Performance Analysis of Medium Access Control Protocol for Body Sensor Networks

Ph.D. Dissertation

㧎㼊 ㎒㍲ ⍺䔎㤢䋂⯒ 㥚䞲 ⰺ㼊 㩧⁒ 㩲㠊 䝚⪲䏶䆲㦮 ㎇⓻ ⿚㍳

Performance Analysis of Medium Access Control Protocol for Body Sensor Networks

BY PERVEZ KHAN

August 2015

Graduate School of Information Technology & Telecommunications Inha University

Performance Analysis of Medium Access Control Protocol for Body Sensor Networks

A DISSERTATION submitted to The Graduate School of Information Technology & Telecommunications of Inha University in partial fulfillment of the requirements for the degree of Doctor of Philosophy By Pervez Khan

Major: Information & Communication Engineering Advisor: Prof. Kyung Sup Kwak

c Pervez Khan 2015

Abstract The rising costs of healthcare and the increase in continuous healthcare monitoring of the aging population throughout the world pose challenges for healthcare and medical monitoring. A wireless body area network (WBAN), with medical sensors attached to a human body continuously sending measurements of human physiological parameters to a remote server or physician, has been shown to be adequate for monitoring the patient’s health status without constraining his or her normal activities.

WBANs must support a

combination of low power, quality of service (QoS), high data rate, reliability, and non-interference, to address the gamut of WBAN applications. The IEEE 802.15.6 standard was officially approved in February 2012 for wireless communications in WBANs. The standard provides efficient communication solutions to ubiquitous healthcare and telemedicine systems, interactive gaming, military services, and portable audio/video systems. The IEEE 802.15.6 standard defines a Medium Access Control (MAC) layer that supports several Physical (PHY) layers, such as Ultra-wideband (UWB), Narrowband (NB), and Human Body Communications (HBC) layers.

The focus of this thesis is the

short-range wireless communication network that is formed between the sensors and the hub in a WBAN, particularly at the medium access control (MAC) layer. The main focus is on the analytical modeling and performance evaluation of contention-based MAC protocols (supported by NB PHY) for a WBAN. Narrowband wireless communication is arguably top suited to a considerable number of healthcare applications, and is thus the focus of this thesis. The contributions of the dissertation i

are divided into two main parts. The first part of the dissertation focuses on the analysis of the IEEE 802.15.6 Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) mechanism. We analytically approach to investigate the behavior of the protocol. We develop a discrete time Markov chain (DTMC) to evaluate the performance measures as throughput, mean frame service and energy consumption of IEEE 802.15.6 CSMA/CA under saturated, non-saturated traffic conditions using a single user priority (homogeneous traffic scenarios).

We also study the

heterogeneous scenarios (multiple user priorities (U Ps )) differentiated by minimum and maximum contention window sizes as shown in the IEEE 802.15.6 standard. We extended the proposed analytical model to consider a portion of the access phases (i.e., EAP1 and RAP1) of the superframe and analyze its impact on the performance of the IEEE 802.15.6 CSMA/CA. While constructing the DTMC, we take into consideration the time spent by a node awaiting the acknowledgement frame, in our DTMC model this state is presented as (i, −1). The analysis is validated against extensive simulation. In the second part of this dissertation we reconstruct the DTMC model by considering the ACK-timeout state (i, −1) as an additional state in our model that colliding nodes check to learn the fate of their transmission, while other contending nodes performing backoff check that slot as usual. We extended the model for an error-prone WBAN channel.

Finally, to reduce the gap between successive contention

window sizes we adopted the Fibonacci backoff procedure and compare the results with the binary exponential backoff procedure mentioned in the IEEE 802.15.6 standard.

ii

DEDICATION

To my Parents This thesis is dedicated to my parents for their endless love, support and encouragement.

iii

Acknowledgement All praise are due to Allah, the Exalted. I would like to express my sincerest gratitude to my advisor and mentor, Professor Kyung Sup Kwak, for giving me the inspiration and flexibility to explore my ideas and research interests. Without his continuous guidance, support and encouragement, this dissertation would not have been possible. I would like to thank him sincerely for everything he did for me during my stay at Inha University. My thanks also extend to Prof. Sang-Jo Yoo (Inha University), Prof. Wonik Choi (Inha University), Prof. Sang Hun Chun (JEI University) and Dr. Baek-Hyun Kim (Korea Railroad Research Institute) for being a part of my dissertation committee. Their insightful suggestions and constructive comments greatly improved the quality of this work. In the beginning semesters of my graduate study, Professor Kyung Sup Kwak and Professor Sang-Jo Yoo introduced me to certain advanced topics associated to protocol analysis in their specific courses. Later on Professor Sang Bang Choi provided us the opportunity to study a very rigorous course that covered a lot of topics on protocol analysis. I express my sincere appreciation to Prof. Choi for enlightening me with such topics which eventually has led me to do the work contained in this dissertation. I am really thankful to all my lab members, for providing me a friendly environment during my stay at Inha University. Many thanks go to Murad Ali, Faiz Ali, Ashraf Ali, Nasre Alam, Asdaque Hussain, Asif Iqbal, M. Sana Ullah Chawdry, Kabir and S.M. Riazul Islam for spending a great time together. My deep appreciation goes to Niamat Ullah and Sana Ullah for the moral support they gave me throughout iv

my PhD journey. I deeply appreciate the financial support of the Inha University (Jungseok International Scholarship) and Juan Rotary Club, Incheon. I am very grateful to my dearest parents and siblings for tolerating my absence during my stay at Inha University. I have no words to express my feelings to thank my dearest parents for their emotional and moral support throughout my life. With deepest appreciations, I would like to dedicate this work to my parents.

Pervez Khan Graduate School of IT & Telecommunications, Inha University, Incheon Date: July 6, 2015

v

Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

i

Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

iii

Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . .

iv

Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vi

List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xi

List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv 1 Introduction

1

1.1

Background . . . . . . . . . . . . . . . . . . . . . . . . . .

1

1.2

Motivations . . . . . . . . . . . . . . . . . . . . . . . . . .

2

1.3

Contributions . . . . . . . . . . . . . . . . . . . . . . . . .

4

1.4

Dissertation Organization . . . . . . . . . . . . . . . . . .

6

2 Basic Concepts of Wireless Body Area Network

7

2.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . .

7

2.2

Wireless Sensor Networks . . . . . . . . . . . . . . . . . .

7

2.3

Wireless Body Area Networks . . . . . . . . . . . . . . . .

8

2.4

Challenges in a WBAN . . . . . . . . . . . . . . . . . . . .

11

2.4.1

Energy Efficiency . . . . . . . . . . . . . . . . . . .

11

2.4.2

Reliability . . . . . . . . . . . . . . . . . . . . . . .

12

2.4.3

Security and Interference . . . . . . . . . . . . . .

12

vi

2.5

2.6

2.7 3

2.4.4

Material Constraints . . . . . . . . . . . . . . . . .

13

2.4.5

Quality of Service . . . . . . . . . . . . . . . . . . .

13

2.4.6

Robustness . . . . . . . . . . . . . . . . . . . . . . .

13

WBAN Applications . . . . . . . . . . . . . . . . . . . . .

14

2.5.1

Monitoring Patients with Cardiovascular Diseases

14

2.5.2

Monitoring Elderly Patients . . . . . . . . . . . . .

15

2.5.3

Cancer Detection . . . . . . . . . . . . . . . . . . .

15

2.5.4

Telemedicine Systems . . . . . . . . . . . . . . . .

15

2.5.5

Diabetes . . . . . . . . . . . . . . . . . . . . . . . .

16

2.5.6

Battlefield . . . . . . . . . . . . . . . . . . . . . . .

17

2.5.7

Asthma . . . . . . . . . . . . . . . . . . . . . . . . .

17

2.5.8

Artificial Retina . . . . . . . . . . . . . . . . . . . .

17

Existing Health-care Projects . . . . . . . . . . . . . . . .

18

2.6.1

CodeBlue . . . . . . . . . . . . . . . . . . . . . . .

18

2.6.2

Ayushman . . . . . . . . . . . . . . . . . . . . . . .

19

2.6.3

MobiHealth . . . . . . . . . . . . . . . . . . . . . .

20

2.6.4

Human++ research program . . . . . . . . . . . .

20

2.6.5

HipGuard System . . . . . . . . . . . . . . . . . .

21

2.6.6

eWatch . . . . . . . . . . . . . . . . . . . . . . . . .

21

2.6.7

UbiMon . . . . . . . . . . . . . . . . . . . . . . . .

22

2.6.8

LifeShirt . . . . . . . . . . . . . . . . . . . . . . . .

22

Summary of the Chapter . . . . . . . . . . . . . . . . . . .

23

The IEEE 802.15.6 standard: An overview

24

3.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . .

24

3.2

The IEEE 802.15.6 standard . . . . . . . . . . . . . . . . .

24

3.3

IEEE 802.15.6 PHY Specifications . . . . . . . . . . . . . .

25

3.3.1

26

NB PHY Specifications . . . . . . . . . . . . . . . .

vii

3.4

3.3.2

HBC PHY Specifications. . . . . . . . . . . . . . .

27

3.3.3

UWB PHY Specifications . . . . . . . . . . . . . .

28

IEEE 802.15.6 MAC Specifications . . . . . . . . . . . . . .

28

3.4.1

29

IEEE 802.15.6 Communication Modes . . . . . . . 3.4.1.1

Beacon

Mode

with

Superframe

Boundaries . . . . . . . . . . . . . . . . . 3.4.1.2

Non-beacon mode with superframe boundaries . . . . . . . . . . . . . . . . .

3.4.1.3

31

Non-beacon mode without superframe boundaries . . . . . . . . . . . . . . . . .

31

3.4.2

IEEE 802.15.6 MAC Frame Format . . . . . . . . .

31

3.4.3

Priority Mapping . . . . . . . . . . . . . . . . . . .

32

3.4.4

IEEE 802.15.6 Access Mechanisms . . . . . . . . .

33

3.4.4.1

33

3.4.4.2

Random Access Mechanism . . . . . . . 3.4.4.1.1

Slotted ALOHA Protocol . . . .

33

3.4.4.1.2

CSMA/CA Protocol . . . . . . .

34

Improvised and Unscheduled Access Mechanism . . . . . . . . . . . . . . . . .

3.4.4.3

Scheduled

and

37

Scheduled-Polling

Access Mechanisms . . . . . . . . . . . .

38

MICS band communication . . . . . . .

38

Summary of the Chapter . . . . . . . . . . . . . . . . . . .

38

3.4.4.4 3.5

30

4 Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

40

4.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . .

40

4.2

Performance Modeling Approaches for MAC Protocols .

41

4.2.1

42

Simulation Approach . . . . . . . . . . . . . . . .

viii

4.3

4.4

4.5

4.6

4.2.2

Real Experimentation Approach . . . . . . . . . .

42

4.2.3

Analytical Approach and Markov Analysis . . . .

42

IEEE 802.15.6-based MAC protocol performance under saturation conditions . . . . . . . . . . . . . . . . . . . . .

44

4.3.1

Related Works . . . . . . . . . . . . . . . . . . . . .

44

4.3.2

General Assumptions . . . . . . . . . . . . . . . .

45

4.3.3

Analytical Model . . . . . . . . . . . . . . . . . . .

47

4.3.4

Performance Metrics . . . . . . . . . . . . . . . . .

54

4.3.5

Results and Discussion . . . . . . . . . . . . . . . .

55

IEEE 802.15.6-based MAC protocol performance under non-saturation conditions . . . . . . . . . . . . . . . . . .

58

4.4.1

Analytical Model . . . . . . . . . . . . . . . . . . .

58

4.4.2

Performance Metrics . . . . . . . . . . . . . . . . .

66

4.4.3

Results and Discussion . . . . . . . . . . . . . . . .

69

IEEE 802.15.6-based MAC protocol performance under different access periods . . . . . . . . . . . . . . . . . . . .

73

4.5.1

Performance Measures and Analytical Modeling .

73

4.5.2

Results and Discussion . . . . . . . . . . . . . . . .

84

Summary of the Chapter . . . . . . . . . . . . . . . . . . .

88

5 Rethinking

the

IEEE

802.15.6-based

WBANs

MAC

Performance Modeling Methodology

89

5.1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . .

89

5.2

IEEE 802.15.6-based MAC protocol performance under non-saturation conditions . . . . . . . . . . . . . . . . . .

90

5.2.1

Assumptions . . . . . . . . . . . . . . . . . . . . .

91

5.2.2

Analytical Model . . . . . . . . . . . . . . . . . . .

92

5.2.3

Homogeneous Networks . . . . . . . . . . . . . .

93

ix

5.2.4

5.3

5.2.3.1

Performance Metrics . . . . . . . . . . . .

99

5.2.3.2

Results and Discussion . . . . . . . . . . 100

Heterogeneous Networks . . . . . . . . . . . . . . 102 5.2.4.1

Performance Metrics . . . . . . . . . . . . 104

5.2.4.2

Results and Discussion . . . . . . . . . . 105

IEEE 802.15.6 MAC protocol performance under Error-prone channel . . . . . . . . . . . . . . . . . . . . . 109

5.4

5.3.1

Analytical Model . . . . . . . . . . . . . . . . . . . 110

5.3.2

Performance Metrics . . . . . . . . . . . . . . . . . 110

5.3.3

Results and Discussion . . . . . . . . . . . . . . . . 111

IEEE 802.15.6 MAC protocol performance under Different backoff algorithms . . . . . . . . . . . . . . . . . 113

5.5

5.4.1

Analytical Model . . . . . . . . . . . . . . . . . . . 113

5.4.2

Performance Metrics . . . . . . . . . . . . . . . . . 115

5.4.3

Results and Discussion . . . . . . . . . . . . . . . . 116

Summary of the Chapter . . . . . . . . . . . . . . . . . . . 117

6 Conclusions and Future Works

118

6.1

Summary and Conclusions . . . . . . . . . . . . . . . . . 118

6.2

Future Works . . . . . . . . . . . . . . . . . . . . . . . . . 121

Bibliography

121

x

List of Figures 1.1

Layout of access phases with superframe boundaries . .

4

2.1

A glimpse of WBAN and it’s Framework . . . . . . . . . .

10

2.2

A real-time telemedicine infrastructure . . . . . . . . . .

16

2.3

Artificial Retina for Blind People . . . . . . . . . . . . . .

18

2.4

CodeBlue architecture for emergency response . . . . . .

19

2.5

HipGuard System . . . . . . . . . . . . . . . . . . . . . . .

21

2.6

Life Shirt . . . . . . . . . . . . . . . . . . . . . . . . . . . .

22

3.1

IEEE 802.15.6 PHY and MAC layers . . . . . . . . . . . . .

25

3.2

IEEE 802.15.6 frequency bands . . . . . . . . . . . . . . .

26

3.3

PPDU structure for NB PHY (the indicated lengths are in bits) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

27

3.4

PPDU structure for HBC PHY . . . . . . . . . . . . . . . .

28

3.5

PPDU frame structure for UWB PHY . . . . . . . . . . . .

29

3.6

IEEE 802.15.6 MAC frame format . . . . . . . . . . . . . .

32

3.7

Slotted Aloha access illustration . . . . . . . . . . . . . . .

35

3.8

IEEE 802.15.6 CSMA/CA mechanism . . . . . . . . . . .

37

4.1

IEEE 802.15.6 CSMA/CA flowchart for DTMC in Figure 4.2 46

xi

4.2

DTMC Model for the CSMA/CA behavior in saturated traffic conditions . . . . . . . . . . . . . . . . . . . . . . . .

4.3

48

Normalized saturation throughput for homogeneous network . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

56

4.4

Head-of-line delay for homogeneous network . . . . . . .

57

4.5

DTMC

Model

for

the

CSMA/CA

behavior

in

non-saturated traffic conditions . . . . . . . . . . . . . . . 4.6

Normalised system throughput for non-saturated homogeneous network . . . . . . . . . . . . . . . . . . . .

4.7

67

Per class normalised throughput for non-saturated heterogeneous network . . . . . . . . . . . . . . . . . . . .

4.9

66

Head-of-line delay for non-saturated homogeneous network . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.8

60

68

Normalized system throughput for non-saturated heterogeneous network . . . . . . . . . . . . . . . . . . . .

69

4.10 Head of line delay for non-saturated heterogeneous network . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

71

4.11 Energy consumption for non-saturated heterogeneous network . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

72

4.12 IEEE 802.15.6 CSMA/CA flowchart for DTMC in Figure 4.13 . . . . . . . . . . . . . . . . . . . . . . . . . . .

74

4.13 DTMC Model for the CSMA/CA behavior under non-saturated conditions and different access periods . .

75

4.14 Per class normalised throughput; where EAP length is half of RAP . . . . . . . . . . . . . . . . . . . . . . . . . . .

84

4.15 Normalized system throughput; where EAP length is half of RAP . . . . . . . . . . . . . . . . . . . . . . . . . . .

85

4.16 Head of line delay; where EAP length is half of RAP . . .

86

xii

4.17 Energy consumption; where EAP length is half of RAP .

87

5.1

IEEE 802.15.6 CSMA channel access diagram . . . . . . .

91

5.2

DTMC

Model

for

the

CSMA/CA

behavior

in

non-saturated traffic conditions . . . . . . . . . . . . . . . 5.3

94

Normalized system throughput in the homogenous case (U P0 ) for different network sizes . . . . . . . . . . . . . . 100

5.4

Mean frame service time in the homogenous case (U P0 ) for different network sizes . . . . . . . . . . . . . . . . . . 101

5.5

Per class normalised throughput for non-saturated heterogeneous network . . . . . . . . . . . . . . . . . . . . 106

5.6

Normalized system throughput for non-saturated heterogeneous network . . . . . . . . . . . . . . . . . . . . 107

5.7

Head of line delay for non-saturated heterogeneous network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

5.8

Energy consumption for non-saturated heterogeneous network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

5.9

Normalized system throughput in the homogenous case (U P0 ) for error-prone network . . . . . . . . . . . . . . . . 111

5.10 Mean frame service time in the homogenous case (U P0 ) for error-prone network . . . . . . . . . . . . . . . . . . . 112 5.11 Normalized system throughput in the homogenous case (U P0 ) for different backoff algorithms . . . . . . . . . . . 114 5.12 Mean frame service time in the homogenous case (U P0 ) for different backoff algorithms . . . . . . . . . . . . . . . 115

xiii

List of Tables 3.1

Contention window bounds for CSMA/CA and contention probability thresholds for prioritized-based slotted Aloha access . . . . . . . . . . . . . . . . . . . . . .

33

4.1

Contention window bounds for CSMA/CA . . . . . . . .

54

4.2

Narrowband "channel seventh" parameters . . . . . . . .

55

xiv

Chapter 1

Introduction 1.1 Background Wireless Sensor Network is an emerging technology that enables information gathering in several different scenarios ranging from wild life monitoring to industrial, medical, critical infrastructure and military application. A wireless sensor network consists of group of sensor nodes using wireless links to perform distributed sensing task. These nodes are typically provided with embedded microprocessor and a very small amount of memory. The application of WSN are a lot and diverse, but it actually involves some kind of monitoring, tracking, and controlling mechanisms. In a typical application a WSN is scattered in a region where it is meant to collect data through its sensor nodes. Wireless sensor networks can be effectively used in healthcare monitoring systems to enhance the quality of healthcare services. For example, patients equipped with medical sensors, be an easy and fast way to diagnose the patient’s status and to consult the doctor without restricting their movements.

1

2

Chapter 1. Introduction

A wireless body area network (WBAN) is a logical set comprising of tiny and smart wireless medical sensors (which are wearable or implanted into the human tissues) and a common hub.

These

medical sensors are capable of measuring, processing, and forwarding important physiological parameters such as the heart rate, blood pressure, glucose level, body and skin temperature, oxygen saturation, and respiration rate, as well as records such as electrocardiograms and electromyograms.

This enables health professionals to predict,

diagnose, and react to critical and adverse events earlier than ever.

The IEEE 802.15 Working Group established Task Group 6

(TG6) in November 2007 to develop a communication standard known as IEEE 802.15.6.

The purpose of the group is to define a

communication standard optimized for short-range and low-power on-body/in-body nodes to serve a variety of medical, entertainment and consumer electronics applications.

WBANs must support a

combination of reliability, low power, quality of service (QoS), non-interference and high data rate to address the gamut of WBAN applications. The IEEE 802.15.6 standard for wireless communications in WBANs was approved in February 2012. The standard provides efficient communication solutions to ubiquitous healthcare and telemedicine systems, interactive gaming, military services, and portable audio/video systems.

1.2 Motivations The Medium Access Control (MAC) protocol provides a control mechanism to allow packet transmission through a shared wireless channel. The IEEE 802.15.6 supports two communication modes: 1)

Chapter 1. Introduction

3

beacon communication mode, where the hub transmits beacons for resource allocation and synchronization, 2) non-beacon communication mode, where the scheduled/unscheduled allocations and polling are used [1].

In the beacon communication mode, the beacons are

transmitted in the beginning of each superframe. As illustrated in Figure 1.1, in a beacon communication mode each superframe is divided into different access phases (APs). A superframe includes exclusive AP1 (EAP1), random AP1 (RAP1), management AP1 (MAP1), exclusive AP2 (EAP2), random AP2 (RAP2), management AP2 (MAP2), and an optional B2 frame followed by a contention access phase (CAP). The EAPs are used for life-critical traffic while the RAPs and CAP are used for regular traffic. Each AP, except RAP1, may have zero length [2]. In IEEE 802.15.6, the contention-based access methods for obtaining allocations are either carrier sense multiple access/collision avoidance (CSMA/CA) if a narrowband physical layer (PHY) / ultra-wideband (UWB) PHY is chosen or slotted ALOHA if UWB PHY is used [2]. The mathematical analysis helps to obtain an in-depth understanding of a protocol behavior.

