a method that selectively and
dynamically changes the membership of nodes and
rearranges their traffic to maximize overall sleeping
intervals without causing delay to data delivery. As well as,
a temporary membership change scheme and a traffic
scheduling algorithm is proposed in order to reduce overall
power consumption. Fig. 1. Schematic diagram of the sensor nodes components

The paper [19] presented a framework to reduce power B. Receiver host architecture:
consumption of M2M / IoT devices by buffering data, The receiver host architecture, presented in Fig. 2,
which keeps IoT devices and gateway nodes in sleep mode incorporates a fixed-point digital signal processor (DSP)
as long as possible. Potsh et al [20] proposed a measurement [24] based on an advanced modified Harvard architecture
solution for tracking the highly dynamic power which provides an arithmetic logic unit (ALU) with a high
consumption of modern wireless embedded systems found degree of parallelism, application-specific hardware logic,
in many IoT devices. on-chip memory, and additional on-chip peripherals.
In [22], a network assistance platform was proposed for The GPS module provides the time synchronization to
saving power consumption of IoT devices. The platform the receiver host. Once immersed, the acoustic pulses
enables a server and smart devices to monitor and control coming from the sensor nodes are received through the
operation modes of the IoT devices. Besides, researchers acoustic receiver chain. The DSP computes the Time of
presented in [23] a comparison of the expected lifetime for Arrival (ToA) of received pulses, interprets the values of
IoT devices operating in several wireless networks: IEEE battery level, temperature and water depth received from
802.15.4/e, Bluetooth Low Energy (BLE), IEEE 802.11 each sensor node, and sends the computed data to the Flash
Power Saving Mode (PSM), IEEE 802.11ah, as well as in MCU which transfers it to the local server through TCP/IP.
new emerging long-range technologies such as LoRa and
SIGFOX.
III. PROPOSED ARCHITECTURE
A. Sensor nodes architecture
The sensor nodes architecture, presented in Fig. 1, is
based on an ultra-low-power consumption flash
microcontroller (Flash MCU) that features a powerful 16-bit
Reduced Instruction Set Computing "RISC" architecture,
16-bit registers, and constant generators that contribute to
maximum code efficiency.
The Flash MCU manages the communication with the
water sensors, the GPS module and the acoustic transmitter.
The water sensors are immersed underwater to measure the
physical characteristics such as the temperature of water and
the depth of the seafloor. A Time synchronization is Fig. 2. Schematic diagram of the receiver host components
required for many reasons: (1) Coordination and
collaboration of sensor nodes need a common timescale. (2) Inter-processor communications were implemented
Sensor nodes need synchronization to manage sleep and between the Flash MCU and the DSP based on the Host
wake up times. (3) Many scheduling algorithms need time Port Interface link (HPI) [25]: A high speed parallel port
synchronization. The GPS module is used to provide the through which the Flash MCU can directly access to DSP
time synchronization to sensor nodes on the surface. The memory space. An HPI software has been developed on the
measured battery level and the sensed temperature and Flash MCU to perform inter-processor communications
depth are transmitted through water to a distant receiver by between the Flash MCU and DSP. The main C functions
means of an acoustic transmitter that incorporates a piezo- that were developed are:
ceramic sensor to transform electrical signals to mechanical
vibrations transmitting through water. HPI_init(): Initialize HPI link.
HPI_read(): Read a 16-bits from DSP memory.
HPI_write(): Write a 16-bits to DSP memory.
HPI_isr(): Handle interrupts from DSP. hardware configuration is not required, TDMA provides a
great deal of flexibility and tthe sensor nodes can be turned
HPI_dspInt(): Interrupt DSP through HPI link. off during idle periods, there
thereby prolonging the battery life.
C. GPS Time Synchronization However, a strict time synchron
synchronization is required amongst
all sensor nodes, which implies that sensor nodes must
The hardware architecture of GPS synchronization, remain synchronized even during node failures or node
shown in Fig. 3, is based on the Flash MCU and a GPS movement in order to maintain the transmissions schedule
schedule.
module. An embedded software has been developed on the
Flash MCU to manage the interface to the GPS receiver Hence, the TDMA is adopted for the actual sy system. The
through Universal Asynchronous Receiver/Transmitter time synchronization is provided by GPS. Moreover, the
(UART) link. This allows to synchronize the clock of each acoustic communications
cations presents a large propagation
sensor in reference to the UTC time, received from GPS delays, thus, each time slot must be long enough to avoid
module.. Because of the disability of GPS underwater, the collisions,, which increases the turn
turn-around time depending
process of GPS time synchronization is performed in
i Air, the number of sensor nodes. Fig. 4 shows an example of
before launching sensor nodes into the water. implementation of TDMA scheduling protocol for five
sensor nodes. The turn-around
around time is defined as the period
between two successive transmittions of the same sensor
node.

