Mobility Modeling for Efficient Data Routing in Wireless Body Area Networks

Transcription

1 Mobility Modeling for Efficient Data Routing in Wireless Body Area Networks By Mr. Muhammad Moid Sandhu CIIT/FA12-REE-049/ISB MS Thesis In Electrical Engineering COMSATS Institute of Information Technology Islamabad Pakistan Fall, 2014

2 Mobility Modeling for Efficient Data Routing in Wireless Body Area Networks A Thesis Presented to COMSATS Institute of Information Technology, Islamabad In partial fulfilment of the requirement for the degree of MS (Electrical Engineering) By Mr. Muhammad Moid Sandhu CIIT/FA12-REE-049/ISB Fall, 2014 ii

3 Mobility Modeling for Efficient Data Routing in Wireless Body Area Networks A Graduate Thesis submitted to Department of Electrical Engineering as partial fulfilment of the requirement for the award of Degree of M.S (Electrical Engineering). Name Mr. Muhammad Moid Sandhu Registration Number CIIT/FA12-REE-049/ISB Supervisor: Dr. Nadeem Javaid, Assistant Professor, Center for Advanced Studies in Telecommunications (CAST), COMSATS Institute of Information Technology (CIIT), Islamabad Campus, October, iii

4 Final Approval This thesis titled Mobility Modeling for Efficient Data Routing in Wireless Body Area Networks By Mr. Muhammad Moid Sandhu CIIT/FA12-REE-049/ISB Has been approved For the COMSATS Institute of Information Technology, Islamabad External Examiner: Dr. Muhammad Sher Dean, Faculty of Basic and Applied Sciences, IIU, Islamabad Supervisor: Dr. Nadeem Javaid Assistant Professor, Center for Advanced Studies in Telecommunications (CAST), CIIT, Islamabad HoD: Dr. Shahid A. Khan Professor, Department of Electrical Engineering, CIIT, Islamabad iv

5 Declaration I Mr. Muhammad Moid Sandhu, CIIT/FA12-REE-049/ISB herebyxdeclare that I havexproduced the work presented inxthis thesis, duringxthe scheduledxperiod of study. I also declare that I havexnot taken anyxmaterial from anyxsource exceptxreferred toxwherever due that amountxof plagiarism isxwithin acceptablexrange. If a violationxof HEC rulesxon research hasxoccurred in thisxthesis, I shall be liablexto punishablexaction under the plagiarismxrules of the HEC. Date: Signature of the student: Mr. Muhammad Moid Sandhu CIIT/FA12-REE-049/ISB v

6 Certificate It is certified that Mr. Muhammad Moid Sandhu, CIIT/FA12-REE-049/ISB has carried out all the work related to this thesis under my supervision at the Department of Electrical Engineering, COMSATS Institute of Information Technology, Islamabad and the work fulfills the requirements for the award of the MS degree. Date: Supervisor: Dr. Nadeem Javaid Assistant Professor Head of Department: Dr. Shahid A. Khan Professor, Department of Electrical Engineering, vi

7 This thesis is dedicated to my parents. For their endless love, support and encouragement. vii

8 ACKNOWLEDGMENT Foremost, I would like to express my sincere gratitude to my supervisor Dr. Nadeem Javaid for the continuous support of my MS thesis and research, for his patience, motivation, enthusiasm, and immense knowledge. His guidance helped me in all the time of research and writing of this thesis. Besides my supervisor, I would like to thank the rest of my thesis committee members for their encouragement and insightful comments. My sincere thanks also goes to other faculty members of department of electrical engineering for their continuous support and guidance. Last but not the least, I would like to thank my family: my parents, for giving me love and supporting me spiritually throughout my life. Mr. Muhammad Moid Sandhu CIIT/FA12-REE-049/ISB viii

