Energy-Efficient Battery-Aware MAC protocol for

Wireless Sensor Networks

by

Yamen Nasrallah

A Thesis submitted to the Faculty of Graduate and Postdoctoral Studies In partial

fulfillment of the requirements for the degree of

Master in Applied Science

in

Electrical and Computer Engineering

Ottawa-Carleton Institute for Electrical and Computer Engineering

School of Electrical Engineering and Computer Science

University of Ottawa

Ottawa, Ontario, Canada

© Yamen Nasrallah, Ottawa, Canada, 2012

 Abstract

Wireless sensor networks suffer from limited power resources. Therefore, managing the energy

constraints and exploring new ways to minimize the power consumption during the operation of

the nodes are critical issues. Conventional MAC protocols deal with this problem without

considering the internal properties of the sensor nodes’ batteries. However, recent studies about

battery modeling and behaviour showed that the pulsed discharge mechanism and the charge

recovery effect may have a significant impact on wireless communication in terms of power

saving. In this thesis we propose two battery-aware MAC protocols that take benefit of these

factors to save more energy and to prolong the lifetime of the nodes/network without affecting

the throughput. In both protocols we measure the remaining battery capacity of the node and use

that measurement in the back-off scheme. The first protocol gives the nodes with higher

remaining battery capacity more priority to access the medium, while the other one provides

more medium access priority to the nodes with lower remaining battery capacity. The objective

is to investigate, through simulations, which protocol reduces the power consumption of the

nodes, improve the lifetime of the network, and compare the results with the CSMA-CA

protocol.

ii
 Acknowledgements

First and foremost I want to thank my supervisor Professor Hussein Mouftah. It has been an

honor to be his student during the masters’ program, and it will be a great pleasure to continue

my doctoral study under his supervision. I am grateful for all his contributions of time, ideas, and

funding to make my masters’ experience productive and stimulating.

I offer my sincerest gratitude to my co-supervisor, Dr. Mouhcine Guennoun , who has supported

me throughout my thesis with his patience and knowledge. I highly appreciate his

encouragement, effort, guidance and support. Without him this thesis would not have been

completed in such short time. One simply could not wish for a better or friendlier supervisor.

Special thanks also to my cousin and brother Mounib Khanafer for the valuable suggestions and

advises that he provided.

My deepest appreciation goes to my family for their unflagging love and support throughout my

life. I am indebted to my father who worked hard to support the family and spare no effort to

provide the best possible environment for me to grow up and achieve this scientific level. I have

no suitable expression that can sincerely describe the everlasting love that my mother covered

me with in my whole life. I have no appropriate word that can deeply express my feelings and

gratefulness to her. My warmest thankfulness goes to my brother and sisters, Bilal, Maysaa and

Manal. I am so blessed to have you in my life.

I would like to convey my heartfelt thanks to my loving, supportive, encouraging and patient

wife Sajida Khanafer. I will never forget your support and sacrifices at every stage of my study.

iii
To my babies, Wared and Rawand, who represent the source of my happiness, the origin of my

joy, and to whom I dedicate my whole life, I offer a heart full of endless care and love.

Last but not least, Thanks be to God, the merciful and the compassionate, for providing me the

guidance and the power in every step in my life. May your name be exalted, honored and

glorified.

iv
 Table of contents

Abstract ..................................................................................................................................... ii

Acknowledgements ................................................................................................................... iii

List of tables ............................................................................................................................. ix

List of figures ..............................................................................................................................x

List of Acronyms ..................................................................................................................... xii

List of notations ........................................................................................................................ xv

Chapter 1 Introduction ...........................................................................................................1

1.1 Background...................................................................................................................1

1.2 Motivation and Objective ..............................................................................................4

1.3 Thesis contributions ......................................................................................................6

1.4 Thesis outline ................................................................................................................7

Chapter 2 State-of-the-art of Energy-Efficient MAC Protocols ...............................................9

2.1 Introduction ..................................................................................................................9

2.2 Models of real battery behaviours ............................................................................... 10

2.3 Power-Aware MAC protocols ..................................................................................... 16

2.3.1 TDMA-based energy-efficient MAC protocols ......................................................... 17

2.3.2 CSMA-based energy-efficient MAC protocols .......................................................... 20

v
 2.4 Battery-aware energy-efficient MAC protocol............................................................. 25

2.5 Summary .................................................................................................................... 35

Chapter 3 Battery Characteristics and CSMA-CA Modeling ................................................ 36

3.1 Introduction ................................................................................................................ 36

3.2 Charge Recover Mechanism........................................................................................ 37

3.2.1 Introduction .............................................................................................................. 37

3.2.2 Basics of a battery..................................................................................................... 37

3.2.3 Battery discharge behaviour ...................................................................................... 39

3.2.3.1 Constant current discharge ................................................................................. 39

3.2.3.2 Pulsed current discharge and charge recovery mechanism .................................. 40

3.2.4 Markov chain representation ..................................................................................... 42

3.2.5 WSNs and battery operations .................................................................................... 44

3.3 CSMA-CA Overview .................................................................................................. 44

3.3.1 Introduction .............................................................................................................. 44

3.3.2 CSMA-CA Communication Modes .......................................................................... 45

3.3.2.1 Beacon enabled mode ........................................................................................ 46

3.3.2.2 Non-beacon enabled mode ................................................................................. 47

3.3.3 CSMA-CA operation ................................................................................................ 48

3.3.3.1 CSMA-CA in a beacon enabled network ............................................................ 49

3.3.3.2 CSMA-CA in a non-beacon enabled network ..................................................... 51

vi
 3.3.4 Markov-chain Modeling ........................................................................................... 52

3.3.4.1 System description ............................................................................................. 53

3.3.4.2 Mathematical equations...................................................................................... 54

3.4 Summary .................................................................................................................... 58

Chapter 4 Battery-Aware CSMA-CA Protocol ..................................................................... 59

4.1 Introduction ................................................................................................................ 59

4.2 System description ...................................................................................................... 60

4.3 Algorithm ................................................................................................................... 61

4.4 Results and validation ................................................................................................. 68

4.4.1 Average charge recovery .......................................................................................... 68

4.4.2 Average lifetime of a node ........................................................................................ 69

4.4.3 Lifetime of the network ............................................................................................ 70

4.5 Summary .................................................................................................................... 72

Chapter 5 Battery-Aware Energy-Efficient MAC Protocols ................................................. 74

5.1 Introduction ................................................................................................................ 74

5.2 Our proposed protocols ............................................................................................... 75

5.2.1 MAC-HV Protocol.................................................................................................... 77

5.2.1.1 Flow chart .......................................................................................................... 78

5.2.1.2 Algorithm .......................................................................................................... 79

5.2.2 MAC-LV Protocol .................................................................................................... 82

vii
 5.2.2.1 Flow chart .......................................................................................................... 83

5.2.2.2 Algorithm .......................................................................................................... 83

5.3 Simulation and performance analysis .......................................................................... 86

5.3.1 Simulation Parameters .............................................................................................. 88

5.3.2 Channel utilization .................................................................................................... 89

5.3.3 Percentage of collisions ............................................................................................ 90

5.3.4 Channel idle time ...................................................................................................... 91

5.3.5 Average energy recovered......................................................................................... 92

5.3.6 Average lifetime of a node ........................................................................................ 92

5.3.7 Lifetime of the network ............................................................................................ 93

5.4 Summary .................................................................................................................... 95

Chapter 6 Conclusions and Future Work .............................................................................. 97

6.1 Summary and Concluding Remarks ............................................................................ 97

6.2 Future Research .......................................................................................................... 98

Bibliography ........................................................................................................................... 102

viii
List of tables

Table 2-1. Contents of the control message ............................................................................... 19

Table2-2BatteryTable ...................................................................................................................33

Table 2-3 Election Control Frame ............................................................................................. 32

Table 3- 1 Piecewise constant function of the charge units delivered ......................................... 44

Table 5-1 Simulation parameters ............................................................................................... 88

Table 5-2 Simulation parameters ............................................................................................... 89

ix
List of figures

Figure 2-1. Binary pulsed discharge .......................................................................................... 12

Figure 2- 2. Generalized pulsed discharge ................................................................................. 12

Figure 2- 3 Markov chain representation of a real battery behavior ............................................ 14

Figure 2- 4 EMAC frame format ............................................................................................... 18

Figure 2-5. The S-MAC duty cycle: the arrows indicate transmitted and received messages ...... 21

Figure 2-6 D-MAC tree structure and the sequence receive-transmit-sleep ................................ 22

Figure 2-7 Duty cycles of S-MAC and T-MAC protocols compared to CSMA-CA protocol .... 24

Figure 2-8. Example of BAMAC(k) Operation .......................................................................... 27

Figure 2-9. a) Wakeup/sleep schedules of S-MAC, DSMAC, T-MAC and BEL-MAC protocols

b) communication process of BEL-MAC protocol..................................................................... 29

Figure 2-10 Virtual Grid............................................................................................................ 32

Figure 2-11. Proposed battery model ......................................................................................... 34

Figure3-1. Battery basic components ........................................................................................ 37

Figure 3-2 Battery basic ............................................................................................................ 39

Figure 3-3 Discharge time with different current intensities....................................................... 40

Figure 3-4 Operation of the pulsed current discharge. ................................................................ 41

Figure 3-5 Pulsed current discharge........................................................................................... 42

Figure 3- 6 Discrete Markov chain representation of a real battery behaviour ............................ 43

Figure 3- 7 Zigbee protocol Stack ............................................................................................. 45

Figure 3-8 Superframe structure ................................................................................................ 47

Figure 3-9 a. Non-beacon network communication, b. Beacon network communication ............ 48

x
Figure 3-10 CSMA-CA operations in beacon and non-beacon enabled modes ........................... 52

Figure 3-11 Discrete Markov chain representation of CSMA-CA protocol ................................ 53

Figure 4-1 Real battery behaviour of a node during CSMA-CA protocol execution ................... 61

Figure 4-2 Battery nominal voltage at every stage of CSMA-CA protocol................................. 62

Figure 4-3 Charges recovered with the new scheme .................................................................. 69

Figure 4-4 Average lifetime of a node ....................................................................................... 70

Figure 4-5 Total Lifetime of the network................................................................................... 71

Figure 4-6 Number of remaining alive nodes (N=25) ................................................................ 71

Figure 4-7 Number of remaining alive nodes (N=50) ................................................................ 72

xi
List of Acronyms

ACK Acknowledgment

BAMAC Battery Aware Medium Access Control

BE Beacon Exponent

BEL-MAC Battery-Aware Energy-Efficient Low-Latency Medium Access Control

BO Beacon Order

BRANS Battery and Redundancy Aware Node Scheduling protocol

CAP Contention Access Period

CCA Clear Channel Assessment

CDMA Code Division Multiple Access

CFP Contention Free Period

CR Communication Request

CSMA-CA Carrier Sense Multiple Access with Collision Avoidance

CTS Clear To Send

CW Contention Window

ECF Election Control Frame

EMAC Energy efficient Medium Access Control

xii
FFD Fully Function Devices

GTS Guaranteed Time Slots

IP Internet Protocol

L-MAC Low-Latency Medium Access Control

LPDM Low Power Distributed Medium access control

MAC Medium Access Control

NB Number of Back-offs

PAMAS Power Aware Multi Access with Signaling

PAN Personal Area Network

RF Radio Frequency

RFD Reduced Function Devices

RTS Request To Send

S-MAC Sensor Medium Access Control

SN-MAC Sensor Network - Medium Access Control

SO Superframe Order

STEM Sparse Topology and Energy Management

TA Time-out period Activity

xiii
TC Traffic Control

TDMA Time Division Multiple Access

T-MAC Timeout Medium Access Protocol

WLAN Wireless Local Area Network

WSN Wireless Sensor Network

xiv
List of notations

Ni , Vi Battery nominal voltage

Ti Battery theoretical voltage

RNi,Ti Recharge probability function

xv
Chapter 1 Introduction

1.1 Background

Wireless sensor networks (WSNs) are composed of tens to hundreds of small sensory nodes that

are able to communicate wirelessly with each other in a multi-hop fashion. WSNs are used to

provide a wide range of applications in various fields such as the medical domain, the

transportation system, the environmental studies etc. A typical sensor-node device has a small

processor to process data and control packets, a memory to save information needed by the node,

an RF system that assures the communication with other nodes, and a small battery that

represents the main (and most often the only ) source of power.

Designing and developing new effective protocols to overcome the different obstacles presented

in the protocol layers has been the focus of the researchers in the past few years. However, how

to conserve the power resources to prolong the lifetime of a node (and therefore the lifetime of

the network) has always appeared as a big challenge. It is in fact the most important factor that

should be taken into account while designing new protocols for WSNs, due to the limited

capacity of batteries and the difficulty of frequent battery recharging or replacement in many

situations.

Multiple key reasons stand behind the waste of energy in the sensor nodes such as (inspired from

[MINE09]):

1
  The interference: it causes the reception of corrupted packets and the nodes then will

waste energy trying to decode it.

 Overhearing: A node, located in the range of another transmitting node, receives packets

that seek another destination. The Energy is wasted due to receiving and processing the

wrong packets.

 Collision: This is a very common reason. When packets collide, the node has to

retransmit them. So the node loses energy in the first transmission, in discarding the

remaining of the packets, and in retransmitting them.

 Packet overhead: the lost of energy increases with the increase in the size of the packet

overhead

 Idle listening: this is the major cause of energy dissipation, because usually the nodes in

the wireless sensor networks spend most of their time in an idle state sensing the medium

to check its availability.

A variety of protocols and techniques have been developed from different perspectives to

mitigate the effect of these problems. For example, the cross-layer techniques ([FRAN11],

[BOUA09]) in which the protocols residing in the different layers share some crucial information

of the node, which lead to a better performance in terms of power consumption. Another

approach is to reduce the amount of data transmitted by the mean of software-based solutions

such as compression and data aggregations. There are also routing solutions based on energy

efficient schemes, and others that consider the network topology and the power control such as

fine-tuning the transmission power while maintaining network connectivity.

However in this report we focus mainly on schemes developed in the MAC layer protocol due to

its particular relevancy to our work.

2
Most of the existing MAC protocols that try to maximize the lifetime of the WSN are based on

switching off the Radio Frequency (RF) system while the node is not operating. All these MAC

protocols fall under the title of power-aware protocols. Their basic mechanism is to let the nodes

periodically listen to the medium and then sleep, instead of keeping them awake and listening all

the time even in the absence of any activity like packets’ transmission or reception. Therefore

they aim to inactivate the node for a portion of time to save more power and extend its life.

However, these protocols do not take into account the characteristics of the node’s battery. Only

few studies available in the literature considered tackling the power saving challenge by

including the battery’s behavior in their MAC protocols.

The majority of the protocols that deal with power conservation in WSNs assume an unrealistic

model of a battery. They assume that the battery is an ideal source of energy that drains power in

a linear manner. They presume that the discharge of the batteries happens in a constant rate. In

other words, they assume that the battery is just defined by a number of charge units and at every

transmission of a packet a constant number of charges will be taken out. However, in a realistic

situation, the battery behaviour is more complicated, the discharge process is nonlinear, and

many other factors affect its operation.

Studying in depth the properties of the real battery is out of the scope of this report. However, it

is important to briefly list some of the essential battery characteristics that will be used in our

schemes:

 The remaining voltage of a battery depends on the intensity of the discharge current.

 The battery discharge is faster when the drained current is high (and vice versa).

3
  The capacity of a battery will increase if the battery is allowed to rest for some time

(Charge recovery effect).

 The recovery process depends on the idle time duration.

 The recovery process depends on the state of the charge of the battery [CHIA99c],

[DATT05].

Particularly, we shed light on two effects: The pulsed current discharge and the charger recovery

mechanism due to their importance in our work as we will see later. We will discuss them in

further details in chapter 3.

1.2 Motivation and Objective

Only few research studies, about energy-efficient MAC schemes, that take into consideration the

real operation of the battery in their implementation have been submitted. Generally, the

researchers tend to develop new duty-cycle mechanisms in the MAC sub-layer to conserve more

energy.