Moreover, based on low level details, it

validates the technical capabilities of a protocol and helps how to fine-tune it in different scenarios to come up with the desired line of performance. These motivated us to come out with the performance analysis of the contention-based MAC protocols for WBAN. The IEEE 802.15.6 CSMA/CA mechanism is different in important aspects from the CSMA/CA mechanism of other wireless standards. The backoff mechanism is not binary exponential, and the contention window doubles only when the retry counter is an even number. In addition to busy channel the node will also lock the backoff counter if it

4

Chapter 1. Introduction

is not allowed to access the medium during the current AP or the current AP length is not long enough for a frame transmission. These differences require changes in the typical discrete time markov chain (DTMC) adopted for the CSMA/CA mechanism of previous standards presented in [3], [4], [5], and [6] for IEEE 802.11; in [7], [8], [9] and [10] for IEEE 802.11e; in [11], and [12] for IEEE 802.15.4; and in [13] for IEEE 802.15.3c. UP7

All UPs

CSMA/Slotted CSMA/Slotted Aloha Aloha

Polling Mechanisms

B

EAP1

RAP1

MAP1

UP7 All UPs CSMA/Slotted CSMA/Slotted Aloha Aloha

EAP2

RAP2

Polling Mechanisms

All UPs CSMA/Slotted Aloha

B2 MAP2

CAP

Beacon period (superframe) n

Figure 1.1: Layout of access phases with superframe boundaries

1.3 Contributions The main objective of this dissertation is an in-depth investigation, analysis, and performance evaluation of the IEEE 802.15.6 MAC protocols. This research focuses on the DTMC-based modeling of the IEEE 802.15.6-based CSMA/CA protocol that assists in achieving a better understanding of the protocol behavior and serves as a platform for future researches for the betterment of the protocol. The IEEE 802.15.6 is a new standard on WBANs for short-range, extremely low power wireless communication with high data rates in the near locality of, or inside a human body. CSMA/CA using an alternative binary exponential backoff procedure and prioritized slotted ALOHA are the two contention-based channel access schemes defined by the IEEE 802.15.6 standard. The standard supports QoS differentiation through user priorities and access phases. In this study,

Chapter 1. Introduction

5

we develop an analytical model for the estimation of performance metrics like normalized throughput, mean frame service time and energy consumption of the CSMA/CA protocol as described in the IEEE 802.15.6 standard, deploying a Markov chain model under saturated, non-saturated, homogeneous and heterogeneous traffic scenarios. While constructing the DTMC, we take into consideration the time spent by a node awaiting the acknowledgement frame, in our DTMC model this state is presented as (i, −1). In IEEE 802.15.6 MAC protocols, a superframe also includes different access phases specified for different user Priorities and access methods. We also consider a portion of the access phases of the superframe and analyze its impact on the performance of the IEEE 802.15.6 CSMA/CA. Our results show that the deployment of EAP is not necessary in a typical WBAN; in fact, IEEE 802.15.6 CSMA/CA employing different access phases degrades the overall system throughput performance and results in higher delay for non-emergency nodes and hence more energy per packet consumed. We extended the DTMC model by considering the ACK-timeout state (i, −1) as an additional state in our model that colliding nodes check to learn the fate of their transmission, while other contending nodes performing backoff check that slot as usual. Wireless channels are not ideal, therefore we extended the model for an error-prone WBAN channel.

Finally, to reduce the gap between successive contention

window sizes we adopted the Fibonacci backoff procedure and compare the results with the binary exponential backoff procedure mentioned in the IEEE 802.15.6 standard. To validate the models, a custom-made simulator that closely follows the CSMA/CA procedure of IEEE 802.15.6 has been developed in the C++ programming language.

6

Chapter 1. Introduction

1.4 Dissertation Organization The rest of the dissertation is organized as follows: Chapter 2: In this chapter we briefly describe Wireless Sensor Networks being the basis of healthcare systems.

Wireless Body

Area Networks as a building block of the healthcare system and healthcare applications are introduced. The currently available projects which intent to provide a human body monitoring system have been described. Chapter 3: In this chapter we presented the key features of the IEEE 802.15.6 standard. Starting from the fundamental details, we have provided deep insight into the MAC and PHY layers specification of IEEE 802.15.6. We review different communication modes and access mechanisms and explain the NB, UWB, and HBC specifications in detail. Chapter 4: In this chapter the IEEE 802.15.6-based WBANs MAC protocol performance is investigated under saturated, non-saturated and different access phases conditions by developing analytical and simulation models. Chapter 5: The DTMC model presented in this chapter is different from the model presented in Chapter 4. In the case of collision the ACK-timeout state presented in this model can be utilized by the other contending nodes as a backoff check slot. We extended the model for an error-prone channel and Fibonacci backoff procedure for WBANs. Chapter 7: Eventually, we conclude the dissertation and suggest some future works.

Chapter 2

Basic Concepts of Wireless Body Area Network 2.1 Introduction This chapter briefly describe WSNs as the basis of healthcare systems. WBANs as a building block of the healthcare system and healthcare applications are introduced in this chapter. The currently available projects which intent to provide a body monitoring healthcare system are described. The wireless technologies which these projects employ and their shortcomings are described.

We presented the research

describes in this dissertation toward an efficient healthcare monitoring system. Finally, a short summary conclude this chapter.

2.2 Wireless Sensor Networks Recent

advances

technology,

in

integrated

microelectro-mechanical circuits

and

systems

wireless

(MEMS)

communication

technologies [14, 15] have allowed the establishment of a large 7

Chapter 2. Basic Concepts of Wireless Body Area Network

8

scale, low power, multifunctional, and low cost wireless networks. A wireless sensor network consists of group of sensor nodes using wireless links to perform distributed sensing task. These nodes are typically provided with embedded microprocessor, a very limited amount of memory, a wireless transceiver, an antenna, one or more sensors and small irreplaceable power source. These sensor nodes are self-organizing and are densely deployed in the area (to be monitored) to sense and transmit data towards the base station. WSNs have great potential for many applications in scenarios ranging from wild life monitoring to industrial, medical, critical infrastructure and military application.

The application of WSN are a lot and diverse, but it

actually involves some kind of monitoring, tracking, and controlling mechanisms. In a typical application a WSN is scattered in a region where it is meant to collect data through its sensor nodes. WSNs have and will continue to play a vital role in our daily lives.

2.3 Wireless Body Area Networks The rising costs of healthcare and the increase in continuous healthcare monitoring of the aging population have led to the concept of novel wireless-sensor-technology-driven body-monitoring networks.

As

WSN strictly depends on the particular applications and scenarios that are investigated. WBAN therefore, constitute a particular branch of WSNs that must fulfill all the requirements of wireless body-monitoring systems. These wireless body-monitoring networks comprise small and intelligent wireless medical sensors attached to the clothes or the body of a human being or even implanted in human (or animal) tissues. These medical sensors are capable of measuring, processing, and forwarding

Chapter 2. Basic Concepts of Wireless Body Area Network

9

important physiological parameters such as the heart rate, blood pressure, glucose level, body and skin temperature, oxygen saturation, and respiration rate, as well as records such as electrocardiograms and electromyograms. This enables health professionals to predict, diagnose, and respond to adverse incident earlier than ever. A WBAN has been shown to be adequate for dealing with emergency cases, where it continuously sends the patient’s physiological information to a remote server or physician to help maintain optimum health status. Integration of WBANs with a telemedicine system has the potential to enable the provision of a wide range of assistance to patients, medical personnel, and society through continuous monitoring in the ambulatory environment, early detection of abnormal conditions, supervised restoration, and knowledge discovery through data mining of all the gathered information [16]. A WBAN provides long-term healthcare monitoring system under natural physiological states, without restricting normal activities of a patient [17]. Moreover, this technology reduces the amount of time a doctor requires to identify a problem, decreases the amount of paperwork required, and eliminates duplication of patient records [18].

A conceptual view of medical

WBAN is shown in Figure 2.1. The depicted WBAN contain few sensors to monitor vital bodywide health information and send it to a remote server using a PDA. The concept of a WBAN was introduced by Van Dam et al. in 2001 and attracted the interest of several researchers. The currently available WLAN/WPAN standards such as IEEE 802.15.4, IEEE 802.15.3, and IEEE 802.11 are not suitable for WBANs since they do not support the medical (nearness to body tissue) and relevant communication regulations for some application environments [19]. They also do not

10

Chapter 2. Basic Concepts of Wireless Body Area Network

sustain the combination of low power, reliability, QoS, high data rate, and noninterference required to broadly address the spectrum of body area network applications [2]. Owing to the lack of an appropriate wireless technology that satisfies all the requirements of WBANs, the IEEE 802.15 Working Group formed Task Group 6 (TG6) in November 2007 to develop a communication standard known as IEEE 802.15.6. The purpose of the group is to establish a communication standard optimized for short-range and low-power on-body/in-body nodes to serve a variety of medical, entertainment and consumer electronics applications. The standard provides efficient communication solutions to ubiquitous healthcare and telemedicine systems, interactive gaming, military services, and portable audio/video systems. WBANs must support a combination of reliability, high data rate, low power, QoS, and non-interference to address the gamut of WBAN applications. WBAN EEG

Satellite

EMG

Physician

ECG GPRS

BP

PDA/Phone Medical Information Database

Motion Thermometer

Home Computer

Figure 2.1: A glimpse of WBAN and it’s Framework

Emergency

Chapter 2. Basic Concepts of Wireless Body Area Network

11

2.4 Challenges in a WBAN In the present-day literature, WBAN is deemed to be a special type of WSN with different application requirements. Many challenges of a WBAN are similar to those of WSN. However, Traditional sensor networks do not specifically tackle all the challenges associated with the human body. Due to the special features of the environment in which the WBAN operates (human body) the data loss is more significant. The sensors which are either attached on the surface of human body or implanted into the tissue must be very tiny in size to support invisible monitoring of the patients. Also, the tiny size of the WBAN sensors heavily affects the energy resources of the devices. The challenges in healthcare application may includes: energy efficiency, reliability, security and interference, material constraints, robustness and QoS [20].

2.4.1 Energy Efficiency As most wireless networks based devices are battery operated therefore, energy efficiency arguably the most important consideration in almost every area of application of wireless sensor networks. But the tiny size of the WBAN sensors heavily affects the energy resources of the devices. The power supply recharge and replacement of the devices are often impossible or difficult. As a typical alkaline battery provides about 50 watt-hours of energy, therefore in a full active mode a node can’t operate more than a month. Thus, sensors need to run for a long lifetime. In practice, for many medical applications, the device (such as heart pacemakers) should work for several months/years without any replacement. To deal with the power issues in a WBAN, the developers have to design a better power management schemes and scheduling

Chapter 2. Basic Concepts of Wireless Body Area Network

12

algorithms [21].

2.4.2 Reliability Contrary to the classical WSNs, there are no redundant sensor nodes in a WBAN. The missing information from a node often cannot be compensated by other nodes and therefore, all nodes in the network must be reliable, and accurate. The importance of reliability is also obvious when one considers that a WBAN monitoring the critical symptoms of people in poor health; one missed vital signal could be the difference between life and death.

2.4.3 Security and Interference One of the most important issues, especially for medical systems is Security and interference.

Physiological data collected by the

sensor network is the health information, which is of personal nature. Therefore, It is important to keep the information from being approached by unauthorized entities.

This confidentiality can be

achieved by encrypting the information during transmission. Data Authenticity is also one of the security requirements. This property is also very important because absence of this property may lead to; an illegal entity disguise as a legal one, reports false data, and can also gives wrong instructions to the other medical sensors possibly causing serious harm to the host [22]. The WBAN coordinator must also support a large number of devices to coexist in a single network without having any interference with each other. Things become more complicated when multiple people wearing WBANs come into range of each other, and hence making the coordination more difficult.

Chapter 2. Basic Concepts of Wireless Body Area Network

13

2.4.4 Material Constraints Another issue for wireless sensor networks application to healthcare is material constraints. For an implanted medical sensor, the materials, size, and shape of sensor must be harmless to the human tissue. For example, the designed of the retina prosthesis sensor must be small enough to fit within an eye. Also chemical reactions with body tissue and the disposal of the sensor are of extreme importance [23].

2.4.5 Quality of Service WBAN applications are very sensitive and hence QoS issues in WBAN require more attention and focus and should be taken up more seriously [24].

For example, when WBAN is used to monitor a patient’s

health condition, information have to be delivered instantly. Delays in delivering critical or emergency alarms may be catastrophic.

2.4.6 Robustness In WSNs hundreds to thousands of sensor-nodes cover large areas and a group of sensor nodes are responsible to monitor a same physical or environmental condition, thus offering a considerable degree of redundancy in WSNs. In contrast to WSNs, WBANs cover an area limited to the human body and offer no redundancy. Therefore, in a conventional WBAN all nodes must be highly reliable, robust, and accurate.

Chapter 2. Basic Concepts of Wireless Body Area Network

14

2.5 WBAN Applications WBAN has a great potential for various applications including ubiquitous healthcare and telemedicine systems, interactive gaming, military services, and portable audio/video systems. WBANs support both on-body and in-body applications.

On-body medical and

non-medical applications include monitoring blood pressure, heart rate, respiration, temperature, establishing a social network, searching forgotten things and assessing soldier fatigue. In-body applications may include monitoring pacemakers, control of bladder function, monitoring implantable cardiac defibrillators, and restoration of limb movement [25].

The following part presents some of the WBAN

applications:

2.5.1 Monitoring Patients with Cardiovascular Diseases The Cardio Vascular Disease (CVD), representing 30% of all the world’s deaths is the main cause of death throughout the globe. The World Health Organization has reported a death toll of more than 17 millions people due to coronary heart disease or strokes in 2004. While by 2030 this number will raise to almost 23.6 millions. A WBAN is a key technology to prevent abnormal conditions like occurrence of CVD, and atrial fibrillation, and can be used for ambulatory health monitoring. The corresponding medical staff can do treatment preparation in advance as they receive vital information regarding heart rate and irregularities of the heart while monitoring the health status of the patient.

Chapter 2. Basic Concepts of Wireless Body Area Network

15

2.5.2 Monitoring Elderly Patients According to the U.S. Census Bureau, the number of adults aged 65 and over is expected to be doubled from 35 million in 1990 to nearly 70 million by 2025. So according to the same trend, the worldwide population aged 65 and over is also expected to be more than 761 million in 2025. In addition, one study has founded that almost one third of the U.S. adults, were serving as a caregivers mostly to an elderly parents. WBANs offer an invaluable healthcare, non-intrusive and regular monitoring for elderly people (who often feel depressed and lonely) by detecting any abnormal situation and alerting neighbors, family or the nearest hospital [26].

2.5.3 Cancer Detection Cancer is nowadays one of the huge threats for human life. With rising numbers each year, cancer is the second major cause of death in the United States. A set of miniaturised sensors capable to detect nitric oxide (released by cancer cells) can be seamlessly integrated in WBAN. These sensors have the capability to differentiate cancerous cells, between different types of cells. This allows physician to diagnose tumors without biopsy.

2.5.4 Telemedicine Systems Telemedicine is the way to provide better health care facilities to the people of the underprivileged unprocurable areas.

Existing

telemedicine systems either deploy dedicated wireless channels to transfer data to the remote nodes, or energy efficient protocols such as Bluetooth that are prone to interference by other different

Chapter 2. Basic Concepts of Wireless Body Area Network

16

devices working in the identical frequency band. These characteristics limit prolonged health monitoring.

Integration of WBANs with a

telemedicine system has the potential to enable the provision of a wide range of assistance to patients, medical staff, and society through nonstop monitoring in the ambulatory environment, early detection of abnormal conditions, supervised restoration, and knowledge discovery through data mining of all the gathered information [16]. Figure 2.2 shows a real time telemedicine framework for patient rehabilitation.

Figure 2.2: A real-time telemedicine infrastructure

2.5.5 Diabetes Worldwide, more than 246 million people are suffering from diabetes, and this number is expected to mount to 380 million by 2025. The US national institute of health (NIH) reported 15.7 million people had the diabetes in 1999 in the US alone. Diabetes can yield other complicated diseases like heart disease, stroke, high blood pressure, blindness,

Chapter 2. Basic Concepts of Wireless Body Area Network

17

kidney disease, and amputations. A WBAN could provide a more accurate, consistent, and less invasive modus operandi by monitoring glucose levels, transmit the results to a fixed terminal or wireless PDA, and by injecting insulin automatically when a threshold glucose level is reached [27]. Frequent monitoring enables an appropriate dosing of medicines and minimizes the risk of blindness, fainting and deficit of circulation and other complications.

2.5.6 Battlefield WBANs can be used to link soldiers in a combat zone, and report their movements to the commander, i.e., firing, running, and digging. However, the soldiers must have a secure and reliable communication channel in order to stop trapping.

2.5.7 Asthma A WBAN can assist millions of patients affected from asthma by monitoring allergic particles in the atmosphere and by providing real time response to the physician. Chu et al. [28] developed a GPS based device that triggers an alarm in case of revealing information allergic to the patient.

2.5.8 Artificial Retina WBANs can also serve blind people with no vision or limited vision. By using an implanted retina prosthesis chips within a human eye can help blind people to see at a reasonable level. Figure 2.3 shows an artificial retina for blind people [29].

Chapter 2. Basic Concepts of Wireless Body Area Network

18

Figure 2.3: Artificial Retina for Blind People

2.6 Existing Health-care Projects There are a number of research projects throughout the world which have focused on the design and implementation of monitoring systems for patients having some chronic diseases and disabilities. The wireless technologies in the field of wireless short-range connectivity used in these projects are the IEEE 802 family of WLANs, WPANs, Zigbee and Bluetooth [30, 31]. But the IEEE 802.15.4/zigbee network has been the most recommended approach in the existing healthcare projects before the release of IEEE 802.15.6 standard. A brief overview of these existing projects are given below

2.6.1 CodeBlue CodeBlue is a sensor networks based medical research project being developed at Harvard University. CodeBlue is a distributed

Chapter 2. Basic Concepts of Wireless Body Area Network

19

Figure 2.4: CodeBlue architecture for emergency response WBAN including a two-lead ECG, pulse oximeter, and a specialized motion-analysis sensor used for transmitting vital signs and geolocation information using the IEEE 802.15.4 platform. This project includes in-hospital emergency care, and pre-hospital care, stroke patient rehabilitation and disaster response. Research from this project has capabilities for real-time triage decisions, resuscitative care, and long term patient observations [32].

It also offers services for location

tracking, handoff, credential establishment, and aggregation of sensor data.

A simple query interface can able the emergency medical

technicians to demand data from groups of patients. CodeBlue can scale across a wide range of network densities. It is designed to operates on a range of wireless devices, from motes to PDA and PC systems. [23, 33]. CodeBlue architecture for emergency response is shown in Figure 2.4.

2.6.2 Ayushman Ayushman, being developed by the IMPACT lab at Arizona State University is a health monitoring infrastructure based on ZigBee real-time sensor network. Ayushman provides a medical monitoring

Chapter 2. Basic Concepts of Wireless Body Area Network

20

system that is dependable, energy-efficient, secure and collects real-time health data in diverse scenarios, from home based monitoring to disaster relief.

It is also designed to be a testbed which allow

researchers to test their communication protocols and systems in a realistic environment.

2.6.3 MobiHealth Mobihealth is a project using Bluetooth and Zigbee for intra-BAN communication

and

Universal

Mobile

Telecommunications

System (UMTS) or General Packet Radio Service (GPRS) wireless communication technology for transferring data between the BANs and a remote healthcare server. It is a European based project which use BAN-Based sensors and wireless telephony technology to create a generic platform for home healthcare[32] . MobiHealth aims to provide a constant monitoring system to patients outside the hospital domain [34]. MobiHealth targets, improving the quality of healthcare services by enabling advanced services in the areas of remote assistance, disease diagnosis, disease prevention, and physical state monitoring [23]. Therefore, a patient doesn’t need to stay in the hospital for short or long periods of health monitoring. With the MobiHealth BAN a patient can be free to run his daily life activities.

2.6.4 Human++ research program The Human++ research project by IMEC-NL aims to develop key technologies and components for future WBANs health monitoring applications.

This project will provide medical, lifestyle, assisted

living, sports and entertainment functions. It combines expertise in

Chapter 2. Basic Concepts of Wireless Body Area Network

21

wireless ultralow power communications, packaging, 3D integration technologies, MEMS energy scavenging techniques and lowpower design techniques.

The system is highly power efficient, two AA

batteries are able to work for 3 months using the 2.4 GHz ISM band.

2.6.5 HipGuard System HipGuard is a prototype developed for patients recovering from hip surgery. This system monitors patient’s hip rotation and position with embedded wireless sensors. Real-time alarm warnings can be sent to Wrist Unit of the patient if hip rotations are false [35]. Figure 2.5 shows the HipGuard system.

Figure 2.5: HipGuard System

2.6.6 eWatch The eWatch is a wristwatch wearable sensing, computing, and notification platform [36]. ewatch can be used for applications such as

Chapter 2. Basic Concepts of Wireless Body Area Network

22

elderly monitoring, fall detection and context aware notification. An ewatch system can sense and query emergency events. The ewatch system can use its network capabilities to call for help upon no response from the user. The ewatch can also notify a patient’s certain medication.

2.6.7 UbiMon UbiMon is a DTI funded project which aims to provide a continuous monitoring system for patient in order to capture transient events [37, 38]. A number of biosensors were developed such as a 3-lead ECG, 2-lead ECG strip, and SpO2. Furthermore, a PDA is use to analyze and display the collected sensor information.

Figure 2.6: Life Shirt

2.6.8 LifeShirt LifeShirt is a completely noninvasive and comfortable smart garment that provides the most complete picture of the healthcare system. It enables researchers and healthcare professionals to accurately monitor over 30 vital life-sign functions of the real-world settings where people

Chapter 2. Basic Concepts of Wireless Body Area Network

23

live and work [32, 39] . LifeShirt collects patient data using integrated sensors including ECG, and respiratory bands. records physical activities and postures.

