2011 Faible Tension Faible Consommation (FTFC)

Evolution of wireless sensor networks and necessity of

power management technique
Rym Cheour1, Khawla Lahmar2, Mohamed Abid3
CESlab, National School of Engineers of Sfax, Sfax, Tunisia
1
[email protected]
2
[email protected]
3
[email protected]

Abstract—Wireless sensor networks interfere in a growing lifetime, significant research has been undertaken to increase
number of applications ranging from simple environmental lifetime while still meeting functional requirements. There are
monitoring like temperature detection to complex calculation many techniques to reduce power consumption [1]. This paper
such as video processing. This last type of application requires a presents an overview of some of the key areas and research in
high load at the sensor level and leads to a problem of wireless sensor networks related to energy savings.
optimization. It is critical to design algorithms and protocols in The remainder of the paper is organized as follows: we use
such a way to use minimal energy. This paper provides a survey examples of recent work to portray the state of art of the
of WSNs technologies, main applications and hardware
evolution those networks. In particular, we discuss the
evolutions. In the other hand, we will provide an insight about
the latest trends of the technique to reduce the power progress made in the fields of application, the architecture and
consumption within the WSN that could possibly make this the platforms in the second section. Besides we give a glance
emerging technological area more useful than ever. about the progress known by the architecture and the
Keywords—Wireless Sensor Networks; Power consumption; platforms of WSN. We have proposed also many innovative
Platforms; Applications; Dynamic Voltage and Frequency applications of wireless sensor networks. We also attempt an
Scaling; Dynamic Power Management. investigation also into pertaining energy constraints. Section 3
states the deployment of the power management technique to
meet the best of our needs. We conclude in section 4.
I. INTRODUCTION
Sensor area is growing at an accelerated pace, attracting more II. EVOLUTION OF WSN
and more people to its use, and is becoming one of the Wireless sensor network (WSN) is an exciting new
subjects of predilection of researchers. The imminent progress technology that has attracted extensive interest. The WSN is
made in MEMS (Micro Electro Mechanical systems) coupled subject for various resource constraints which are memory,
with wireless communication and digital electronic has led to processing ability, bandwidth, and mainly energy efficiency
the massive deployment of tiny size, low cost, low power, [2]. To overcome those many possible architecture and new
multifunctional and high performance sensor nodes[1]. Thus, platforms have been defined offering more reliability and
the network technologies of wireless sensor have become a broadening the spectrum of deployment of these networks.
global trend in communication, mobility and research of We try to provide next an overview of the recent advances of
flexible implementation. The first challenge of a wireless architecture, in the platforms and the field of application.
sensor network is to ensure optimal performance for users and
A. Hardware architecture
to save the energy consumed by sensor nodes [2]. Widespread
in hostile environments and equipped with limited resources, A Wireless Sensor Network is a self-configuring network of
this network sees reducing its performances and confines itself small sensor nodes communicating among themselves using
to some applications. Once deployed, it is often impossible or radio signals, and deployed in quantity to sense, monitor and
inappropriate to recharge sensor nodes or replace their understand the physical world [4]. A sensor node is typically
batteries [3] [2]. made up of four basic components which are the sensing unit,
Recently, wireless sensor networks (WSN) are expected to the processing unit, the transceiver section and the power
carry out high functionalities in various scopes. It’s more supply unit as illustrated in figure 1. The motes or the sensor
nodes have to provide information anytime and anywhere to a
evident now to support heterogeneous WSN’s applications
final user, by collecting, processing, analysing, and
with different sensors: temperature, pressure, image, sound,
disseminating data [5]. The recent advances of wireless sensor
video instead of sampling temperature and sending the value
networks create new opportunities for innovative applications,
to a base station. Unfortunately, energy remains the major
but poses new technical challenges for constructing such
handicap in front of this progress due to limit capacity of applications. In fact, new types of sensors are now
node’s batteries. Therefore, the energy optimization becomes commercially available and could be broadly classified as
more and more a necessity for sustaining a sufficiently long pressure, temperature, light, biological, chemical, strain,
network lifetime [4]. Since most systems require much longer

978-1-61284-647-7/11/$26.00 ©2011 IEEE 75

fatigue, tilt sensors, humidity, vibration, microphone, camera generation of platforms that are equipped with a camera, a
and so on[6]. microphone, an earpiece for dedicated applications that
integrate audio or video treatments. Also treatment has
become increasingly complex and requires more resources
and mainly better power management than sleep-policies. In
this context, new platforms (imate2) are equipped with
processors with management policy dynamic frequency and
voltage (DVFS). Indeed, the functioning of a processor is not
constant; it consists of treatment periods that require
computing variable power.