In this thesis, we propose two new battery-aware energy-efficient MAC protocols based on two

different approaches. These protocols exploit the internal properties of the battery to extend the

lifetime of a node and to improve also the performance in terms of throughput and collision

percentage. In both protocols, we introduce new wakeup/sleep schedules that include the actual

battery voltage in their computations.

We call the first battery-aware MAC protocol: MAC-LV. This protocol gives priority to nodes

with low battery voltage to access the medium before the nodes with higher battery voltage. In

this case, the nodes with higher remaining voltage remain in the sleep mode longer than the

4
nodes with lower remaining voltage, and thus they will save more energy and they will recover

more charges.

The second battery-aware energy-efficient protocol we propose is MAC-HV. The operation of

this protocol is exactly the opposite. With this approach, the nodes with high battery residual

capacity have the right to access the channel before the nodes with lower battery voltage. As a

result the weaker nodes, in terms of power, have the opportunity to rest more and then to restore

more power than the other nodes.

Eventually, the performance of these protocols will be studied and compared with each other as

well as with the performance of the conventional CSMA-CA protocol through simulations.

We summarize the objectives of this thesis with the following points:

 To introduce the battery characteristics, precisely the charge recovery mechanism, in the

design of a new MAC protocol for energy saving purpose.

 To verify that incorporating the capacity recovery effect of the battery in the

implementation of a MAC protocol will add extra benefits to its performance.

 To provide a new active/idle scheme function of the current remaining voltage of the

battery.

 To study the influence of giving the precedence right of channel access to the nodes with

higher battery power on the overall performance of the network.

 To study the impact of allowing the nodes with lower battery voltage to gain access to the

medium before the nodes with higher battery capacity on the general performance of the

network.

5
  And to reach a conclusion about which approach is the best in terms of some

performance metrics: is it the usual CSMA-CA protocol, or the approach that permits the

powerful nodes to relax the most, or the approach that allows the weak nodes to rest

more.

1.3 Thesis contributions

This thesis intends to provide innovative means to address the energy-efficiency problem in the

wireless sensor networks. The design and implementation of these schemes led to new

significant original work and novel results.

The contributions of this proposed study are as follows:

 The verification, through analytical study, of the importance and the benefits of including

the real behaviour of the battery in the design of the MAC protocol in the wireless sensor

networks has been achieved.

 New Back-off schemes that use the current battery state in their computations have been

defined.

 A novel battery-aware MAC protocol that gives priority to powerful nodes (in terms of

battery capacity) to use the channel has been designed and implemented.

 A novel battery-aware MAC protocol that allows weaker nodes to have the precedence in

accessing the medium has been developed.

 A new comprehensive computer simulation program has been developed in C language to

implement the proposed protocols

6
1.4 Thesis outline

This thesis consists of six chapters. Chapter 1 introduces the background of the energy-efficient

protocols in WSNs, and focuses mainly on the protocols designed in the MAC sub-layer. It

shows the background of the battery-aware MAC schemes by introducing the realistic behaviour

of batteries. It also presents the motivation, the objectives and the contributions of this thesis.

Chapter 2 provides a review of the literature on the related work. It is basically composed of

three main sections: the first section shows the studies about battery operations and properties. It

presents many battery models that reflect the real internal behaviour of the battery and verifies

the advantages of the pulsed current discharge and the charge recovery mechanism.

In the second section, we discuss the existing duty-cycle energy-efficient MAC protocols. And

the third section talks about the few battery-aware MAC protocols reported to the literature.

Chapter 3 talks about two essential subjects which we used to develop our protocols. It consists

of two main parts. The first part describes briefly the battery, its main parameters, its functions,

the pulsed current discharge and the capacity recovery effect. The second part discusses the

CSMA-CA protocol. It talks about its communication modes, its operations, and finally it

represents it in a discrete Markov chain model.

Chapter 4 verifies the advantage of considering the internal battery properties and effects in the

design of a MAC protocol in WSNs. In fact, it proves through analytical study, that modifying

the CSMA-CA protocol by exploiting the charge recovery mechanism of the battery, has

extended the lifetime of the nodes and the network.

7
Chapter 5 explains our proposed protocols MAC-HV and MAC-LV in details. It defines the new

back-off functions which depend on the actual battery voltage. It presents the algorithms and

flowchart of each protocol. And finally it shows and discusses the simulation results in terms of

many performance metrics

Chapter 6 concludes our current research study and explores the future work we intend to target.

1.5 List of Publications

1. Yamen Nasrallah, Mouhcine Guennoun, Hussein T. Mouftah, Energy-Efficient

Battery-Aware MAC Protocol for Wireless Sensor Networks, 2012 IEEE Wireless

Communications and Networking Conference (WCNC 2012), Paris, France, April

2012.(to appear)

2. Y. Nasrallah, M. Guennoun and H.T. Mouftah, “Energy Efficient Battery Aware MAC

protocol for WSN”, presented at 2011 WiSense Workshop, University of Ottawa, Ottawa,

September 2011.

8
Chapter 2 State-of-the-art of Energy-

Efficient MAC Protocols

2.1 Introduction

Designing new MAC protocols to reduce the power consumption and to prolong the lifetime of

the nodes in WSNs attracted the attention of many researchers. However, in most of the

proposed work, the authors try to find the best duty-cycle schemes that effectively use the current

energy, and the optimal ways to reduce the power spent during sending and receiving. They are

known as power-aware MAC protocols. Their main drawback is that they do not consider the

state of node battery in their design[YE04a], [CHIA02], [CHIA00], [DAM03].

The aim of this chapter is to review the existing MAC protocols in the literature that tackle the

energy savings and power consumptions challenges in WSNs. This will includes those which

take into consideration the battery characteristics and those which do not.

This chapter is divided into three main sections (in addition to the introduction and the

summary): the first section discusses new battery representations that reflect its real functionality

and reveal its actual advantage in a communication system. The second section talks about

power-aware MAC protocol. The third section reviews a few existing battery-aware MAC

protocol.

9
2.2 Models of real battery behaviours

Studying in depth the operation of the real battery, the internal chemical reactions and the details

of its technical characteristics and properties is out of the scope of this project. However, since

our proposed work is based on some important effects that take place in the battery, which

influence positively the communication procedures in the sensor networks, we are obliged to

give a brief overview about some studies reported in the literature related to these effects.

Basically these effects are: the pulsed current discharge and the charge recovery mechanism.

We note that an overview section about the batteries operation and properties will be discussed

in the Chapter 4.

The handbook of batteries [REDD10] is a modern, advanced and complete guide about batteries

from theoretical perspectives to industrial applications. It covers the principles of the internal

operations, it describes in details the electrochemical reactions, and it classifies the batteries to

different categories based on different criteria. It is one of the best references about batteries

nowadays.

The following research works [LAFO90], [PODH94], ] [FULL94], [NELS97] and [DOYL97]

prove that draining the battery power with a sequence of current impulses is way better, in terms

of lifetime and performance, than with a constant current discharge.

In what follows we will review some papers that propose new schemes to model the battery

behavior taking into account the charge recovery capability. In [CHIA99b] an investigation is

conducted about finding new ways to improve the energy efficiency of communication protocols

that take advantage of the charge recovery mechanism.

10
The study starts with an overview about the battery behavior. It focuses mainly on the effect of

the pulsed current discharge. In a communication system, two different battery models with

pulsed discharge modes were assumed: the binary pulsed discharge and the generalized pulsed

discharge. In both models the packet arrival process is considered to be Bernoulli.

The binary pulsed discharge process is illustrated in Figure 2-1. (N is the nominal capacity, Ns is

a dummy state added to represent the start of the discharge, T is the theoretical capacity). The

system transmits only one packet, if available, in each time slot. If there is no packet to send, the

system recovers one charge unit. In case of a transmission, the system uses power during the

whole duration of the time slot.

The probability that one packet arrives during one time slot (and thus the probability of a packet

transmission a time slot) is defined as a1. As a result, the charge recovery occurs with a

probability of 1 – a1 (in case of no transmission). The authors calculated the total estimated

number of packet that can be transmitted before the theoretical capacity of the battery is

exhausted, and then they determined the ratio of the mean transmitted packet under binary pulsed

discharged and constant current discharge. The objective is to analyze the gain when varying the

probability a1.

The results showed that the performance of the system with binary pulsed current discharge

surpass the performance of the system with the constant current discharge, especially when the

probability of packet arrival is small (less than 50%), and that’s obvious because nodes will have

more time to rest in this case and thus more chance to recover charges.

11
 Figure 2-1. Binary pulsed discharge

In the generalized pulsed discharge, the node is allowed to send multiple packets in a time slot.

The node relaxes in the remaining period of the time slot after completing the packets

transmission. The system is presented in the Figure 2-2.

Figure 2- 2. Generalized pulsed discharge

ak : probability of the arrival of burst of packets in a time slot.

M is the maximum number of packet per burst.

The same concept of study followed in the previous case is applied in this system. A new gain

fraction of packet arrival between the generalized discharge system and constant discharge

system is calculated. And the results show a significant improvement, however as the number of

packet transmitted increase the performance becomes worse.

12
We think that this research was one of first attempts of introducing the battery behaviour in the

study of a communication system. It proved that such approach can be very useful and can

achieve remarkable results in the future. However, it was not applied on a specific

communication protocol; it was only based on an analytical scheme. We also think that it is

incorrect to assume a fixed recharge probability; in fact the probability of recharge is variable

and depends on the actual battery voltage.

A new model of the battery behaviour that takes into consideration the pulsed current discharge

is presented in [CHIA99a]. It adopts the same Markov chain representation of the generalized

pulsed current discharge discussed in [CHIA99b], however the main distinction is the definition

of the charge recovery function: it is not a fixed parameter anymore, it is variable and depends on

the battery state and on its discharge capacity.The new model was then applied on a new battery

management technique based on the traffic shaping algorithm (leaky bucket) to get the most out

of the battery capacity.

The authors identify the recovery probability as a decrementing exponential function of the

battery current voltage and of its discharge operation. They adopt it from [ALZI97] where three

recovery parameters were introduced γ1, γ2, and γ3 . While the cell is discharged, the slope of the

exponential function ranges over the value of these parameters.

They also assume that at each time slot, the battery may lose "i" charge units if it receives and

sends "i" packets. If no activities were detected, the battery state may remain unchanged or it

may increment by one charge unit.

13
 Figure 2- 3 Markov chain representation of a real battery behavior

The recovery probability function at state j after transmitting k packets is defined as the

following:

- N is the nominal capacity ( we will talk about it in Chapter 3)

- αN and αc are parameters that depends on the recovery capability of the battery.

- αN is assumed constant

- αc is a piecewise constant function of the number of packets already transmitted, that

changes value in correspondence with γc (c=1,2,3)

- ai is the probability that a burst of i packets arrives in one time slot

- a0 is the probability to remain in state N when the initial state of the battery is N.

ai = rN(k).

14
The authors choose the value of parameter so that the discharge process is as close as possible to

the real operation of a battery.

Similar to [CHIA01], the authors define the ratio of the average number of packet transmitted

during the discharge process over the initial maximum capacity of the battery (Gp); and then they

study the performance under tow different arrival processes: Bernoulli arrival process, and

Poisson arrival process.

With Bernoulli arrival process, the probability of the arrival of one packet in one time slot is

fixed and equal to q. However with Poisson arrival process the probability of i packets arrival is

defined by = .
!

The results show the real benefit of pulsed current discharge, the performance increases with the

higher recovery capabilities. It also proves that the current intensity has a big impact on the

recovery process whatever the internal properties of the battery are.

The second step the authors have discussed is applying the previous discharge model on the

leaky buck technique [SCHW96]. Their goal is to be able to return the battery voltage level to its

initial value under different traffic load. They define a threshold state of the battery B. When the

battery capacity reaches this threshold it stops transmitting packets, and the remaining packets

will be stored in a data buffer big enough to not lose any of them. During the recovery period,

the battery recovers one charge unit in one time slot. In the absence of any packet in the queue, it

continues recovering. However, when new packets arrive, the node starts transmitting as long as

the battery level is greater than B.

15
They study the performance, with Bernoulli arrival process, in terms of throughput, average

packet delay and average number of queued packets. The pro is that the battery capacity can be

maximized, but this is at the cost of packet transmission delay. However, by choosing the

suitable size for the token in accord to the battery recovery properties, we can attain an optimized

solution.

In [CHIA01], the authors revisited the earlier work with same processes, functions, assumptions

and schemes but with a new mathematical approach. They investigate the real battery activities

and the benefit of the pulsed current discharge through stochastic models. And they achieve the

same results attained previously.

We limit our survey in this subsection to this end, because the majority of the other papers

reported to the literature that discuss this subject delve deep into batteries internal functions and

material characteristics [RAHM10], [SUBR09], [PODH94] and [FULL94], which is far away

from our main field of interest.

In the next subsection, we discuss power-efficient MAC protocols. We mainly talk about

protocols based on new duty-cycle schemes aimed to reduce the power dissipation.

2.3 Power-Aware MAC protocols

In general, MAC protocols can be classified as either contention-based or contention free

protocols. TDMA is an example of a contention-free MAC protocols and CSMA is an example

of contention-based MAC protocols. It is true that the use of the TDMA technique in the MAC

layer will prevent the collisions of packet and will solve the hidden node problem; however it

encounters various disadvantages (documented in [YE04b]) such as the need of a centralized

16
node, a very precise synchronization clock, a very tight schedule with high degree of

concurrency etc. which make it a bad option for wireless sensor networks. On the other hand,

what concern us in this report are the MAC protocols that aim to extend the operation life of a

network. The conventional contention based MAC protocols for wireless networks such as

WLAN are not suitable for WSNs, because they require the nodes to listen to the channel

continuously [SELV08] and as a result consume too much energy. In what follows we will focus

on the CSMA-CA based protocols designed to reduce the idle listening time using new duty

cycle techniques, however we will first present briefly two energy efficient TDMA-based

protocol.

2.3.1 TDMA-based energy-efficient MAC protocols

The Energy efficient Medium Access Protocol (EMAC) [NIEB03] is a TDMA-based protocol

that aims to decrease the energy consumption of a sensor node. The authors identified major

sources that cause the waste of energy, and then they identified numerous novel mechanisms to

mitigate their effects.

This protocol divides time into frames and each frame into time slots and then it assigns one time

slot to each node. The node has possession of the timeslot and controls it alone without any

competition with any other node. For efficient use and distribution of these timeslots, the

network is composed into cells; each cell consists of neighbor nodes. Two nodes located in two

different cells cannot communicate with each other, because the range of communication of one

node does not reach the second. Therefore a common time slot can be used by these nodes.

The timeslot in its turn is divided to three sections: Communication Request (CR), Traffic

Control (TC) and Data section. In the CR section, the node controlling the timeslot accepts

17
requests from other node. For example the nodes can ask for data or inform the owner node that

there are data available for it (very similar to the RTS packet in CSMA-CA based protocols). In

the CR period a small probability of collision exists.

Each node retains a schedule table where it saves the schedule of the neighbor nodes located in

its cell. In TC section the owner transmits its schedule and broadcasts important information to

the neighbors such as the timeslots of surrounding nodes, and it also tells them about what kind

of communication will take place in the data section.

In the data section, the actual data will be transmitted to the target.

In the CR period, the nodes that do not have a request for the current slot owner will keep their

transceiver in a low power state. When a time slot is not controlled by any node, all nodes will

remain in sleep state. If a node is not addressed in the TC section nor its request was approved, it

will resume a standby state during the entire data section [VAN06].

Figure 2- 4 EMAC frame format

This protocol shows good results in terms of energy efficiency; however that is in the expense of

latency.

The authors of EMAC protocol developed a new protocol called: Lightweight Medium Access

Control protocol (L-MAC) for wireless sensor networks based on ideas derived from the EMAC

protocol. It is also an energy-efficient TDMA-based scheme.

18
Similar to EMAC, the frame is composed of time slots and one time slot is assigned to one node

to control over it. However, unlike EMAC the time slot in this case has only two sections: the

control message section and data unit section.

In the control section, the node transmits a control message defined in the table 2-1.