It also tracks and

Figure 2.6 shows smart

LifeShirt system for healthcare monitoring.

2.7 Summary of the Chapter In this section, we presented our research goals toward an applicable and efficient healthcare system.

We pointed out the challenges in

a WBAN in healthcare perspective.

We introduced the structure

and applications of a human body healthcare system. The existing healthcare projects were examined to have a closer look at the problem considered in this dissertation.

Chapter 3

The IEEE 802.15.6 standard: An overview 3.1 Introduction In this chapter, we present some basic features of the IEEE 802.15.6-based WBANs standard.

Starting from the fundamental

details, we provide deep insight into the MAC and PHY layers. We review different communication modes and access mechanisms and explain the NB, UWB, and HBC specifications in detail. A detailed description of the PHY and MAC characteristics of this standard is available in [2].

3.2 The IEEE 802.15.6 standard IEEE 802.15.6 [40], defines a communication standard for a WBAN which is low power, highly reliable, and short range, wireless communication in the near locality of, or inside a human body. The standard is designed to assist the advanced medical and entertainment 24

Chapter 3. The IEEE 802.15.6 standard: An overview

25

NB PHY

MAC

UWB PHY

HBC PHY

Figure 3.1: IEEE 802.15.6 PHY and MAC layers applications. The standard defines a MAC layer that supports several Physical (PHY) layers, such as Ultra-wideband (UWB), Narrowband (NB), and Human Body Communications (HBC) layers, as illustrated in Figure 3.1. The proper selection of PHYs or frequency bands is one of the key issues to be considered in the development of WBANs [41]. Generally, the communication authorities in different countries can regulate the available frequencies for WBANs. Figure 3.2 shows the available frequency bands for WBANs [42]. The following sections present the PHY specifications of IEEE 802.15.6.

3.3 IEEE 802.15.6 PHY Specifications IEEE 802.15.6 PHY is responsible for (1) activation/deactivation of the radio transceiver, (2) clear channel assessment (CCA), and (3) data reception and transmission. The standard supports three operational PHYs, two of which are mandatory and one is optional. The two

Chapter 3. The IEEE 802.15.6 standard: An overview

26

Figure 3.2: IEEE 802.15.6 frequency bands mandatory PHYs are HBC and UWB PHYs, whereas, the NB PHY is considered optional. The following sub-sections present the NB, UWB, and HBC PHY specifications of IEEE 802.15.6.

3.3.1 NB PHY Specifications The NB PHY is an optional physical layer, which is responsible for the following tasks [2]: • Activation and deactivation of the radio transceiver. • Clear channel assessment (CCA). • Data transmission and reception. The NB PHY operates in seven distinct frequency bands and can offer a variable number of bit rates, channels, and modulation schemes. The different frequency bands are—

402-405 MHz, 420-450 MHz,

863-870 MHz, 902-928 MHz, 950-958 MHz, 2360 to 2400 MHz, and 2400-2483.5 MHz [43]. As depicted in Figure 3.3, the Physical-layer Protocol Data Unit (PPDU) encapsulates the Physical-layer Service Data Unit (PSDU) in its frame and appends several control fields that are used to synchronise the transmission and identify the transmission parameters. The preamble of the Physical-layer Convergence Protocol

27

Chapter 3. The IEEE 802.15.6 standard: An overview

(PLCP) is a concatenation of two sequences and is used for coarse time synchronisation, carrier-offset recovery, packet detection and fine timing synchronisation. The PLCP header consists of several fields that convey the PHY parameters to the receiver. The PSDU consists of a MAC header, a MAC frame body, and the FCS. RATE

Reserved LENGTH Reserved

3

1

8

PLCP Preamble

1

BURST MODE

SCRAMBLER SEED

1

1

PHY Header

HCS

BCH parity

MAC Header

MAC Frame Body (0~255 Bytes)

FCS

15

4

12

56

variable length

16

PLCP Header

Sequence #1

Sequence #2

63

27

PSDU

Figure 3.3: PPDU structure for NB PHY (the indicated lengths are in bits)

3.3.2 HBC PHY Specifications. HBC PHY supports two operation modes —high QoS mode and default mode, depending on the application. HBC PHY uses Electric Field Communication (EFC) technology and operates in two frequency bands centred 16 MHz and 27 MHz with the bandwidth of 4 MHz [44]. Similar to the NB PHY, the HBC packet structure encapsulates the PSDU in the packet after adding control bits and error correction and detection bits, as depicted in Figure 3.4. According to the standard, a maximum of 64 nodes may be connected to a hub or LDPU simultaneously.

28

Chapter 3. The IEEE 802.15.6 standard: An overview

PLCP Preamble

SFD/RI

PLCP Header

PSDU

Figure 3.4: PPDU structure for HBC PHY

3.3.3 UWB PHY Specifications The UWB PHY is used to provide a MAC layer data interface under the control of Physical Layer Convergence Protocol (PLCP). Compared to the two aforementioned PHY specifications, UWB PHY intents to achieve high performance, low complexity, and low power consumption. UWB PHY operates in two frequency bands; low band and high band. UBW PHY supports 11 channels; three in the low band (channels 0-2) and eight in the high band (channels 3-10). UWB PHY supports Impulse Radio UWB (IR-UWB) and wideband Frequency Modulation UWB (FM-UWB) technologies. A hub can only implement one of these technologies, but the devices can implement either FM or IR-UWB or both technologies. As shown in Figure 3.5, the UWB PHY supports two operational modes, high Quality of Service (QoS) operational mode and default mode, where the first one is designated for high-priority medical applications and the second mode is used for medical and non-medical applications. The PPDU for a UWB frame consists of the Synchronisation Header (SHR), Physical-layer Header (PHR), and PSDU. The content of the PSDU depends on the operation mode. The maximum number of retransmissions for a UWB PHY, is set to four.

3.4 IEEE 802.15.6 MAC Specifications According to the IEEE 802.15.6 standard, all the nodes and the hubs in the network are organized into logical sets, referred to as Body

29

Chapter 3. The IEEE 802.15.6 standard: An overview

FCS

BCH parity bits high QoS mode

or MAC Header

MAC Frame Body

FCS

MPDU SHR

PHR

default mode

PSDU

MAC Header

MAC Frame Body

FCS

BCH parity bits

MPDU

Figure 3.5: PPDU frame structure for UWB PHY Area Networks (BANs). A BAN may consist of only one hub, and up to mMaxBANSize (mMaxBANSize is often set to 64) number of sensor nodes connected to it.

The sensor nodes are connected to

the hub, over the wireless medium, in a star network topology. The standard supports both one-hop and two-hop communications for the WBANs. In a one-hop star topology, the exchange of frames takes place directly between the sensor nodes and the hub. In a two-hop star topology, relay-capable nodes can be used to exchange packets between the hub and the sensor nodes. A hub divides time a-xis into multiple superframes of equal length. Each superframe is subdivided into a number of allocation slots that are used for data transmission. The following sections present the communication modes, MAC frame format, and access mechanisms defined in the IEEE 802.15.6 standard.

3.4.1 IEEE 802.15.6 Communication Modes IEEE 802.15.6 network can operate in one of the following three access modes.

Chapter 3. The IEEE 802.15.6 standard: An overview

3.4.1.1

30

Beacon Mode with Superframe Boundaries

In this mode, the hub transmits a beacon on the medium at the beginning of each active superframe to provide time-referenced allocations. The active superframes may be followed by several inactive superframes whenever there is no scheduled transmission.

Each

superframe is further divided into access phases (APs), as indicated in Figure 1.1. A superframe includes exclusive AP1 (EAP1), random AP1 (RAP1), management AP1 (MAP1), exclusive AP2 (EAP2), random AP2 (RAP2), management AP2 (MAP2), and an optional B2 frame followed by a contention access phase (CAP). Each access phase, except RAP1, may have zero length.

To provide a zero-length CAP, the

hub shall not transmit a prior B2 frame unless it needs to provide a group acknowledgment. If EAP1 has a nonzero length, it starts immediately after the preceding beacon. The EAPs can only be used by the highest-priority nodes such as those reporting emergencies or medical events, whereas the RAPs and the CAP can be used by any regular priority node. In an MAP, the hub may arrange scheduled uplink/downlink allocation intervals and scheduled/unscheduled bilink allocation intervals and may provide Type I or Type II polled allocation intervals.

In the case of improvised and unscheduled

transfers, the nodes wait for a post or poll frame from the hub, while in the scheduled transfers case, the nodes use their assigned slots for data transmission . The difference between the Type I and Type II access phases lies in the units used to request reservations. In Type I, the device requests allocation intervals in terms of time, while in Type II, the device requests allocation intervals in terms of the number of frames.

Chapter 3. The IEEE 802.15.6 standard: An overview

3.4.1.2

31

Non-beacon mode with superframe boundaries

In this access mode, the hub operates during the MAP periods only. The hub transmits the superframe structure via T-Poll frames, and the entire superframe duration is covered by either a Type I or a Type II access phase, but not by both phases. 3.4.1.3

Non-beacon mode without superframe boundaries

In this access mode, each node establishes its own time base independently. The hub grants unscheduled Type II polled or posted allocation or a combination of both, which allows the sensor to transmit only a limited number of frames.

3.4.2 IEEE 802.15.6 MAC Frame Format A MAC frame is an ordered sequence of fields delivered to or from the physical layer service access point (PHY SAP). Figure 3.6 shows the general MAC frame format consisting of a fixed-length (56-bit) header, variable length frame body, and a fixed-length (18-bit) Frame Check Sequence (FCS). The MAC frame body ranges from 0 to 255 bytes. The MAC header further consists of 8 octets frame control, 1 byte recipient Identification (ID), 1 byte sender ID, and 1 byte WBAN ID fields. The frame control field carries control information including beacon, acknowledgement, type of frame and other control frames. The sender and recipient ID fields contain the address of the sender and recipient of the data frame, respectively. The BAN-ID field contains information of the active WBAN. The first 1 byte field in the MAC frame body carries information required for replay detection and nonce construction. The frame payload field carries data frames. The last

32

Chapter 3. The IEEE 802.15.6 standard: An overview

32-bit Message Integrity Code (MIC) carries information about the integrity and authenticity of the frame. 7

Octets:

MAC Header

0-255

MAC Frame Body Variable Length:0-255 bytes

MHR

FRAME CONTROL

RECIPIENT ID

2

x

SENDER ID

FCS

FTR

BAN ID

Octets: 4

1

1

1

Figure 3.6: IEEE 802.15.6 MAC frame format

3.4.3 Priority Mapping The User Priorities (UPs) for accessing the medium is differentiated by 8 different access categories. UP values are referenced in categorizing medium access of management and data type frames. The type of payloads in the frame determine these prioritizing values. These traffic designation are typed as Emergency or medical implant event report, High-priority medical data or network control, Medical data or network control, Voice (VO), Video (VI), Excellent effort (EE), Best effort (BE), and Background (BK). The UPs are prioritized by the values of the minimum and maximum Contention Windows (CW) and Contention Probability (CP). The predefined relationships between Contention Window bounds, CWmax and CWmin, and UP for CSMA/CA, and between contention probability (CP) thresholds CPmax and CPmin and UP for slotted Aloha access are depicted in Table 3.1.

33

Chapter 3. The IEEE 802.15.6 standard: An overview

Table 3.1: Contention window bounds for CSMA/CA and contention probability thresholds for prioritized-based slotted Aloha access User Priority CSMA/CA Slotted Aloha CWmin CWmax CPmax CPmin 7 6 5 4 3 2 1 0

1 2 4 4 8 8 16 16

4 8 8 16 16 32 32 64

1 1 2 3 8 3 8 1 4 1 4 1 8 1 8

1 4 3 16 3 16 1 8 1 8 3 32 3 32 1 16

3.4.4 IEEE 802.15.6 Access Mechanisms In IEEE 802.15.6, the access to the shared medium is provided using various mechanisms. These are divided into four categories; scheduled access, improvised and unscheduled access, random access, and medical implant communications service (MICS) band access. The following sections briefly describe these access mechanisms. 3.4.4.1

Random Access Mechanism

In IEEE 802.15.6, two random access protocols are proposed. The contention based random access methods for obtaining allocations are either CSMA/CA if a NB PHY / UWB PHY is chosen or slotted ALOHA if UWB /HBC PHY is used [2]. The following sections briefly describe the slotted ALOHA and CSMA/CA protocols of IEEE 802.15.6 standard. 3.4.4.1.1

Slotted ALOHA Protocol

The IEEE 802.15.6 standard uses a particular kind of slotted Aloha

34

Chapter 3. The IEEE 802.15.6 standard: An overview

as a MAC choice. It is designed to tackle with a limited number of users. This protocol behavior is different from the conventional one as it attempt to resolve contention through reduction of the re-transmission probability in a distinct way. In the slotted ALOHA protocol, the nodes access the channel using predefined UP values as given in Table 3.1. These priority values are used to classify the high-priority and low-priority traffic. Initially, the Collision Probability (CP) is selected according to the UPs. The nodes obtain contended allocation if z ≤ CP, where z is randomly selected from the interval [0, 1]. To obtain a new contended allocation for the first time transmission or retransmission of a packet a node shall set its CP as follows [45]. 1. If the node did not obtain any contended allocation previously, it shall set the CP to [CP ]max [UP]. If the node succeeded, it shall set the CP to [CP ]max [UP]. 2. If the node failed in the last contended allocation it had obtained, a) It shall keep the CP unchanged if this was the m-th time the node had failed consecutively, where m is an odd number; b) It shall halve the CP if this was the n-th time the node had failed consecutively, where n is an even number [46]. c) If halving the CP makes the new CP value smaller than [CP ]min [UP] value, then the new CP value is set to [CP ]min [U P ]. 3.4.4.1.2

CSMA/CA Protocol

This section briefly summarizes the operational mechanism of CSMA/CA of IEEE 802.15.6 MAC. For a more detailed presentation, the reader can refer to the IEEE 802.15.6 standard [2].

To employ

the CSMA/CA mechanism, a node with a new packet to transmit

Chapter 3. The IEEE 802.15.6 standard: An overview

35

Figure 3.7: Slotted Aloha access illustration shall maintain a Contention Window (CW) to detect a new contended allocation, where CW ∈ (CW min, CW max) and a backoff counter ∈ [1, CW ]. As given in Table 3.1, the values of CWmin and CWmax are selected according to the UPs. The high-priority traffic will have a small contention window compared to that of low-priority traffic, which increases the probability of accessing the channel to report emergency events. The contending node having a packet for transmission shall set its backoff counter to a sample of an integer random variable uniformly distributed over interval [1, CW ] to minimize the probability of collision. CW is a contention window chosen by a node having a packet for transmission as follows: - If the node does not obtain any contended allocation previously, or if it succeeds in a data frame transmission, or the node after transmitting a frame requires no acknowledgement, it will set the CW to CWmin . - If the node fails, that is, if the node does not receive the expected acknowledgement for its last frame transmission, then it shall keep the CW unchanged if this is the mth time the node has failed consecutively, where m is an odd number; otherwise, the CW is doubled. - If doubling the CW results in a value that exceeds CWmax [U Pi ], the node will set the CW to CWmax [U Pi ].

Chapter 3. The IEEE 802.15.6 standard: An overview

36

After choosing the contention window, the node starts its carrier sensing at the beginning of the next pCSMAslot to determine the current state of the channel. Each pCSMAslot has a fixed duration specified by pCSMASlotLength. The very first portion of pCSMAslot, which is equal to 63/symbol-rate in time length, corresponds to pCCATime (physical CCA), while the latter portion of pCSMAslot is used by the contending node to transmit its frame to the transport medium when its backoff counter reaches zero. The node will decrement its backoff counter by one for each idle pCCATime. Further, the node will lock the backoff counter whenever it detects any transmission on the channel during pCCATime and will unlock it when the channel has been idle for pSIFS. The node will also lock the backoff counter if it is not allowed to access the medium during the current AP or the current AP length is not long enough for a frame transmission. When the backoff counter reaches zero, the node then transmits. Figure 3.8 shows an example of the CSMA/CA protocol for a non emergency node. As shown in the figure, the node unlocks the backoff counter in RAP1. However, the contention fails and the value of CW remains unchanged because CW does not change for an odd number of failures. In the following CAP period, the backoff counter is set to five; however, it is locked at two because the time between the end of the slot and the end of the CAP is not sufficient to accommodate the data frame transmission and Nominal Guard Time (GTn ). The backoff counter is then unlocked in the RAP2 period. This time, the value of CW is doubled because there is an even number of contention failures. The backoff counter is set to eight and is unlocked. Once the backoff counter reaches zero, the data are transmitted and the value of CW is set to CWmin.

37

Chapter 3. The IEEE 802.15.6 standard: An overview

Backoff counter decrements TD

GTn

TD

Backoff counter(=0) Backoff counter(=2) is unlocked

SIFS

Contention fails 1st time counter(=5) CW is now reset to 5 is unlocked over [1,CW] and locked

D

Slot Slot Slot Slot Slot Slot Slot Slot

No enough time is left; backoff counter (=2) is locked. Backoff

Slot Slot SIFS

CW=Cwmin=8; backoff counter is set to 3 over[1,CW] and unlocked

SIFS

Backoff counter(=0)

Slot Slot Slot Slot

SIFS

Slot Slot Slot

Data arrives

GTn

RAP1

CAP

RAP2 D

TD

TD

GTn

Backoff counter(=8)

D

Backoff counter(=0)

Contention fails 2nd time CW=16(doubled); backoff counter is reset to 8 over [1,CW] and locked

Contention succeeds. CW is reset to 2 over [1,CW] and locked

Figure 3.8: IEEE 802.15.6 CSMA/CA mechanism 3.4.4.2

Improvised and Unscheduled Access Mechanism

As discussed above, the hub may use improvised access to send poll (i.e.

a hub instruction) or post (i.e.

a data request from the

hub) commands to a node without prereservation or advance notice in beacon or non-beacon modes with superframe boundaries or in non-beacon mode without superframe. These commands are used to initiate the transactions of one or more data frames by the nodes or hub outside the scheduled allocation interval. The polls are used to grant Type I or Type II polled allocation to the nodes, while the posts are used to send management frames. The Type I polled allocation starts after the duration of pSIFS and stops at the end of the allocated slot in the current superframe. Similarly, the Type II polled allocation starts after the duration of pSIFS and stops after all of the data frames are sent by the polled node. The hub may also use an unscheduled access mechanism to obtain an unscheduled bilink allocation. The unscheduled bilink allocation may be (1) one-periodic, where frames are exchanged between the nodes and hub every superframe, or (2) multiple-periodic (m-periodic), where frames are exchanged every m

Chapter 3. The IEEE 802.15.6 standard: An overview

38

superframes thus allowing the devices to sleep between m superframes. An m-periodic bilink allocation is suitable for low-duty cycle nodes because nodes in m-periodic allocation sleep between m superframes. 3.4.4.3

Scheduled and Scheduled-Polling Access Mechanisms

Unlike unscheduled allocation, the scheduled access mechanism is used to obtain scheduled uplink and scheduled downlink allocations only in beacon or non-beacon mode with superframes. In addition, the scheduled polling is used to obtain scheduled bilink, polled and posted allocations, but not in non-beacon mode without superframes. These allocations may be one-periodic or m-periodic; however, neither of these allocations is allowed in a single WBAN at the same time. The nodes consider the superframe periods (with allocated slots) as the wakeup periods. The uplink and downlink allocations are used to send management and data frames to and from the hub, respectively. 3.4.4.4

MICS band communication

In the MICS band, a hub shall operate with or without superframes. The hub may choose a new channel only when required and an implant shall communicate as a node with a hub. The hub and the node may perform Unconnected mutual discovery or Connected mutual discovery before their exchange of data or management type frames.

3.5 Summary of the Chapter This chapter presented the most important features of the IEEE 802.15.6 standard. A deep explanation of PHY, and MAC layers specifications of the standard was presented. Different communication modes and

Chapter 3. The IEEE 802.15.6 standard: An overview

39

access mechanisms were explained. The NB, HBC, and UWB PHY specifications were reviewed in terms of frame structure, and other key parameters. In addition, this chapter could be used to quickly understand different features of the IEEE 802.15.6 standard and to analyse its potential for different applications.

Chapter 4

Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols 4.1 Introduction In this chapter, we examine the IEEE 802.15.6-based WBAN MAC protocols performance under saturation,

non-saturation traffic

conditions using both accurate analytical and simulation models. We also consider a portion of the access phases of the superframe and analyze its impact on the performance of the IEEE 802.15.6 CSMA/CA MAC protocols. The existing Markov chain-based analysis of IEEE 802.15.6 do not consider the time spent by a node awaiting the acknowledgement frame, until time-out occurs. This work remains distinct as, while constructing the DTMC, we take into consideration the time spent by a node awaiting the acknowledgement frame, in our Markov model this state is presented as (i, −1). We study important

40

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

41

network performance descriptors such as normalized throughput mean frame service time and energy consumption under saturation , non-saturation (both homogeneous and heterogeneous scenarios). This chapter is organized as, Section 4.2 briefly describes different performance modeling approaches for MAC protocols. In Section 4.4 we investigate the IEEE 802.15.6-based WBANs MAC protocol performance under saturation conditions through analytical and simulation models. Section 4.5 provides the performance evaluation of the IEEE 802.15.6-based WBANs under non-saturation conditions. Section 4.5 provides the performance of the IEEE 802.15.6-based WBAN MAC protocols under different access phases of the superframe. Finally, Section 4.6 concludes the most important research findings developed in this chapter.

4.2 Performance Modeling Approaches for MAC Protocols MAC protocols performance evaluation can be carried out by means of software simulations, analytical models or employing a testbed. The simulation are usually time consuming and may only provide reasonable understanding of a protocol under specific conditions. On the other hand, analytical modeling provides simplified but robust estimation of the protocol. While developing the stochastic models of MAC protocols performance evaluation, various assumptions and approximations are considered. There are mainly three performance modeling techniques used in this area, and are briefly discussed below.