Figure 1 A sensor node components
C. Applications of Sensor Networks
Today, the lack of appropriate tools and methodology that are
able to handle the imminent progress made in WSN is an WSN have become an equipment key in many industrial
obstacle to potential uptake. That’s why this progress requires applications. Indeed, WSN have proved their efficiency in
inherent evolution in the hardware architecture of the motes. monitoring, tracking, or controlling phenomena where wiring
The nodes forming the WSN are now equipped with more is difficult or cost-prohibitive. In terms of areas of application,
processors and are more powerful (4Mhz mica, 12Mhz with sensor networks have been extremely successful, because they
Imote2) and a greater memory. As it is shown in Table1, new hold a potential which revolutionized many sectors of our
platforms have 32Mbyte as RAM memory and the same economy and our daily life, such as the monitoring and
capacity as flash memory. In addition, they support dynamic preservation of the environment, the industrial manufacturing,
voltage and frequency scaling (DVFS) technique thanks to the passing through the automation in the sectors of transport and
range of frequency of processor. health, modernization of the medicine, the agriculture, the
telematics, and the logistics [10]. Also they invade military
Table 1. Evolution of hardware resources of motes and many other different scoops that were unthinkable several
decades ago such as cellular phones, GPS units, television
WeC Dot Mica2 Imote1 Imote2 Shimmer Cookies Egs [9]
[7] [8] satellite, remote sensing and the internet [5]. Far away from
2008 2008 2010
simple application, WSN set up today applications more
development 1999 2001 2003 2005 2007
complex with heterogeneous sensors. We have attempt next to
Processor 4 7 12-48 13-400 8 96
give a glance about the latest trends of application of WSN.
(MHz)
1) Habitat Monitoring: As an example, we cite an
48KB
+external
4 Mbit 256 KB
+external
application for firefighters proposed in the project GEODE,
Flash 8K 16K 128K 512K 32MB
memory
2GB
memory
2GB
comprising nodes for sampling temperature, smoke detection
4 Kbytes and motion as well as cameras in a building. On the other
RAM 0.5K 1K 4K 64K 32MB 10KB + 62 52KB hand, we can also equip firefighters with their nodes to
Kbytes
Bluetooth ETRX2 CC2520 : determine areas where there are people in danger or simply
and TELEGE 802.15.4
Transceiver RFM ChipCon Bluetooth Zeegbee 802.15.4 SIS and 2 determining the output once it is in the building on fire[11].
(WML- Bluetooth 2) Agriculture: In addition, WSNs have the potential for
C46A, modules
CC2420)
MSP430 ADUC84 ARM widespread application in precision agriculture, particularly in
ARM XScale
processor 1+
Spartan
Cortex
M3 the areas of crop and irrigation management, planned
µC Amtel 7TDMI PXA271 3FPGA fertilization and pest control, variable rate chemical input
application and modeling crop performance[12].
3) Civil infrastructures: Once deployed in offshore
The heterogeneity of the nodes in the WSN as well as the fields, the use a low-cost WSN presents a major advantage
power resources, help not only to improve the capabilities of when it comes to feeding measurement and communication
the WSN, but also to limit its lifetime. The lifetime of a node equipment data to central units. The WSN can be retrofitted
may be determined by the battery life. The network lifetime is on offshore platforms that have hundreds of pumps, fans, and
defined as the maximum number of times a certain data other motor driven devices requiring permanent
collection function or task can be carried out without any node maintenance[13].
running out of energy [4]. This is why implementing dynamic 4) Medicine: Bio-sensors are implanted in the human
power management techniques become more and more body to monitor the patient's physiological parameters such as
obligatory. heart beat or blood pressure. The data so collected is sent
regularly to alert the concerned doctor on detection of an
anomaly[2].