Fig. 3. Schematic diagram of the UART link between the GPS and the
Flash MCU.

IV. IMPLEMENTATION OF LOW POWER COMMUNICATION


COMMUNICATI
PROTOCOLS Fig. 4. Diagram showingng the allocation of TDMA slot. Example of five
sensor nodes. The start of turn-around
round time is provided by the GPS time.
A. Acoustic Communication Protocol The TDMA slot number progresses in clock-wise.
Acoustic communication protocols were designed to
provide collision-free
free and fair channel usage. Exhaustive B. Proposed power saving solutions
studies have been performed on three most popular The following three solutions were implemented in the
communication protocols [26] [27] [28]] [29]
[ [30]. (1) actual system.
Frequency Division Multiple Access
ccess (FDMA), (2) Time
Division Multiple Access (TDMA), (3) Code C Division 1) Low power MCU: Reducing the power consumption
Multiple Access (CDMA). of the embedded microcontroller is a key. However,
choosing a low-power
power MCU requires investigation beyond
The FDMA gives advantage of algorithmic simplicity datasheet bullet points. Most MCU silicon vendors have
and efficiency when dealing with a small number of sensor
low-power
power offerings, but choosing the best one for your
nodes in uniformly constant traffic. However, this scheme is
not suitable for underwater sensor nodes,, given the limited application is usually challenging
challenging. The Flash MCU was
bandwidth of underwater acoustic channels and the chosen to comply the lowest
lowest-power features, including
existence of prevalent fading. power-down
down modes, clocking systems, event event-driven
capability, on-chip
chip peripherals
peripherals, pin leakage current, and
The CDMA enables multiple nodes to operate processing efficiency. Temperature and water depth sensors
concurrently over the entire frequency band. Different nodes
were also selected to meet the low power specifications.
access the channel based on unique pseudo seudo noise
noi (PN)
codes. This scheme presents some algorithms
algorithm complexity 2) MQTT Communication protocol
protocol: The receiver host
and a power control algorithm is necessary to reduce the board implements one of the low power communication
output power level of each sensor node in order to establish protocols between the Flash MCU and the local server
server: The
reliable acoustic data transfer without creating excessive (Message Queue Telemetry Transport) MQTT messaging
interference and to improve the battery autonomy.
autonomy protocol, known as a lightweight protocol that optimizes the
The TDMA divides a time interval into multiple time power consumption at the MCU level [31].
slots and uniquely assigns each time slot to an individual 3) Duty cycle: The sensor node operating cycle was
node. In this scheme, data is buffered by each sensor node divided into three modes: (1) Data acquisition mode, during
and transmitted during its assigned time slot. Since which the Flash MCU collects and buffers measurements
from sensors. (2) Data Transmission, when sensor node consumption: The current consumption was measured
transmits its own data in the corresponding
nding TDMA Slot. (3) during the three states: (1) The Flash MCU reads data from
Deep sleep mode, during which the sensor node enters in a temperature sensor. (2) The FFlash MCU transmits data. (3)
low power mode and the majority of peripherals are The Flash MCU enters in deep sleep sleep. Fig. 7 shows the
obtained battery autonomy measurements depending the
switched off. As shown in Fig. 5, thehe current consumption
variation of the turn-around
around time
time. The experimentation
level is commonly higher during Transmission, and very parameters are listed in Table I.
lower during deep sleep mode.
TABLE I. BATTERY AUTONOMY EXPERIMENTATION PARAMETERS

During sleep current 1 milli


milli-Ampers
Transmission current 150 milli-Ampers
Period of transmission 5 seconds
Battery Specifications 3x
3x1.5V, 2050Ah AA Alkaline