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

42

4.2.1 Simulation Approach Simulation is a fast and easy technique to analyze a system. Simulation can help in predicting and analyzing the performance of the system. Most of the computer based systems can easily be analyzed via simulation.

However, a system defined by a simulation usually

provides specific features under optimal environments and can not provide details on a protocol behavior. The commonly used network simulators are OPNET, OMNet++, NS-2, QualNet, and GloMoSim. Moreover, custom made simulator based on C/C++ and other programming languages are also used.

4.2.2 Real Experimentation Approach The real experimentation method usually confirms the results obtained by other modeling Approaches (simulation/analytical or both) and are mostly preferred over other methods. This method usually requires more efforts, time, hard work and even more budget. As compare to the other methods, the result of this method can be changed but it is difficult to diagnose such changes. An experimental evaluation can help to identify which characteristics need to be included into the model.

4.2.3 Analytical Approach and Markov Analysis Analytical approach is a way to describe a system mathematically by applying tools such as probability and queuing theories. Then numerical methods are used to get the insight from the developed model. Analytical modeling would be preferable for relatively small and simple systems. In this case, the model is mathematically tractable and requires less computational efforts. Analytical modeling can be

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

43

cost-effective. It can provide a conceptual view of the components in the system interacting with one another. However, many simplifying assumptions during the modeling process can lead to an improper representation of the real system. The mathematical analysis can be further intensified by the real experimentation and/or simulations. The analytical method generally provides an insight into the effects of different parameters, their ranges, and their interactions. The analytical method can detect design errors that can not be founded easily by using real experiment or simulation. One of the techniques commonly used for mathematical analysis of MAC protocols is the Markov analysis. The Markov analysis is the most widely used one in the performance modeling of MAC protocols. The state transition probabilities must be found to solve the Markov chain model. To find out the state transition probabilities of the Markov chain, the traffic is considered to be Bernoulli/Poisson so that the memoryless characteristic is maintained, or the nodes are assumed as always having at least one packet in the queue waiting for transmission.

Even with such assumptions,

a common issue is the complexity associated with the transition probability matrix, especially for large number of states. The state space of the Markov chain model increases with the complexity of the protocol to be studied as well as with the increasing number of nodes in the system. A standard practice should be to cross validate the analytical results with simulation.

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

44

4.3 IEEE 802.15.6-based MAC protocol performance under saturation conditions In this section, we develop a two-dimensional Markov chain model in order to model the backoff procedure of the CSMA/CA mechanism of IEEE 802.15.6 under saturation condition. The CSMA/CA protocol descriptions are given at Section 3.4.4.1.2. We consider the saturation throughput as it indicates the maximum load that the system can carry in stable conditions.

4.3.1 Related Works Since the IEEE 802.15.6 standard has recently been released, there have been very few probabilistic works in the literature that analyze the CSMA/CA mechanism of the IEEE 802.15.6 standard. However, many researchers have analyzed the CSMA/CA protocol of various other communication standards in their articles. Performance analyses of the CSMA/CA mechanism for various IEEE wireless communication standards were presented in [3–6] for IEEE 802.11; in [7], [47], [8], and [10] for IEEE 802.11e; in [48], [11], [49], and [12] for IEEE 802.15.4; and in [13] for IEEE 802.15.3c. Because the IEEE 802.15.6 CSMA/CA mechanism is different from the CSMA/CA mechanisms of other wireless technologies, these analytical models are not appropriate for the IEEE 802.15.6 standard. In [50], the authors present numerical formulas to determine the theoretical throughput and delay limits of IEEE 802.15.6-based networks. They aim to optimize the packet size and to determine the upper bounds of IEEE 802.15.6 networks for different WBAN applications. They assume a collision-free network with no UPs. In [51] and [52], the authors study the performance of

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

45

IEEE 802.15.6 CSMA/CA only under saturation conditions. The results indicate that the medium is accessed widely by the high-user-priority nodes, while the other nodes starve. The authors in [53] propose an analytical model to evaluate the performance of a contention-based IEEE 802.15.6 CSMA/CA mechanism under saturated conditions for heterogeneous WBAN scenarios. However, in most real-world IEEE 802.15.6 networks, the saturation assumption is not likely to hold, and the traffic is mostly non-saturated. In [19] the authors develop an analytical model for performance evaluation of the IEEE 802.15.6 standard under non-saturation regime. They only calculated the mean response time of the data frames in the network.

4.3.2 General Assumptions First, we assume a star-topology single-hop WBAN and a hub. The traffic flows between the nodes and the WBAN’s hub. The total number P of nodes in the network can be obtained as N = 7i=0 ni , where ni is the number of nodes in a class. We consider two nodes in each class for heterogeneous scenarios. Given that our objective is to investigate the performance of the CSMA/CA mechanism, we ignore activities in the contention-free access phases (i.e., MAP1 and MAP2). In the proposed analytical model, we consider that the lengths of EAP2, RAP2, and CAP are set to 0. All nodes within the network are synchronized to support the contention-based mechanism of IEEE 802.15.6. The standard defines a MAC layer that supports several PHY layers. Here, we consider the CSMA/CA MAC mechanism running in the NB PHY, as described by the standard. The NB PHY operates in seven different frequency bands and offers a variable number of channels, bit rates, and modulation schemes. One of these seven frequency bands is used for an implantable

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

46

Start

UPi node has a packet to transmit

BC=rand (1,CW ) i,j

Collision Counter

j=0

Backoff Counter (BC)=rand (1,CW ) i ,min

BC=rand (1,CW ) i,max

Y Freeze the BC CWi,j > CWi,max

Y

N

Channel is Busy?

N

New CW Size; j 2 CW =2 CW i,j i,min





BC = BC-1

BC=0?

N

Y Transmit Packet and Wait for ACK

    

N j=j+1 N ACK rece ived within tim eout period?

Y

Figure 4.1: IEEE 802.15.6 CSMA/CA flowchart for DTMC in Figure 4.2 WBAN, whereas the other six are used for a wearable WBAN. The focus of this analysis is on the seventh band (or the sixth band of a wearable WBAN) of the NB PHY layer of 2400-2483.5 MHz, because it is a commonly used, free Industrial, Scientific, and Medical (ISM) band. The packet payload has been assumed to be constant and is equal to

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

47

1020 bits, which is the average value of the largest allowed payload size for the NB PHY. We assume that transmissions errors are only due to collisions, and each node has a fixed probability of collision when it attempts to transmit, irrespective of its history. We consider that the nodes access the medium without any RTS/CTS mechanism. We do not consider any retry limit in our model.

4.3.3 Analytical Model In order to analyze the CSMA/CA performance of the IEEE 802.15.6 MAC protocol, we introduce a DTMC model for the activity of a sensor node under saturation modes. A node is saturated when its frame arrival rate exceeds or is equal to its frame departure rate. Here, we assume a fixed number of nodes in saturated traffic conditions (i.e., each node always having packet available for transmission). We assume a star-topology single-hop typical WBAN with N homogeneous nodes (in each other’s sensing range) and a hub. We have assumed the lengths of EAP1 and RAP1 access modes to be one RAP and can be access by all the UPs. Other assumptions remain same as described in Section 4.3.2. The Markov chain for a sensor node deployed with a CSMA/CA mechanism under saturation condition is shown in Figure 4.2. Each node needs to wait for a random backoff time before transmission. Let b(t) be the stochastic process representing the backoff time counter for a given sensor node. The backoff time counter of each contending node decrements after each successful pCCAtime, and the counter is stopped when the medium is sensed busy. Given that the value of the backoff counter of each contending node also depends on its transmission attempts, each transmission attempt leads the node to a new backoff window called the backoff stage.

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

1,0

1

1,1

1

1,2

1

1/W1 1 1,W1-1 1

48

1,W1

1

(1-γ)

1,-1

γ/W2 2,0

1

2,1

1

2,2

1

1

2,W2-1

1

2,W2

1 (1-γ)

2, -1

γ/W3

1 (1-γ)

i-2,-1

γ/Wi-1 i-1,0

1

i-1,1

1 i-2,1 1

1 i-1,W

i-1-1

1

i-1,Wi-1

1

(1-γ)

i-1,-1

γ/Wi i,0

1

i,1

1

i,2

1

1 i,Wi-1

1

i,Wi

1 (1-γ)

i,-1

γ/Wi+1

1 (1-γ) m-1,-1

γ/Wm m,0 (1-γ)

1

1

m,1

1

m,2

1

1 m,Wm-1

1 m,Wm

γ/Wm

m,-1

Figure 4.2: DTMC Model for the CSMA/CA behavior in saturated traffic conditions

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

49

Let s(t) be the stochastic process representing the backoff stage of the node at time t . For mathematical convenience, the abbreviated notations (i, k) are used to represent the random processes s(t) and b(t), respectively. The backoff stage i starts at 1 and can reach a maximum value of m . Once the backoff stage reaches the maximum value m, it is not increased for a packet re-transmissions. A contending node, after reaching a maximum backoff stage m will continue to try in that backoff stage until the packet is successfully transmitted. Counter k is initially chosen uniformly between [1, W ] , where W is initially set to CWmin , and then its value increases in a non-binary exponential manner, as explained in Section 3.4.4.1.2. The contention window size for a node during a particular backoff stage is calculated as Wi = 2⌊i/2⌋ CWmin . The two special states in our Markov chain are (i, 0) , the state of transmission (at backoff stage i), which can either be successful or colliding, and (i, −1) , the timeout slot (at backoff stage i). The timeout slot is a single pCSMAslot needed for the nodes to know the status (success/collision) of their transmitted packet. Let Ptr be the probability that there is at least one transmission in the time slot under consideration and β be the probability that a node transmits in a generic slot, Ptr is given by Ptr = 1 − (1 − β)N

(4.1)

The collision probability γ can be expressed as follows: γ = 1 − (1 − β)N −1

(4.2)

Let Ts and Tc be the average durations for which the medium is sensed to be busy owing to a successful and a collision transmission, respectively.

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

50

Ts and Tc can be computed as

Ts = Tc = T(M AC+P HY )overhead + TP ayload

(4.3)

Let Estate be the expected time spent per state of the Markov chain by a tagged node. Estate is computed as follow Estate = (1 − Ptr ).δ + Ptr .(1 − γ).Ts + Ptr .γ.Tc + Ptr .(1 − γ)Tack (4.4) δ is the length of a pCSMAslot mentioned in the standard. Ps be the probability that a transmission occurring on the medium is successful and is given by Ps = N β(1 − β)N −1

(4.5)

It is possible to model the two-dimensional stochastic processes s(t) and b(t) depicted in Figure 4.2 with a discrete time Markov chain having the following one-step transition probabilities among them:   P r((i, k − 1)|(i, k)) = 1,       P r((i, −1)|(i, 0)) = 1,        P r(i + 1, k)|(i, −1) = Wγi+1 ,  

1 ≤ k ≤ Wi 1≤i≤m 1 ≤ i ≤ m − 1,

1 ≤ k ≤ Wi+1      P r(1, k)|(i, 0) = (1 − γ). W11 , 1 ≤ i ≤ m,       1 ≤ k ≤ W1      1 ≤ k ≤ Wm P r((m, k)|(m, −1)) = Wγm ,

(4.6)

The first equation in (4.6) reflects the fact that, after each successful pCCAtime, the backoff counter is decremented as shown in Figure 4.1. The second equation reflects the fact that the nodes involved in a

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

51

transmission (at a state (i, 0)) wait for an ACKtimeout period to know the status (success/collision) of their transmitted packet. Upon an unsuccessful transmission, the node chooses another random backoff value uniformly distributed in the range 1...Wi+1 , and this is shown in the third transition probability of equation (4.6). The fourth case deals with the situation that after a successful transmission, another packet is generated, and the node takes a new backoff for the new packet. Finally, the last case models the fact that once the backoff stage reaches value m, it is not increased in a subsequent packet retransmission. With b(i, k) denoting the stationary distribution of the Markov chain in states (i, k), we now show how to obtain a closed-form solution for the Markov chain depicted in Figure 4.2. The main quantities of interest are the two special states in our Markov chain, (i, 0) and (i, −1). The first quantity of interest is the probability that a node transmits in a generic slot, regardless of the backoff stage. This probability is denoted by β and is expressed as β=

m X

b(i, 0)

(4.7)

i=1

The stationary probability of being in the ACKtimeout state (i,-1) can be expressed as

b(i, −1) = 1b(i, 0)

1≤i≤m

(4.8)

Therefore Equation (4.8) can be written as

β=

m X

b(i, −1)

(4.9)

i=1

The stationary distributions

PW1 −1 k=1

b(1, k) + b(1, W1 ) represents the

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

52

topmost row of the Markov chain and is simplified as W 1 −1 X

b(1, k) + b(1, W1 ) = (1 − γ)

m X

γ i (1 − γ)β

i=1

k=1

PWm −1

Similarly, The stationary distribution

k=1

W1 + 1 2

(4.10)

b(m, k) + b(m, Wm ),

represents the lowermost row of the Markov chain and can be expressed as W 1 −1 X

b(m, k) + b(m, Wm ) = γ{b(m − 1, −1) + b(m, −1)}

k=1

The stationary distribution be expressed as m−1 i −1 X WX

b(i, k) +

i=2 k=1

m−1 X

Pm−1 PWi −1 i=2

b(i, Wi ) = γ

k=1

b(i, k) +

m−1 X

Pm−1

{b(i − 1, −1)

i=2

i=2

Wm + 1 (4.11) 2

i=2

b(i, Wi ) can

Wi + 1 } 2

(4.12)

Similarly, sum of the remaining stationary distributions of the Markov chain is given by m X i=1

b(i, 0) +

m X

b(i, −1) = 2β

(4.13)

i=1

To find the normalized equation, Wi m X X

b(i, k) = 1

(4.14)

i=1 k=−1

Let us sum the stationary distributions of (4.10), (4.11), (4.12), and (4.13) that give

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

W 1 −1 X

b(1, k) + b(1, W1 ) +

k=1

WX m −1

b(m, k) + b(m, Wm ) +

k=1 m−1 X

b(i, Wi ) +

m X

m−1 i −1 X WX

b(i, k)+

i=2 k=1 m X

b(i, −1) = 1

b(i, 0) +

i=1

i=1

i=2

53

(4.15)

=⇒ (1 − γ)

m X i=1

m−1

(1 − γ) X i W1 + 1 γ (Wi+1 + 1)+ +γ β{ γ (1 − γ)β 2 2 i

i=1

γ m (Wm + 1)} + 2β = 1 (4.16)

=⇒ β{(1 − γ)2

m X

γi

i=0

Wi+1 + 1 Wm + 1 + (1 − γ)γ m+1 + 2} = 1 2 2 (4.17)

=⇒ β =

1 2 + (1 −

γ)2

i Wi+1 +1 i=0 γ . 2

Pm

+ (1 − γ).γ m+1 . Wm2+1

(4.18)

Equations (4.2) and (4.18) represent a nonlinear system with two unknowns β and γ, which can be solved by using a contraction-mapping method in MATLAB. The values of β and γ can then be used to estimate the desired performance metrics such as normalized throughput and mean frame service time.

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

54

Table 4.1: Contention window bounds for CSMA/CA User Priority

CWmin

CWmax

7

1

4

6

2

8

5

4

8

4

4

16

3

8

16

2

8

32

1

16

32

0

16

64

4.3.4 Performance Metrics In this Section, we calculate different performance metrics such as normalized throughput and head of line delay of the saturated homogeneous network to study behaviour of the system. Let η be the network throughput, defined as the fraction of time for which the medium is used to successfully transmit payload bits. It can be computed as η=

Ptr (1 − γ) × Tpayload Estate

(4.19)

Tpayload is the mean payload duration. We are also interested in the calculation of the mean frame service time E[T ], which is defined as the time duration between the instant that the packet arrives at the head of the queue and the time when the packet is successfully acknowledged by the receiver. The mean frame

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

55

Table 4.2: Narrowband "channel seventh" parameters Slottime

145µs

pSIF S

75µs

pCCA

105µs

pCSM AM ACP HY T ime

40µs

M ACHeader

56 bits

M ACF ooter

16 bits

P HY Header

31 bits

P ayload

1020 bits

P LCP Header

91.9(kb/s)

P SDU

971.4(kb/s)

Ptx

29.9mW

Prx

24.5mW

Pbo

24.5mW

Psleep

37µW

service time can be expressed as m−1 X Wi+1 γ m Wm γ γi + Ts + Estate + Estate E[T ] = δ. 1−γ 2 2(1 − γ)

(4.20)

i=0

where Wi represents the number of backoff slots in a particular backoff stage.

4.3.5 Results and Discussion To validate the accuracy of the developed analytical model, we have compared its results with an event-driven custom-made simulation program written in the C++ programming language.

We run the

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

simulator 30 times and then take the average.

56

For each run the

simulation time is 50s. The simulation closely follows the CSMA/CA mechanism of the IEEE 802.15.6 standard. The values of the parameters used to obtain our results, for both the analytical model and the simulation, are summarized in Table 4.2. These parameters are specified for a narrowband PHY in the IEEE 802.15.6 standard. The packet payload has been assumed to be constant and is equal to 1020 bits, which is the average value of the largest allowed payload size for the NB PHY. All the results show that the analytical result curves coincide with the simulation results. In all the plots in this section, we used standard markers to represent the data collected from the simulations and different type of lines to refer to the analytical results.

0.9

(UP0_Sim) (UP0_Ana) (UP2_Sim) (UP2_Ana) (UP4_Sim) (UP4_Ana) (UP6_Sim) (UP6_Ana) (UP7_Sim) (UP7_Ana)

0.8

Normalized Throughput

0.7

0.6

0.5

0.4

0.3

0.2

0.1 1

2

3

4

5

6

7

8

Number of Nodes

Figure 4.3: network

Normalized saturation throughput for homogeneous

The normalized system throughput and the mean frame service

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

57

5

4.5

x 10

Head-of-line Delay (Microsecond)

4 3.5 3 2.5

(UP0_Sim) (UP0_Ana) (UP2_Sim) (UP2_Ana) (UP4_Sim) (UP4_Ana) (UP6_Sim) (UP6_Ana) (UP7_Sim) (UP7_Ana)

2 1.5 1 0.5 0 1

2

3

4

5

6

7

8

Number of Nodes

Figure 4.4: Head-of-line delay for homogeneous network time performance in a non-saturated homogeneous scenario, which are given by (4.19) and (4.20), are illustrated in Figure 4.3 and Figure 4.4 respectively. For homogeneous scenario we consider a group of nodes where all the nodes are of same user priority, i.e. all of the nodes have same CWmin and CWmax values. In Figure 4.3 and Figure 4.4 we have considered different user priority classes for the homogeneous scenario. The simulation validates the analytical predictions of our model. We plot simulation data using the standard markers and numerical results with solid lines.

The

throughput depends on the number of nodes in the network and the values of CWmin and CWmax . The greater the network size, the lower is the throughput. The throughput also depends on two other parameters CWmin and CWmax . For a specific user priority, we see that the throughput decreases(or

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

58

increases up to a peak point) and the mean frame service time increases with an increase in the network size. We see that, the throughput drastically decreases for high priority users and slowly decreases for low priority users. Similarly, the mean frame service time increases more quickly for the high priority users than for low priority users as network size increases. This drastic decrease in the throughput and quick increase in the mean frame service time for high priority users are due to the small gap between CWmin and CWmax values, which causes more collisions as network size increases. From these results, we can optimize the number of nodes to achieve better throughput with a reasonable delay.

4.4 IEEE 802.15.6-based MAC protocol performance under non-saturation conditions In this section, we extend the analytical model proposed in 4.2 for the non-saturated case. we develop a two-dimensional Markov chain in order to model the backoff procedure of the CSMA/CA mechanism of IEEE 802.15.6 under non-saturation condition.

The CSMA/CA

protocol descriptions given at Section 3.4.4.1.2 are same. We consider the non-saturation traffic conditions, as in most real WBAN networks the saturation assumption is not likely to hold valid. We assume that packets arrive at the MAC in a Poisson manner with rate λ per node.

4.4.1 Analytical Model In order to analyze the CSMA/CA performance of the IEEE 802.15.6 MAC protocol, we introduce a DTMC model under non-saturation modes, as shown in Figure 4.5. We consider Poisson packet arrival at

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

59

the rate of λ packets/microsecond. We assume that a sensor node can have only one packet at a time so that if it has a packet to transmit, then no other packet is generated. Eight user priorities in the WBAN, U Pi where i ∈ {0, 1, 2, 3, 4, 5, 6, 7} are differentiated by CWmin and CWmax , as depicted in Table 4.1. U P7 has been given an aggressive priority as compared to the other U Ps . The contention window size for a U Pi node during the j th backoff stage is calculated as Wi,j = 2⌊j/2⌋ CWi,min . We assume a star-topology single-hop WBAN with N heterogeneous nodes. The total number of nodes in the network can be obtained as P N = 7i=0 ni , where ni is the number of nodes in a class. We consider two nodes in each class. We have assumed the lengths of EAP1 and

RAP1 access modes to be one RAP and can be access by all the UPs. Other assumptions remain same as described in Section 4.3.2. Let Ptr be the probability that there is at least one transmission in the time slot under consideration and βi be the probability that a node of class i transmits in a generic slot, Ptr is given by

Ptr = 1 −

7 Y

(1 − βi )ni

(4.21)

i=0

The collision probability for a class i node can be obtained as follows: γi = 1 − (1 − βi )(ni −1)

7 Y

(1 − βj )nj

(4.22)

j=0,j6=i

Let Ts and Tc be the average durations for which the medium is sensed to be busy owing to a successful and a collision transmission, respectively. Ts and Tc can be computed as

Ts = Tc = T(M AC+P HY )overhead + TP ayload

(4.23)

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

60

1-q

ι

q

1,0 (1-γ).q

1

1,1

1

2,1

1

1,2

1

1

2,2

1

1

1/W1 1,W1-1 1

1,W1

1 1,-1

γ/W2 2,0

1

2,W2-1

1

2,W2

1 (1-γ).q

2, -1

γ/W3

1 (1-γ).q

(1-γ).(1-q)

i-2,-1

γ/Wi-1 i-1,0 (1-γ).q

1

i-1,1

1 i-2,1 1

1 i-1,W

i-1-1

1

i-1,Wi-1

1 i-1,-1

γ/Wi i,0

1

i,1

1

1

i,2

1

1 i,Wi-1

m,2

1

γ/Wm 1 m,Wm-1 1 m,Wm

1

i,Wi

1 (1-γ).q i,-1

γ/Wi+1

1 (1-γ).q m-1,-1

m,0 (1-γ).q

1

1

m,1

γ/Wm

m,-1

Figure 4.5: DTMC Model for the CSMA/CA behavior in non-saturated traffic conditions

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

61

Let Estate,i be the expected time spent per state of the Markov chain by a tagged node of class i. We compute Estate,i as follow

Estate,i = (1−Ptr ).δ+

7 X

Ps,i .Ts +Tc (1−

7 X

Ps,i )+Ptr .(1−γ)Tack (4.24)

i=0

i=0

δ is the length of a pCSM Aslot mentioned in the standard. Let q be the probability that a packet is available to the MAC of a node in a given slot and λ be the packet arrival rate. q is determined by the equation below q = 1 − e−λEstate,i

(4.25)

Ps,i be the probability that a transmission occurring on the medium by a class i node is successful and can be computed as ni −1

Ps,i = ni βi (1 − βi )

7 Y

(1 − βj )nj

(4.26)

j=0,j6=i

A sensor node deployed with a CSMA/CA mechanism needs to wait for a random backoff time before transmission. Let b(t) be the stochastic process representing the backoff time counter for a given sensor node.