9 ABSTRACT Mobility Modeling for Efficient Data Routing in Wireless Body Area Networks In recent years, Wireless Body Area Networks (WBANs) have achieved significant attention due to their potential applications in health care. In these networks, mobility models of human body and routing protocols largely affect the network lifetime. In this thesis, our main contribution is the proposition of a mobility model for the analysis of mobile human body while the other contributions are three proposed energy efficient routing protocols for WBANs. Mobility models play significant role in analysis of WBANs as they provide information about the distance between node and sink at any time instant. The distance between node and sink affects energy consumption, delay and path loss. In subject to more realistic scenarios, we propose mathematical models for five different postures; standing, sitting, walking, running, and laying. Nodes have different movement pattern in all of these postures. Now coming towards the first proposed routing protocol; Forwarding data Energy Efficiently with Load balancing (FEEL), in which a forwarder node is selected which reduces the transmission distance between node and sink, thereby reducing the energy consumption of nodes. In order to minimize propagation delay, Electro Cardio Graphy (ECG) and glucose level measuring nodes directly send their data to the sink. FEEL protocol is applicable for continuous monitoring of patients. However, continuous monitoring of patients is unnecessary in some applications like, temperature monitoring, etc. So, we also propose Reliable Energy Efficient Critical data routing (REEC) for critical data transmission in WBANs. In REEC, two forwarder nodes are selected on the basis of cost function and are used for relaying the data towards sink. In order to overcome the unbalanced load problem on forwarder nodes, the selection of forwarder nodes is rotated in each round. We also propose a novel routing protocol for Balanced Energy Consumption (BEC) and enhancing the network lifetime in WBANs. In BEC, relay nodes are selected based on a cost function. The nodes send their data to their nearest relay nodes to route it to the sink. Furthermore, the nodes send only critical data when their energy becomes less than a specific threshold. In order to distribute the load uniformly, relay nodes are rotated in each round based on a cost function. Simulations show improved results of our proposed protocols as compared to the selected existing protocols in terms of stability period, network lifetime and throughput. ix

10 LIST OF PUBLICATIONS 1. M. M. Sandhu, N. Javaid, M. Jamil, Z. A. Khan, M. Imran, M. Ilahi, M. A. Khan, Modeling Mobility and Psychological Stress based Human Postural Changes in Wireless Body Area Networks, Computers in Human Behavior, DOI: /j.chb , S. Ahmed, M. M. Sandhu, N. Amjad, A. Haider, M. Akbar, A. Ahmad, Z. A. Khan, U. Qasim, N. Javaid, imod LEACH: improved MODified LEACH Protocol for Wireless Sensor Networks, Journal of Basic and Applied Scientific Research, 3(10)25-32, A. Haider, M. M. Sandhu, N. Amjad, S. H. Ahmed, M. J. Ashraf, A. Ahmed, Z. A. Khan, U. Qasim, N. Javaid, REECH-ME: Regional Energy Efficient Cluster Heads based on Maximum Energy Routing Protocol with Sink Mobility in WSNs, Journal of Basic and Applied Scientific Research, 4(1) , N. Amjad, M. M. Sandhu, S. H. Ahmed, M. J. Ashraf, A. A. Awan, U. Qasim, Z. A. Khan, M. A. Raza, N. Javaid, DREEM-ME: Distributed Regional Energy Efficient Multi hop Routing Protocol based on Maximum Energy with Mobile Sink in WSNs, Journal of Basic and Applied Scientific Research, 4(1) , M. M. Sandhu, N. Javaid, M. Akbar, F. Najeeb, U. Qasim, Z. A. Khan, FEEL: Forwarding Data Energy Efficiently with Load Balancing in Wireless Body Area Networks, The 28 th IEEE International Conference on Advanced Information Networking and Applications (AINA-2014), Victoria, Canada. 6. M. M. Sandhu, M. Akbar, M. Behzad, N. Javaid, Z. A. Khan, U. Qasim, REEC: Reliable Energy Efficient Critical data routing in wireless body area networks, The 9 th International Conference on Broadband and Wireless Computing, Communication and Applications (BWCCA 2014), Guangzhou, China. 7. M. M. Sandhu, M. Akbar, M. Behzad, N. Javaid, Z. A. Khan, U. Qasim, Mobility Model for WBANs, The 9 th International Conference on Broadband and Wireless Computing, Communication and Applications (BWCCA 2014), Guangzhou, China. 8. Mohsin Raza Jafri, Muhammad Moid Sandhu, Kamran Latif, Zahoor Ali khan, Ansar Ul Haque Yasar, Nadeem Javaid, Towards Delay-Sensitive Routing in Underwater Wireless Sensor Networks, The 5 th International Conference on Emerging Ubiquitous Systems and Pervasive Networks (EUSPN-2014), Halifax, Nova Scotia, Canada. 9. Ashfaq Ahmad, Muhammad Babar Rasheed, Muhammad Moid Sandhu, Zahoor Ali Khan, Ansar Ul Haque Yasar, Nadeem Javaid, Hop Adjusted Multi-chain Routing for Energy Efficiency in Wireless Sensor Networks, The 5 th International Conference on Emerging Ubiquitous Systems and Pervasive Networks (EUSPN-2014), Halifax, Nova Scotia, Canada. x