Description Size (bytes)

Identification 2
Current Slot Number 1
Occupied Slots 4
Distance to Gateway 1
Collision in Slot 1
Destination ID 2
Data size (bytes) 1
Total 12
Table 2-1. Contents of the control message

During this section, the node which owns the time slot exchanges important information with the

neighbors. For example, it informs them the id of the current time slot, the distance in hops to the

gateway, the size of the data that will be sent etc. It also sends an indication to the destination

node about the data length so it prepares to receive the data in the next section. It also uses this

section to maintain synchronization with other nodes.

All the nodes, except the one indicated as a destination, will switch to sleep mode to save energy.

And after completing the data transfer, both nodes (the source and the destination) power off

their transceiver.

The main advantage over EMAC protocol is that LMAC tries to minimize the overhead of the

physical layer. And in contrast to EMAC, LMAC is a contention free protocol.

19
Simulation results were compared to EMAC and S-MAC protocols (S-MAC will be discussed

later in this section). LMAC protocol was able to extend more the network lifetime compared to

EMAC and S-MAC respectively [VAN04].

Z-MAC [RHEE05] is a hybrid TDMA-CSMA protocol. It is designed to take the advantages of

both TDMA and CSMA-based schemes and to avoid their weaknesses.

Under low contention conditions, Z-MAC operates like a CSMA-based protocol and thus it

accomplishes high throughput and low latency communication among the nodes. In contrast, it

functions as a TDMA-based protocol under high contention situations, and as result its

performance achieves high channel utilization and nearly eliminates collisions among two-hop

neighbors.

Z-MAC succeeded to solve the problems of synchronization, network topology variations and

time varying channel conditions. However, it requires a long initial setup phase in which the

nodes should execute multiple tasks in order to establish their links.

There are many other proposals made for TDMA in wireless sensor networks [GOBR09],

[ARIS02], [LI04], and [VENK06] , but we restrict our review to what we presented so far,

because the solutions based on TDMA alone has been neglected in wireless ad hoc networks for

its failure to resolve their fundamental challenges, such as scalability, energy constraints etc.

2.3.2 CSMA-based energy-efficient MAC protocols

The Sensor Medium Access protocol (S-MAC) proposed in [YE04a] is a contention based

protocol that assumes a fixed duty-cycle scheme. Its goal is to shorten the duration of the idle

time to save more energy and to decrease the number of collisions in the network.

20
It switches periodically the transmission system between two modes: sleep and active. In fact,

the frame in S-MAC is divided into two parts: an active part and a sleeping part. During the

sleeping part the radio of the node is turned off, and it keeps all the packets ready to be sent in a

queue. These packets are sent out once the node switches to the active mode, and therefore the

energy wasted on idle listening is reduced. The active-sleep schedule created initially by the

node is broadcasted to the neighbors; this schedule should be updated periodically among

different nodes, thus synchronization is always required.

To overcome the hidden node problem, S-MAC protocol applies the RTS/CTS/DATA/ACK

handshake scheme.

The main advantage of the S-MAC protocol is the energy savings due to the long sleeping

period. But this comes at the expense of the network throughput and packets latency. Moreover,

a fixed duty cycle scheme does not adapt to the traffic variation in sensor networks.Another

problem is the data forwarding interruption which takes place when packets need to go over

multiple hops to reach their destination.

Figure 2-5. The S-MAC duty cycle: the arrows indicate transmitted and received messages

D-MAC protocol [LU04] was designed to avoid the data forwarding problem. It is an energy

efficient and low latency protocol based on data gathering trees where the flow of packets is

21
directed in one predetermined way from source nodes to one coordinator or sink node which is

the only destination in the network.

The tree architecture is divided into levels (Figure2-6). Each level consists of a set of nodes that

receive data from the nodes of lower level and forward them to the nodes in the higher level. The

frame is divided into three main modes: transmitting, receiving and sleeping, and the schedule

proposed is designed in a way so when a node at a given level is sending data, the neighbor node

in the upper level will be in the receiving mode. The nodes at each successive level up the tree

follow a receive-transmit-sleep sequence shifted to the right [SHEL08].

Figure 2-6 D-MAC tree structure and the sequence receive-transmit-sleep

This protocol comprises multiple advantages. The latency is reduced because the nodes involved

in the communication wake up in a sequence manner from bottom up to the destination. Another

22
pro is that the nodes which are not on the path of communication operate in a low duty cycle, and

thus they save energy.

The main shortcoming is that D-MAC protocol can be applied only on this type of networks

(unidirectional gathering trees).

A similar approach has been adopted in [WANG07] [L-MAC protocol] with almost the same

network assumptions and topology. However, the author implements an adaptive sleeping

scheme (inspired form a protocol called T-MAC which we will discuss later), to mitigate the

effect of unbalanced traffic problem. In fact, he introduces a time interval within which if a node,

in receiving mode, does not sense any communication, it changes to sleep mode.

In contrast to S-MAC protocol, the Timeout MAC protocol (T-MAC) [DAM03] introduces a

new adaptive duty-cycle mechanism that dynamically terminates the active mode. Its main goal

is to reduce the idle listening time to save more energy.

To achieve this goal, T-MAC protocol allows the node to send the entire messages it possesses in

bursts of different sizes, and to sleep after that. The key feature of T-MAC protocol is the

introduction of a new time out policy that helps to adapt to the variations in traffic by ending

dynamically the active period.

A Time Out (TA) interval is defined. The node terminates its active time if within the period of a

TA no activities were detected, such as transmitting or receiving DATA/ACK packets, listening

to the medium during a collision, sending an RTS or CTS signal. In this case the node assumes

that no neighbor is willing to communicate so it switches to sleep mode. On the other hand, if a

node is involved in a communication, it starts a new time-out period after it completes the

communication.

23
Figure 2-7 shows a basic comparison between the operation of S-MAC, T-MAC and CSMA-CA

protocols

Figure 2-7 Duty cycles of S-MAC and T-MAC protocols compared to CSMA-CA protocol

The results of T-MAC protocol exceed those with S-MAC protocol, especially in reducing the

energy consumption of the nodes. In addition to that, the dynamic aspect of its design makes it

more suitable to WSNs then S-MAC which uses a fix duty-cycle scheme.

Other protocols, such as PAMAS [SING98], LPDM [GUO01] and STEM [SCHU02] use a radio

system with dual channels: the data channel and the signaling channel. The data channel remains

most of the time in a sleep mode. It just wakes up for data transmission once it receives a wakeup

signal by the control (signaling) channel.

We note finally that there are so many other protocols that propose new duty cycle schemes

([JURD08], [ANAS09b] and [YUAN11]), and novel energy efficient approaches ([SUN08],

[ALIP09], and [YIN10]), others developed based on the conventional S-MAC protocol

([HAMA10] and [EZZE09] for example), trying to enhance its performance or to avoid some of

its drawbacks. But since this is not the main concern of our study, we will limit this subsection to

this point.

24
2.4 Battery-aware energy-efficient MAC protocol

In [JAYA04] the authors propose a new battery-aware MAC protocol for ad hoc wireless

networks, based on a novel distributed scheduling scheme to increase the lifetime of the nodes

and the network by exploiting the charge recovery effect of the battery. They refer to this

protocol as BAMAC(k). k represents the number of packet that the node should send once it

gains access to the medium.

Their main goal is to ensure a near round-robin scheduling and a uniform discharge of their

batteries to outperform the conventional IEEE 802.11 CSMA-CA protocol.

The proposed scheduling scheme depends on the current residual battery voltage. It schedules the

nodes, using a round-robin scheduler, in a way that the back-off period decreases with the

decrease of the remaining battery capacity. In order to achieve this target, the authors suggested

that each node saves a table in which it keeps the following information about the batteries of its

neighbor nodes: the remaining theoretical voltage, the remaining nominal voltage and the last

time the battery was used. The table is arranged in descending order of its theoretical capacity of

the nodes. Every node sends the current battery information to the neighbor nodes with every

packet: RTS, CTS, DATA, and ACK.

They define the back-off period as:

− = [0, (2 ∗ ) − 1] ∗ ∗( + + )

Where: - CWmin is the minimum size of the contention window.

- rank is the position of that entry in the information table

25
 - TSIFS and TDIFS the short inter-frame spacing and the DCF inter-frame spacing

durations

- Tt is the longest possible time required to transmit a packet successfully including the

RTS-CTS-Data-ACK handshake

To demonstrate the operation of the protocol, an example was introduced. It is illustrated in

Figure 2-8. A and B are the source and the destination nodes. In the first step, A sends an RTS

signal to B. The battery information of A are attached to RTS. The nodes C, I and B update their

battery table. In step two, B sends back a CTS signal including the battery information of A and

of itself, as a result the nodes D, I and A add the new entries to their battery table. Same

procedure occurs with the Data and the ACK packets. After step 4, node B will have the right to

access the channel before the other nodes because its rank is the smallest.

Their results showed that this protocol outperforms IEEE 802.11 in terms of power consumption

and the lifetime of the nodes and network.

26
 Figure 2-8. Example of BAMAC(k) Operation
In our opinion, this approach was very successful, however there is a network security question

since all essential information about the batteries are exchanged between the nodes. Other

drawbacks are noticed: the big overhead attached to every packet sent, all the nodes need to

continue listening to the medium all the time so they can update their battery table, additional

memory and processing capabilities are required.

BEL-MAC protocol [DHAN05] introduces a new duty-cycle scheme that exploits the internal

properties of the battery to minimize the end-to-end delay and maximize the network’s lifetime.

27
The throughput and the latency of BEL-MAC were compared with S-MAC, D-SMAC, T-MAC

and CSMA-CA outcomes, and the results showed that BEL-MAC is the best among them.

In this study, it is assumed that the node is equipped with two batteries: primary and secondary.

The secondary battery is always active and it is used to activate the primary battery, which is

shut off most of the time (during the sleep mode). It is also assumed that the sensor node

periodically switches between sensing and computing modes all the time.

BEL-MAC protocol is designed in a way to keep the remaining voltage level in the battery as

close as possible to the initial value of the nominal voltage. In order to achieve this goal, the

author introduces new dynamic active slots in every frame. After the contention period, the node

that succeeded to access the medium sends the entire packets it possesses. However the node is

not permitted to continue transmitting in multiple time slots. The node saves the remaining

packets and toggles to sleep state by putting off the primary battery, while the secondary battery

stays awake and listens to the medium. This sleeping period allows the main battery to recover

some charge units.

The major difference between BEL-MAC and S-MAC and D-SMAC protocols is that the former

insert an optimum number of additional active slots to achieve a compromise between the

latency and the energy recovery.

The first figure of Figure 2-9 shows the difference between the various protocols in term of

Wakeup/sleep schedule. As we can see, in S-MAC protocol the node is active just for a short

period of time and sleeps most of the time which will increase the latency. D-SMAC protocols

uses more time in transmitting then in sleeping. In T-MAC protocol, the nodes transmit all the

packets once it has access to the medium and thus the battery will be discharge so quickly. And

28
finally BEL-MAC protocol adds, in an optimal way, some active time slots in the frames to gain

the advantages of S-MAC and T-MAC and neglects their disadvantages

Figure 2-9. a) Wakeup/sleep schedules of S-MAC, DSMAC, T-MAC and BEL-MAC protocols
b) communication process of BEL-MAC protocol

In the second figure of Figure 2-9 the authors show the communication process in a one-hop

network. A cluster head is connected to four nodes. As we can see, a node can transmit only in

on active time slot in each frame and relaxes for the remaining time.

The simulation demonstrates promising results, however there are some disadvantages related to

this protocol: the extra size of the node and its extra cost due to the usage of an extra battery.

An interesting study about exploiting the battery recovery effect in the sensor networks was

described in [CHAU10]. The author conducted experiments on a sensor network test-bed. The

results demonstrated a considerable increase in the lifetime of the node due to the charge

recovery effect, and it provided evidence that there is a time saturation threshold after which the

node will not recover any charge units.

29
A remarkable protocol that exploits the battery charge recovery property to extend the lifetime of

the network is designed in [SELV08]. It is the Battery and Redundancy Aware Node Scheduling

protocol [BRANS]. The interesting factor that distinguishes this protocol from the previous ones

is the introduction of the node level redundancy with respect to sensing. It schedules the sensor

nodes (Wake up/sleep) based on the sensing level redundancy information and exploits the

battery charge recovery mechanism in order to attain an increased network lifetime with

controlled latency [SELV08].

The basic assumptions and variables that characterize this protocol are summarized as follows:

- The sensing coverage area of the node is defined as a circle with radius Rs.

- The communication range of the node is determined by a circle with a radius Rc

- Every node is at the same time a source of information and a relay that forwards the

data sent by neighbor nodes to the sink node.

- Redundancy with respect to the sensing should be guaranteed. To assure that, the

authors assume the operation of a large number of nodes distributed randomly in a

precise area.

- Every node is equipped with two batteries. The big battery, denoted the primary

battery, is responsible to supply the node with the required power to be fully

functional. The small battery, denoted the secondary battery, is used only to generate

an activation signal to wake up the primary battery on termination of a timer.

- All nodes are aware of their locations.

As describe in BEL-MAC protocol, the primary battery switches to sleep mode whenever the

active period is expired. Sleep mode means a complete shutoff of the battery, which will allow it

to recuperate a portion of its charges.

30
The BRANS protocol is implemented in 4 phases: Virtual Grid Foundation, Initialization,

Periodic active node election and routing neighbor updating. It starts by dividing the area where

the nodes are deployed, into virtual grids with equal surfaces. The size of each square unit of this

grid depends on the application, it could be for example 2Rs * 2Rs. After exchanging the

location information, the nodes will be able to identify their neighbor nodes which are located

within their range of sensing (a grid) (Check Figure2-10). The nodes which reside in the same

grid are assumed redundant. Then each node will create a neighbor table that contains

information about their neighbors.

In the second phase, in each grid the node with the lowest node id remains active while the other

go to sleep mode. The sink node communicates then with all active nodes to create a routing

table (Table 2-2).

In the third phase, every active node in a grid should choose a successor node to be active. In

fact, in every predefined period of time all the nodes wake up and wait for an Election Control

Frame (ECF) (Table 2-3)to update their neighbor table and to be aware of the next active node.

The node with nominal capacity higher than a predetermined threshold will be chosen to be the

next active node. And finally the routing table should be updated based on the new active nodes

in all grids that form the actual network.

The simulation results showed great increase in the lifetime of the network compared to IEEE

802.11 CSMA-CA and T-MAC. The implementation of the protocol is quite interesting;

however the latency is still a problem that needs to be resolved especially in high network loads.

We should consider also the additional size and cost of the nodes due to the dual battery setup.

31
Exchanging crucial information among nodes presents a big challenge in term of security. And

finally the use of this protocol is limited to a network with redundant nodes.

Figure 2-10 Virtual Grid

Table 2-2 Battery Table Table 2-3 Election Control Frame

Q-Length: length of queue in MAC buffer

SN-MAC protocol [WATF10] is a recent MAC protocol that makes use of the chemical

properties of the battery to increase the lifetime of the network. The authors assume a cluster-

based network, in which the nodes are assembled in groups characterized by the presence of one

cluster head which controls and manages the internal and external communications. The cluster

head of every set of nodes is unique, i.e. that it does not circle and switch places with other nodes

for the sake of load distribution in terms of energy. On the other hand, all cluster heads are in

turn organized in multiple levels, each level is one hop away from the other, and it is selected

32
based on the battery state of the nodes and based on information transmitted in the packets issued

by a priority function at each cluster head. The highest level is connected directly to the main

base station. As a consequence the network will be considered virtually hierarchal.

CDMA-based communication scheme is applied inside of each cluster, where every new node

joining the cluster gets a code that allows it to transmit at the same time and frequency with other

nodes. Conversely, the cluster heads communicates with the base station based on an adaptive

TDMA schedule mechanism.

The main objective behind the design of this protocol is to increase the channel utilization and

reduce the end-to-end packet transmission delay in addition to extending the life duration of the

network.

The author designed a new battery model represented in a discrete Markov chain process. It is

composed of ”M” states, starting from the state M where the battery is considered full and

ending with the state 0 (zero) where the battery is considered empty. This model assumes the

following:

- The time frame is split into time slots.

- Prec_i : This is the probability that a node, in state “i”, returns to state M if it remains

idle for one time slot

- Prec_total = 1. This is the probability that a node, in state “i”, returns to sate M if it

remains idle for Δ time slots.