The backoff time counter of each contending node

decrements after each successful pCCAtime, and the counter is stopped when the medium is sensed busy. Given that the value of the backoff counter of each contending node also depends on its transmission attempts, each transmission attempt leads the node to a new backoff window called the backoff stage. Let s(t) be the stochastic process representing the backoff stage of the node at time t. It is possible to model the two-dimensional stochastic processes s(t) and b(t) depicted

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

62

in Figure 4.5 with a discrete time Markov chain having the following one-step transition probabilities among them:

                                                    

P r((i, k − 1)|(i, k)) = 1,

1 ≤ k ≤ Wi

P r((i, −1)|(i, 0)) = 1,

1≤i≤m

P r(i + 1, k)|(i, −1) =

γ Wi+1 ,

1 ≤ i ≤ m − 1, 1 ≤ k ≤ Wi+1

P r(1, k)|(i, −1) = q.(1 − γ). W11 ,

1 ≤ i ≤ m,

(4.27)

1 ≤ k ≤ W1 P r(l|(i, −1)) = (1 − γ)(1 − q),

1≤i≤m

P r(l|l) = 1 − q P r((1, k)|l) = q. W11 , P r((m, k)|(m, −1)) =

1 ≤ k ≤ W1 γ Wm ,

1 ≤ k ≤ Wm

The first equation in (4.27) reflects the fact that, after each successful pCCAtime, the backoff counter is decremented. The second equation reflects the fact that after a transmission, the nodes involved in the current transmission (at a state (i, 0)) wait for an ACKtimeout period to know the status (success/collision) of their transmitted packet. Upon an unsuccessful transmission, the node chooses another random backoff value uniformly distributed in the range 1...Wi+1 , and this is shown in the third transition probability of equation (4.27). The fourth case deals with the situation that after a successful transmission, another packet is generated, and the node takes a new backoff for the new packet. The fifth case models the fact that after a successful transmission, the node has no packet to transmit and so enters the idle state. The node remains in the idle state until a new packet arrives, when the node takes a new random backoff value in the range 1..W1 (first backoff stage); these are

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

63

depicted in the sixth and seventh expressions, respectively. Finally, the last case models the fact that once the backoff stage reaches value m, it is not increased in a subsequent packet re-transmission. For mathematical convenience, the abbreviated notations (i, k) are used to represent the random processes s(t) and b(t), respectively. The backoff stage i starts at 1 and can reach a maximum value of m. Once the backoff stage reaches the maximum value m, it is not increased for a packet retransmissions. A contending node, after reaching a maximum backoff stage m will continue to try in that backoff stage until the packet is successfully transmitted. Counter k is initially chosen uniformly between [1, W ] , where W is initially set to CWmin , and then its value increases in a non-binary exponential manner, as explained in Section 3.4.4.1.2. The state (i, 0) in our Markov chain is the transmission state at a backoff stage i. With b(i, k) and b(l) we now show how to obtain a closed-form solution for the Markov chain depicted in Figure 4.5. The main quantity of interest is the probability that a node transmits in a generic slot, regardless of the backoff stage. We denote βi : i ∈ {0, 1, 2, 3, 4, 5, 6, 7} as the transmission probability by a U Pi node. This probability can be expressed as

βi =

m X

b(i, 0)

(4.28)

i=1

The stationary probability of being in the ACKtimeout state (i,-1) can be expressed as

b(i, −1) = 1b(i, 0)

1≤i≤m

Therefore Equation (4.29) can be written as

(4.29)

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

βi =

m X

64

(4.30)

b(i, −1)

i=1

The stationary distributions

PW1 −1 k=1

b(1, k) + b(1, W1 ) represents the

topmost row of the Markov chain and is simplified as W 1 −1 X

b(1, k) + b(1, W1 ) = (1 − γi )

m X

γij (1 − γi )βi

j=1

k=1

Similarly, The stationary distribution

PWm −1 k=1

W1 + 1 2

(4.31)

b(m, k) + b(m, Wm ),

represents the lowermost row of the Markov chain and can be expressed as W 1 −1 X

b(m, k) + b(m, Wm ) = γi {b(m − 1, −1) + b(m, −1)}

k=1

The stationary distribution be expressed as m−1 i −1 X WX

b(i, k) +

m−1 X

Pm−1 PWi −1 i=2

b(i, Wi ) = γi

i=2

i=2 k=1

k=1

m−1 X i=2

b(i, k) +

Wm + 1 (4.32) 2

Pm−1

{b(i − 1, −1)

i=2

b(i, Wi ) can

Wi + 1 } 2

(4.33)

Similarly, sum of the remaining stationary distributions of the Markov chain is given by m X i=1

b(i, 0) +

m X i=1

1 b(i, −1) + b(l) = βi {2 + (1 − q)(1 − γi )} q

(4.34)

The stationary distribution b(l) takes into consideration the situation where the queue of the node is empty and is waiting for a packet to arrive. To find the normalized equation,

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

Wi m X X

65

(4.35)

b(i, k) + b(l) = 1

i=1 k=−1

Let us sum the stationary distributions of (4.31), (4.32), (4.33), and (4.34) that give

W 1 −1 X

b(1, k) + b(1, W1 ) +

W 1 −1 X

b(m, k) + b(m, Wm ) +

m−1 X

b(i, Wi ) +

m X

b(i, 0) +

m X

b(i, −1) + b(l) = 1

i=1

i=1

i=2

b(i, k)+

i=2 k=1

k=1

k=1

m−1 i −1 X WX

(4.36)

m−1

(W1 + 1) (1 − γi )βi X j =⇒ (1 − γi )βi . γi (Wi,j+1 + 1)+ + γi . { 2 2 j=1

γim (Wi,m

=⇒ βi =

(4.37)

1 + 1)} + βi {2 + (1 − q)(1 − γi )} = 1 q

1 2+

1 q (1

− q)(1 − γi ) + (1 − γi )2

(1 − γi ).γim+1 .

Wi,m +1 2

Pm

j=0 γi

j . Wi,j+1 +1 2

+ (4.38)

Equations (4.22) and (4.38) represent a nonlinear coupled system with 16 unknown variables of γi and βi , which can be solved by using a contraction-mapping method in MATLAB. The values of γi and βi can then be used to estimate the desired performance metrics such as normalized throughput, mean frame service time and energy consumption by using (4.39), (4.41), and (4.42) respectively.

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

66

0.7

Normalized Throughput

0.6

0.5

N = 3: (Sim) N = 6: (Sim) N = 8: (Sim) N = 10: (Sim) N = 3: (Ana) N = 6: (Ana) N = 8: (Ana) N = 10: (Ana)

0.4

0.3

0.2

0.1

1

2

3

4

5

6

7

8

9

Arrival Rate (Pkts / Microsecond)

10 -5

x 10

Figure 4.6: Normalised system throughput for non-saturated homogeneous network

4.4.2 Performance Metrics In this Section, we calculate different performance metrics such as normalized throughput, mean frame service time and energy consumption of the non-saturated network to study behaviour of the system. The per-class throughput for a U Pi is defined as the fraction of time for which the medium is used to successfully transmit the payload bits. The normalized per-class throughput can be written as ηi =

Ps,i × Tpayload , Estate,i

i = 0..7

Thus, normalized system throughput can be obtained as

(4.39)

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

67

4

4

x 10

Head-of-line Delay (Microsecond)

3.5

3

2.5

N = 3: (Sim) N = 6: (Sim) N = 8: (Sim) N = 10: (Sim) N = 3: (Ana) N = 6: (Ana) N = 8: (Ana) N = 10: (Ana)

2

1.5

1

0.5

0 1

2

3

4

5

6

7

Arrival Rate (Pkts / Microsecond)

8

9

10 -5

x 10

Figure 4.7: Head-of-line delay for non-saturated homogeneous network

η=

7 X

ηi × pi , i = 0, ..., 7

(4.40)

i=0

Where, pi =

ni N

The mean frame service time of a node belonging to U Pi can be obtained as E[Ti ] = δ.

m i −1 X Wi,j+1 γ mi Wi,m γi γij . + Ts + Estate,i + Estate,i i 1 − γi 2 2(1 − γi ) j=0

(4.41) Energy is quite critical in WBANs, and therefore, in addition to the throughput and the mean frame service time, we are also interested in calculating the energy consumption.

We estimate the energy

consumption on a per-node per-packet basis. The expression for the

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

68

mean frame service time E[Ti ] in (4.41) represents the time elapsed from the arrival of the packet until its successful delivery. The service time of the packet might contain a number of unsuccessful transmissions, with the associated backoff intervals. Denoting by Ptx , Prx , Pbo , and Psleep the power consumed by the transceiver of a node during transmission, reception, backoff, and sleep respectively, we derive an estimate of the energy consumption EAV G,i for a class of U Pi node on a per-node per-packet basis as follows:

0.16

Normalized per class throughput

0.14

UP0: (Simulation) UP2: (Simulation) UP6: (Simulation) UP5: (Simulation) UP7: (Simulation) UP0: (Analytical) UP2: (Analytical) UP6: (Analytical) UP5: (Analytical) UP7: (Analytical)

0.12

0.1

0.08

0.06

0.04

0.02 1

2

3

4

5

6

7

Arrival Rate (Pkts / Microsecond)

8

9

10 -5

x 10

Figure 4.8: Per class normalised throughput for non-saturated heterogeneous network

EAV G,i = 1/λ × Psleep + δ. Estate,i

m i −1 X j=0

γij .

γi × Prx + Ts × Ptx + Pbo × 1 − γi

γ mi Wi,m Wi,j+1 + Pbo × Estate,i i 2 2(1 − γi )

(4.42)

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

69

0.65

System Throughput: (Simulation) System Throughput: (Analytical)

0.6

Normalized System Throughput

0.55 0.5 0.45 0.4 0.35 0.3 0.25 0.2

1

2

3

4

5

6

7

8

9

10 -5

Arrival Rate (Pkts / Microsecond)

x 10

Figure 4.9: Normalized system throughput for non-saturated heterogeneous network

4.4.3 Results and Discussion To validate the accuracy of the developed analytical model, we have compared its results with an event-driven custom-made simulation program written in the C++ programming language. simulator 30 times and then take the average.

We run the

For each run the

simulation time is 50s. The simulation closely follows the CSMA/CA mechanism of the IEEE 802.15.6 standard. The values of the parameters used to obtain our results, for both the analytical model and the simulation, are summarized in Table 4.2. These parameters are specified for a narrowband PHY in the IEEE 802.15.6 standard. The packet payload has been assumed to be constant and is equal to 1020 bits, which is the average value of the largest allowed payload size for the NB PHY. In Figure 4.6 and Figure 4.7, we have considered the U P0 class for

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

70

homogeneous scenario. We plot simulation data using the standard markers and numerical results with solid lines. For a given number of nodes, we see that the throughput (up to a peak point) and the mean frame service time increase with an increase in the arrival rate. We see that, after a peak point, the throughput drastically decreases for large values of N and slowly decreases for small values of N just before the saturation point. Similarly, the mean frame service time increases more quickly for large values of N than for small values of N as λ increases. This drastic decrease in the throughput and quick increase in the mean frame service time for large values of N are due to the large number of nodes, which causes more contentions and hence more collisions. We also note that for all values of N, the maximum throughput is shown to be higher than the saturated throughput, and hence, the model has accurately captured an interesting feature of non-saturated operation. From these results, we can optimize the number of nodes to achieve better throughput with a reasonable delay. Figure 4.8 shows the normalized throughput performance of different U Ps against the traffic load in the heterogeneous scenario. We plot simulation data using the standard markers and numerical results with different lines. All these curves show that classes with smaller CWmin and CWmax have a higher priority in accessing the channel and hence higher throughput performance because smaller values of CWmin and CWmax reduce the average backoff time before a transmission attempt. The maximum throughput is shown to be higher than the saturated throughput, but for lower-priority nodes, after a peak point, the throughput decreases more drastically than for the high-priority nodes as λ increases. This is because with a low arrival rate, very few nodes have packets to transmit, but when the arrival rate increases,

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

71

4

5

x 10

4.5

Head-of-line Delay (Microsecond)

4 3.5 3

UP0: (Simulation) UP2: (Simulation) UP6: (Simulation) UP5: (Simulation) UP7: (Simulation) UP0: (Analytical) UP2: (Analytical) UP6: (Analytical) UP5: (Analytical) UP7: (Analytical)

2.5 2 1.5 1 0.5 0 1

2

3

4

5

6

7

Arrival Rate (Pkts / Microsecond)

Figure 4.10: network

8

9

10 -5

x 10

Head of line delay for non-saturated heterogeneous

the number of medium accesses decreases more for the lower-priority nodes. Figure 4.9 shows the overall network throughput performance of the network. The network consists of five different user priority classes, where each class having the same number of nodes (i.e, two nodes per class) but having different combination of CWmin and CWmax values. The mean frame service time performance (in microseconds) in a non-saturated heterogeneous scenario, which is given by Equation (4.41), is illustrated in Figure 4.10 as a function of the arrival rates. For a given U Pi , we see that the mean frame service time increases with an increase in the arrival rate. The mean frame service time increases quickly for low-priority classes than for high-priority classes as λ increases, because smaller values of CWmin and CWmax reduce the

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

72

-4

Energy Consumption (J / Packet)

x 10

2

UP0: (Simulation) UP2: (Simulation) UP6: (Simulation) UP5: (Simulation) UP7: (Simulation) UP0: (Analytical) UP2: (Analytical) UP6: (Analytical) UP5: (Analytical) UP7: (Analytical)

1

0 1

2

3

4

5

6

7

Arrival Rate (Pkts / Microsecond)

8

9

10 -5

x 10

Figure 4.11: Energy consumption for non-saturated heterogeneous network average backoff time. Figure 4.11 shows the average energy consumption of a U Pi node on a per-node per-packet basis against the arrival rate (packets/microsecond).

It is clear that the energy consumption

for a higher user priority is very low as compared to that for a low user priority, as is the case for the mean frame service time in Figure 4.10. This is understandable in light of the fact that the longer frame service time is attributed to the longer periods of backoff and unsuccessful transmissions, and thus, the associated energy consumption also increases until a successful transmission occurs.

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

73

4.5 IEEE 802.15.6-based MAC protocol performance under different access periods In this section, we extend the analytical model proposed in Figure 4.5 to consider a portion of the access phases (i.e., EAP1 and RAP1) of the superframe and analyze its impact on the performance of the IEEE 802.15.6 CSMA/CA. Our results show that the deployment of EAP is not necessary in a typical WBAN; in fact, IEEE 802.15.6 CSMA/CA employing different access phases degrades the overall system throughput performance and results in higher delay for non-emergency nodes and hence more energy per packet consumed.

4.5.1 Performance Measures and Analytical Modeling In order to analyze the CSMA/CA performance of the IEEE 802.15.6 MAC protocol under different access periods, we introduce a DTMC model under non-saturation modes, as shown in Figure 4.13.

We

consider Poisson packet arrival at the rate of λ packets/microsecond. We assume that a sensor node can have only one packet at a time so that if it has a packet to transmit, then no other packet is generated. Eight user priorities in the WBAN, U Pi where i ∈ {0, 1, 2, 3, 4, 5, 6, 7} are differentiated by CWmin and CWmax , as depicted in Table 4.1. U P7 has been given an aggressive priority as compared to the other U Ps . Moreover, U P7 nodes also have a separate AP for transmission. Here we consider a portion of the access phases (i.e., EAP1 and RAP1) of the superframe and analyze its impact on the performance of the IEEE 802.15.6 CSMA/CA. The contention window size for a U Pi node during the j th backoff stage is calculated as Wi,j = 2⌊j/2⌋ CWi,min . We assume a star-topology single-hop WBAN with N heterogeneous nodes. The total

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

74

Start

UPi node has a packet to transmit

BC=rand (1,CW ) i,j

Collision Counter

j=0

Backoff Counter

BC=rand (1,CW

(BC)=rand (1,CW ) i ,min

i,max

)

Y CW > CW i,j i,max

Freeze the BC

N

Current time is outside of suitable access phase?

Y Y

N New CW Size;

Channel is Busy?

N Y

No enough time left in the suitable access phase(s) to transmit the packet?

j  2 CW =2  CW i,j i,min

BC = BC-1

N BC=0?

N

N

Y Transmit Packet and Wait for ACK

Figure 4.12: Figure 4.13

j=j+1

ACK received within timeout period?

Y

IEEE 802.15.6 CSMA/CA flowchart for DTMC in

number of nodes in the network can be obtained as N =

P7

i=0 ni , where

ni is the number of nodes in a class. We consider two nodes in each class. Other assumptions remain same as described in Section 4.3.2. Let Ptr be the probability that there is at least one transmission in the time slot under consideration and βi be the probability that a node of class i transmits in a generic slot, Ptr is given by

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

75

1-q

ι

q

1,0 (1-γ).q

fi

1,1

fi

1,2

fi

1/W1 fi 1,W -1 fi 1

1,W1

1 1-fi

1-fi

1,-1

2,0

fi

2,1

fi

2,2

1-fi

1-fi

fi

γ/W2 fi 2,W -1 fi 2

2,W2

1

(1-γ).q

2, -1

1-fi

1-fi

1-fi

1-fi

γ/W3

(1-γ).(1-q)

1 (1-γ).q i-2,-1

γ/Wi-1 i-1,0 (1-γ).q

fi

i-1,1

fi

i-2,1

fi

fi i-1,W

i-1-1

fi

i-1,Wi-1

1 1-fi

i-1,-1

1-fi

1-fi

1-fi

γ/Wi i,0

fi

1

1-fi

(1-γ).q i,-1

i,1

fi

i,2

fi

fi

1-fi

i,Wi-1

fi

1-fi

i,Wi

1-fi

γ/Wi+1

1 (1-γ).q m-1,-1

γ/Wm m,0 (1-γ).q

m,1

fi

m,2

fi

fi

m,Wm-1

1 m,-1

Figure 4.13:

fi

1-fi

1-fi

1-fi

fi m,W m

γ/Wm

1-fi

DTMC Model for the CSMA/CA behavior under

non-saturated conditions and different access periods

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

Ptr = 1 −

7 Y

(1 − βi )ni

76

(4.43)

i=0

The collision probability for a class i node can be obtained as follows: (ni −1)

γi = 1 − (1 − βi )

7 Y

(1 − βj )nj

(4.44)

j=0,j6=i

Let Ts and Tc be the average durations for which the medium is sensed to be busy owing to a successful and a collision transmission, respectively. Ts and Tc can be computed as

Ts = Tc = T(M AC+P HY )overhead + TP ayload

(4.45)

Let Estate,i be the expected time spent per state of the Markov chain by a tagged node of class i. We compute Estate,i as follow

Estate,i = (1−Ptr ).δ+

7 X i=0

Ps,i .Ts +Tc (1−

7 X

Ps,i )+Ptr .(1−γ)Tack (4.46)

i=0

δ is the length of a pCSM Aslot mentioned in the standard. Let q be the probability that a packet is available to the MAC of a node in a given slot and λ be the packet arrival rate. q is determined by the equation below q = 1 − e−λEstate,i

(4.47)

Ps,i be the probability that a transmission occurring on the medium by a class i node is successful and can be computed as

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

Ps,i = ni βi (1 − βi )ni −1

7 Y

(1 − βj )nj

77

(4.48)

j=0,j6=i

Let Xeap and Xrap be the mean number of slots in EAP1 and RAP1, respectively, and can be computed as follow Xeap =

eap Estate,i

and

Xrap =

rap Estate,i

(4.49)

where eap and rap are the duration of the access phases EAP1 and RAP1, respectively. In a given pCSM A slot the backoff counter of a node should be locked till beginning of the next eligible AP if there is no enough time for a packet transmission during the current AP. This probability is represented as

eˆi =

    

1 rap−Ts

0≤i≤6

1 eap+rap−Ts

i=7

(4.50)

Therefore, for a U Pi node the probability to decrement the backoff counter during RAP1 is given by

fi =

Q7

j=0 (1

− βj )nj × (1 − eˆi ) (1 − βi )

i = 0..6

(4.51)

The probability that a U P7 node decrement the backoff counter during EAP1 or RAP1 is given by

f7 =

Xrap

Q7

j=0 (1

− βj )nj (1 − eˆ7 )

(Xeap + Xrap )(1 − β7 )

+

Xeap (1 − β7 )n7 (1 − eˆ7 ) (Xeap + Xrap )(1 − β7 )

(4.52)

Let ηi be the normalized per-class throughput, defined as the

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

78

fraction of time for which the medium is used to successfully transmit payload bits. It can be computed as

ηi =

   Ps,i ×Tpayload×Xrap

0≤i≤6

eap+rap

  Ps,i ×(Xrap +Xeap )×Tpayload

(4.53)

i=7

eap+rap

Tpayload is the mean payload duration. Thus, normalized system throughput can be obtained as

η=

7 X

(4.54)

ηi × pi , i = 0, ..., 7

i=0

Where, pi =

ni N

Let E[Ti ] be the duration between the instant that the packet arrives at the head of the queue of a class i node and the time when the packet is successfully acknowledged by the receiver. The mean frame service time can be expressed as

E[Ti ] =

                

γi + Ts + Estate,i (δ. 1−γ i m

Pmi −1 j=0

γij .