11 TABLE OF CONTENTS 1 Introduction 1 2 Related Work and Background Mobility-supporting Adaptive Threshold-based Thermal-aware Energyefficient Multi-hop ProTocol (M-ATTEMPT) for WBANs Stable Increased-throughput Multi-hop Protocol for Link Efficiency (SIMPLE) in WBANs On Increasing Network Lifetime (OINL) in body area sensor networks using global routing with energy consumption balancing FEEL: Forwarding Data Energy Efficiently with Load Balancing in Wireless Body Area Networks Motivation Radio Model FEEL: Proposed Protocol Deployment of Nodes Start-up Phase Selection of Forwarder Node Scheduling Phase Data Transmission Phase Energy Consumption Analysis Simulation Results and Analysis Network Lifetime Throughput Residual Energy Path Loss REEC: Reliable Energy Efficient Critical data routing in Wireless Body Area Networks Motivation Radio Model REEC: Proposed Protocol xi

12 4.3.1 Deployment of Nodes Start-up Phase Forwarders Selection Phase Scheduling Phase Data Transmission Phase Energy Consumption Analysis Experiments and Discussions Stability Period and Network Lifetime Throughput Residual Energy Path Loss BEC: A Novel Routing Protocol for Balanced Energy Consumption in Wireless Body Area Networks Motivation Analysis of Energy Consumption BEC: The Proposed Protocol Radio Model Placement of Nodes Start-up Phase Routing Phase Scheduling Phase Data Transmission Phase Experiments and Discussions Stability Period and Network Lifetime Network Throughput Residual Energy Path Loss Mobility Modeling for Wireless Body Area Networks Motivation Mobility Modeling Standing Sitting Walking Running Laying Impact of Mobility in WBANs Energy Consumption xii

13 6.3.2 Delay Path loss Implementation of Mobility Model in the Routing Protocols Energy Consumption Analysis Multi-hop Technique Data Transmission using Forwarder Nodes Initialization phase Forwarders selection phase Scheduling phase Data transmission phase Simulation Results and Analysis Network lifetime Throughput Residual energy Delay Path loss Energy consumption Conclusion and Future Work 73 8 References 75 9 List of Publications 83 xiii

14 LIST OF FIGURES 1.1 Components of a sensor node Deployment of nodes on the human body in FEEL Contents of HELLO message in FEEL Comparison of stability period and network lifetime for case Comparison of stability period and network lifetime for case Comparison of network throughput (aggregated) for case Comparison of network throughput (aggregated) for case Comparison of residual energy for case Comparison of residual energy for case Comparison of path loss for case Comparison of path loss for case Deployment of nodes on the human body in REEC Flowchart of REEC Comparison of stability period and network lifetime in REEC, SIM- PLE and M-ATTEMPT Comparison of network throughput (aggregated) in REEC, SIM- PLE and M-ATTEMPT Comparison of residual energy in REEC, SIMPLE and M-ATTEMPT Comparison of path loss in REEC, SIMPLE and M-ATTEMPT Placement of nodes on the human body and mechanism for path selection in OINL and BEC Format of the HELLO packet in BEC Comparison of stability period and network lifetime in BEC and OINL Comparison of network throughput in BEC and OINL Comparison of residual energy in BEC and OINL Comparison of path loss in BEC and OINL Markov model for posture pattern selection xiv

15 6.2 Human body in sitting position Human body in walking position Human body in running position Human body in laying position Value of η e and η k Effect of distance on energy consumption of nodes Effect of distance on delay Effect of distance on path loss Placement of nodes on the human body Network flow tree in multi-hop routing scheme Network flow tree in forwarder based routing scheme Comparison of number of dead nodes in multi-hop and forwarder based routing techniques Comparison of stability period and network lifetime in multi-hop and forwarder based routing techniques Comparison of packets sent to sink (aggregated) in multi-hop and forwarder based routing techniques Comparison of dropped packets (aggregated) in multi-hop and forwarder based routing techniques Comparison of received packets (aggregated) in multi-hop and forwarder based routing techniques Comparison of residual energy in multi-hop and forwarder based routing techniques Comparison of delay in multi-hop and forwarder based routing techniques Comparison of path loss in multi-hop and forwarder based routing techniques Comparison of energy consumption in multi-hop and forwarder based routing techniques Comparison of average energy consumption in multi-hop and forwarder based routing techniques xv