3500
3000

Autonomy (Hours)
2500
2000
Fig. 5. Schematic diagram showing the implementation of the duty cycle 1500
mechanism on three the sensor nodes. The Horizontal axis is divided 1000
following the TDMA slots allocations. The vertical axis illustrates the 500
variation of the current consumption during each level.
0

V. EXPERIMENTS
Turn around time (hours)
A. Temperature monitoring
The laboratory setup of temperature monitoring system Fig. 7. Battery autonomy measurements plots (Horizontal axis) depending
the TDMA turn-around time:: from 1 hour to 48 hours (Vertical axis).
contains a Flash MCU, a temperature sensor and a local
server. The Flash MCU collects the sensed
ensed data from the We notice that the battery autonomy increases when the
temperature sensor, and sends measurements to the server turn-around
around time increases. The current consumption of the
through MQTT protocol. Fig. 6 illustrates the plot of a set of sensor nodes depends significantl
significantly to the amount of current
temperature measurements stored by the server database. consumed during transmission.
sion. This confirms the benefits of
These measurements were qualified and validated in implementing the duty cycle modemode, which minimizes the
reference to a calibrated temperature sensor. transmission current. The autonomy becomes less sensitive
to the TDMA turn-around
around variation where values of the
Measurements have been performed to qualify the
turn-around exceed 10 hours
hours. An autonomy of 3000 hours
benefits of using the MQTT protocol compared to the t
could be obtained.
Hypertext Transfer Protocol (HTTP).. The obtained results
showed that the MQTT reduces (by a factor of five) the time C. Battery level measurements
required for transmitting data from the Flash MCU to the
server, which reduces notably the power consumption of the The objective of these measurements is to qualify the
receiver host. ability of the sensor node to monitor the voltage level
supplied by its batteries power supply. Fig. 8 shows the
measured battery discharge rate
rate.

Fig. 6. Plot of temperature measurements stored in the server database.


database
measured by a single temperature sensor during around 25 minutes.

B. Power consumption measurements


Fig. 8. Battery discharge rate obtained with 3 x 1.5Volts 2050Ah AA
Experiments have been performed in order to qualify the batteries, mounted in parallel. The turn
turn-around time was set to one minute.
performance of sensor nodes in terms of power
Based on the obtained measured battery discharge rate,
and with taking into consideration
ation the requirements
requiremen of the
Flash MCU in terms of power supply voltage, a low battery
alarm was implemented on the sensor node in order to
notify the user when the battery voltage of the sensor node
goes low, down to 3.3Volts.. In this case, the sensor node
no
can be automatically turned to a permanent deep sleep
mode.
D. Sea trials setup
The developed system was tested during sea trials, trials
performed in BOUREGREG MARINA, located at the
mouth of the Bouregreg River, on the shore of SALE,
Morocco. This location provides ides an easy-to-access
easy
environment for experimental tests (Fig. 8). The sensor node Fig. 10. Schematic diagram
iagram showing the receiver host setup.
and the receiver host were spaced with a maximum range of
500 meters. E. Water depth measurements
Fig. 11 shows a set of water depth measurements logs,
obtained at the Marina of Bouregrag, by means of an
ultrasonic water depth sensor, connected to the sensor node,
and immersed at 520 centimeters.