- Φ the energy wasted in the battery during the process of transmission

33
 Figure 2-11. Proposed battery model

The question here is to evaluate Δ in order to take decisions on how long to remain idle. After

calculating Δ, the algorithm will use its value to compute the node’s priority in being selected as

a routing candidate towards the base station.

We will briefly present the basic design and setup steps followed by SN-MAC protocol:

1- Initial phase after deployment: this is the first step, it includes neighbor discovering,

communication establishment, node levels identification, codes spreading and formation

of clusters

2- Second phase is the implementation of the CDMA and TDMA protocols among nodes

and clusters.

SN-MAC offers the following properties: it can be applied on any type of cluster-based

networks, it can adopt any spreading code mechanism, it is capable of overcome the challenges

derived from the collisions, nodes overhearing, idle listening and protocol overhead. It can be

simply integrated with different upper layer protocol due to its flexible and proper cross-layer

design.

However it presents some drawbacks: it is only applicable on cluster-based networks and it does

not consider a realistic model of the battery.

34
2.5 Summary

In this chapter, recent works in the MAC sub-layer about reducing the power consumption and

increasing the energy savings have been discussed. We have started the survey with an overview

about the importance and the benefits of incorporating the real battery behaviour in the design of

a communication protocol. We have proceeded with a study about the current energy-efficient

power-aware MAC protocols, in which the focal point was about developing new duty-cycle

schemes that force the nodes to switch to sleep mode whenever it is inactive. And we terminated

our review by presenting the MAC protocols that consider the battery state to restore more

energy and to extend the lifetime of the network.

35
Chapter 3 Battery Characteristics and

CSMA-CA Modeling

3.1 Introduction

In this chapter, we will describe two important subjects related to our study: the battery

characteristics and the IEEE 802.15.4 CSMA-CA. In fact, in the next chapters, we will build on

top of these concepts to provide our new propositions.

This chapter is organized in four main sections: The current introduction presents its first section.

In the second section we explain the charge recovery mechanism; we start by an overview about

the battery operation, we then focus on the pulsed current discharge and the charge recovery

mechanism, and finally we present the recharge probability function which will be used in our

proposed protocols.

The third section provides detailed description of CSMA-CA protocol; it includes its operation,

its modes of communication and its algorithm. The most important part in this section is the

representation of CSMA-CA protocol in a discrete Markov chain model. In this subsection, we

study analytically (through mathematical equations) the operation of the CSMA-CA protocol,

and later (in Chapter 4) we will develop on top of this model our new enhanced schemes.

And finally, section five concludes this chapter.

36
3.2 Charge Recover Mechanism

3.2.1 Introduction

Without any doubt, energy resources and power consumptions present a major challenge in

wireless sensor networks. The majority of the applications in wireless sensor networks require

the usage of sensors that rely on batteries as their main source of power. Therefore,

understanding the characteristics of the batteries, knowing their discharge behaviours and

exploiting their special effects and mechanisms may lead to promising results in terms of

prolonging its life and reducing its discharge rate.

We will start this section by talking about the main properties of a battery, and then we will

mainly focus on two important effects: the charge recovery mechanism and the pulsed current

discharge. These two concepts are the bases on which we developed our new idea.

3.2.2 Basics of a battery

The battery is an electrical device that stores chemical energy and

provides electrical energy. It is an electro-chemical container

usually composed of one or multiple cells organized as an array,

that converts the stored chemical power into useful electrical

power.

Every cell consists of two electrodes: an anode and a cathode, and

of an electrolyte. Figure3-1. Battery basic

components

37
The main role of the electrolyte is to separate the two electrodes and to allow the transfer of

electrons between them.

A battery is characterized by the following parameters

 Theoretical capacity: we refer to it as T.

It is the maximum voltage that can be extracted from a battery. It is directly related to the type

and amount of material contained in the battery. So two batteries possess two different

substances will have dissimilar theoretical capacities because of their different internal electrical

characteristics. And two batteries with the same kind of materials but with different sizes will

have different theoretical capacities since the amount of materials (or their weights) is not equal.

 Standard or Nominal capacity: we refer to it as N.

It is the energy that a battery can provide under standard load conditions at a specific constant

current discharge. The nominal capacity can never go beyond the theoretical capacity.

 Actual Capacity:

It is the energy that the battery delivers under a given load [DHAN05], and it is usually used as a

metric to judge the battery efficiency for constant loads.

38
3.2.3 Battery discharge behaviour

The following description is based on the information

presented in [CHIA99b]. In the absence of any current, the

concentration of the active material in the electrolyte is

uniform and equal to its average value. In the presence of

current, and due to an electrochemical reaction, the active

species will be consumed at the electrolyte-electrode interface

and will be replaced by other new active materials that move

Figure 3-2 Battery basic
from the electrolyte solution to the electrode. discharge process

3.2.3.1 Constant current discharge

In constant current discharge, and as the intensity of the discharge current grows, the

concentration of the active material around the electrode will drop. As a consequence, the state

of the charge of the electrode as well as the battery voltage decreases. And very quickly the

electrolyte next to the cathode will run out of active materials that assure the electrochemical

conversion, and thus the battery will expire in a relatively short period of time.

At this point, even though a portion of the active material still exists in the battery, the battery

will not be able to provide more electrical energy due to the distance of these active materials

from the cathode. In fact, the theoretical capacity in this case has not been exhausted, but the

nominal capacity is considered zero.

It is important to mention that if the discharge current rate increases, the battery will be drained

faster due to the fast decrement of the active material concentration around the electrodes.(see

Figure3-3 taken from [CHIA99b])

39
 Figure 3-3 Discharge time with different current intensities

3.2.3.2 Pulsed current discharge and charge recovery mechanism

With pulsed current discharge, the battery is drained by a train of current impulses. So

periodically, the battery will be utilized for a short period of time and then it takes a rest time.

With pulsed current discharge, we can overcome some of the drawbacks presented by the

constant current discharge.

In fact, when the battery is allowed to stay idle for sometime after usage, the distribution of the

active materials in the electrolyte may return to be uniform and the concentration may get back

to its average value. Therefore, more active species will be involved in the electrochemical

conversion, and as a result more electrical power can be extracted from the battery. So, in idle

time the nominal voltage of the battery increases, however the theoretical capacity remains the

same and always greater than the nominal voltage.

Figure 3-4, taken from [JAYA04], shows how the pulsed current discharge operates. We can see

that after every current impulse, the battery recovers a portion of its charges, although the new

energy level is slightly lower the previous level.

40
However we should note that the relation between the numbers of charge recovered and the

remaining voltage is directly proportional. It means that the battery will be able to recharge more

if the remaining voltage is higher, and the amount of charges recovered will decrease with lower

level of battery voltage. And that's because with low voltage, the battery contains a small amount

of active materials, which reduces the supply of charges.

Figure 3-4 Operation of the pulsed current discharge

In a) the battery is full. It discharges in b). It switches to an idle state in c) and recover charges

in d). In e) the battery loses more charges and it is incapable of recovering more charges. And

finally it gets empty in f).

41
 Figure 3-5 Pulsed current discharge

In Figure 3-5 we can see the pulsed current discharge. The battery drains power by a sequence of

current impulses followed by an inactive period. As we can notice in the lower part of Figure 3-5

the battery recuperates most of its power. However with time, and with the reduction of voltage

level, the battery recovers less and less charges.

3.2.4 Markov chain representation

The behaviour of the battery under pulsed current discharge may be presented in a discrete time

Markov process (Figure 3-6). Every state is identified by the couple <Ni, Ti> where Ni is the

current nominal voltage and Ti is the current theoretical voltage. The initial state would be

<N, T> where N and T are the maximum values of the nominal and the theoretical voltage, and

42
the last state would be <0, Tmin>, where the nominal voltage is null and the remaining

theoretical voltage is Tmin, which can be equal to zero in an ideal case.

In every state there are three possible transitions:

 The battery may lose one charge unit with a probability pi.

 The battery capacity may stay the same with a probability qi.

 The battery may recover one charge unit with a probability ri.

Figure 3- 6 Discrete Markov chain representation of a real battery behaviour

With pulsed current discharge, if the battery loses one charge unit, its state will change from (Ni,

Ti) to (Ni-1, Ti-1). When the battery reaches the idle time, it will recover one charge unit in one

idle time unit with a probability defined by the function RNiTi [JAYA04]

∗( ) ∅( )
= (1 < < ,
1 < < )
= 0 ℎ

Where, g is a constant value and Φ(Ti) is a piecewise constant function of the number of charge

units delivered. Table 3-1defines the values of the piecewise function Φ(Ti).

This function clearly shows that the charge recovery effect is higher when the battery has higher

remaining capacity and decreases with the decrease in the remaining battery capacity.

43
 Ti Φ(Ti)
200 to 196 0.0
195 to 101 0.0025
100 to 6 0.008
5 to 0 15.6
Table 3- 1 Piecewise constant function of the charge units delivered

3.2.5 WSNs and battery operations

Batteries, as they are the main source of power in most of the cases, are perhaps the most

important devices in wireless sensor nodes. That’s why it is wise to study their activities and to

try to exploit all the benefits they may offer.

In fact, The operation of a sensor node can take the most advantage of pulsed current discharge

effect, because a sensor node usually sends its packets, then it goes back to its back-off period

waiting to access the medium again. During this period, the node will be in idle or sleep mode,

and thus the battery will rest, and it will be able to gain additional capacity.

We note finally that we will use the recharge probability function RNiTi and the piecewise

constant function Φ(Ti) defined in the previous part, in our new protocol.

3.3 CSMA-CA Overview

3.3.1 Introduction

The primary set of protocols that deals with low power and low data rate wireless sensor

networks is defined in the Zigbee standard. This standard is mainly designed to provide some

key features for sensor networks such as the simplicity of implementation, low cost of device

44
installation and maintenance, reliability of data transfer, and most importantly low power

consumptions and thus efficient use of the battery so it can stay alive for long time (months or

even years).

The Zigbee protocol stack is illustrated in the following figure

Figure 3- 7 Zigbee protocol Stack

The actual Zigbee standards are located in the Network and the Application layers, and the IEEE

802.15.4 standard is implemented in the physical and MAC layer [ZIGAL08]

A WSN can be composed of tens to hundreds of devices; one or multiple source nodes may send

packets to one or multiple destination nodes at the same time. To avoid collisions and

miscommunications between the various nodes, the Carrier Sense Multiple Access with

Collision Avoidance (CSMA-CA) algorithm (which is already implemented in IEEE 802.15.4

MAC sub layer) is used. Basically, every source node will have to compete to access the

transmission channel with other nodes that decide to send data in its range of communication.

This competition for channel access is totally controlled and managed by CSMA/CA protocol.

3.3.2 CSMA-CA Communication Modes

There are two main communication modes provided by the MAC sub-layer of IEEE 802.14.5

CSMA-CA protocol: the beacon enabled mode and the non-beacon enabled mode.

45
3.3.2.1 Beacon enabled mode

In the beacon enabled data transfer model, the communication between the nodes is executed in a

superframe structure which is presented in Figure 3-8. A superframe is usually bordered by two

interval periods called beacons, and it is divided into two sections: the active period and the

inactive period. The length of the time interval between two beacons is:

aBaseSuperframeDuration * 2BO . The “aBaseSuperframeDuration” parameter is defined as the

minimum superframe duration. BO is the Beacon Order. The value of BO can vary from 0 to 14.

The length of the active period section of the superframe is defined by:

aBaseSuperframeDuration*2SO. The parameter SO is Superframe Order. The value of SO can

vary between 0 and 14.

The active mode of the superframe is in turn divided into two sections: the Contention Access

Period (CAP) and the Contention-Free Period (CFP). In the CFP period, the coordinator is totally

in control, it gives the exclusive access rights on a channel to the nodes. It may dedicate a portion

of the active superframe to one node. These free time portions are called Guaranteed Time Slots

(GTSs). It may assign up to seven GTS for a node, and each GTS may take more than one slot

period. If a node wants to transmit, it has to request the GTS from the coordinator in a specified

time slot, then the coordinator will check if a free GTS is available (not used by another node). If

it is free, then the coordinator updates the superframe structures and assigns a GTS as long as the

minimum CAP length is not reached.

In the CAP period, the slotted CSMA-CA protocol will be used to arrange the communication

between two devices in the network. In this period, all the nodes have the right to access the

channel by competing with other nodes using CSMA-CA. If a node cannot finish its transmission

46
before the CAP ends, it has to cancel its transmission. The total time slots allocated for the CFP

and the CAP periods is 16 time slots.

During the inactive period, the coordinator and the other devices can be switched to sleep mode

to reduce power consumption.

Figure 3-8 Superframe structure

3.3.2.2 Non-beacon enabled mode

The other mode of communication is the non-beacon enabled mode. In this case there are no

beacons to be transmitted; there is no superframe structure to be used. It simply consists of a

CAP section in which the nodes will contend to access the channel using unslotted CSMA-CA

protocol.

47
 Figure 3-9 a. Non-beacon network communication, b. Beacon network communication

In the beacon enabled network, the coordinator indicates in the beacon that the data is pending.

The device periodically listens to the beacon and transmits a MAC command request using

slotted CSMA/CA if necessary.

In the non-beacon enabled network, a device transmits a MAC command request using unslotted

CSMA-CA. If the coordinator has its pending data, it will transmit the data frame using unslotted

CSMA/CA. Otherwise a data frame with zero length payload will be transmitted.

3.3.3 CSMA-CA operation

It is important to mention first that there are two versions of CSMA-CA algorithm. First one is

used in non-beacon enabled networks and the second one is used in beacon enabled networks. In

both cases, there are basic parameters and variables defined in the implementation of the

algorithm.

 The basic time unit is the back-off period: it is identified by 20 symbols. The actual

duration of the back-off period depends on the frequency band in which Zigbee is

48
 operating. In beacon enabled networks, this period is related to the internal clock of

the coordinator.

 The Beacon Exponent (BE): this variable is related to how many back-off periods a

node shall wait before attempting to access the channel. The waiting period may be a

random value between 0 to 2BE – 1.

 The Number of Back-offs (NB): it is the number of times CSMA-CA algorithm is

required to back-off before trying to access the channel. If this number reaches a

certain threshold defined by the algorithm, the communication terminates with a

failure

 The Contention Window (CW): it is the length of the contention window, it defines

the number of back-off periods that needs to be clear of activity before transmission

can be launched. This parameter is only used with slotted CSMA-CA protocol

(Beacon enabled network)

3.3.3.1 CSMA-CA in a beacon enabled network

As we mentioned before, superframes are used to perform the communication between the

nodes, and packets will be exchanged during the active section using slotted CSMA/CA

algorithm.

The operation of CSMA-CA algorithm in the CAP period is schematically shown in Figure3-10.

It first initializes the parameter NB and CW to the value 0 and 2 respectively. The parameter

macBattLifeExt indicates whether the node is operating on battery power or not. If it is the case,

the parameter BE will be set to 2 or to the constant macMinBE, whichever is less.

49
Next step is to locate the boundary of the next back-off period, because all remaining operations

need to be synchronized to back-off time units. A random waiting time k within the range of 0 to

2BE-1 back-off periods will be generated. And then at each back-off period the value of k will be

decremented by 1. Once k reaches zero, the device listens to the channel to make sure that the

medium is clear before attempting to transmit any frame. This procedure is known as Clear

Channel Assessment (CCA). CCA must be repeated two successive times. If the channel was

found busy, the values of NB and BE will be incremented by one (BE should not exceed a

predefined value macMaxBE) and the value of CW will be set to 2, and then it checks the

number of retries, if it is above to a certain predefined threshold macMaxCSMABack-offs, the

communication will be terminated with a channel access failure status. If the number of retries is

less or equal to macMaxCSMABack-offs, the algorithm returns to the point where the node waits

for a new random number of back-off periods.

If the channel is found idle in both CCA, the node starts transmitting. But in case the channel is

found busy in the second CCA, the algorithm checks whether the remaining time within the CAP

area of the current superframe is sufficient. If it is, the algorithm proceeds, if not it simply stops

and waits the next superframe and resumes the execution from the point it stopped.