16 LIST OF TABLES 3.1 Energy Parameters of Transceivers Simulation Parameters Improvement in Percentage for case Improvement in Percentage for case Energy Parameters of Transceivers Simulation Parameters Improvement in Percentage Energy Parameters of Transceivers Distances of nodes from the sink Simulation Parameters Improvement in Percentage Simulation Parameters xvi

17 Chapter 1 Introduction 1

18 Nowadays, traditional health care systems are facing challenges due to increase in the elderly population and limited financial resources. The total health care budget of Pakistan is Rs billion for the year [1] and is expected to increase in the upcoming years. This appeals scientists and researchers to find the best and economical solutions for health care. Remote monitoring of patients vital signs presents a solution to the increasing cost of health care. Therefore, monitoring of human body and surrounding environment is important, especially for patients, athletes, and soldiers. Wireless Body Area Network (WBAN) is a subfield of Wireless Sensor Networks (WSNs) in which different vital parameters of human body are monitored. WBAN is used to solve theproblems relatedto health care. It consists ofsmall, low power, and intelligent nodes deployed on/in/around the human body for monitoring and diagnosis (note: we use the term sensors, nodes, and sensor nodes interchangeably in this document). The components of a node are shown in fig These nodes collect data from the human body and transmit via single-hop or multi-hop mechanism to sink which further sends the collected data to medical server. The medical specialist at a remote place can access the patients data. Nodes provide flexibility in terms of data gathering and are cost effective. WBAN provides long term health monitoring without affecting routine activities [2]. Sensor Unit ADC Processor Memory Protocols Transceiver Power Source Figure 1.1: Components of a sensor node There are a number of applications of WBANs including real time health monitoring of patients. They are also used to monitor the soldiers in the field. The sensors placed on the body measure different physiological parameters and send data to the concerned authorities. Interactive gaming is an emerging application of WBANs. The players can physically move their limbs and the sensors placed on the body send data to the gaming device. It provides enhanced entertainment. The sensors used in WBANs have limited energy. It is difficult to replace or recharge the batteries very often. Therefore, it is necessary to use minimum energy in order to increase the stability period and network lifetime. Other performance parameters used in WBANs are network throughput, delay, pathloss, etc. There are different routing protocols used to enhance the network lifetime. We propose a high throughput and reliable routing protocol for WBANs having in- 2

19 creased stability period called Forwarding data Energy Efficiently with Load balancing (FEEL). We deploy eight nodes at different positions on the human body. Two cases are considered for the placement of sink. In the first case, sink is placed on the chest while in the second case, sink is placed on the wrist. Two sensors measuring ECG and glucose levels communicate directly to the sink. They possess critical data which is sent to the sink immediately without any delay. The other six nodes communicate to the sink via forwarder node. All nodes are homogeneous and have same specifications. This scheme uses energy efficiently and increases the stability period and throughput of the network. FEEL is suitable for continuous monitoring of patients. However, some applications require only critical data. So, we propose Reliable Energy Efficient Critical data routing (REEC) for efficient monitoring of patients in WBANs. The proposed protocol selects two forwarders which collect the data of other nodes, aggregate it, and route it to the sink. REEC routes only critical data of the patients. We define critical data as the abnormal data that demands immediate medical aid and treatment of the patient. For long term health monitoring, we propose a new routing protocol for Balanced Energy Consumption (BEC) in WBANs. In BEC, relay nodes are selected based on a cost function. The nodes send their data to their nearest relay nodes to route it to the sink. The nodes closer to the sink send their data directly to it. Furthermore, the nodes send only critical data when their energy becomes less than a specific threshold. In order to distribute the load uniformly, relay nodes are rotated in each round based on a cost function. The proposed protocols (FEEL, REEC and BEC) assume that human body is static. On the other hand, several mobility models are proposed in literature for WSNs and ad hoc networks. However, they are not suitable for WBANs due to their different movement patterns. In general, the movements of nodes can be classified into two categories; single and group mobility. In the former case, there is no correlation between the movements of different nodes. In this scenario, nodes move regardless the mobility pattern of other nodes in the network. In the latter approach, however, nodes move in a group having a particular relationship between them. In this case, nodes move relative to a reference which decides the movement pattern of other nodes. We propose a new mobility model for WBANs which considers different postures of human body. There are different posture transition probabilities from one state to another. We consider five different postures; standing, walking, running, sitting, and laying. In each of these postures, nodes placed on human body have 3