B. Platforms of Sensor Networks
5) Environment monitoring: These sensors are very
The WSN platforms (mica2, micaz, etc.) used to contain just a useful for environmental applications also. As a network, they
temperature or pressure sensor and they were restricted to the have been used to detect environmental hazards such as
transmission of the captured information to a third element. earthquakes and floods. Some motes have even deployed to
Recently, the proliferation of the WSN has led to new analyse remote locations, observing the motion of a tornado,

76
or detecting fire in a forest [5]. Sensors that are used also for is proportional to the consumption, the techniques of code’s
satellite oceanographic observations, records the amount of optimization can help to minimize this consumption by
incident energy returned from the imaged surface. This reducing the number of access to the external memory. But as
application provides a wealth of information on a diverse the optimization of these codes is manually difficult to
range of geophysical and biological parameters and implement, we can use other techniques for automatic
phenomena [14]. optimization that intervenes at the level of the code generator.
6) Logistic: The WSN nodes are attached to goods Among the automatic optimization techniques, we can cite
which are mostly food due to their perishable nature. The inlining and loop method (copy the code in a function where it
goods are loaded from a warehouse to a freight vehicle, in is mentioned instead of using a procedure of function call) and
which their nodes need to self-organize and form a network of the unrolling technique(Copy the code several times)[3].At the
nodes, which can deliver information of the goods’ state to the level of communication, we can cite low power listening,
outside world using a gateway [15]. communication scheduling, power-aware routing etc. At the
operating system level, the Operating System (OS) for sensor
III. TECHNIQUE TO REDUCE THE ENERGY CONSUMPTION nodes could implement a low-power task-scheduling which
WSN are subject to many resource constraints which are take advantage of nonlinear battery effects to reduce energy
mainly closely related to the energy consumption. The nodes consumption. At the computation level, the operating system
contain a conventional battery, supplemented by a renewable powers down the processor between events when idle which
source that generates power using scavenging techniques reduces considerably the wasted energy[19].
(vibration, solar, EM, piezoelectric, radioactive, etc.)[16].
C. Hybrid
However, a sensor node powered by 2 AA batteries can last
for up to three years with a 1% low duty cycle working mode At the software level power management solutions are
[4]. targeted to minimize the communications since transmitting
To extend the autonomy of a system, only two methods exist: and listening to messages is energy expensive, and to create
increasing the capacity of the on-board energy or reducing the sleep/wake-up schedules for nodes or particular components
consumption of the system. The first solution is a subject of of nodes [2]. Indeed, this power manager allows us to adapt
research related to the field of batteries. However, it is always the energy resources in an automatic way and to reduce the
difficult to increase the capacity of a battery without power consumption of the WSN. Dispose of this utility offers
increasing the weight, volume and price. Notwithstanding, the various advantages such as a structure in which nodes can
battery technology is not progressing fast enough to satisfy intelligently manage their own power and a module which has
those requirements [2]. The second solution has also led to a global view across the network, the power and the sensor
several researches in the field of electronics and computer nodes states. Each component in the node can have different
science. Different techniques can minimize the energy power modes. A task can be active, slowed down, or in sleep
consumption of WSN. In the literature, there are several mode: radio can transmit, receive or in standby or other mode
methods that allow us to get a system with low consumption. [20]. Each sleep state of the node corresponds to a particular
Typically, they are divided into three categories. combination of components of power. An interesting aspect of
this concept is to take advantage of the static and dynamic
A. Hardware Level resources management algorithms. Those policies refer
In fact, the power being proportional to the square of the mainly to the DPM (Dynamic Power Management) technique
voltage, we can think that the most effective one is to lower it and to DVFS (Dynamic Voltage and Frequency Scaling).