Wi,j+1 2

γi i Wi,m Estate,i 2(1−γ ) × Xeap i) eap+rap Pmi −1 j Wi,j+1 γi + Ts + Estate,i j=0 γi . 2 δ. 1−γ i m γi i Wi,m Estate,i 2(1−γi )

+ 0≤i≤6 +

i=7 (4.55)

Energy is quite critical in WBANs, and therefore, in addition to the throughput and the mean frame service time, we are also interested in calculating the energy consumption.

We estimate the energy

consumption on a per-node per-packet basis. The expression for the mean frame service time E[Ti ] in (4.55) represents the time elapsed from the arrival of the packet until its successful delivery. The service time

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

79

of the packet might contain a number of unsuccessful transmissions, with the associated backoff intervals. Denoting by Ptx , Prx , Pbo , and Psleep the power consumed during transmission by the transceiver of a node, reception, backoff, and sleep respectively, we derive an estimate of the energy consumption EAV G,i for a class of U Pi node on a per-node per-packet basis as follows:

EAV G,i

 γi × Prx + Ts × Ptx + Pbo × (1/λ ×Psleep +δ. 1−γ  i   P  mi −1 j Wi,j+1   Estate,i j=0 γi . 2 + Pbo ×    mi  γi Wi,m ) × Xeap Estate,i 2(1−γ = i)  eap+rap   γi   1/λ × Psleep +δ. 1−γ × Prx + Ts × Ptx + Pbo ×  i   m P  γi i Wi,m W j m −1 E i γ . i,j+1 +P ×E state,i

j=0

i

2

bo

state,i 2(1−γi )

0≤i≤6 i=7 (4.56)

A sensor node deployed with a CSMA/CA mechanism needs to wait for a random backoff time before transmission. Let b(t) be the stochastic process representing the backoff time counter for a given sensor node. The backoff time counter of each contending node decrements after each successful pCCAtime, and the counter is stopped when the medium is sensed busy. Given that the value of the backoff counter of each contending node also depends on its transmission attempts, each transmission attempt leads the node to a new backoff window called the backoff stage. Let s(t) be the stochastic process representing the backoff stage of the node at time t. It is possible to model the stochastic processes s(t) and b(t) depicted in Figure 4.13 with a discrete time Markov chain having the following one-step transition probabilities among them:

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

                                                          

P r((i, k − 1)|(i, k)) = fi ,

1 ≤ k ≤ Wi

P r((i, −1)|(i, 0)) = 1, P r(i + 1, k)|(i, −1) =

80

1≤i≤m γ Wi+1 ,

1 ≤ i ≤ m − 1, 1 ≤ k ≤ Wi+1

P r(1, k)|(i, −1) = q.(1 − γ). W11 ,

1 ≤ i ≤ m,

1 ≤ k ≤ W1 P r(l|(i, −1)) = (1 − γ)(1 − q),

(4.57)

1≤i≤m

P r(l|l) = 1 − q P r((1, k)|l) = q. W11 , P r((m, k)|(m, −1)) =

1 ≤ k ≤ W1 γ Wm ,

P r((i, k)|(i, k)) = 1 − fi ,

1 ≤ k ≤ Wm 1 ≤ k ≤ Wi

The first equation in (4.57) reflects the fact that, after each successful pCCAtime, the backoff counter is decremented as depicted in Figure 4.12. The second equation reflects the fact that after a transmission, the nodes involved in the current transmission (at a state (i, 0)) wait for an ACKtimeout period to know the status (success/collision) of their transmitted packet. Upon an unsuccessful transmission, the node chooses another random backoff value uniformly distributed in the range 1...Wi+1 , and this is shown in the third transition probability of equation (4.57). The fourth case deals with the situation that after a successful transmission, another packet is generated, and the node takes a new backoff for the new packet. The fifth case models the fact that after each successful transmission, there is no packet available to the MAC of the node and so enters the idle state. The node remains in the idle state until a new packet arrives, when the node takes a new random backoff value in the range 1..W1 (first backoff stage); these are depicted

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

81

in the sixth and seventh expressions. The second last case models the fact that once the backoff stage reaches value m, it is not increased in a subsequent packet re-transmission. Finally, the last case reveals that the backoff counter is locked whenever a node detects any transmission on the channel during pCCAT ime, or if it is not allowed to access the medium during the current access phase, or the current AP length is not long enough for a frame transmission. For mathematical convenience, the abbreviated notations (i, k) are used to represent the random processes s(t) and b(t), respectively. The backoff stage i starts at 1 and can reach a maximum value of m. Once the backoff stage reaches the maximum value m, it is not increased for a packet retransmissions. A contending node, after reaching a maximum backoff stage m will continue to try in that backoff stage until the packet is successfully transmitted.

Counter k is initially

chosen uniformly between [1, W ] , where W is initially set to CWmin , and then its value increases in a non-binary exponential manner, as explained in Section 3.4.4.1.2. The state (i, 0) in our Markov chain is the state of transmission (at backoff stage i), which can either be successful or colliding. With b(i, k) and b(l) we now show how to obtain a closed-form solution for the Markov chain depicted in Figure 4.13. The main quantity of interest is the probability that a node transmits in a generic slot, regardless of the backoff stage. We denote βi : i ∈ {0, 1, 2, 3, 4, 5, 6, 7} as the transmission probability by a U Pi node. This probability can be expressed as

βi =

m X

b(i, 0)

(4.58)

i=1

The stationary probability of being in the ACK_timeout state (i,-1)

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

82

can be expressed as

b(i, −1) = 1b(i, 0)

1≤i≤m

(4.59)

Therefore Equation 4.59 can be written as

βi =

m X

(4.60)

b(i, −1)

i=1

The stationary distributions

PW1 −1 k=1

b(1, k) + b(1, W1 ) represents the

topmost row of the Markov chain and is simplified as W 1 −1 X

m

b(1, k) + b(1, W1 ) =

X j W1 + 1 1 γi (1 − γi )βi (1 − γi ) fi 2

(4.61)

j=1

k=1

Similarly, The stationary distribution

PWm −1 k=1

b(m, k) + b(m, Wm ),

represents the lowermost row of the Markov chain and can be expressed as W 1 −1 X

Wm + 1 1 γi {b(m − 1, −1) + b(m, −1)} fi 2 k=1 (4.62) Pm−1 Pm−1 PWi −1 The stationary distribution i=2 i=2 b(i, Wi ) can k=1 b(i, k) + b(m, k) + b(m, Wm ) =

be expressed as m−1 i −1 X WX

b(i, k) +

m−1 X i=2

i=2 k=1

m−1 1 X Wi + 1 b(i, Wi ) = γi {b(i − 1, −1) } (4.63) fi 2 i=2

Similarly, sum of the remaining stationary distributions of the Markov chain is given by m X i=1

b(i, 0) +

m X i=1

1 b(i, −1) + b(l) = βi {2 + (1 − q)(1 − γi )} q

(4.64)

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

83

The stationary distribution b(l) takes into consideration the situation where the queue of the node is empty and is waiting for a packet to arrive. To find the normalized equation, Wi m X X

(4.65)

b(i, k) + b(l) = 1

i=1 k=−1

Let us sum the stationary distributions of (4.61), (4.62), (4.63), and (4.64) that give

W 1 −1 X

b(1, k) + b(1, W1 ) +

k=1

W 1 −1 X

b(m, k) + b(m, Wm ) +

i=2

b(i, Wi ) +

b(i, k)+

i=2 k=1

k=1

m−1 X

m−1 i −1 X WX

m X

b(i, 0) +

m X

b(i, −1) + b(l) = 1

i=1

i=1

(4.66)

m−1

=⇒

1 1 (W1 + 1) (1 − γi )βi X j γi (Wi,j+1 + 1)+ + .γi . { (1 − γi )βi . fi 2 fi 2 j=1

1 γim (Wi,m + 1)} + βi {2 + (1 − q)(1 − γi )} = 1 q (4.67)

=⇒ βi =

1 P j Wi,j+1 +1 2 + 1q (1 − q)(1 − γi ) + f1i (1 − γi )2 m j=0 γi . 2 W +1 m+1 i,m 1 . 2 fi (1 − γi ).γi

+ (4.68)

Equations (4.44) and (4.68) represent a nonlinear coupled system with 16 unknown variables of γi and βi , which can be solved by

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

84

using a contraction-mapping method in MATLAB. The values of γi and βi can then be used to estimate the desired performance metrics such as normalized throughput, mean frame service time and energy consumption by using (4.53), (4.55), and (4.56) respectively.

0.16

UP0: (Simulation) UP2: (Simulation) UP6: (Simulation) UP5: (Simulation) UP7: (Simulation) UP0: (Analytical) UP2: (Analytical) UP6: (Analytical) UP5: (Analytical) UP7: (Analytical)

Normalized Per Class Throughput

0.14

0.12

0.1

0.08

0.06

0.04

0.02 1

2

3

4

5

6

7

8

9

10 -5

Arrival Rate (Pkts / Microsecond)

x 10

Figure 4.14: Per class normalised throughput; where EAP length is half of RAP

4.5.2 Results and Discussion To validate the accuracy of the developed analytical model, we have compared its results with an event-driven custom-made simulation program written in the C++ programming language. simulator 30 times and then take the average.

We run the

For each run the

simulation time is 50s. The simulation closely follows the CSMA/CA mechanism of the IEEE 802.15.6 standard. The values of the parameters used to obtain our results, for both the analytical model and the

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

85

0.5 System Throughput: (Simulation) System Throughput: (Analytical)

Normalized System Throughput

0.45

0.4

0.35

0.3

0.25

0.2

0.15

0.1 1

2

3

4

5

6

7

Arrival Rate (Pkts / Microsecond)

8

9

10 -5

x 10

Figure 4.15: Normalized system throughput; where EAP length is half of RAP simulation, are summarized in Table 4.2. These parameters are specified for a narrowband PHY in the IEEE 802.15.6 standard. The packet payload has been assumed to be constant and is equal to 1020 bits, which is the average value of the largest allowed payload size for the NB PHY. In Figure 4.14 all the curves show that classes with smaller CWmin and CWmax have a higher priority in accessing the channel and hence higher throughput performance, because smaller values of CWmin and CWmax reduce the average backoff time before a transmission attempt. From Figure 4.14 and Figure 4.8 it is clear that the IEEE 802.15.6 CSMA/CA employing different access phases degrades the normalized throughput performance of the nodes other than U P7 nodes. This is because nodes other than U P7 are unable to transmit in the EAP period and hence their performance degrade. While U P7 has the same number

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

86

of nodes for all the results so its performance is same even with the use of an EAP period. The duration of EAP is taken 0.3 seconds and the duration of RAP is 0.6 seconds in case when EAP is half in length of RAP period. Figure 4.15 show the overall network throughput for the access phases scenario. The network consists of five different user priority classes, where each class having the same number of nodes but having different combination of CWmin and CWmax values. From Figure 4.15 and Figure 4.9 it is clear that IEEE 802.15.6 CSMA/CA employing access phases degrades the overall system throughput performance. 5

2.5

x 10

Head-of-line Delay (Microsecond)

2

1.5

UP0: (Simulation) UP2: (Simulation) UP6: (Simulation) UP5: (Simulation) UP7: (Simulation) UP0: (Analytical) UP2: (Analytical) UP6: (Analytical) UP5: (Analytical) UP7: (Analytical)

1

0.5

0 1

2

3

4

5

6

7

Arrival Rate (Pkts / Microsecond)

8

9

10 -5

x 10

Figure 4.16: Head of line delay; where EAP length is half of RAP The mean frame service time performance in a non-saturated heterogeneous scenario is illustrated in Figure 4.16 as a function of the arrival rates. For a given U Pi , we see that the mean frame service time increases with an increase in the arrival rate. The mean frame service

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

87

time increases quickly for low-priority classes than for high-priority classes as λ increases, because smaller values of CWmin and CWmax reduce the average backoff time. From Figure 4.16, and Figure 4.10, it is clear that the IEEE 802.15.6 CSMA/CA employing different access phases maximizes the mean frame service time of the nodes other than U P7 . This is because nodes other than U P7 are unable to transmit in the EAP period and hence their mean frame service time increases. While U P7 has almost the same performance for all the results even with the use of an EAP period. From these results, we can optimize the length of the access phases and number of nodes to achieve a reasonable delay. -3

1

x 10

0.9

Energy Consumption (J / Packet)

0.8 0.7 0.6

UP0: (Simulation) UP2: (Simulation) UP6: (Simulation) UP5: (Simulation) UP7: (Simulation) UP0: (Analytical) UP2: (Analytical) UP6: (Analytical) UP5: (Analytical) UP7: (Analytical)

0.5 0.4 0.3 0.2 0.1 0 1

2

3

4

5

6

7

Arrival Rate (Pkts / Microsecond)

8

9

10 -5

x 10

Figure 4.17: Energy consumption; where EAP length is half of RAP Figure 4.17 shows the average energy consumption of a U Pi node on a per-node per-packet basis for the access phases scenario against the arrival rate (packets/microsecond).

It is clear that the energy

consumption for a higher user priority is very low as compared to

Chapter 4. Performance Evaluation of IEEE 802.15.6-based WBAN MAC Protocols

88

that for a low user priority, as is the case for the mean frame service time in Figur 4.16. This is understandable in light of the fact that the longer frame service time is attributed to the longer periods of backoff and unsuccessful transmissions, and thus, the associated energy consumption also increases until a successful transmission occurs.The more energy consumption for lower user priority classes in the case of different access phases happens due to the higher mean frame service time of the nodes.

4.6 Summary of the Chapter This chapter presented the IEEE 802.15.6-based WBAN MAC protocols performance under saturation, non-saturation traffic conditions using both accurate analytical and simulation models. In Section 4.5 we also consider a portion of the access phases of the superframe and analyze its impact on the performance of the IEEE 802.15.6 CSMA/CA MAC protocols. Our results show that the IEEE 802.15.6 CSMA/CA mechanism utilizes the medium poorly under high traffic loads. In addition, the use of different access phases degrades the overall system throughput performance, resulting in higher delay for non-emergency nodes and hence more energy per packet consumed.

Chapter 5

Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 5.1 Introduction The IEEE 802.15.6 CSMA/CA mechanism is different from the CSMA/CA mechanism of other wireless standards in important aspects.

In IEEE 802.15.6 CSMA/CA, successive transmissions are

separated by at least one slot gap, as shown in Figure 5.1. To learn the fate of a transmission, the nodes involved in transmission take an entire slot right after the end of immediate past transmission. Therefore, we add an additional state in our model that colliding nodes check to learn the fate of their transmission, while other contending nodes performing backoff check that slot as usual. The DTMC model 89

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 90

presented in this chapter is different from the model presented in Figure 4.5. The ACK-timeout state (i, −1) in Figure 4.5 can not be utilize by the other contending nodes as a backoff check slot. We study important network performance descriptors such as normalized throughput mean frame service time and energy consumption under non-saturation (both homogeneous and heterogeneous scenarios). Our analysis is validated by computer simulation. This chapter is organized as,

Section 5.2 investigates the

IEEE 802.15.6-based WBANs MAC protocol performance under non-saturation conditions and by considering an additional state in our model that colliding nodes check to learn the fate of their transmission, while other contending nodes performing backoff check that slot as usual. Section 5.3 provides the performance evaluation of the IEEE 802.15.6-based WBANs MAC protocols under non-saturation conditions and error-prone channel.

Section 5.4 model the DTMC

presented in Figure 5.2 for Fibonacci backoff procedure.

Finally,

Section 5.5 concludes the most important research findings developed in this chapter.

5.2 IEEE 802.15.6-based MAC protocol performance under non-saturation conditions In this section we investigates the IEEE 802.15.6-based WBANs MAC protocol performance under non-saturation conditions. We considering an additional state in our model that colliding nodes check to learn the fate of their transmission, while other contending nodes performing backoff check that slot as usual. Our results show an improvement in the normalized per class throughput performance over the throughput

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 91

performance of the DTMC model presented in Chapter 4. Idle Collided transmission Successful transmission Channel Status

Channel slot boundary where nodes participating in the immediate past collision can not start TX despite taking smallest possible backoff value (=1) after collision, but other nodes can possibly start TX. Time out slot for colliding node A

Tc

A

2

1

4

0

3

Backoff value taken by A after collision

Time out slot for colliding node B Ts

Tc

B

2

1

1

0

0

New Backoff value taken by B following collision C

Ts 4

3

2

0

1

Backoff decrement by C during timeout of A and B

Tc = Cntl Ovhd+Payload Ts = Cntl Ovhd+Payload+SIFS+ACK

Figure 5.1: IEEE 802.15.6 CSMA channel access diagram

5.2.1 Assumptions We

consider

Poisson

packets/microsecond.

packet

arrival

at

the

rate

of

λ

We assume that a sensor node can have

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 92

only one packet at a time so that if it has a packet to transmit, then no other packet is generated. We assume a star-topology single-hop (because of the small WBAN environment) typical WBAN with N homogeneous nodes in each other’s sensing range. In the proposed analytical model, we consider that the lengths of CAP, EAP2, and RAP2, are set to 0. We have assumed the lengths of EAP1 and RAP1 access modes to be one RAP; this is because EAP phases are not necessary for a typical WBAN, as high-priority nodes already enjoy an aggressive priority by obtaining a random backoff value from a very small contention window. We assume that transmissions errors are only due to collisions. We consider that the nodes access the medium without any RTS/CTS mechanism. We do not consider any retry limit in our model.

5.2.2 Analytical Model In order to analyze the CSMA/CA performance of the IEEE 802.15.6 MAC protocol, we introduce a DTMC model for the activity of a sensor node under non-saturation modes, as shown in Figure 5.2. Each node needs to wait for a random backoff time before transmission. Let b(t) be the stochastic process representing the backoff time counter for a given sensor node. The backoff time counter of each contending node decrements after each successful pCCAtime, and the counter is stopped when the medium is sensed busy. Given that the value of the backoff counter of each contending node also depends on its transmission attempts, each transmission attempt leads the node to a new backoff window called the backoff stage. Let s(t) be the stochastic process representing the backoff stage of the node at time t . For mathematical convenience, the abbreviated notations (i, k) are used to represent the

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 93

random processes s(t) and b(t), respectively. The backoff stage i starts at 1 and can reach a maximum value of m . Once the backoff stage reaches the maximum value m, it is not increased for a packet retransmissions. A contending node, after reaching a maximum backoff stage m will continue to try in that backoff stage until the packet is successfully transmitted. Counter k is initially chosen uniformly between [1, W ] , where W is initially set to CWmin , and then its value increases in a non-binary exponential manner, as explained in Section 3.4.4.1.2. The contention window size for a node during a particular backoff stage is calculated as Wi = 2⌊i/2⌋ CWmin . The two special states in our Markov chain are (i, 0) , the state of transmission (at backoff stage i), which can either be successful or colliding, and (i, −1) , the timeout slot (at backoff stage i). The timeout slot is a single pCSMAslot needed for the nodes involved in a colliding transmission (at state (i, 0)) to know the status (success/collision) of their transmitted packet (e.g., no ACK forthcoming). However, all other contending nodes find this slot idle and can perform a backoff countdown in this slot.