In the CFP portion, the communication is totally controlled and managed by the coordinator; the

communication from/to the nodes should be performed by requesting the right of access to the

channel from the coordinator. Therefore the algorithm CSMA-CA will not be used; instead the

coordinator will assign some free slots to nodes that need to transmit, within the area of the CFP

portion.

50
3.3.3.2 CSMA-CA in a non-beacon enabled network

The operation of CSMA-CA algorithm in the beaconless network is schematically shown

Figure3-10. It is very similar to the algorithm in the CAP portion of a beacon enabled network.

The algorithm starts by initializing the parameters NB and BE to 0 and macMinBE, and then it

waits for a random number of back-off periods units within the range 0 – 2BE-1. After that it

needs to assure that the medium is not in usage by other nodes, therefore it waits two CCA

periods and listens to the medium. If it is idle, the node may transmit, however if the channel is

busy, the NB and BE parameters will be incremented by one. If the value of NB becomes greater

than macMaxCSMABack-offs then the transmission will fail, if not the node waits again for a

new random number of back-off periods and repeats the whole process.

51
 Figure 3-10 CSMA-CA operations in beacon and non-beacon enabled modes

3.3.4 Markov-chain Modeling

This is the most important part of this section. We will first represent the operation of the

CSMA-CA algorithm in a Markov chain-based model. Then we will determine the mathematical

equations that define it.

52
The Markov chain-based model of the MAC sub-layer proposed by Pollin et al. [POLL08] is a

very accurate model that succeeded to represent the main properties, characteristics and

operation of CSMA-CA, that’s why our analytical study will be based on it.

3.3.4.1 System description

Figure 3-11 Discrete Markov chain representation of CSMA-CA protocol

In Pollin’s study, a network composed of a fix number of devices (N) is assumed. This network

is supposed to be in a saturated traffic conditions, it means that every node of the network is

always considered having packets to be transmitted.

Two stochastic processes are defined:

53
  c(t): the stochastic process representing the delay line and transmission duration

counters of the device

 s(t) : the stochastic process representing the delay line stages or the transmission

stage.

These processes describe all the possible states that a node could experience to transmit frames.

If the value of s(t) is in the range of {0...NB}, it represents the delay line stage. However if

s(t) = -1, it represents the transmission stage.

When the decrementing delay counter reaches zero, the values c=-1 and c=-2 correspond to the

first CCA and the second CCA respectively.

An important assumption to represent these processes in a Markov chain-based model, is that the

probability to start sensing the medium is independent of all other nodes.

The main goal here is to determine a set of equations that uniquely defines the network

operation. We start by studying the behaviour of a single node. We define first the following

variables:

 Φ: the probability that a node attempts its CCA for the first time within a slot.

 α : the probability of assessing the channel busy during the first CCA.

 β : the probability of assessing the channel busy during the second CCA.

3.3.4.2 Mathematical equations

The two-dimensional Markov chain is characterized by the following transition probabilities:

 { , | , + 1} = 1, ≥ 0 ... eq 3.1

54
 ( )( )
 {0, | , 0} = , < ... eq 3.2

( )
 { , | − 1, 0} = , ≤ , ≤ − 1... eq 3.3

( )( )
 {0, | , 0} = + / , ... eq 3.4

Eq 3.1 presents the condition to decrement the delay line counter per slot.

Eq 3.2 presents the case where the channel was sensed idle two consecutive times (during the

first and the second CCA) and the node was able to transmit packets.

Eq 3.3 presents the case where the channel was sensed busy in one of the CCA time slots, and

the node selects a new state in the next delay level

Eq 3.4 gives the probability of starting a new transmission attempt when leaving the last delay

line, following a successful or failed packet transmission attempt.

Pf is the probability of transmission failure. In fact the number of transmission attempts is

limited, after exceeding this number the node will move to a state of transmission failure.

Wi is the delay window. Its initial value is W0 = 2aMinBE , and then this value will be doubled

every state until it reaches the maximal value : Wi = Wmax = 2aMaxBE (aMaxBE-aMinBE ≤ i ≤

NB).

Between W0 and Wmax, Wi is defined as: = ∗2 … 3.5

Assume now that the Markov chain’s steady-state probability is , = {( ( ), ( )) = ( , )}

for ∈ {−1, } ∈ {−2, max(L − 1, W − 1)}

, = lim → ( ( ) = , ( ) = ).

55
Our aim is to determine the system of equations that define the Marko chain model. The

variables that defines this system are Φ, α and β.

The node can be in one of the following states:

 Back-off stage : where the node waits before listening to the channel to check if it is

busy or not

 First CCA: once the back-off counter reaches zero, the node checks if the medium is

free for the first time

 Second CCA: if in the first CCA the channel was free, the node will listen for the

second time to verify that no other node is transmitting

 The node is transmitting: the duration of transmission is measured in slots. L is the

total number of slots.

The sum of all probabilities should be equal to one.

, + , + , + , = 1… 3.6

Using the equations eq 3.1 to eq 3.4 we can get:

, ∗ ( + (1 − ) ) = , 0 < ≤ … eq 3.7

−
, = (1 − )(1 − ) , + = 0 … eq 3.8

−
, = , > 0… 3.9

56
The probability of the transmission failure Pf is calculated as:

= , ( − + )… 3.10

From eq 3.5 we can get:

, = [( + (1 − ) ) ] ∗ , 0 < ≤ … eq 3.11

In order to calculate eq 3.6 we will substitute the parameter Wi and the variables bi,k by their

values from the equations: eq 3.5, eq 3.7 to eq 3.11, the equation eq 3.6 becomes:

, 1−
[ 1 + 2(1 − ) + 2 ]∗
2 1−

, −
+ 2 ∗ ∗
2 1−

, 1 − [2 ]
+ ∗ = 1 … eq 3.12
2 1−2

Where x = (1-α) * (1-β); and y = α + (1-α)*β.

Eq 3.12 is the first equation to define the two-dimension Markov chain, what remains are the

equations related to Φ, α and β.

The equation that derives the probability of the node to attempt its CCA for the first time within

a slot is:

= , … eq 3.13

Using eq 3.11 and eq 3.13 we obtain:

57
 =∑ ∗ , … 3.14 .

Eq 3.14 represents the second equation to define the system.

The calculation and the explanation required to determine the probabilities α and β are too long

and they are out of the scope of this report, therefore we will just use the equations presented in

[POLL08].

= ∗ [ 1 − (1 − ) ] ∗ = 1 − 1 − ... eq 3.15
∗

1
= 1− ∗ [1 − (1 − ) ] =1− 1− … eq 3.16
1 1−
1+
1 − (1 − )

If we replace the value of b0,0 (eq 3.12) in eq 3.14, we obtain a system of three non-linear

equations (eq 3.14, eq 3.15, and eq 3.16) that defines the network operation point. The values of

these three variables (α, β and Φ) are sufficient to determine all the performance metrics of the

network, such as throughput, collision percentage, energy consumption, end-to-end delay, energy

recovered and the lifetime of the network as we are going to prove later in the next chapter.

3.4 Summary

Battery characteristics and IEEE 802.15.4 CSMA-CA protocol are the basis of our new work. In

this chapter we have overviewed their theories and their Markov chain representation. In the next

chapter we will use them develop the battery-aware CSMA-CA protocol.

58
Chapter 4 Battery-Aware CSMA-CA Protocol

4.1 Introduction

In this chapter, we will introduce the basics of our novel approach, and we will prove that the

modifications we suggest to add on the CSMA-CA protocol will improve its performance by

increasing the average lifetime of the node and the total life duration of the network.

In fact the network lifetime is a key characteristic for evaluating sensor networks; it includes

especially the availability of nodes, the sensor coverage, and the connectivity of the nodes

[DIET09]. That’s why many researchers investigated via simulations and experiments the

estimation of WSNs lifetime [NGUY11], [KERA10], [BARB08].

We will extend the previous Markov chain representation to define our new model that includes

the charge recovery mechanism. Then we will validate our idea by solving numerically the

mathematical equations in C language and we will demonstrate that the performance of the

algorithm has been improved in many aspects.

This chapter consists of five main parts. The first part is the current introduction. The second part

is about the system description. We present our algorithm in the third part. We show and discuss

our results in part four and we summarize in part 5.

59
4.2 System description

As we previously explained in the preceding chapter, a battery may gain some charge units if it

remains unused for some period of time, which means that the node power will grow if this node

is in idle or sleep mode (back-off period).

According to the conventional CSMA-CA, the node has Wi possible idle state, and in each state

there is a probability that the remaining voltage of the node battery will increase due to the

recovery capacity effect. This probability is defined by:

∗( ) ∅( )
= (1 < < ,
1 < < )
= 0 ℎ

Then the node listens to the medium. Depending on the availability of the channel, the battery

may lose energy in the first CCA, in the second CCA and/or in the L time slots of packet

transmission, before going back to the next level of back-off stage where it will probably recover

more charges.

Figure 4-1 illustrates the behaviour of the node battery when applying CSMA-CA algorithm.

We assume that during this back-off period the losses of battery power are very minimal so we

neglect them. Therefore we assume that the battery at every state during this period may keep the

same voltage value or it may recover one charge unit. However the voltage level cannot exceed

at any time the current nominal neither the theoretical value of the battery voltage.

60
 Figure 4-1 Real battery behaviour of a node during CSMA-CA protocol execution

In Figure 4-1, the initial state of the battery at the beginning of the back-off period is <Vi,Ti>.

After one time slot, the battery may recover one charge unit and its state changes to

<Vi+1,Ti+1> or it may stay in its initial state without gaining any new charges <Vi,Ti>. Again

the same process repeats after each time slot in the back-off phase, the battery voltage level may

increase with an additional charge unit or it may keep the current voltage value. Then it loses one

charge unit after CCA1 and CCA2 (if the channel was clear), and drops by L charge units when

the node transmits the packet.

4.3 Algorithm

We assume first that the discharge rate of the node battery in each of the following state is one

charge unit per state: the first CCA, the second CCA and each of the L states.

61
 Figure 4-2 Battery nominal voltage at every stage of CSMA-CA protocol

In Figure 4-2 we summarize our algorithm. For the sake of simplicity, we only illustrate one

cycle of operation and we assumed the best case scenario in which the channel is sensed free in

two consecutive CCA, and in which the node succeeded to transmit its packets.

We assume that our cycle starts at a point where the node is at the first CCA. We assume that Vi

is the nominal voltage of the battery at that state. The node will then check the medium, if the

medium is free, the new node state will be in the second CCA and the battery voltage will drop

by one charge unit. The new nominal voltage will be: (Vi-1) charge units.

The probability to reach the second CCA is (1-α). The node will sense the channel again, and

there is a probability of (1-β) that the channel is free; if it is, the node will jump then to the

transmission state and the battery voltage will drain an additional charge unit. The new nominal

voltage will be: (Vi-2) charge units.

62
The node will start transmitting the packet. The duration of the transmission is assumed to be L

time slots, at each slot the capacity of the battery will decrease one charge unit. At the end of the

transmission the new nominal voltage will be: (Vi - 2 -L) charge units.

The node will return to the back-off period in one of these three cases:

 After the first CCA, if channel was sensed busy.

 After the second CCA, if channel was sensed busy.

 or After transmitting the packets

There are Wi states, and the probabilities to be in one of these state are equal and uniform (1/Wi),

Therefore, in order to simplify the calculation, we assume that the number of charges that can be

recovered during this period is the average value:

W ∗( ) ∅( )
∗ R , = ∗
2 2

The new nominal voltage will be: (Vi - 2 -L +ER) charge units, where ER is the energy

recovered.

What follows is our detailed algorithm.

% Definition of the parameters

N = 50; % Number of nodes in the network

aMinBE = 3; % Minimal value of BE
aMaxBE = 8; % Maximal value of BE
macMAxCSMABack-offs = 5; % Maximum number of Back-offs before the transmission is
%considered a failure

L = 14; % 14 time slots duration of transmission

g = 0.025; % constant related to the characteristics of the materials in
%the battery

63
Vmax=65,000; %Initial value of the nominal voltage: 65000 charge units
Tmax=100,000; % Initial value of the theoretical voltage: 100,000 charge
%units

% Variable Declaration
Int i; % index

V[i]; % Remaining nominal voltage in the battery

% V[0] is assumed to be the initial nominal voltage (Vmax)

T[i]; % Remaining theoretical voltage in the battery

%T[0] is assumed to be the initial theoretical voltage (Tmax)

PhiT[i] % The variable that defines the piecewise constant function

% included in the recharge probability function

Alpha; Betta, Phi; % the variables that define the Markov chain model
ELost; % Energy lost at each state
ER; % Energy recovered
Per; %Recover probability function

p1, p2, p3; % Variables to calculate the probabilities

BE; % Back-off Exponent
NB; % Number of back-offs
W0, W; %Back-off parameter

TimeSim[i] % Simulation time

TR; % Number of retries in case of a collision
fail; % Boolean variable. In case of transmission failure,
%fail=1;

% Body of the algorithm

1. Solve the nonlinear equation system to calculate Alpha, Betta and Phi;

2. Define the function that helps to determine PhiT

function ValueofV(double v){

if v is greater than 0.975 then return 0;

if v is between 0.5 and 0.975 then return 0.0025;
if v is between 0.025 and 0.5 then return 0.008;
if v is less than 0.025 then return 15.6;

64
3. Initialize the parameters
i = 0;
V[0]=Vmax;
T[0]=Tmax;
BE=aMinBE;
TR=0;
W0=2^BE;
i=0;

4. Calculation

While V[i]>=0 do { % in each cycle we do the following

i = i+1
The node loses one charge unit at the first CCA
Elost=1;

Choose a random value of p1 between 0 and 1;

Compare p1 to (1-Alpha);

If p1<=1-Alpha then
{
The node is in the second CCA and it loses one charge unit.
So increment Elost by one;
Elost = Elost + 1;

Then choose another random value p2 between 0 and 1;

Compare p2 to (1-Betta);
If p1<=1-Betta then
{
The node is transmitting and it loses L charge units;
So increment Elost by L;
Elost = Elost + L;

Then we check if there is a collision during the transmission

If collision then
{
Increment transmission failure counter
TR = TR + 1;

Initialize the parameter BE and NB

BE = aMinBE;
NB = 0;

}
}

65
 else
p2 is less than 1-Betta, in another word if the channel was not clear at the second CCA,
so the node failed to transmit,

Update fail
fail = 1;

} else
p1 is less than 1-Alpha, in another word if the channel was not clear at the first CCA,
so the node failed to transmit,

Update fail
fail =1;

In case of a transmission failure

If fail==1 then
{
Increment NB by one
NB = NB + 1;

Then check if NB exceeds macMaxCSMABack-offs

If NB>maxMaxCSMABack-offs then
{
Reset the values of NB and BE
BE=aMinBE;
NB=0;
}
Else
If it is less then macMaxCSMABack-offs
Increment the value of BE by one as long as it is less then aMaxBE
Update the value of BE
BE = min(BE+1,aMaxBE);
}

Reset the parameter fail

fail =0;

Update the value of W0

W0 = 2^BE;

The new nominal voltage of the node battery is decreased by Elost,

Update V[i]
V[i] = V[i-1] – Elost;

66
The new theoretical voltage of the node battery is decreased by Elost,
Update T[i]
T[i] = T[i-1] – Elost;

Whether the node failed at the first CCA, failed at the second CCA, failed during transmission
(collision), or transmitted successfully, it is now at the back-off stage,

Set the back-off period W, it is a random value between 1 and W0

W=random();

Now at every state of the back-off period, there is a probability Pre to gain back one charge unit,
At every state of the back-off period do the following
for W states do
{
Calculate the function Phik[i] to be able to determine the probability of recharge
Update Phik[i],
Phik[i]=ValueofV(T[i]/T[0]);

Calculate Pre
Pre = exp(-g*(V[0] – V[i]) – Phik[i]);

Create a probability p3 and compare it to Pre,

if p3 is greater than Pre, then the node recuperate one charge unit

If p3 < Pre then {

Increment the energy recovery variable ER by one
ER = ER + 1;

The sum of the current nominal energy and the energy recovered should be less then
Vmax and less then Tmax,

If ( (V[k]+ER <= V[0]) && (V[k]+ER <= T[k]) ) then

{
The new battery capacity is
V[k] = V[k] + ER;
}
}

}%end of the while loop

67
4.4 Results and validation

To validate our scheme, we implemented our mathematical model in C language. To reveal the

benefits of introducing the charge recovery mechanism into the plain CSMA-CA protocol, we

calculate the following performance metrics and we compare them with the conventional

CSMA-CA: Average charge recovery, average lifetime of a node and total lifetime of the

network.