[16]. However, lowering this parameter also leads to a
1) DPM: The basic idea of DPM is to stop devices
reduction of the threshold voltage Vth which increases the
when they are not required and to wake them up when they
leakage current and therefore the static power. Because our
are. Test-bed developed in cooperation with the start-up
system is often in standby mode, it is therefore necessary to
company SeNet s.r.l. for agricultural monitoring has shown
find a compromise between the reduction of the supply
that in sleeping mode, node consumption is only 0.5 mA,
voltage and the threshold voltage. On the other hand, we can
whereas during the activity mode the overall consumption is
reduce the output capacity at the layout level, to the sizing
30 mA[17]. Currently, the DPM is done thanks to the ability
gates level and also at the architectural level [17]. There is
of the hardware to support mechanisms of sleep state ranging
also the opportunity to try to reduce the frequency of
from total activity of the system up to full sleep
activation at the architectural level partitioning the circuit
implementation or disconnection of the system. Indeed, idle
block, enable those who are necessary for the application and
state transitions and implementation cost is running a bit
put the rest in standby mode. It is possible to add solar cells or
expensive in the point of view of energy. In addition, the node
scavenge energy from motion or wind[16].
must have a progressive quality scalability of energy so that
B. Software Level its lifetime will be extended if the application requires, this is
Numerous system software techniques have achieved to achieved at the expense of accuracy[3].
extend the WSN lifetime such as duty-cycling subsystems, 2) Dynamic voltage and frequency scaling: Approaches
batching operation, reducing redundancies, low-power to deal with the reduction of energy consumption are various.
software approaches[18].Given that the number of instruction We will give a synthesis of those methods. Pillali article [16]

77
introduced the dynamic management of tension as a key for within the nodes. We focused also in the DVFS and DPM
the physical characteristics of the processor to reduce the techniques.
energy dissipation by minimizing the supply voltage and the
frequency. These algorithms have proved their effeicency by REFERENCES
offering a considerable energy gain, while ensuring the [1] D. Puccinelli and M. Haenggi, ―Wireless sensor networks: Applications
performance of the real time embedded system. and challenges of ubiquitous sensing‖ IEEE Circuits Syst. Mag., vol. 5,
no. 3, pp. 19–29, Third Quarter, 2005.
Notwithstanding this, we can not deny their impact in the [2] I. F. Akylidiz, W. L. Su, Y. Sankarasubramaniam and E. Cayirci,
execution time. Thus, taking into account the time constraints "Wireless Sensor Networks: A Survey on Sensor Networks," IEEE
is necessary. Consequently, he introduced a new algorithm Communications Magazine, Vol. 40, No. 8, 2002, pp. 102-114.
"Real Time Dynamic Voltage Scaling" which modifies the [3] R.Cheour, S. Bilavarn, M. Abid, ―EDF scheduler technique for
wireless sensors networks: case study‖, in Proc. Fourth International
operating system real time and task management. The article Conference on Sensing Technology, Lecce, Italy, June, 2010, paper 93,
written by Sinha and Chandrakasan [21] proposes the use of p.530.
power aware methodology. They employ a micro incorporated [4] Yang Yu, Viktor K Prasanna, B. Krishnamachari, ―Information
operating system to reduce the energy consumption of the processing and routing in wireless sensor networks‖, chapter1:
Introduction to Wireless Sensor Networks, pp.204, Dec 2006
node by exploiting the node sleep state and active power [5] Ituen, G. Sohn, ―The Environmental Applications of Wireless Sensor
management. Networks”. International Journal of Contents, 3, pp :1-7. 2007.
[6] N. Jain and D. P. Agrawal, "Current Trends in Wireless Sensor
The appearance of variable voltage processors has led to Network Design", International Journal of Distributed Sensor
greater autonomy and energy savings. Dynamic voltage and Networks, vol. 1, January 2007, pp.101-122
frequency scaling (DVFS) is an effective technique for [7] http://www.shimmer-research.com/r-d/ hardware, February 2011
[8] Y. E. Krasteva, J. Portilla, J. M. Carnicer, E. de la Torre, T.