5.2.3 Homogeneous Networks Let Ptr be the probability that there is at least one transmission in the time slot under consideration and β be the probability that a node transmits in a generic slot, Ptr is given by Ptr = 1 − (1 − β)N

(5.1)

The collision probability γ can be expressed as follows: γ = 1 − (1 − β)N −1

(5.2)

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 94

1-q ι

q 1

1/W1 1 1,W1-1 1 1,W1

1

1/W2 1 2,W 2-1 1

(1-γ).q 1,0

1

1,1

1

1,2

γ 1,-1 (1-γ).q 2,0

1

2,1

1

2,2

2,W2

γ

(1-γ).(1-q)

2, -1

1/W3

γ i-2,-1

(1-γ).q i-1,0

1/Wi-1 1

1 i-2,1 1

1 i-1,W

i,1

1

m,1

1

i-1,1

i-1-1

1

i-1,W i-1

γ i-1,-1

i,2

1

1/Wi 1 i,Wi-1 1

m,2

1

1/Wm 1 m,Wm-1 1 m,Wm

(1-γ).q i,0

1

i,W i

γ i,-1

1/Wi+1

γ m-1,-1

(1-γ).q m,0

γ

1

1/Wm

m,-1

Figure 5.2: DTMC Model for the CSMA/CA behavior in non-saturated traffic conditions

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 95

Let Ts and Tc be the average durations for which the medium is sensed to be busy owing to a successful and a collision transmission, respectively. Ts and Tc can be computed as

Ts = T(M AC+P HY )overhead + TP ayload + TpSIF S + TACK

(5.3)

Tc = T(M AC+P HY )overhead + TP ayload Let Estate be the expected time spent per state of the Markov chain by a tagged node. Estate is computed as follow

Estate = (1 − Ptr ).δ + Ptr .(1 − γ).Ts + Ptr .γ.Tc

(5.4)

δ is the length of a pCSMAslot mentioned in the standard. Let q be the probability that a packet is available to the MAC of a node in a given slot and λ be the packet arrival rate. q is determined by the equation below q = 1 − e−λEstate

(5.5)

Ps be the probability that a transmission occurring on the medium is successful and is given by Ps = N β(1 − β)N −1

(5.6)

It is possible to model the two-dimensional stochastic processes s(t) and b(t) depicted in Figure 5.2 with a discrete time Markov chain having the following one-step transition probabilities among them:

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 96

   P r((i, k − 1)|(i, k)) = 1, 1 ≤ k ≤ Wi       P r((i, −1)|(i, 0)) = γ, 1≤i≤m       P r(i + 1, k)|(i, −1) = W1i+1 , 1 ≤ i ≤ m − 1,       1 ≤ k ≤ Wi+1       P r(1, k)|(i, 0) = q.(1 − γ). 1 , 1 ≤ i ≤ m, W1   1 ≤ k ≤ W1       P r(l|(i, 0)) = (1 − γ)(1 − q), 1≤i≤m       P r(l|l) = 1 − q       1 ≤ k ≤ W1 P r((1, k)|l) = q. W11 ,       P r((m, k)|(m, −1)) = 1 , 1 ≤ k ≤ Wm Wm

(5.7)

The first equation in (5.7) reflects the fact that, after each successful pCCAtime, the backoff counter is decremented. The second equation reflects the fact that after a failed transmission, the nodes involved in a colliding transmission (at a state (i, 0)) wait for an ACKtimeout period to know the status (success/collision) of their transmitted packet. Upon an unsuccessful transmission, the node chooses another random backoff value uniformly distributed in the range 1...Wi+1 , and this is shown in the third transition probability of equation (5.7). The fourth case deals with the situation that after a successful transmission, another packet is generated, and the node takes a new backoff for the new packet. The fifth case models the fact that after a successful transmission, the node has no packet to transmit and so enters the idle state. The node remains in the idle state until a new packet arrives, when the node takes a new random backoff value in the range 1..W1 (first backoff stage); these are depicted in the sixth and seventh expressions. Finally, the last case models the fact that once the backoff stage reaches value m, it is not increased in a subsequent packet retransmission.

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 97

With b(i, k) and b(l) denoting the stationary distribution of the Markov chain in states (i, k) and l, respectively. We now show how to obtain a closed-form solution for the Markov chain depicted in Figure 5.2. The main quantities of interest are the two special states in our Markov chain, (i, 0) and (i, −1). The first quantity of interest is the probability that a node transmits in a generic slot, regardless of the backoff stage. This probability is denoted by β and is expressed as

β=

m X

(5.8)

b(i, 0)

i=1

The second quantity of interest is the timeout slot (i, −1), needed for the nodes involved in a colliding transmission to know the status (success/collision) of their transmitted packet. The stationary probability of being in state (i, −1) can be expressed as b(i, −1) = γ.b(i, 0) The stationary distributions

PW1 −1 k=1

(5.9)

1≤i≤m

b(1, k) + b(1, W1 ) represents the

topmost row of the Markov chain and is simplified as W 1 −1 X

b(1, k) + b(1, W1 ) = (1 − γ).β.

k=1

W1 + 1 2

(5.10)

PWm −1 The stationary distribution b(m, k) + b(m, Wm ), k=1 Pm−1 Pm−1 PWi −1 i=2 b(i, Wi ) can be expressed as i=2 k=1 b(i, k), and WX m −1

b(m, k) + b(m, Wm ) +

m−1 i −1 X WX i=2 k=1

k=1

m−1 X

(1 − γ)β = { 2

i=1

b(i, k) +

m−1 X

b(i, Wi )

i=2

γ i (Wi+1 + 1) + γ m (Wm + 1)}

(5.11)

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 98

Similarly, sum of the remaining stationary distributions of the Markov chain is given by m X

b(i, 0) +

m X i=1

i=1

1 b(i, −1) + b(l) = β{1 + γ + (1 − q)(1 − γ)} q

(5.12)

The stationary distribution b(l) takes into consideration the situation where the queue of the node is empty and is waiting for a packet to arrive. To find the normalized equation, Wi m X X

(5.13)

b(i, k) + b(l) = 1

i=1 k=−1

Let us sum the stationary distributions of (5.10), (5.11), and (5.12) that give W 1 −1 X

b(1, k) + b(1, W1 ) +

k=1

+

m−1 i −1 X WX

b(i, k) +

i=2 k=1

+

m X

WX m −1 k=1 m−1 X

b(m, k) + b(m, Wm ) b(i, Wi ) +

b(i, 0)

(5.14)

i=1

i=2

b(i, −1) + b(l) =

m X

Wi m X X

b(i, k) + b(l) = 1

i=1 k=−1

i=1

P i m =⇒ (1 − γ)β. (W12+1) + (1−γ)β { m−1 i=1 γ (Wi+1 + 1) + γ (Wm + 1)} + 2

β{1 + γ + 1q (1 − q)(1 − γ)} = 1

=⇒ β =

1 1+γ+

1 q (1

− q)(1 − γ) + (1 − γ)

(1 − γ).γ m . W2m +

1−γ m+1 2

Pm−1 i=0

γ i . W2i+1 + (5.15)

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 99

Equations (5.2) and (5.15) represent a nonlinear system with two unknowns β and γ, which can be solved by using a contraction-mapping method in MATLAB. The values of β and γ can then be used to estimate the desired performance metrics such as normalized throughput and mean frame service time. 5.2.3.1

Performance Metrics

In this Section, we calculate different performance metrics such as normalized throughput and head of line delay of the non-saturated homogeneous network to study behaviour of the system. Let η be the network throughput, defined as the fraction of time for which the medium is used to successfully transmit payload bits. It can be computed as η=

Ptr (1 − γ) × Tpayload Estate

(5.16)

Tpayload is the mean payload duration. We are also interested in the calculation of the mean frame service time E[T ], which is defined as the time duration between the instant that the packet arrives at the head of the queue and the time when the packet is successfully acknowledged by the receiver. The mean frame service time can be expressed as

E[T ] = δ.

m−1 X Wi+1 γ m Wm γ γi + Ts + Estate + Estate 1−γ 2 2(1 − γ)

(5.17)

i=0

where Wi represents the number of backoff slots in a particular backoff stage.

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 100

0.8

0.7

Normalized Throughput

0.6

0.5

0.4

N = 3: (Simulation) N = 6: (Simulation) N = 8: (Simulation) N = 10: (Simulation) N = 15: (Simulation) N = 3: (Analytical) N = 6: (Analytical) N = 8: (Analytical) N = 10: (Analytical) N = 15: (Analytical)

0.3

0.2

0.1

0 1

2

3

4

5

6

7

8

9

Arrival Rate (Pkts / Microsecond)

10 −5

x 10

Figure 5.3: Normalized system throughput in the homogenous case (U P0 ) for different network sizes 5.2.3.2

Results and Discussion

To validate the accuracy of the developed analytical model, we have compared its results with an event-driven custom-made simulation program written in the C++ programming language. simulator 30 times and then take the average.

We run the

For each run the

simulation time is 50s. The simulation closely follows the CSMA/CA mechanism of the IEEE 802.15.6 standard. The values of the parameters used to obtain our results, for both the analytical model and the simulation, are summarized in Table 4.2. These parameters are specified for a narrowband PHY in the IEEE 802.15.6 standard. The packet payload has been assumed to be constant and is equal to 1020 bits, which is the average value of the largest allowed payload size for the NB PHY. All the results show that the analytical model is extremely accurate as

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 101

the analytical result curves coincide with the simulation results. In all the plots in this section, we used standard markers to represent the data received from the simulations and different type of lines to refer to the analytical results. For homogeneous scenario we consider a group of nodes having the same user priority, i.e. all the nodes have same CWmin and CWmax values. In Figure 5.3 and Figure 5.4 we have considered the U P0 class for homogeneous scenario. 4

3.5

x 10

Head−of−line Delay (Microsecond)

3

2.5

2

N = 3: (Simulation) N = 6: (Simulation) N = 8: (Simulation) N = 10: (Simulation) N = 15: (Simulation) N = 3: (Analytical) N = 6: (Analytical) N = 8: (Analytical) N = 10: (Analytical) N = 15: (Analytical)

1.5

1

0.5

0 1

2

3

4

5

6

7

Arrival Rate (Pkts / Microsecond)

8

9

10 −5

x 10

Figure 5.4: Mean frame service time in the homogenous case (U P0 ) for different network sizes The normalized system throughput and the mean frame service time performance in a non-saturated homogeneous scenario, which are given by (5.33) and (5.35), are illustrated in Figure 5.3 and Figure 5.4 respectively, as a function of the arrival rates. The simulation validates the analytical predictions of our model. For a given number of nodes,

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 102

we see that the throughput (up to a peak point) and the mean frame service time increase with an increase in the arrival rate. We see that, after a peak point, the throughput drastically decreases for large values of N and slowly decreases for small values of N just before the saturation point. Similarly, the mean frame service time increases more quickly for large values of N than for small values of N as λ increases. This drastic decrease in the throughput and quick increase in the mean frame service time for large values of N are due to the large number of nodes, which causes more contentions and hence more collisions. We also note that for different values of N, the saturated throughput is less than the maximum throughput, and hence, the model has accurately captured an interesting feature of non-saturated operation. From these results, we can optimize the number of nodes to achieve better throughput with a reasonable delay.

5.2.4 Heterogeneous Networks Next, we will extend the performance measures of the analytical modeling for the heterogeneous scenario. Eight user priorities in the WBAN, U Pi where i ∈ {0, 1, 2, 3, 4, 5, 6, 7} are differentiated by CWmin and CWmax , as depicted in Table 4.1. U P7 has been given an aggressive priority as compared to the other U Ps . U P7 has two types of priorities; the first is a very small contention window size, and the second is a separate access phase (i.e., EAP), here we do not consider the EAP for U P7 . The contention window size for a node i belonging to U Pi during the j th backoff stage is calculated as Wi,j = 2⌊j/2⌋ CWi,min . The total P number of nodes in the network can be obtained as N = 7i=0 ni , where ni is the number of nodes in a class. We consider two nodes in each class. We denote βi : i ∈ {0, 1, 2, 3, 4, 5, 6, 7}, as the transmission probability by

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 103

a U Pi node. This probability can be expressed as βi =

1 1 + γi + 1q (1 − q)(1 − γi ) + (1 − m +1 1−γi i W (1 − γi ).γimi . 2i,m + 2

γi )

Pm−1 j=0

Wi,j+1 2

γi j .

+

(5.18)

where γi : i ∈ {0, 1, 2, 3, 4, 5, 6, 7} is the collision probability for a class i node and can be obtained as ni −1

γi = 1 − (1 − βi )

7 Y

(1 − βj )nj

(5.19)

j=0,j6=i

The expected time spent per state of the Markov chain by a tagged node of class i can be calculated as Estate,i = (1 − Ptr ).δ +

7 X

Ps,i .Ts + Tc (1 −

7 X

Ps,i )

(5.20)

i=0

i=0

where Ptr is the probability that there is at least one transmission in the considered time slot and can be obtained as Ptr = 1 −

7 Y

(1 − βi )ni

(5.21)

i=0

Ps,i is the probability that a transmission occurring on the medium by a class i node is successful and can be computed as Ps,i = ni βi (1 − βi )ni −1

7 Y

(1 − βj )nj

(5.22)

j=0,j6=i

Equations (5.15) and (5.28) represent a nonlinear coupled system with 16 unknown variables of βi and γi , which can be solved by using a contraction-mapping method in MATLAB. The values of βi and γi can then be used to measure the desired performance parameters such

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 104

as the normalized throughput, mean frame service time and energy consumption by using Equations (5.23), (5.25), and (5.26) respectively. 5.2.4.1

Performance Metrics

In this Section, we calculate different performance metrics such as normalized throughput, mean frame service time and energy consumption of the non-saturated heterogeneous network to study behaviour of the system. The per-node throughput for a U Pi is defined as the fraction of time for which the medium is used to successfully transmit the payload bits. The normalized per-class throughput can be written as ηi =

Ps,i × Tpayload , Estate,i

i = 0..7

(5.23)

Normalized system throughput can be obtained as

η=

7 X

ηi × pi , i = 0, ..., 7

(5.24)

i=0

Where, pi =

ni N

In the case of the heterogeneous traffic scenarios, the mean frame service time of a node belonging to U Pi can be obtained as

E[Ti ] = δ.

m i −1 X Wi,j+1 γ mi Wi,m γi γij . + Ts + Estate,i + Estate,i i 1 − γi 2 2(1 − γi ) j=0

(5.25) Energy is quite critical in WBANs, and therefore, in addition to the throughput and the mean frame service time, we are also interested in calculating the energy consumption.

We estimate the energy

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 105

consumption on a per-node per-packet basis. The expression for the mean frame service time E[Ti ] in (5.25) represents the time elapsed from the arrival of the packet until its successful delivery. The service time of the packet might contain a number of unsuccessful transmissions, with the associated backoff intervals. Denoting by Ptx , Prx , Pbo , and Psleep the power consumed by the transceiver of a node during transmission, reception, backoff, and sleep respectively, we derive an estimate of the energy consumption EAV G,i for a class of U Pi node on a per-node per-packet basis as follows:

EAV G,i = 1/λ × Psleep + δ. Estate,i

m i −1 X j=0

5.2.4.2

γi × Prx + Ts × Ptx + Pbo × 1 − γi

γij .

γ mi Wi,m Wi,j+1 + Pbo ∗ Estate,i i 2 2(1 − γi )

(5.26)

Results and Discussion

To validate the accuracy of the developed analytical model, we have compared its results with an event-driven custom-made simulation program written in the C++ programming language. simulator 30 times and then take the average.

We run the

For each run the

simulation time is 50s. The values of the parameters used to obtain our results, for both the analytical model and the simulation, are summarized in Table 4.2. The packet payload has been assumed to be constant and is equal to 1020 bits. For heterogeneous scenario we consider five representative classes of users, where each class having two nodes. All the classes have different CWmin and CWmax values as shown in Table 4.1.

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 106

0.18

Normalized−per−class−throughput

0.16

UP0: (Simulation) UP2: (Simulation) UP6: (Simulation) UP5: (Simulation) UP7: (Simulation) UP0: (Analytical) UP2: (Analytical) UP6: (Analytical) UP5: (Analytical) UP7: (Analytical)

0.14

0.12

0.1

0.08

0.06

0.04 1

2

3

4

5

6

7

Arrival−rate (Pkts/Microsecond)

8

9

10 −5

x 10

Figure 5.5: Per class normalised throughput for non-saturated heterogeneous network Figure 5.5 shows the normalized throughput performance of different U Ps against the traffic load in the heterogeneous scenario. We plot simulation data using the standard markers and numerical results with different lines. All these curves show that classes with smaller CWmin and CWmax have a higher priority in accessing the channel and hence higher throughput performance because smaller values of CWmin and CWmax reduce the average backoff time before a transmission attempt. The saturated throughput is shown to be less than the maximum throughput, but for lower-priority nodes, after a peak point, the throughput decreases more drastically than for the high-priority nodes as λ increases. This is because with a low arrival rate, very few nodes have packets to transmit, but when the arrival rate increases, the number of medium accesses decreases more for the

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 107

0.65 0.6

System Throughput: (Simulation) System Throughput: (Analytical)

Normalized System Throughput

0.55 0.5 0.45 0.4 0.35 0.3 0.25

1

2

3

4

5

6

7

Arrival Rate (Pkts / Microsecond)

8

9

10 −5

x 10

Figure 5.6: Normalized system throughput for non-saturated heterogeneous network lower-priority nodes. DTMC model presented in Figure 5.2 showed an improvement in the normalized per class throughput performance over the throughput performance of the DTMC model presented in Figure 4.5. It is due to the fact that the ACK-timeout state (i, −1) is utilized by the other contending nodes as a backoff check slot. Figure 5.6 shows the overall network throughput performance of the network. The network consists of five different user priority classes, where each class owning the same number of nodes (i.e, two nodes per class) but having different combination of CWmin and CWmax values. These results show that the DTMC model presented in Figure 5.2 having good performance in terms of system throughput as compared with the DTMC model in Figure 4.5. The mean frame service time performance (in microseconds)

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 108

60,000

Head−of−line−delay (Microsecond)

50,000

40,000

UP0: (Simulation) UP2: (Simulation) UP6: (Simulation) UP5: (Simulation) UP7: (Simulation) UP0: (Analytical) UP2: (Analytical) UP6: (Analytical) UP5: (Analytical) UP7: (Analytical)

30,000

20,000

10,000

1

2

3

4

5

6

7

Arrival−rate (Pkts/Microsecond)

8

9

10 −5

x 10

Figure 5.7: Head of line delay for non-saturated heterogeneous network in a non-saturated heterogeneous scenario, which is given by Equation (5.25), is illustrated in Figure 5.7 as a function of the arrival rates. For a given U Pi , we see that the mean frame service time increases with an increase in the arrival rate. The mean frame service time increases quickly for low-priority classes than for high-priority classes as λ increases, because smaller values of CWmin and CWmax reduce the average backoff time. Figure 5.8 shows the average energy consumption of a U Pi node on a per-node per-packet basis against the arrival rate (packets/microsecond).

It is clear that the energy consumption

for a higher user priority is very low as compared to that for a low user priority, as is the case for the mean frame service time in Figure 5.7. This is understandable, because longer frame service time is attributed to the longer periods of backoff and unsuccessful transmissions, and thus,

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 109 −4

Energy Consumption (J/Packet)

x 10

2

UP0: (Simulation) UP2: (Simulation) UP6: (Simulation) UP5: (Simulation) UP7: (Simulation) UP0: (Analytical) UP2: (Analytical) UP6: (Analytical) UP5: (Analytical) UP7: (Analytical)

1

0 1

2

3

4

5

6

7

Arrival Rate (Pkts / Microsecond)

Figure 5.8: network

8

9

10 −5

x 10

Energy consumption for non-saturated heterogeneous

the associated energy consumption also increases until a successful transmission occurs.

5.3 IEEE 802.15.6 MAC protocol performance under Error-prone channel We have considered an ideal channel for WBAN in the previous works. However, WBANs experience unsuccessful transmission caused by unreliable link. Hence, in this section, we extend the analytical model proposed in Figure 5.2 to consider an error-prone channel and analyze its impact on the performance of the IEEE 802.15.6 CSMA/CA. Our results show a degradation in the performance, which is obvious due to channel errors.

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 110

5.3.1 Analytical Model In this section we will model the DTMC presented in Figure 5.2 for error-prone channels. We assume a single-hop network which operates in a 2.4 GHz ISM band having the bit error rate set to 10−3 . Each bit has the same bit error probability. Other assumptions remain the same as presented in Section 5.2.2. The frame may not be received correctly if the channel is noisy, so much so atleast one of the bits are erroneously received. Probability of such an event can be given by γe as γe = 1 − (1 − ǫb )L

(5.27)

where L is the frame length in bits and ǫb is the Bit Error Rate (BER). Thus the probability of failure, γ presented in Equation 5.2 can now be expressed as γ = 1 − (1 − β)N −1 (1 − ǫb )L

(5.28)

Equations (5.28) and (5.15) represent a nonlinear system with two unknowns β and γ, which can be solved by using a contraction-mapping method in MATLAB. The values of β and γ can then be used to measure the desired performance parameters such as normalized throughput and mean frame service time.

5.3.2 Performance Metrics The normalized system throughput for non-saturated homogeneous network in error-prone channel can be obtained as

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 111

η=

Ptr (1 − γ) × Tpayload Estate

(5.29)

Tpayload is the mean payload duration. The mean frame service time can be expressed as

E[T ] = δ.

m−1 X Wi+1 γ γ m Wm γi + Ts + Estate + Estate 1−γ 2 2(1 − γ)

(5.30)

i=0

where Wi represents the number of backoff slots in a particular backoff stage. 0.7

Normalized System Throughput

0.6

(ideal(UP0)_Sim) (ideal(UP0)_Ana) (ber(UP0)_Sim) (ber(UP0)_Ana)

0.5

0.4

0.3

0.2

0.1

0 1

2

3

4

5

6

7

8

Number of Nodes

Figure 5.9: Normalized system throughput in the homogenous case (U P0 ) for error-prone network

5.3.3 Results and Discussion We consider that all the nodes in the network having the same user priority. Here we choose U P0 priority class. The simulation results

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 112 4

6

x 10

Head-of-line Delay (Microsecond)

5

(ideal(UP0)_Sim) (ideal(UP0)_Ana) (ber(UP0)_Sim) (ber(UP0)_Ana)

4

3

2

1

0 1

2

3

4

5

6

7

8

Number of Nodes

Figure 5.10: Mean frame service time in the homogenous case (U P0 ) for error-prone network are taken from our custom built discrete event C++ program. We run the simulator 30 times and then take the average. For each run the simulation time is 50s.

The values of the parameters used to

obtain our results, for both the analytical model and the simulation, are summarized in Table 4.2. The packet payload has been assumed to be constant and is equal to 1020 bits, which is the average value of the largest allowed payload size for the NB PHY. The arrival process of packets follow Poisson process at a rates λ = 0.0001 packets/µs. BER is set to 10−3 for all the nodes. The normalized non-saturation system throughput for user priorities U P0 is shown in Figure 5.9. In case of ideal channel, the maximum throughput is around 0.65, and it is 0.23 while considering error-prone channel. The large gap in the throughput performance

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 113

is due the high BER. Figure 5.10 shows the mean frame service time in homogeneous case for U P0 class. The mean frame service time increases more quickly for the non-ideal channel than that of ideal channel as network size increases.

5.4 IEEE 802.15.6 MAC protocol performance under Different backoff algorithms The Backoff algorithm Plays an important role to avoid collision in wireless networks.