Fig. 9. Map image showing the location of the experimental setup

Schematic diagrams of the experimental


erimental platform are
shown in Fig. 9 and Fig. 10. The sensor node electronic Fig. 11. Experimental
xperimental measurement of water depth. The obtained depth
board was set on the deck, while its watertight
water acoustic value is read at the vertical axis in centimeter, the horizontal axis gives the
transmitter ceramic was immersed at 2 meters depth. The number of measurements. The true depth (reference) is estimated around
520 centimeters.
receiver host electronic
lectronic board, connected to a Laptop, was
also set on the deck, 280 meters far away from the sensor Globally,, the obtained measurements are slightly close
node. The acoustic receiver antenna was immersed at a to the true depth (reference) an
and the accuracy was estimated
depth of 2 meters. the sensor node and the receiver host around +/- 140 centimeters.. However, the implementation
were put in line of sight. The water sensors of the sensor of a more accurate depth sensor is the subject of our future
node is immersed at about 520 centimeters. works.
VI. CONCLUSION
Reducing the power consumption of the embedded
applications is becoming more and more import
important. This
paper proposed a real time, low power and low cost IoT
embedded system architecture, applied to wireless
underwater sensing.. The proposed system is battery
powered and implements a set of embedded interfaces that
facilitate its integration to und
underwater monitoring systems
like Autonomous Underwater Vehicles and Sonar diving. So
far, we focalized on some of the most important points of
IoT platforms and sensor networks
networks: Water depth and
temperature sensing, synchronization between sensors using
Fig. 1 Schematic diagram
iagram showing the sensor node test setup. GPS and the implementation of power saving mechanisms
in order to minimize the power consumption of the realized
system. The proposed design integrates a combination of an
ultra-low-power MCU, the MQTT communication protocol
and the duty cycle mechanism. The Laboratory and sea [14] R. P. Dick, G. Lakshminarayana, A. Raghunathan, and N. K. Jha,
Power analysis of embedded operating systems. Proceedings of the
water tests results illustrated the ability of the system to 37th Conference on Design Automation, pp. 312–315.
gather the sensed depth and temperature measurements to a [15] T. K. Tan, A. Raghunathan, and N. K. Jha, Energy macromodelling
local host server, installed on a Raspberry PI unit. In our of embedded operating systems. ACM Transactions on Embedded
future work, we are intending to develop a complete IoT Computing Systems 4, 231 (2005)
platform, that integrates both easy access to data by the end [16] Hyvärinen, J. Methods for Energy Modelling of Embedded Operating
user, and the scalability to add more sensors such as System.
underwater cameras, underwater GPS, and more. [17] Simunicy, T., Benini, L., & De Micheliy, G. Energy-E cient Design
of Battery-Powered Embedded Systems.
ACKNOWLEDGMENT [18] Kim, T., & Chang, J. M. (2017). Enhanced Power Saving Mechanism
for Large-Scale 802.11 ah Wireless Sensor Networks. IEEE
Authors would like to thank the Bouregrag Marina staff, for Transactions on Green Communications and Networking.
giving us permission and support to conduct sea trials. [19] Taneja, M. (2014, September). A framework for power saving in IoT
networks. In Advances in Computing, Communications and
REFERENCES Informatics (ICACCI, 2014 International Conference on (pp. 369-
375). IEEE.
[1] Mike Mitchel, “Choosing An Ultralow-Power MCU”, Texas
Instrument, Jun 2004. [20] Pötsch, A. Berger and A. Springer, "Efficient analysis of power
consumption behaviour of embedded wireless IoT systems," 2017
[2] Akyildiz, I.F.; Pompili, D.; Melodia, T. Underwater acoustic sensor IEEE International Instrumentation and Measurement Technology
networks: Research challenges. Ad Hoc Netw. 2005, 3, 257-279 Conference (I2MTC), Turin, 2017, pp. 1-6
[3] R. Es-sadaoui, J. Khallaayoune and T. Brizard, "A Real Time [21] Martinez, B., Montón, M., Vilajosana, I., & Prades, J. D. (2015). The
Embedded System Architecture for Autonomous Underwater Sensors power of models: Modeling power consumption for IoT
Localization" Transactions on Machine Learning and Artificial devices. IEEE Sensors Journal, 15(10), 5777-5789.