4.4.1 Average charge recovery

It is the average amount of charges recovered by the network. In the usual CSMA-CA protocol,

the capacity recovery effect is not considered and thus no charges are recovered. However, in our

proposed model, the nodes of the network are able to re-gain a portion of the charges when they

become idles.

Figure 4-3 illustrates the average quantity of charges recuperated by the nodes in networks with

different sizes. We note that the initial nominal and theoretical voltages are 65,000 charge units

and 100,000 units respectively when the battery of the node is full. Therefore the maximum

number of charges we can recover is the difference between these two values which equal to

35,000 charge units. The average number of charges recovered with the network consisting of 50

nodes is 11,356 charge units which is equivalent to 32.5% of the total amount.

68
 12000

10000
Charges Recovered

8000

6000

4000

2000

0
0 10 20 30 40 50 60
Number of Nodes

Figure 4-3 Charges recovered with the new scheme

4.4.2 Average lifetime of a node

The lifetime of the node is the period of time during which the node has power. It starts with the

first moment the node is activated and ends when the battery is completely drained out of energy.

We proved in the previous part that the number of charges used by the nodes is greater with the

new scheme than with CSMA-CA, therefore we expect that the average lifetime of the node to

be extended accordingly.

In Figure 4-4 we show the average lifetime of a node in different networks, and we compare the

results of the new scheme with CSMA-CA. At N=50 nodes, the average lifetime of a node with

CSMA-CA is 335 seconds, and 392 seconds with the new scheme. This presents an enhancement

of 17.2%.

69
 450.00
Node Average Lifetime (seconds)
400.00
350.00
300.00
250.00
200.00 CSMA_CA
150.00 New Scheme
100.00
50.00
0.00
0 10 20 30 40 50 60
Number of Nodes

Figure 4-4 Average lifetime of a node

4.4.3 Lifetime of the network

We define the lifetime of the network as the duration of time in which at least one node is still

alive. It begins when all the nodes are activated and running and terminates when the last node of

the network expires.

Figure 4-5 clearly demonstrates that the network with our proposed scheme survive longer then

the network with CSMA-CA protocol. In fact, there is an improvement of 17.2%.

70
 450
400
Network Lifetime (seconds)

350
300
250
200 CSMA-CA
150 New Scheme
100
50
0
0 10 20 30 40 50 60
Number of Nodes

Figure 4-5 Total Lifetime of the network

We achieve the same conclusion in Figure 4-6 and Figure 4-7 which compute the number of

remaining alive nodes in case of N=25 nodes and N=50 nodes.

The results presented in Figure 4-6 show that all the nodes of the network become inactive after

259 seconds and 293 seconds with CSMA-CA and with the new scheme respectively. So, the

number of alive nodes has incremented by 13%.

30
Number of Remaining Nodes

25

20

15
CSMA-CA
10 New Scheme

0
0 50 100 150 200 250 300 350
Time in Seconds

Figure 4-6 Number of remaining alive nodes (N=25)

71
With a bigger network composed of 50 nodes, the improvement is higher. The number of

remaining alive nodes over time has increased by 17.2% (see Figure 4-7)

60
Number of Remaining Nodes

50

40

30
CSMA-CA
20 New Scheme

10

0
0 100 200 300 400 500
Time in Seconds

Figure 4-7 Number of remaining alive nodes (N=50)

4.5 Summary

In this chapter, we demonstrated that incorporating the battery characteristics, precisely the

pulsed current discharge and the energy recovery mechanism, in the design of CSMA-CA

protocol had a great impact on the overall lifetime of the network and on the average life

duration of the nodes. In fact, we verified through analytical simulation, that the total lifetime of

the network has raised by more than 17% compared to the plain CSMA-CA, after considering

the battery effects. Moreover, we proved that the nodes could recover more than 32% of the

unused charges.

These results present evidence that new approaches can be developed in the MAC sub-layer of

the WSNs protocol stack to save more energy, to optimize the usage of power and to increase the

lifetime of the network. We refer to these approaches as "Battery-aware MAC protocols". In the

72
following chapter, we will propose two new Battery-aware MAC protocols, in which the MAC

sub-layer take into account the battery state in its computation to attain superior outcomes.

73
Chapter 5 Battery-Aware Energy-Efficient

MAC Protocols

5.1 Introduction

As we stated earlier in this thesis, only few battery-aware MAC protocols that consider battery

characteristic factors have been reported. The majority of the existing MAC protocols that deal

with the power consumption matter and the energy conservation subject are founded on the

development of new duty-cycle schemes in the MAC sub-layer, without any connection to the

actual status of the battery. Moreover, all these protocols do not consider the real

charge/discharge operations that take place in the battery, and they don’t take advantage of the

existing effects and mechanisms that occur during these processes.

For example, in a large network, where the number of sensor nodes exceeds 50, the node’s idle

time can be higher that 97% of the time. This means that the node transmits packet in less than

3% of the time and rests for the remaining period. Therefore, taking into consideration the pulsed

current discharge effect and the charge recovery mechanism of the battery will firstly show a

realistic performance of the nodes, secondly it will allow a better modeling of the wireless sensor

network behaviour, and thirdly it will demonstrate enhanced results in terms of power savings

and network lifetime.

74
In this chapter, we propose two battery-aware MAC schemes, and investigate their advantages

and disadvantages in the context of the CSMA-CA protocol. The target is to exploit the capacity

recharge effect in order to extend the lifetime of the node/network, and enhance the throughput.

Therefore, we propose two new back-off methods that consider the battery’s current capacity in

their functionalities. In another word, the back-off period defined in CSMA-CA protocol will be

changed in our proposed protocols; it will be related to the remaining battery voltage at the

actual time of operation, and thus the back-off duration before attempting to access the channel,

will depend on the current battery state.

And eventually we explore, through simulations, the performance of our approaches compared to

each other, and compared to the conventional CSMA-CA protocol in terms of node/network

lifetime, throughput, percentage of collisions, and energy recovered.

5.2 Our proposed protocols

Based on the results we attained in the previous chapter, we know now that introducing the

probability of battery recharge in the functionality of the CSMA-CA protocol will improve its

performance in terms of power recovery and lifetime extension. So the question is how to exploit

this conclusion to design a new enhanced protocol?

In addition to that, we know also from the previous chapters two important things:

 The amount of charges that the battery gains, depends on the duration of the resting

time. The longer the battery remains unused (in idle or sleep mode), the higher is the

amount of charges it recovers

75
  The amount of charges that the battery recovers, depends on the current remaining

voltage. The battery recuperates more charges in case of high residual voltage than

with small remaining energy.

The abovementioned points led us to think about the following:

In order to achieve better network performance, which nodes should have higher priority to rest

more? Should we give priority to the nodes with higher residual voltage to relax longer than the

nodes with lower voltage level? Or should it be the opposite way?

In fact, based on the first point, we should offer more relaxing time to the nodes with low battery

level so they can recover more charges and live longer. However, the second point implies that

these nodes will not be able to gain enough power due the lack of active materials in the

batteries.

Conversely, the second point suggests that we should provide the nodes with higher residual

capacity the priority to rest longer, because their batteries will be capable of gaining the

maximum amount of charges. However, the other nodes with low battery power will be

intensively used and they will be exhausted so quickly, then the overall operation of the network

will be affected.

Based on this brief discussion, we propose two battery-aware MAC protocols:

 The first protocol let the nodes with lower battery power to stay idle longer than the

nodes with higher battery power. It means that the nodes with higher remaining

voltage will access the medium faster. We refer to it as MAC-HV.

76
  The second protocol allows the nodes with higher remaining voltage to relax more

than the nodes with lower remaining voltage. It means that the nodes with lower

remaining voltage will access the medium faster. We refer to it as MAC-LV.

And then we will study the performance of both protocols, explore their impacts on the overall

operation of the network, and finally compare their results with the IEEE 802.14.5 CSMA-CA

protocol.

5.2.1 MAC-HV Protocol

In this protocol, we propose that the nodes with higher remaining voltage in their batteries have

the priority to access the medium before the nodes with lower remaining voltage. In other words,

with this protocol, the nodes with lower remaining voltage will stay in idle or sleep mode longer

than the nodes with higher remaining voltage.

Consequently, we need to define a new back-off function that takes into account the current

residual voltage in the battery, and use it to extend the idle period of the nodes with lower

residual voltage, and thus give them low priority to access the medium. We refer to this back-off

function as BP-HV.

Assume that V is the initial value of the voltage when the battery is full, and Vi is the current

remaining value of the voltage after some time of operation. We define the back-off period

function of MAC-HV as:

− = (1: 2 + 2 ∗ (1 − ))

77
It is clear here that when Vi is closer to V (high voltage) the value of the BP-HV decreases. This

means that the node with higher value of voltage will wait less time to access the medium than

the other nodes. And if Vi is much smaller than V the value of BP-HV increases (low voltage),

so the node with low value of voltage will wait longer to access the medium.

It is important to note that V and Vi represent the nominal voltage of the battery, not its

theoretical voltage.

5.2.1.1 Flow chart

The follwoing flow chart illustrates how MAC-HV protocol operates in the case of CSMA-CA

protocol. We highlighted in green the blocks where we include our modifications.

78
 Figure 5-1 Flow chart of MAC-HV protocol

5.2.1.2 Algorithm

The protocol MAC-HV operates as follows:

% First step is the initialisation

If a node wants to transmit a packet then

79
Initialise the local parameters, BE, NB and CW.

BE = macMinBE;

NB = 0;

CW = 2;

% Second step (after locating the back-off period boundary in case of slotted CSMA-CA) is to

read the current battery voltage

Read the current remaining voltage of the battery;

Vi = current voltage;

% Third step is to calculate the back-off period and stay idle during this period

Calculate Back-off-period

Back-off -period=random(1 : 2BE + 2BE * (1-Vi/V));

While (back-off-period > 0) do

Stay idle;

Decrement back-off-period;

Back-off-period = back-off-period – 1;

% forth step is to perform the double clear channel assessment

Perform CCA1;

If the channel is idle then

80
Decrement CW by one

CW = CW -1;

If CW = 0 then

Transmit the packets

Else

Return to the forth step

Else

Update the parameter values.

BE = min (BE + 1, aMaxBE);

NB = NB + 1;

CW = 2;

Check now if the number of retries is higher than the threshold value

If NB > macMaxCSMABack-off then

Packet transmission has failed;

Else

81
 {

Return to the second step;

% End of the algorithm

5.2.2 MAC-LV Protocol

This protocol grants access to the medium to the nodes with low remaining capacity, these nodes

will have the precedence right to use the channel before the nodes with higher power. This

means that, in this approach, the nodes with higher remaining capacity will have the chance to

relax more than the other nodes.

As a consequence, we should identify a new back-off function that considers the actual battery

voltage in its computation, and that allows the low power nodes to transmit their packets before

the high power nodes. We refer to this back-off function as BP-LV.

BP-LV is defined as:

− = (1: 2 + 2 ∗ )

Vi and V are always the current nominal voltage and the initial nominal voltage respectively

In this case, BP-LV decreases when Vi decreases and increases when it does. A node with weak

battery (Vi is too small) will guarantee access to the medium before the nodes with high battery

voltage.

82
5.2.2.1 Flow chart

We demonstrate the protocol MAC-LV functions in the next flow chart. The green blocks

present the changes we applied to the conventional CSMA-CA protocol. The main difference

with MAC-HV protocol is the calculation of the back-off function.

Figure 5-2 Flow chart of the MAC-LV protocol

5.2.2.2 Algorithm

The algorithm of the protocol MAC-LV is very similar to the algorithm of MAC-HV described

in details in the previous section. The main difference is the calculation of the back-off period.

83
% First step is the initialisation

If a node wants to transmit a packet then

Initialise the local parameters, BE, NB and CW.

BE = macMinBE;

NB = 0;

CW = 2;

% Second step (after locating the back-off period boundary in case of slotted CSMA-CA) is to

read the current battery voltage

Read the current remaining voltage of the battery;

Vi = current voltage;

% Third step is to calculate the back-off period and stay idle during this period

Calculate Back-off-period

Back-off -period=random(1 : 2BE + 2BE * Vi/V);

While (back-off-period > 0) do

Stay idle;

Decrement back-off-period;

84
Back-off-period = back-off-period – 1;

% forth step is to perform the double clear channel assessment

Perform CCA1;

If the channel is idle then

Decrement CW by one

CW = CW -1;

If CW = 0 then

Transmit the packets

Else

Return to the forth step

Else

Update the parameter values.

BE = min (BE + 1, aMaxBE);

NB = NB + 1;

85
CW = 2;

Check now if the number of retries is higher than the threshold value

If NB > macMaxCSMABack-off then

Packet transmission has failed;

Else

Return to the second step;

% End of the algorithm

5.3 Simulation and performance analysis

In this section we study the performance of our proposed schemes and compare them with the

performance of the conventional CSMA-CA protocol. We implemented a simulator for the

energy efficient battery aware MAC protocols in the C programming language. We verified first

the accuracy of the simulator by running it over multiple networks, with different sizes, using the

conventional CSMA-CA protocol, and we found that the outcomes are totally compatible.

We obtained our results by executing the simulator over ten networks with the following sizes: 5,

10, 15, 20, 25, 30, 35, 40, 45 and 50 nodes.

The simulator stops only when all the nodes of the network run out of power.

86
We assume the following:

 Each node with enough power can participate in the communication process

 In case of packet transmission, two possible actions may occur: the transmission is

completed successfully or there was a collision during the transmission.

 If the transmission was successful, an ACK packet will be generated by the

destination

The performance is studied in terms of the following parameters: channel utilization, percentage

of collisions, average idle time, average energy recovered by a node, average lifetime of a node,

the total lifetime of the network, and the number of remaining alive nodes over time.

We define these metrics as follows

 The channel utilization (or the throughput) is the fraction of the total successful

transmission time over the total time of the simulation.

 The collision is the fraction of the total collision time over the total time of the

simulation.

 The average idle time is the fraction of the time when the channel was idle over the

total time of the simulation.

 The average energy recovered by a node is the sum of energy recovered by all nodes

divided by the number of nodes

 The average lifetime of a node is the ratio of the total life time of all nodes in the

network over the total number of nodes

 The total lifetime of the network is the period of time that starts at the beginning of

the simulation and ends when the last node of network expires

87
  The number of remaining alive nodes over time is the number of nodes in the network

that are still involved in the communication over time.

The energy consumed by a node during the simulation includes the following:

 The energy consumed during the clear channel assessments (CCA 1 and CCA 2)

 The energy consumed during the successful transmission

 The energy consumed during failed transmission (collision)

 The energy consumed during receiving data

 The energy consumed during ACK

The energy consumed during idle time was neglected.

5.3.1 Simulation Parameters

Our simulation parameters are adopted from [JAYA04], [CHIA99c] and [POLL08], and listed

in Table 5-1and Table 5-2.

Parameter Value
g (Battery parameter) 0.05/mJ
Theoretical capacity (c.u) 100,000 charge units
Nominal capacity (c.u) 65,000 charge units
Theoretical capacity (j) 0.4 Joules
Nominal capacity (j) 0.2 Joules
Energy recovery 0.05 mJ /idle timeslot
Table 5-1 Simulation parameters

88
 Rx 40
Power Tx 30
Consumed CCA 40
(mW) Sleep 0.8
1 timeslot 0.32 ms (80 bits)
Packet Length (L) 14 timeslots
Durations ACK Packet Length (LACK) 2 timeslots
Simulation Time 320 s
802.15.4 macMaxCSMABack-offs 5
Parameter macMinBE 3
Settings macMaxBE 8
Table 5-2 Simulation parameters
We note that every time a node sends a packet, it loses one charge unit. On the other hand, every

time a node rests for one time slot, there is a probability RNiTi that it will recover one charge unit.