The IEEE 802.15.6 MAC layer employs the

Alternative Binary Exponential Backoff (ABEB) function to compute the backoff delay for each node. To reduce the gap between successive contention window sizes we adopted the Fibonacci backoff procedure. The performance of FIB algorithm is compared against the ABEB function.

5.4.1 Analytical Model In this section we will model the DTMC presented in Figure 5.2 for Fibonacci backoff procedure. We follow the same assumptions presented in Section 5.2.2. The number of backoff slots in a particular backoff stage for Fibonacci backoff scheme is represented by Wf and can be computed as Wf = [P hif − (phi)f ]/Sqrt[5]

(5.31)

where P hi = 1 + Sqrt[5]/2 and phi = 1 − Sqrt[5]/2 Let βf be the normalized normalized equation for the Fibonacci

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 114

backoff procedure and can be computed as

1

βf =

1+γ +

1 q (1

− q)(1 − γ) + (1 − γ)

(1 − γ).γ m . W2m +

1−γ m+1 2

Pm−1 i=0

γf .

Wf +1 2

(5.32)

+

Equations (5.2) and (5.32) provide pairs of nonlinear coupled equations with two unknowns β and γ. These can be solved by using a contraction-mapping method in MATLAB. The values of β and γ can then be used to measure the desired performance parameters such as normalized throughput and mean frame service time. 0.75 (Fabo_Sim) (Fabo_Ana) (IEEE 802.15.6_Sim) (IEEE 802.15.6_Ana)

0.7

Normalized Throughput

0.65 0.6 0.55 0.5 0.45 0.4 0.35

2

3

4

5

6

7

8

9

10

Number of Nodes

Figure 5.11: Normalized system throughput in the homogenous case (U P0 ) for different backoff algorithms

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 115 4

3.5

x 10

(Fabo_Sim) (Fabo_Ana) (IEEE 802.15.6_Sim) (IEEE 802.15.6_Ana)

Head-of-line Delay (Microsecond)

3

2.5

2

1.5

1

0.5

0 2

3

4

5

6

7

8

9

10

Number of Nodes

Figure 5.12: Mean frame service time in the homogenous case (U P0 ) for different backoff algorithms

5.4.2 Performance Metrics In this Section, we calculate different performance metrics such as normalized throughput and head of line delay of the non-saturated homogeneous network to study behaviour of the system. Let η be the network throughput, defined as the fraction of time for which the medium is used to successfully transmit payload bits. It can be computed as η=

Ptr (1 − γ) × Tpayload Estate

(5.33)

Tpayload is the mean payload duration. Let E[Tb ] defined as the time duration between the instant that the packet arrives at the head of the queue and the time when the packet is successfully acknowledged by the receiver, be the mean frame service

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 116

time for the binary exponential procedure of the IEEE 802.15.6. The mean frame service time can be expressed as m−1 X Wi+1 γ m Wm γ γi + Ts + Estate + Estate E[Tb ] = δ. 1−γ 2 2(1 − γ)

(5.34)

i=0

where Wi represents the number of backoff slots in a particular backoff stage for the binary exponential procedure of IEEE 802.15.6. Similarly the mean frame service time for Fibonacci backoff procedure represented by E[Tf ] can be expressed as

E[Tf ] = δ.

m−1 X Wf +1 γ γ m Wm γi + Ts + Estate + Estate 1−γ 2 2(1 − γ)

(5.35)

i=0

where Wf represents the number of backoff slots in a particular backoff stage for the Fibonacci backoff procedure.

5.4.3 Results and Discussion We consider that all the nodes having the same user priority in the network. Here we choose U P0 priority class. The simulation results are yielded from our custom made C++ program. We run the simulator 30 times and then take the average. For each run the simulation time is 50s. The values of the parameters used to obtain our results, for both the analytical model and the simulation, are summarized in Table 4.2. The packet payload has been assumed to be constant and is equal to 1020 bits, which is the average value of the largest allowed payload size for the NB PHY. The arrival process of packets follow Poisson process at a rates λ = 0.0001 packets/µs. This subsection depict a comparison between both binary

Chapter 5. Rethinking the IEEE 802.15.6-based WBANs MAC Performance Modeling Methodology 117

exponential backoff procedure of IEEE 802.15.6 and Fibonacci backoff procedure performance in terms of nomalized throughput and mean frame service time. As shown in Figure 5.11 and Figure 5.12, the Fibonacci backoff procedure performance metrics show an improvement as the network size increases.

5.5 Summary of the Chapter In this chapter a new DTMC model for the CSMA/CA scheme of IEEE 802.15.6 is proposed where the ACK-timeout state (i, −1) is modeled as an additional state that colliding nodes check to learn the fate of their transmission, while other contending nodes performing backoff check that slot as usual. DTMC model presented in Figure 5.2 showed an improvement in the normalized per class throughput performance over the throughput performance of the DTMC model presented in Figure 4.5. It is due to the fact that the ACK-timeout state (i, −1) is utilized by the other contending nodes as a backoff check slot. Our analysis is validated by computer simulation. In Section5.3 we have extended the model presented in Figure 5.2 for an error-prone WBAN channel. The performance results are degraded due to channel error which are obvious. Finally, in Section 5.4 we modeled the DTMC presented in Figure 5.2 for Fibonacci backoff procedure. We follow the same assumptions presented in Section 5.2.2.

The respective

performance metrics showed an improvement as the network size increases.

Chapter 6

Conclusions and Future Works In this chapter, we summarize and conclude findings of our dissertation and also include some related potential research directions in this chapter.

6.1 Summary and Conclusions In this dissertation, we have presented the analytical modeling and performance evaluation of contention-based MAC protocols (supported by NB PHY) for a WBAN. We have developed a DTMC model to evaluate the performance measures as throughput, mean frame service and energy consumption of IEEE 802.15.6 CSMA/CA under saturated and non-saturated traffic conditions. While constructing the DTMC, we take into consideration the time spent by a node awaiting the acknowledgement frame. We extended the proposed analytical model to consider a portion of the access phases (i.e., EAP1 and RAP1) of

118

Chapter 6. Conclusions and Future Works

119

the superframe and analyze its impact on the performance of the IEEE 802.15.6 CSMA/CA. The analysis is validated against extensive simulation.

The most important findings of the research can be

summarized as follows: • We first briefly described Wireless Sensor Networks as the basis of healthcare systems.

Wireless Body Area Networks as the

main building block of a healthcare system and their healthcare applications are presented. The currently available projects which aim to provide a human body monitoring system have been described. • We presented the key features of the IEEE 802.15.6 standard. Starting from the fundamental details, we have provided deep insight into the MAC and PHY layers specification of IEEE 802.15.6. We review different communication modes and access mechanisms and explain the NB, UWB, and HBC specifications in detail. • In chapter 4, we have described different performance modeling approaches for MAC protocols.

In Section 4.2 we have

investigated the IEEE 802.15.6-based WBANs MAC protocol performance under saturation conditions by developing analytical and simulation models. The saturation throughput indicates the maximum load that the system can carry in stable conditions. We have provided the performance evaluation of the IEEE 802.15.6-based WBANs under non-saturation regime. We have also presented the performance of the IEEE 802.15.6-based WBAN MAC protocols under different access phases of the superframe. Our results show that the IEEE 802.15.6 CSMA/CA mechanism utilizes the medium poorly under high traffic loads.

120

Chapter 6. Conclusions and Future Works

In addition, the use of different access phases degrades the overall system throughput performance, resulting in higher delay for non-emergency nodes and hence more energy per packet consumed. We can optimize the length of the access phases to achieve better throughput and reasonable delay. Moreover, a smaller priority gap between the UPs will decrease the performance gap between the classes. • In chapter 5, we rethink the model presented in Figure 4.5 by reforming the the ACKtimeout state (i, −1) as an additional state in our model that colliding nodes check to learn the fate of their transmission, while other contending nodes performing backoff check that slot as usual. The DTMC model presented in this chapter is different from the model presented in Figure 4.5. In the case of collision the ACK-timeout state (i, −1) in Figure 4.5 can not be utilized by the other contending nodes as a backoff check slot. Our analysis is validated by computer simulation. DTMC model presented in Figure 5.2 showed an improvement in the normalized per class throughput performance over the throughput performance of the DTMC model presented in Figure 4.5.

It is due to the fact that the ACK-timeout state

(i, −1) is utilized by the other contending nodes as a backoff check slot. We have extended the model presented in Figure 5.2 for an error-prone WBAN channel.

The performance results

are degraded due to channel error which are obvious. Finally, we modeled the DTMC presented in Figure 5.2 for Fibonacci backoff procedure. We follow the same assumptions presented in Section 5.2.2. The respective performance metrics showed an improvement as the network size increases.

Chapter 6. Conclusions and Future Works

121

6.2 Future Works Some of the following relevant future research directions are of importance and deserve further investigation: • We intend to fine-tune the length of the access phases and number of nodes for different user priorities, which will lead to comparatively better system throughput and minimum delay. • In the future, we plan to extend the DTMC model by considering all the access phases of the superframe. • The wireless networks considered in this dissertation are single-hop, a similar framework for multi-hop wireless networks would be an interesting topic. • A light-weight and reliable security mechanisms can be considered as an important research direction for the WBANs. • The IEEE 802.15.6 MAC layer employs the Alternative Binary Exponential Backoff (ABEB) function to compute the backoff delay for each node. In Section 5.4 we adopted the Fibonacci backoff procedure, which outperforms only in high traffic loads. In the future we want to proposed a hybrid backoff mechanism which will lead to better throughput and minimum delay in all kind of traffic load.

Bibliography [1] S. Ullah, M. Mohaisen, and M. A. Alnuem, “A review of ieee 802.15. 6 mac, phy, and security specifications,” International Journal of Distributed Sensor Networks, vol. 2013, 2013. [2] “Part 15.6: Wireless body area networks, IEEE standard for local and metropolitan area networks, IEEE Std 802.15.6, 2012.” [3] G. Bianchi, “Performance analysis of the ieee 802.11 distributed coordination function,”

IEEE Journal on Selected Areas in

Communications, vol. 18, no. 3, pp. 535–547, 2000. [4] C. G. Park, D. H. Han, and S. J. Ahn, “Performance analysis of mac layer protocols in the ieee 802.11 wireless lan,” Telecommunication Systems, vol. 33, no. 1-3, pp. 233–253, 2006. [5] C. H. Foh, J. W. Tantra et al., “Comments on ieee 802. 11 saturation throughput analysis with freezing of backoff counters,” IEEE Communications Letters, vol. 9, no. 2, pp. 130–132, 2005. [6] D. Malone, K. Duffy, and D. Leith, “Modeling the 802.11 distributed coordination function in nonsaturated heterogeneous conditions,” IEEE/ACM Transactions on Networking, vol. 15, no. 1, pp. 159–172, 2007. 122

Bibliography

123

[7] X. Ling, K.-H. Liu, Y. Cheng, X. Shen, and J. W. Mark, “A novel performance model for distributed prioritized mac protocols,” in IEEE GLOBECOM, 2007, pp. 4692–4696. [8] O. M. Abu-Sharkh and A. H. Tewfik, “Toward accurate modeling of the ieee 802.11 e edca under finite load and error-prone channel,” IEEE Transactions on Wireless Communications, vol. 7, no. 7, pp. 2560–2570, 2008. [9] I. Tinnirello and G. Bianchi, “Rethinking the ieee 802.11 e edca performance modeling methodology,” Networking, IEEE/ACM Transactions on, vol. 18, no. 2, pp. 540–553, 2010. [10] I. Inan, F. Keceli, and E. Ayanoglu, “Analysis of the 802.11 e enhanced distributed channel access function,” IEEE Transactions on Communications, vol. 57, no. 6, pp. 1753–1764, 2009. [11] S. Pollin, M. Ergen, S. Ergen, B. Bougard, L. 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, vol. 7, no. 9, pp. 3359–3371, 2008. [12] K. Ashrafuzzaman and K. S. Kwak, “On the performance analysis of the contention access period of ieee 802.15. 4 mac,” IEEE communications letters, vol. 15, no. 9, pp. 986–988, 2011. [13] C. W. Pyo and H. Harada, “Throughput analysis and improvement of hybrid multiple access in ieee 802.15. 3c mm-wave wpan,” IEEE Journal on Selected Areas in Communications, vol. 27, no. 8, pp. 1414–1424, 2009.

124

Bibliography

[14] D. Estrin, “Embedded networked sensing research: Emerging systems challenges,” in NSF workshop on distributed communications and signal processing, 2002. [15] D. Estrin, R. Govindan, J. Heidemann, and S. Kumar, “Next century challenges: Scalable coordination in sensor networks,” in Proceedings of the 5th annual ACM/IEEE international conference on Mobile computing and networking.

ACM, 1999, pp. 263–270.

[16] P. Khan, A. Hussain, and K. S. Kwak, “Medical applications of wireless body area networks.” JDCTA, vol. 3, no. 3, pp. 185–193, 2009. [17] S. Ullah, B. Shen, S. Riazul Islam, P. Khan, S. Saleem, and K. Sup Kwak, “A study of mac protocols for wbans,” Sensors, vol. 10, no. 1, pp. 128–145, 2009. [18] P. Khan, N. Ullah, S. Ullah, and K. S. Kwak, “Seamless interworking architecture for wban in heterogeneous wireless networks with qos guarantees,” Journal of medical systems, vol. 35, no. 5, pp. 1313–1321, 2011. [19] S. Rashwand, J. Misic, and H. Khazaei, “Performance analysis of ieee 802.15. 6 under saturation condition and error-prone channel,” in Wireless Communications and Networking Conference (WCNC), 2011 IEEE. IEEE, 2011, pp. 1167–1172. [20] L. Schwiebert, S. K. Gupta, and J. Weinmann, “Research challenges in wireless networks of biomedical sensors,” in Proceedings of the 7th annual international conference on Mobile computing and networking. ACM, 2001, pp. 151–165.

125

Bibliography

[21] M. I. Ashraf, M. Härkönen, M. Hämäläinen, and J. Riekki, “Health care process management supported by wireless technology,” in Finnish Signal Processing Symposium. Citeseer, 2007. [22] S. Cherukuri, K. K. Venkatasubramanian, and S. K. Gupta, “Biosec: A biometric based approach for securing communication in wireless networks of biosensors implanted in the human body,” in Parallel Processing Workshops, 2003. Proceedings. 2003 International Conference on. IEEE, 2003, pp. 432–439. [23] P. Neves, M. Stachyra, and J. Rodrigues, “Application of wireless sensor networks to healthcare promotion,” 2008. [24] M. Ameen, A. Nessa, and K. S. Kwak, “Qos issues with focus on wireless body area networks,” in Convergence and Hybrid Information Technology, 2008. ICCIT’08. Third International Conference on, vol. 1.

IEEE, 2008, pp. 801–807.

[25] S. Ullah, P. Khan, N. Ullah, S. Saleem, H. Higgins, and K. S. Kwak, “A review of wireless body area networks for medical applications,” arXiv preprint arXiv:1001.0831, 2010. [26] S. Dagtas, Y. Natchetoi, and H. Wu, “An integrated wireless sensing and mobile processing architecture for assisted living and healthcare applications,” in Proceedings of the 1st ACM SIGMOBILE international workshop on Systems and networking support for healthcare and assisted living environments.

ACM, 2007, pp. 70–72.

[27] Y. Zhao, A. Davidson, J. Bain, S. Li, Q. Wang, and Q. Lin, “A mems viscometric glucose monitoring device,” in Solid-State Sensors, Actuators and Microsystems, 2005. Digest of Technical Papers.

126

Bibliography

TRANSDUCERS’05. The 13th International Conference on, vol. 2. IEEE, 2005, pp. 1816–1819. [28] H.-T. Chu, C.-C. Huang, Z.-H. Lian, and J. J. Tsai, “A ubiquitous warning system for asthma-inducement,” in Sensor Networks, Ubiquitous, and Trustworthy Computing, 2006. IEEE International Conference on, vol. 2.

IEEE, 2006, pp. 186–191.

[29] A. Darwish and A. E. Hassanien, “Wearable and implantable wireless sensor network solutions for healthcare monitoring,” Sensors, vol. 11, no. 6, pp. 5561–5595, 2011. [30] P. Johansson, M. Kazantzidis, R. Kapoor, and M. Gerla, “Bluetooth: an enabler for personal area networking,” Network, IEEE, vol. 15, no. 5, pp. 28–37, 2001. [31] C. Otto, A. Milenkovic, C. Sanders, and E. Jovanov, “System architecture of a wireless body area sensor network for ubiquitous health monitoring,” Journal of Mobile Multimedia, vol. 1, no. 4, pp. 307–326, 2006. [32] G. Shobha, R. R. Chittal, and K. Kumar, “Medical applications of wireless networks,” in Systems and Networks Communications, 2007. ICSNC 2007. Second International Conference on.

IEEE, 2007, pp.

82–82. [33] P. Kulkarni and Y. Öztürk, “Requirements and design spaces of mobile medical care,” ACM SIGMOBILE Mobile Computing and Communications Review, vol. 11, no. 3, pp. 12–30, 2007. [34] D.

Konstantas,

“The

mobihealth

project.

ist

project,”

127

Bibliography

IST-2001-36006,

European

Commission:

Deliverable

2.6,

http://www. mobihealth. org, Tech. Rep., 2004. [35] M. Soini, J. Nummela, P. Oksa, L. Ukkonen, and L. Sydänheimo, “Wireless body area network for hip rehabilitation system,” Ubiquitous Computing and Communication Journal, vol. 3, no. 5, pp. 42–48, 2008. [36] U. Maurer, A. Rowe, A. Smailagic, and D. P. Siewiorek, “ewatch: a wearable sensor and notification platform,” in Wearable and Implantable Body Sensor Networks, 2006. BSN 2006. International Workshop on. IEEE, 2006, pp. 4–pp. [37] K. Van Laerhoven, B. P. Lo, J. W. Ng, S. Thiemjarus, R. King, S. Kwan, H.-W. Gellersen, M. Sloman, O. Wells, P. Needham et al., “Medical healthcare monitoring with wearable and implantable sensors,” in Proc. of the 3rd International Workshop on Ubiquitous Computing for Healthcare Applications, 2004. [38] S. DaÇğtas, G. Pekhteryev, Z. SahinoÇğlu, H. Cam, and N. Challa, “Real-time and secure wireless health monitoring,” International Journal of Telemedicine and Applications, vol. 2008, p. 1, 2008. [39] G. Mulligan, “The 6lowpan architecture,” in Proceedings of the 4th workshop on Embedded networked sensors.

ACM, 2007, pp. 78–82.

[40] K. S. Kwak, S. Ullah, and N. Ullah, “An overview of ieee 802.15. 6 standard,” in Applied Sciences in Biomedical and Communication Technologies (ISABEL), 2010 3rd International Symposium on.

IEEE,

2010, pp. 1–6. [41] M. Hernandez and R. Miura, “Coexistence of ieee std 802.15. 6

128

Bibliography

tm-2012 uwb-phy with other uwb systems,” in Ultra-Wideband (ICUWB), 2012 IEEE International Conference on.

IEEE, 2012, pp.

46–50. [42] A. W. Astrin, L. Huan-Bang, and R. Kohno, “Standardization for body area networks,” IEICE transactions on communications, vol. 92, no. 2, pp. 366–372, 2009. [43] R. Chávez-Santiago, K. Sayrafian-Pour, A. Khaleghi, K. Takizawa, J. Wang, I. Balasingham, and H.-B. Li, “Propagation models for ieee 802.15. 6 standardization of implant communication in body area networks,” Communications Magazine, IEEE, vol. 51, no. 8, pp. 80–87, 2013. [44] S. Sarkar, S. Misra, B. Bandyopadhyay, C. Chakraborty, and M. Obaidat, “Performance analysis of ieee 802.15. 6 mac protocol under non-ideal channel conditions and saturated traffic regime.” [45] M. S. Chowdhury, P. Khan, J. Jung, and K. S. Kwak, “Modeling slotted aloha of wban in non-saturated conditions,” KSII Transactions on Internet and Information Systems (TIIS), vol. 8, no. 6, pp. 1901–1913, 2014. [46] M. Chowdhury, K. Ashrafuzzaman, and K. Kwak, “Modelling ieee 802.15. 6 slotted aloha in heterogeneous condition,” Electronics Letters, vol. 50, no. 5, pp. 415–416, 2014. [47] J. Villalón,

P. Cuenca,

and L. Orozco-Barbosa,

“On the

capabilities of ieee 802.11 e for multimedia communications over heterogeneous 802.11/802.11 e wlans,” Telecommunication Systems, vol. 36, no. 1-3, pp. 27–38, 2007.

129

Bibliography

[48] B. Lauwens, B. Scheers, and A. Van de Capelle, “Performance analysis

of

unslotted

csma/ca

in

wireless

networks,”

Telecommunication Systems, vol. 44, no. 1-2, pp. 109–123, 2010. [49] S.-Y. Lee, Y.-S. Shin, K.-W. Lee, and J.-S. Ahn, “Performance analysis of extended non-overlapping binary exponential backoff algorithm over ieee 802.15. 4,” Telecommunication Systems, vol. 55, no. 1, pp. 39–46, 2014. [50] S. Ullah and K. S. Kwak, “Throughput and delay limits of ieee 802.15. 6,” in IEEE WCNC, 2011, pp. 174–178. [51] S. Rashwand and J. Misic, “Performance evaluation of ieee 802.15. 6 under non-saturation condition,” in IEEE GLOBECOM, 2011, pp. 1–6. [52] S. Rashwand, J. Misic, and H. Khazaei, “Ieee 802.15. 6 under saturation: Some problems to be expected,” Journal of Communications and Networks, vol. 13, no. 2, pp. 142–148, 2011. [53] S. Rashwand and J. MišIć, “Effects of access phases lengths on performance of ieee 802.15. 6 csma/ca,” Computer Networks, vol. 56, no. 12, pp. 2832–2846, 2012.