reducing CPU energy. The DVFS tries to combine the Riesgo, ‖Remote HW-SW Reconfigurable Wireless Sensor Nodes‖,
performance and the lifetime of the battery. Lowering only the Centro de Electronica Industrial, Universidad Politecnica de Madrid,
operating frequency can reduce the power consumption but [9] Q.Wang, T.Schmid, W.Hofer, P.Dutta, A.Terzis, J.Gil Ko ―Egs: A
the energy consumption remains the same because the Cortex M3-based Mote Platform” , IEEE Communications Society
Conference on Sensor, Mesh and Ad Hoc Communication and
computation needs more time to finish. Lowering the supply Networks (SECON) Boston, MA,2010.
voltage can reduce a significant amount of energy because of [10] Holger Karl and Andreas Willig. ―Protocols and Architectures for
the quadratic relation between power and voltage. Lowering Wireless Sensor Networks‖. Publisher : John Wiley & Sons,
the supply voltage and operating frequency reduces the power Chichester, 2005.
[11] GEODE project :http://geodes.ict.tuwien.ac.at/ February 2011
and energy consumption further[19]. A number of modern [12] López Riquelme, J. A., F. Soto, J. Suardíaz, P. Sánchez, A. Iborra, and
microprocessors such as Intel’s XScale and Transmeta’s J. A. Vera 2009. ―Wireless Sensor Networks for precision horticulture‖
Cruso are equipped with the DVFS functionality[3]. The first in Southern Spain. Computers and Electronics in Agriculture, 68(1),
feature of this technique provides high performance only for a pp. 25-35.
[13] Aakvaag, N., Mathiesen, M & Thonet, G. 2005. ―Timing and Power
short time reduced, while the rest of the time a low CPU issues in Wireless Sensor Networks :an Industrial test Case‖. Proc. of
power is largely sufficient [16] [20]. Most micro-processor the International Conference on Parallel Processing Workshops
systems are characterized by a time-varying computational (ICPPW’05), 8 p,2005.
load. DVFS exploits the CMOS property that a linear [14] Mathioudakis, N. White and N. Harris; ―Wireless Sensor Networks:
applications utilizing satellite Links‖, The 18th IEEE International
reduction in the supply voltage results in a cubic reduction in Symposium on Personal Indoor and Mobile Radio Communications
the power consumption at the expense of a linear slow down (PIMRC ’07), Athens, Greece 2007.
in the processor frequency [21]. It is better thus to run the [15] Holger Karl and Andreas Willig. ―Protocols and Architectures for
processor at the weakest frequency compatible with the Wireless Sensor Networks‖. Publisher : John Wiley & Sons,
Chichester, 2005.
necessary performance level. When used at a reduced [16] P.Pillai and K. G. Shin. Real-time Dynamic Voltage Scaling for Low
frequency, the processor can operate at a lower supply voltage. Power Embedded Operating Systems. In Proceedings of the 18th
As wireless sensor networks interfere in a growing number of Symposium on Operating System Principles, October 2001.
applications ranging from simple environmental monitoring [17] Kahwla Lahmar, ―Study and validation of a energy simulation
environment in the WSN‖, Master degree, National School of Engineers
like temperature detection to complex calculation such as of Sfax, Tunisia, 2007.
video processing. This last type of application requires a high [18] P. K. Dutta and D. E. Culler. ―System software techniques for low-
load at the sensor level and leads to a problem of optimization. power operation in wireless sensor networks‖. ICCAD, 2005.
It is in this perspective that EDF DVFS technique justifies [19] D. Rakhmatov, S. Vrudhula, C. Chakrabarti, ―Battery conscious task
sequencing for portable devices including voltage/clock scaling‖, in
itself. Proc. of the 39th Design and Automation Conference, New Orleans,
2002.
IV. CONCLUSION [20] John A. Stankovic, ―Wireless Sensor Networks‖, University of
Wireless sensor networks have been widely used in many Virginia, Technical report ,June,19,2006
[21] Amit Sinha, Anantha Chandrakasan ―Dynamic Power
areas. They provide endless opportunities, but at the same Management in Wireless Sensor Networks‖, IEEE Design & Test
time pose several challenges, such as the fact that energy is a of Computers, April 2001.
scarce and usually non-renewable resource. The aim of this
paper is to discuss some of the most relevant issues of WSNs,
from the application, design and technology point of view. We
covered also the techniques reducing the power consumption

78