In the following subsections we show our results along with our analysis and discussions.

5.3.2 Channel utilization

In Figure5-3 we show the performance in terms of channel utilization. We can clearly see that

our proposed schemes outperform the CSMA-CA protocol in terms of this parameter. Also, we

can see that MAC-LV performs much better than MAC-HV. In fact, the CSMA-CA protocol

does not rely in its computations on the battery’s remaining voltage. This is in contrast to the

other two proposed schemes. In these schemes, the battery recovers partially its capacity, and

therefore, it allows the node to conserve more energy and send more packets. We can notice

from Figure 5-3 that the channel utilization under the MAC-LV protocol surpasses that of the

CSMA-CA protocol by 18.75%, while it surpasses the MAC-HV protocol by 13.74%.

89
 Figure 5-3 Channel utilization (Throughput)

5.3.3 Percentage of collisions

Figure 5-4 Percentage of collisions

In Figure 5-4 we illustrate the performance in terms of the percentage of time the channel is in

the collision state. This figure can explain the results we showed for the channel utilization. As

we can see, the percentage of collisions under the CSMA-CA protocol is higher than that of both

MAC-LV and MAC-HV. The fact that MAC-LV and MAC-HV allow the nodes to send more

90
packets, and that they achieve lower percentage of collisions, result in the enhancement we see in

the channel utilization. Figure 5-4 proves that the collisions under the MAC-LV protocol are

reduced by 42.77% compared to the CSMA-CA protocol, while the reduction is by 35.2%

compared to the MAC-HV protocol.

5.3.4 Channel idle time

In Figure 5-5 we show the performance in terms of the achieved channel idle time. This figure

explains the reason behind that that MAC-LV is performing better than MAC-HV. We can

clearly see that the idle time of the nodes under MAC-LV is greater than that under MAC-HV.

This means that the nodes with MAC-LV have the opportunity to rest longer, and therefore,

recover more charges. This enables the nodes to send more packets and extend their lifetime. The

channel idle time under the MAC-LV protocol exceeds that under the CSMA-CA protocol by

21.87%, while it exceeds that under the MAC-HV protocol by 12.6%.

Figure 5-5 Channel idle time

91
5.3.5 Average energy recovered

In Figure 5-6 we show the performance in terms of the average energy recovered per node.

Surprisingly, MAC-HV recovers, on average, more energy that MAC-LV. However, most of this

energy is lost in retransmitting packets, due to the high collision percentage shown in Figure 5-4.

Figure 5-6 Average energy recovered per node

5.3.6 Average lifetime of a node

Figure 5-7 average lifetime of a node

92
In Figure 5-7 we show the performance in terms of the average lifetime of a node. As expected,

the average lifetime of the nodes under the CSMA-CA protocol is shorter than that under our

proposed schemes. Interestingly, we can observe that the average lifetime of the nodes with

MAC-HV goes beyond that with MAC-LV. We can explain this behavior by referring to the

results shown in Figures 5-9 and 5-10 (which we will detail more in the next subsection). In

these figures, we can observe that most of the time, the number of the remaining alive nodes with

MAC-HV is more than that with MAC-LV, even though MAC-LV eventually outperforms

MAC-HV. Consequently, if we calculate the average value of both curves in both graphs, we

will find that the average value of MAC-HV exceeds that of MAC-LV.

According to Figure 5-7, at N = 50, the maximum value of the node’s average lifetime with

CSMA-CA, MAC-LV, and MAC-HV is around 33 seconds, 42 seconds, and 47 seconds,

respectively. Based on these numbers, MAC-LV and MAC-HV are superior to CSMA-CA by

29.56% and 43.08%, respectively. Meanwhile, MAC-HV is superior to MAC-LV by 9.44%.

5.3.7 Lifetime of the network

Figure 5-8 Total lifetime of the network

93
In Figure 5-8 we show the performance in terms of the total lifetime of the network. This figure

proves that MAC-LV can prolong the lifetime of the network more than MAC-HV and CSMA-

CA. In fact, the network lasts around 82 seconds with MAC-LV, around 68 seconds with MAC-

HV, and around 59 seconds with CSMA-CA. This is an improvement by 40.31% for MAC-LV

over CSMA-CA and by 21.14% for MAC-LV over MAC-HV.

Figure 5-9 Number of remaining alive nodes (N=25)

The same conclusion is drawn from Figures 5-9 and 5-10. They clearly show that with the MAC-

LV protocol, the number of remaining alive nodes is superior, especially for bigger networks.

Figure 5-10 Number of remaining alive nodes (N=50)

94
In the case of N=25 (Figure 5-9), all the nodes become disabled after 38s, 44s, and 51s have

passed, with CSMA-CA, MAC-HV and MAC-LV, respectively. Therefore, the number of alive

nodes with MAC-LV is increased by 31.71% compared to CSMA-CA, and it is better by 14.66%

compared to MAC-HV. With bigger networks (Figure 5-10), the enhancement over CSMA-CA

is 40% and over MAC-HV is 20.87%.

Moreover, as we mentioned earlier, the charge recovery mechanism shows that this effect will be

higher (lower) when the battery has higher (lower) remaining capacity. MAC-LV exploits this

effect by giving priority for medium access to the batteries that have low power, and leaving the

nodes with high remaining capacity to rest more. In fact, having a battery with low power

indicates that the remaining quantity of active material in it is very small. Therefore, that battery

will not be able to recover more charges, and so leaving it idle for some period of time will not

add any benefit in terms of conserving energy. In contrast, nodes with high voltage batteries can

recover more charges if they rest longer.

5.4 Summary

In this chapter we have presented our proposed versions of battery-aware energy-efficient MAC

protocol for wireless sensor networks: MAC-HV and MAC-LV. Unlike most of the reported

power-aware and energy-efficiency MAC protocols, our protocols incorporate the actual voltage

of the battery in their computations, and they take advantage of the charge recovery effect that

extend significantly the lifetime of the battery.

With MAC-HV the nodes with higher battery voltage gain the precedence to access the medium

by prolonging the back-off period of the nodes with lower battery voltage, which give them the

chance to rest and recover charges.

95
On the contrary, with MAC-LV the nodes that have low battery capacity have the priority to

used the channel before the nodes with high battery capacity. In this case, the nodes with high

battery power backs-off for longer duration, and thus they earn more energy.

We run both protocols, in addition to the usual CSMA-CA protocol, through extensive

simulations focusing on some important performance metrics such as energy gained, lifetime of

the network, throughput and collision rate. Both protocols proved better performance than

CSMA-CA, and the results confirmed that the best performance is achieved when allowing the

nodes with higher remaining voltage to relax more than the nodes with low remaining voltage.

(MAC-LV is better than MAC-HV).

These results are completely compatible with the conclusions attained in the battery behavior

studies: the charge recovery increase with longer period of relaxation and the amount of charges

recovered augments with higher voltage values.

96
Chapter 6 Conclusions and Future Work

6.1 Summary and Concluding Remarks

In this thesis we propose two novel schemes of battery-aware MAC protocols, namely, MAC-

HV and MAC-LV that take into account the real complex behavior of the sensor node’s battery.

The motivation behind these schemes is to exploit the charge recovery mechanism of the battery

such that more energy is conserved. The direct benefit of that is prolonging the lifetime of the

WSN. Both schemes showed significant improvement over CSMA-CA in terms of throughput,

energy recovered and the average lifetime of the nodes/network.

In these protocols, we identified two new back-off functions that depend on the current

remaining voltage of the battery, we called them BP-HV and BP-LV. The back-off function BP-

HV decreases with higher values of the remaining voltage and increases with smaller values. So

with BP-HV, the nodes with high remaining power back off for shorter duration than other

nodes, and thus they grant access to the channel before the other nodes.

On the contrary, BP-LV function is directly proportional to the battery residual voltage, it

increments when the voltage increments and vice versa. With this back-off function, the nodes

with small current battery capacity wait less during the back-off stage, and therefore they gain

the right to use the medium before the other nodes.

97
Prior to designing the new protocols, we presented an analytical study to demonstrate the gain

that can be offered by the charge recovery concept of the battery. In fact, we applied that concept

to the conventional CSMA-CA protocol, we included the probabilities of gaining or losing

charges by the battery during the different stages (back-off stage, first and second clear channel

assessment and packet transmission/reception/collision stages), and finally we verified that

considering the capacity recovery effect in the communication protocol in WSNs increase the

life duration of the network.

At the end of the thesis, we run our proposed protocols through extensive simulations focusing

on the performance in terms of channel utilization, collisions rate, average energy recovered per

node, and the lifetime of the network. We provide head-to-head comparison of MAC-LV, MAC-

HV and conventional CSMA-CA protocol. We proved that providing longer idle times to nodes

with higher battery capacity leads to enhanced performance. In other words, we showed that

giving the priority to access the medium for the nodes with low residual voltage, while leaving

the other nodes to rest (MAC-LV), offered a superior performance over the other schemes.

6.2 Future Research

This thesis has outlined the development of new algorithms and protocols in the MAC layer of

WSNs. As an extension to the work carried out herein, we believe that potential future research

can be undertaken on the following subjects:

98
  Analytical study

In this thesis, we implemented our protocols and provided results through simulations. The next

normal step is to confirm the results via analytical studies. A new mathematical model for each

proposed protocol should be developed, and then an analytical study should be conducted.

The theory of new mathematical models will be like the one proposed in chapter 3. The same

discrete Markov chain representation of the CSMA-CA protocol proposed in [PARK10] should

be adopted with essential changes. The analytical study will be similar; we expect to apply

modifications to the following parameters and variables:

- The delay window (Wi) will not depend only on the back-off exponent (BE) parameter

anymore, it will also depend on the current battery voltage Vi. So the number of states

during the back-off state will be a function of Vi as well. As a result, the number of back-

off stages with MAC-HV (MAC-LV) will be less with greater Vi (with smaller Vi).

- The probability parameters (α, β, and Φ) that define the Markov chain will not be

constant anymore; they will be functions of Vi.

- The recovery probability function RNi,Ti will have the additional factor (Vi/V) in the case

of MAC-LV and the factor (1- Vi/V) in the case of MAC-HV.

 Collaborative battery-aware MAC protocol

In the current study we did not consider exchanging information between the nodes. Indeed, the

back-off function depends only on local data (current battery capacity). As a matter of fact,

previous studies in many fields of the WSNs proved that sharing some crucial information

99
among neighbor nodes enhance greatly the entire performance of the network especially in terms

of extending the lifetime of the network.

As future work, a new battery-aware MAC protocol based on collaborative work can be designed

with a new back-off scheme that takes into account the state of the batteries in the neighboring

nodes to improve the power control and management, and to increase the energy savings and the

lifetime of the nodes/network.

 Adaptive battery-aware MAC protocol

As we stated previously in this thesis, the concept of packet collisions is one of the aspects that

contribute to the energy expenditure. Our proposed protocols do not take this factor into

consideration.

An enhancement to our schemes will be the development of a new version of the battery-aware

MAC protocol adaptable to the state of the channel. In other words, in the new approach the

nodes will attempt to avoid transmitting packets in case of high channel utilization, and then less

power will be dissipated.

 Battery-aware Routing protocol

The main function of routing protocols in WSNs is to ensure reliable multi-hop communication

among nodes. Selecting the proper routing algorithm that the nodes will use to choose the paths

on which the data will be transmitted, has a great impact on the functionality of the WSN;

because the communication unit consumes most of the battery power during sending/receiving

packets. For example, if long routes have been selected, more energy will be consumed and then

the nodes expire sooner.

100
Multiple routing methods have been reported, some of them are topology based, and some are

based on proactive strategies, others on reactive techniques, and several consider the geographic

location of the node. Most of these protocols do not consider the battery state in their

computations; they do not exploit the pulsed current discharge and the charge recovery effects

(we found only one proposed protocol, very recent, it was published in October 2011 [MA11]).

In the future, a new design of a new routing protocol that takes into consideration the remaining

voltage in the battery can be developed. Similar to our approach in the MAC sub-layer, two

versions can be developed: one algorithm that allow that the nodes with higher voltage to rest

more than the nodes with lower remaining energy, and a second algorithm that operates in an

opposite way.

 Cross-layer design

Another important design we think it is possible to realize in the future, is a novel cross layer

system that combine a battery-aware MAC protocol with a battery-aware routing protocol. In this

system, battery state and effects will be used in a collaborative way between the different layers

to attain better performance at each level of the protocol stack.

101
Bibliography

[ADAM02] M. Adamou and S. Sarkar. “A Framework for Optimal Battery Management for

Wireless Nodes,” in Proc. of IEEE INFOCOM, 2002, vol. 3, pp. 1783-1792.

[ALIP09] C. Alippi, G. Anastasi, M.D. Francesco, M. Roveri, “Energy Management in

Wireless Sensor Networks with Energy-hungry Sensors”, IEEE Instrumentation and

Measurement Magazine , 2009, vol. 12, pp. 16-23.

[ALZI97] J. Alzieu, H. Smimite, C. Glaize, “Improvement of intelligent battery controller:

state-of-charge indicator and associated functions,” Jo. Power Source, 1997, vol. 67.( 1–2), pp.

157–161.

[ANAS09a] G. Anastasi, M. Conti, L. M. Di Franchesco, A. Passarella, “Energy conservation

in wireless sensor networks: A survey”. Ad Hoc Networks, 2009, vol. 7, pp 537-568.

[ANAS09b] G. Anastasi, M. Conti, and M. Di Francesco, “Extending the lifetime of wireless

sensor networks through adaptive sleep,” IEEE Trans. Ind. Informat., 2009, vol. 5, no. 3, pp.

351–365.

[ARIS02] K. Arisha, M. Youssef, and M. Younis. "Energy-aware TDMA-based MAC for

sensor networks". in ProcIEEE Workshop on Integrated Management of Power Aware

Communications, Computing and NeTworking (IMPACCT), 2002.

[BARB08] L. Barboni, M. Valle, “Experimental Analysis of Wireless Sensor Nodes Current

Consumption”, The 2nd International Conference on Sensor Technologies and Applications,

SENSORCOMM , 2008, pp. 401-406.

102
[BASU08] P. Basu and C.-K. Chau, “Opportunistic forwarding in wireless networks with

duty cycling,” in Proc. of ACM Workshop on Challenged Networks (CHANTS), 2008.

Boundary Conditions”, Journal of The Electrochemical Society, 2009, vol. 156 (4), pp. A260-

A271.

[BOUA09] F. Bouabdallah, N. Bouabdallah and R. Boutaba, “Cross-Layer Design for

Energy Conservation in Wireless Sensor Networks“, in Proc. IEEE ICC, 2009, pp. 1-6.

[CHIA00] C. F. Chiasserini and R. R. Rao. "A Distribute Power Management Policy for

Wireless Ad Hoc Networks," in Proc. of IEEE WCNC 2000, vol. 3, pp. 1209-1213.

[CHIA01] C. F. Chiasserini and R. R. Rao. “Improving Battery Performance by Using

Traffic Shaping Techniques.” IEEE Journal on Selected Areas of Communications,, 2001, vol.

19 ( 7), pp. 1385-1394.

[CHIA02] C. F. Chiasserini, I. Chlamtac, P. Monti, and A. Nucci. "Energy-Efficient Design

of Wireless Ad Hoc Networks," in Proc. of Networking, 2002, pp. 376-386.

[CHIA99a] C. F. Chiasserini and R. R. Rao. “A Traffic Control Scheme to Optimize the

Battery Pulsed Discharge,” in Proc. of IEEE MILCOM , 1999, pp. 1419-1423.

[CHIA99b] C. F. Chiasserini and R. R. Rao. “Pulsed Battery Discharge in Communication

Devices,” in Proc. of ACM MOBICOM, 1999, pp. 88-95.

[CHIA99c] C.F. Chiasserini, R.R. Rao. "A model for battery pusled discharge with recovery

effect," in Proc. Wireless Communications and Networking Conference, 1999, vol. 2, pp. 636-

639.

103
[CHAU10] C.K. Chau, F. Quin, S. Sayed, M.H. Wahab, and Y. Yang. "Harnessing battery

recovery effect in wireless sensor networks: experiments and analysis." IEEE Journal on

Selected Areas of Communications, 2010, vol. 28 (7), pp. 1222-1232.