Intelligence. Volume 5, No 4, August 2017, pp. 499-508. doi:
10.14738/tmlai.54.3224. [22] H. Park, E. S. Jung, W. Lee, H. Lee and Y. T. Lee, "Network
assistance platform for saving power consumption of IoT devices and
[4] R. Es-sadaoui, J. Khallaayoune and T. Brizard, "Design and set-top boxes," 2017 IEEE International Conference on Consumer
performance measurement of a digital embedded system Electronics (ICCE), Las Vegas, NV, 2017, pp. 316-318
implementing a I/Q demodulator for AoA range-based localization in
underwater environment" 2017, In Proceedings of the 1st [23] É. Morin, M. Maman, R. Guizzetti and A. Duda, "Comparison of the
International Conference on Smart Digital Environment (pp. 107- Device Lifetime in Wireless Networks for the Internet of Things,"
115). doi: 10.1145/3128128.3128145. ACM. in IEEE Access, vol. 5, no. , pp. 7097-7114, 2017.
doi: 10.1109/ACCESS.2017.2688279.
[5] T. Ahmad, X.J. Li, B-C. Seet, " Parametric Loop Division for 3D
Localization in Wireless Sensor Networks," Sensors 2017, 17(7), [24] Texas Instruments, “TMS320C54x DSP Functional Overview”.
1697; doi:10.3390/s17071697. September 1998 – Revised May 2000.
[6] S. H. Shah and I. Yaqoob, "A survey: Internet of Things (IOT) [25] Texas Instruments, “TMS320C54x DSP Reference Set, Volume 5:
technologies, applications and challenges," 2016 IEEE Smart Energy Enhanced Peripherals”. March 2007.
Grid Engineering (SEGE), Oshawa, ON, 2016, pp. 381-385. doi: [26] T. Watteyne, S. Lanzisera, A. Mehta, and K. Pister, “Mitigating
10.1109/SEGE.2016.7589556 Multipath Fading Through Channel Hopping in Wireless Sensor
[7] Lai Wei et al., "Wireless temperature monitoring system based on Networks,” in IEEE International Conference on CoMmunications
FBG and Zigbee," 2016 15th International Conference on Optical (ICC). Cape Town, South Africa: IEEE, 23-27 May 2010.
Communications and Networks (ICOCN), Hangzhou, 2016, pp. 1-3. [27] T. Watteyne, A. Mehta, and K. Pister, “Reliability Through
doi: 10.1109/ICOCN.2016.7875806 Frequency Diversity: Why Channel Hopping Makes Sense,” in 6th
[8] K. J. Singh and D. S. Kapoor, "Create Your Own Internet of Things: ACM International Symposium on Performance Evaluation of
A survey of IoT platforms.," in IEEE Consumer Electronics Wireless Ad Hoc, Sensor, and Ubiquitous Networks (PE-WASUN),
Magazine, vol. 6, no. 2, pp. 57-68, April 2017. doi: Tenerife, Canary Islands, Spain, 26-30 October 2009.
10.1109/MCE.2016.2640718 [28] IEEE Standard for a Precision Clock Synchronization Protocol for
[9] A. Polianytsia, O. Starkova and K. Herasymenko, "Survey of Networked Measurement and Control Systems, IEEE Instrumentation
hardware IoT platforms," 2016 Third International Scientific- and Measurement Society Std. IEEE Std 1588- 2008, 24 July 2008.
Practical Conference Problems of Infocommunications Science and [29] J. Han and D.-K. Jeong, “A Practical Implementation of IEEE 1588-
Technology (PIC S&T), Kharkiv, Ukraine, 2016, pp. 152-153. doi: 2008 Transparent Clock for Distributed Measurement and Control
10.1109/INFOCOMMST.2016.7905364 Systems,” IEEE Transactions on Instrumentation and Measurement,
[10] A. Raza, A. A. Ikram, A. Amin and A. J. Ikram, "A review of low vol. 59, no. 2, pp. 433–439, February 2010.
cost and power efficient development boards for IoT applications," [30] S. Lanzisera, D. Zats, and K. S. J. Pister, “Radio Frequency Time-of-
2016 Future Technologies Conference (FTC), San Francisco, CA, Flight Distance Measurement for Low-Cost Wireless Sensor
2016, pp. 786-790. doi: 10.1109/FTC.2016.7821693 Localization,” IEEE Sensors Journal, 2011.
[11] Dhouib, Saadia, et al. "Energy and power consumption estimation for [31] "MQTT 3.1.1 specification". OASIS. December 10, 2015. Retrieved
embedded applications and operating systems." Journal of Low April 25, 2017.
Power Electronics. 5.4 (2009), 416-428.
[12] Baynes, K., Collins, C., Fiterman, E., Ganesh, B., Kohout, P., Smit,
C., ... & Jacob, B. (2003). The performance and energy consumption
of embedded real-time operating systems. IEEE transactions on
computers, 52(11), 1454-1469.
[13] Vahdat, A., Lebeck, A., & Ellis, C. S. (2000, September). Every joule
is precious: The case for revisiting operating system design for energy
efficiency. In Proceedings of the 9th workshop on ACM SIGOPS
European workshop: beyond the PC: new challenges for the
operating system (pp. 31-36). ACM.

Das könnte Ihnen auch gefallen