[DAM03] T. V. Dam and K. Langendoen. “An Adaptive Energy-Efficient MAC Protocol

for Wireless Sensor Networks,” in Proc.of ACM SenSys, 2003, pp. 171-180.

[DATT05] S.Datta, B. Eksiri. "Sensor Network Routing Algorithms for Realistic Battery

Models," in Proc. of the Workshop on Information Fusion and Dissemination in Wireless Sensor

Networks,, 2005.

[DHAN05] M. Dhanaraj, S. Jayashree, and C. S. R. Murthy. “A novel battery aware mac

protocol for minimizing energy × latency in wireless sensornetworks,” in Proc. HiPC, 2005, pp.

312-321.

[DIET09] I. Dietrich and F. Dressler, “On the Lifetime of Wireless Sensor Networks,” ACM

Transactions on Sensor Networks (TOSN), 2009, vol. 5, no. 1, pp. 1–39.

[DOYL93] M.Doyle, T. F. Fuller, J. S. Newman. “Modeling of glavanostatic charge and

discharge of lithium/polymer/insertion cell” J. Electrochem. Soc., 1993, vol. 140, pp. 1526-1533.

[DOYL97] M. Doyle, J. Newman, “Analysis of capacity-rate data for lithium batteries using

simplified models of the discharge process,” Jo. Applied Electrochem., 1997, vol. 27 ( 7), pp.

846–856.

[DUTT08] P. Dutta, M. Feldmeier, J. Paradiso, and D. Culler, “Energy metering for free:

Augmenting switching regulators for real-time monitoring,” in Proc. 7th Int. Conf. IPSN, 2008,

pp. 283–294.

104
[EZZE09] T. Ezzedine, M. Miladi, R. Bouallegue. “ An Energy-Latency-Efficient MAC

Protocol for Wireless Sensor Networks”, in Proc. Internation Journal of Electrical and Computer

Engineering, 2009, pp. 816-821.

[FRAN11] D. Francesco, G. Anastasi, M. Conti, S.K. Das, V. Neri, “Reliability and Energy-

Efficiency in IEEE 802.15.4/Zigbee Sensor Networks: An Adaptive and Cross-layer Approach ”,

IEEE Journal on Selected Areas in Communications, 2011, vol. 209 (8), pp. 1508-1524.

[FULL94] T.F. Fuller, M. Doyle, J. Newman, “Relaxation phenomena in lithium-ion

insertion cells,” Jo. Electrochem. Soc., 1994, vol. 141 (4), pp. 982– 990.

[GIBR10] G.Gibran, M.Popa. “A Glance on WSN Lifetime and Relevant Factors for Energy

Consumption”, in Proc. Computational Cybernetics and Technical Informatics (ICCC-

CONTI),2010. pp.523-528.

[GLAT11] P.M. Glatz, L.B. Hormann, C. Steger, R. Weiss, “HANS: Harvesting Aware

Networking Service for Energy Management in Wireless Sensor Networks”, in Proc.

Telecommunications Internation Conference (ICT), 2011, pp. 180-185.

[GOBR09] S. Gobriel, D. Mousse, R. Cleric, “TDMA-ASAP: sensor network TDMA

scheduling with adaptive slot stealing and parallelism”, in Proc.ICDCS , 2009, pp. 458-465.

[GUO01] C. Guo, L.C. Zhong, J.M. Rabaey. “ Low Power Distributed MAC for Ad Hoc

Sensor Radio Networks,” in Proc.of Globecom, 2001, vol. 5, pp. 2944-2948.

[HAMA10] F. Hamady, M. Sabra, Z. Sabra, A. Kayssi, A. Chehab, M. Mansour «

Enhancement of the S-MAC Protocol for Wireless Sensor Networks », in Proc. Energy Aware

Computing (ICEAC), 2010, pp. 1-4.

105
[HALK05] G.P.Halkes, T.Van Dam, K.G. Langendoen. “ Comparing Energy-Saving MAC

protocols for Wireless Sensor Networks” , Mobile Networks and Applications, 2005, vol. 10, pp.

783-791.

[JAYA04] S. Jayashree, B. Manoj, and C. S. R. Murthy. “On using battery state for medium

access control in ad hoc wireless networks,” in Proc. of ACM MOBICOM, 2004, pp. 360-373.

[JURD08] Jurdak, A. G. Ruzzelli, and G. M. P. O’Hare, “Adaptive radio modes in sensor

networks: How deep to sleep?” in Proc. IEEE Commun. Soc. Conf. Ad Hoc Sensor Netw., 2008,

pp. 386–394.

[KERA10] F. Kerasiotis, A.Prayati, C.Antonopoulos, C.Koulamas, G.Papadopoulos,

“Battery Lifetime Prediction Model for a WSN Platform", in Proc. Sensor Tehcnologies

and Applications (SENSORCOMM) internationa conference, 2010, pp. 525-330.

[KIM08] N. Kim, S. Choi, H. Cha, “Automated Sensor-specific Power Management for

Wireless Sensor Networks”, in Proc. IEEE Conference on Mobile Ad Hoc and Sensor Systems

(MASS),2008, pp. 305-314.

[KRED07] K. Kredo II, P. Mohapatra, “Medium access control in wireless sensor networks”,

Computer Networks,2007, vol. 51 (4), pp. 961- 994,

[LAFO90] R.M. Lafollette, D. Bennion, “Design Fundamentals of High Power Density,

Pulsed Discharge, Lead-Acid Batteries. II Modeling,” Jo. Electrochem. Soc., 1990, vol. 137 (12),

pp. 3701–3707.

106
[LAFO95] R.M. Lafollette, “Design and Performance of High Specific Power, Pulsed

Discharge, Bipolar Lead Acid Batteries,” 10th Annual Battery Conference on Applications and

Advances, 1995, pp. 43–47.

[LANG08] K. Langendoen, "Medium Access Control in Wireless Sensor Networks,” in

Medium Access Control in Wireless Networks, 2008, vol. 2, pp. 535–560.

[LI04] J. Li and G. Lazarou. ``A bit-map-assisted energy-efficient MAC scheme for wireless

sensor networks``, In Proc. 3rd Int. Symp. On Information Processing in Sensor Networks

(IPSN), 2004, pages 55-60.

[LU04] G. Lu, B.Krishnamachari, C. Raghavendra. « An Adaptive Energy-Efficient and

Low-Latency MAC for Data-Gathering in Sensor Networks » in Proc of the 4th international

IEEE Workshop on Algorithms for Wireless, Mobile, Ad Hoc and Sensor Networks (WMAN),

April 2004, pp. 224-232

[MA11] C. Ma, Y. Yang, “A battery-aware scheme for routing in wireless ad hoc

networks ”, in Proc. Vehicular tehcnology, IEEE transactions, 2011, vol. 6 (8), pp. 3919-3932.

[MINE09] P. Minet, “Energy efficient routing in Ad Hoc and Sensor Wireless Networks:

Architectures: Algorithms and Protocols”. Bentham Science, 2009.

[MUSU05] R. Musunuri, J.A. Cobb. "Hierarchical-battery aware routing in wireless sensor

networks," in Proc. Vehicular Technology Conference, 2005, pp. 2311-2315.

[NELS97] B. Nelson, R. Rinehart, S. Varley, “Ultrafast pulse discharge and recharge

capabilities of thin-metal film battery technology,” 11th IEEE International Pulsed Power

Conference, 1997, pp. 636–641.

107
[NGUY11] H.A. Nguyen, A. Forster, D. Puccinelli, S. Giordano, , “Sensor node lifetime : An

experimental study”, In Proc. Pervasive Computing and Communications

WorkshopsInternational Conference (PERCOM Workshops), 2011, pp. 202-207.

[NIEB03] T.Nieberg, S.Dulman, P.Havinga, L.van Hoesel and J.Wu, “Collaborative

Algorithms for Communication in Wireless Sensor Networks”, Ambient Intelligence: Impact on

Embedded Systems, Kluwer Academic Publishers, ISBN 1- 4020-7668-1, November 2003

[PARK05] C.Park, K.Lahiri, A.Raghunathan. “Battery Discharge Characteristics of Wireless

Sensor Nodes: An Experimental Analysis”, in Proc. Of the IEEE Conf. on Sensor and Ad-hoc

Communications and Networks (SECON), 2005, pp.430-440.

[PARK10] P. Park, P. D. Marco, C. Fischione, and K. H. Johansson, "Adaptive IEEE

802.15.4 protocol for reliable and timely communications," In Proc. of the 9th ACM/IEEE

International Conference on Information Processing in Sensor Networks (IPSN), 2010, pp. 327-

338.

[PODH94] E.J. Podhala, H.Y. Cheh, “Modeling of Cylindrical Alkaline Cells. VI: Variable

Discharge Conditions,” J. Electrochem. Soc., 1994, vol. 141 (1), pp. 28–35.

[POLL08] S. Pollin, M. Ergen, S. C. Ergen, B. Bougard, L. Van der Perre, I. Moerman, A.

Bahai, P. Varaiya, and F. Catthoor, “Performance Analysis of Slotted Carrier Sense IEEE

802.15.4 Medium Access Layer,” IEEE Transactions on Wireless Communications, 2008, vol. 7

(9), pp. 3359-3371.

108
[RAHM10] J. Rahm, N. Fourty, K. A. Agha, and A. van der Bossche, “A Recursive Battery

Model for Nodes Lifetime Estimation in Wireless Sensor Networks,” in Proc. IEEE Wireless

Communications and Networking Conference (WCNC), 2010, pp. 1-6.

[RAJE03] V. Rajendran, K. Obraczka, and J. J.Garcia-Luna-Aceves.`` Energy-efficient,

collision-free medium access control for wireless sensor networks``, In Proc. of the First ACM

Conference on Embedded Networked Sensor Systems (SenSys), 2003, pp. 181-192.

[REDD10] T.B. Reddy, H.D Linden, Handbook of Batteries, 4th edition, McGraw-Hill, New

York 2010.

[RHEE05] I. Rhee, A. Warrier, M. Aia, and J. Min, “Z-MAC: a hybrid MAC for wireless

sensor networks,” in Proc. of the International conference on Embedded Networked Sensor

Systems, 2005, pp. 90–101.

[SAXE11] S. Saxena, S. Mishra, A. Kumar and D. S. Chauhan, “Efficient Power Utilization

Techniques for Wireless Sensor Networks–A Survey”, International Journal on Computer

Science and Engineering, 2011, vol.:3 (2), pp. 905-925.

[SCHU02] C. Schurgers, V. Tsiatsis, M.B. Srivastava. “STEM: Topology Management for

Energy Efficient Sensor Networks,” in Proc.of IEEE Aerospace Conference, 2002, vol. 3, pp.

1099-1108. [SCHW96] M.Schwartz, Broadband Integrated Networks, Prentice Hall, New

Jersey, 1996.

[SELV08] K. Selvaradjou, M. Dhanaraj, B.Goutham, C.S.R. Murthy. “A New Battery and

Redundancy Aware Node Scheduling Protocol for Wireless Sensor Networs”, in Proc.

109
Communication Systems Software and Middleware and Workshops (COMSWARE), 2008,

pp.784-790

[SEND11] S.Sendra, J. Lloret, M. Garcia, J.F. Toledo, “Power saving and energy

optimization techniques for wireless sensor networks”, Journal of Communications, 2011, vol. 6

(6), pp. 439-459.

[SHEL08] T.R.Sheltami, E.Shakshuki. “Neighbor-Aware Clusterhead with Different Sleep

Scheduling Protocols”, in Proc. Parallel Processing-Workshops Internation Conference, 2008,

pp. 143-147.

[SING98] S. Singh and C. S. Raghavendra. "Power-Aware Multi-Access Protocol with

Signaling for Ad Hoc Networks." ACM Computer Communication Review, 1998, vol. 28 (3), pp.

5-26.

[SOUA11] R. Soua, P. Minet, “A Survey on energy efficient techniques in wireless sensor

networks”, in Proc. Wireless and Mobile Networking Conference (WMNC), 2011, pp. 1-9.

[ SUBR09] V.R. Subramanian, V. Boovaragavan, V. Ramadesigan, M. Arabandi,

“Mathematical Model Reformulation for Lithium-Ion Battery Simulations: Galvanostatic

[SUN08] P. Sun, X. Zhang, Z. Dong, Y. Zhang, ”A Novel Energy Efficient Wireless Sensor

MAC Protocol”, in Proc. of the of the Fourth International Conference on Networked Computing

and Advanced Information Management (ICNCA), 2008, pp. 68-72.

[VAN04] L.F.W van Hoesel, P.J.M. Havinga. “ A Lightweight Medium Access Protocol

(LMAC) for Wireless Sensor Networks: Reducing Preamble Transmissions and Transceiver

110
State Switches”, in Proc. 1st International Workshop on Networked Sensing Systems (INSS),

2004, vol. 51 (4), pp. 205-208.

[VAN06] L. F.W. van Hoesel, P. J. M. Havinga, “Design aspects of an energy-efficient,

lightweight medium access control protocol for wireless sensor networks,” Tech. Rep. TR-CTIT-

06-47, Centre for Telematics and Information Technology, 2006.

[VENK06] Venkatesh Rajendran, Katia Obraczka, J.J. Garcia-Luna-Aceves. ``Energy-

efficient, collision -free medium access control for wireless sensor networks``, Journal Wireless

Networks, 2006, vol. 12(1), pp. 63-78.

[WANG07] L. Wang and K. Liu, “An adaptive energy-efficient and lowlatency MAC

protocol for wireless sensor networks,” in Proc. of the International Conference on Wireless

Communications, Networking and Mobile Computing (WiCOM ), 2007, pp. 2440–2443.

[WATF10] M.K. Watfa, S. Selman, H. Denkilkian. “A Battery-Aware High-Throughput

MAC Layer Protocol in Sensor Networks” , International Journal of Distributed Sensor

Networks,2010.

[YADA09] R. Yaday, S. Varma, and N. Malaviya, “A Survey of MAC Protocols for

Wireless Sensor Networks,” UbiCC Journal, 2009, vol. 4 (3), pp. 827-833.

[YE04a] W. Ye, J. Heidemann, and D. Estrin.. “Medium Access Control with Coordinated

Adaptive Sleeping for Wireless Sensor Networks.” IEEE/ACM Transactions on Networking, vol.

12 (3), pp. 493-506, Jun. 2004.

[YE02] W.Ye, J.Heidemann, D.Estrin. « An Energy-Efficient MAC Protocol for

Wireless Sensor Netowkrs” , in Proc.of IEEE INFOCOM 2002, vol. 3, pp. 1567-1576.

111
[YE04b] W. Ye, J. Heidemann, and D. Estrin. `` Medium access control with coordinated

adaptive sleeping for wireless sensor networks``, IEEE/ACM Trans. Netw. 2004, vol. 12 (3),

pp.493-506.

[YI07] S. Yi, J. Heo , Y. Cho, J. Hong, “PEACH: power-efficient and adaptive clustering

hierarchy protocol for wireless sensor networks” . Computer Communications, 2007, vol. 30

(14–15), pp. 2842–2852.

[YICK08] J. Yick, B. Mukherjee, D. Ghosal, “Wireless sensor network survey”, Computer

Networks, 2008,l vol. 52 (12) , pp. 2292-2330.

[YIN10] C. Yin, Y. Li, D. Zhang, “ DSMAC: An energy-efficient MAC protocol in Event-

driven sensor networks”, in Proc. Advanced Computer Control Internation Conference (ICACC),

2010, pp. 422-425.

[YUAN11] Z. Yuan, L. Wang, Lei Shu, T. Hara, Z. Qin. “A Balanced Energy Consumption

Sleep Scheduling Algorithm in Wireless Sensor Networks”,. In the 7th International Wireless

Communications & Mobile Computing Conference (IWCMC), 2011, pp. 831-835

[ZIGAL04] ZigBee Alliance Document 053474r06; ZigBee Specification, Version 1.0, Dec

2004.

[ZIGAL08] ZigBee Alliance Inc., 2400 Camino Ramon Suite 375, San Ramon CA 94583,

USA, ZigBee Specification, January 2008

112