20 different movement pattern. The nodes placed on the trunk of the body show little movement as compared to nodes placed on limbs. Furthermore, nodes exhibit different movement patterns during routine activities. We model the movement pattern of nodes in different postures and implement the proposed model in two routing protocols of WBANs. We study the impact of human mobility on the functionality of routing protocols in WBANs. The rest of the thesis is organized as follows: chapter 2 contains related work alongwith background and chapter 3 presents the proposed FEEL protocol. Chapter 4 describes the proposed REEC protocol and the proposed BEC protocol is discussed in chapter 5. The proposed mobility model for WBANs is presented in chapter 6. Conclusion alongwith future work is given in chapter 7 and chapter 8 contains references. 4

21 Chapter 2 Related Work and Background 5

22 The routing protocols in WBANs use different mechanisms for data transmission like, single-hop and multi-hop communication. In single-hop communication, nodes send their data directly to the sink. On the other hand, in multi-hop communication, intermediate nodes are used to route data to the sink. A. Ehyaie et al. [3] propose an upper bound on the number of relay nodes, sensors and their distance from sink. The relay nodes are distributed on the human body as a network. The sensors communicate to the relay nodes which further route data to the sink. Authors in [4] give Energy-Aware WBAN Design (EAWD) model. It gives the position and optimum number of relay nodes in WBANs. Relay nodes are responsible for data collection from sensors and routing it towards the sink. They propose integer linear programming for relay nodes for energy efficient routing. Authors in [5] derive a propagation and radio model for energy efficient communication in WBANs. They study energy efficiency on a line and tree topologies using these models. They find that single-hop communication is inefficient in WBANs. A two tier hierarchical architecture for WBANs is presented in[6]. Authors present an interference free routing protocol. Nodes send their data to Cluster Head (CH). This scheme monitors multiple patients and routes their data to the Base Station (BS). In [7], authors present an adaptive routing protocol. The priority and vicinity of nodes is taken into account for the selection of parent node for mobile human body. T. Watteyne et al. [8] formulate a self organization protocol for BANs. Nodes are grouped into clusters which send their data through CH to reduce energy consumption and increase the network lifetime. The protocol shows that clustering based approach is suitable for WBANs. In [9], authors suggest a WBAN protocol for monitoring the patients at home. The home server collects the data from nodes deployed on the human body and routes it to the medical server via internet. A distributed Wireless Body Area Sensor Network (WBASN) for medical supervision is presented in [10]. This system contains three layers: sensor network, mobile computing network, and remote monitoring network. It collects and stores vital signs such as ECG, blood oxygen, body temperature, etc. M. Quwaider et al. [11] present a routing protocol for WBANs, which counts for changes in the network. It uses store and forward mechanism to increase the probability of successful packet transmission. The location based packet routing is developed in this protocol. DARE [12] uses multi-hop scheme to monitor the patients in a ward of the hospital. Sensors attached to the patients send data to the body relay. The body relay aggregates the received data and routs it to the sink. Authors in[13] give THE-FAME to measure the fatigue in the soccer players. They employ a composite parameter for fatigue measurement which consists of a 6

23 threshold parameter for lactic acid and distance covered. The implanted sensor sends the data to the nearest sink deployed at the boundary of the field. Similarly, authors in [14] present a routing protocol for fatigue measurement of a soldier. Three sensors are attached to the body to measure temperature, heartbeat and glucose level in the blood. Different scenarios are considered for the movement of soldier. In [15], virtual groups are formed between doctors and nurses for efficient patient monitoring. Virtual groups are formed and modified according to the requirements of patients and doctors. Authors propose a new metric called Quality of Health Monitoring. G. R. Tsouri et al. [16] propose augmented efficiency for global routing in WBANs. Augmented efficiency is a new link cost, designed for balanced energy consumption in WBANs. Authors propose On Increasing Network Lifetime (OINL) in BANs using global routing with energy consumption balancing. It causes substantial improvement in the network lifetime. Authors in [17] suggest a new cross layer communication protocol for WBANs called Cascading Information retrieval by Controlling Access with Distributed slot Assignment (CICADA). It consumes less energy and is designed for mobile WBANs. Moreover, this protocol forms a network tree in a distributive manner. This tree is used to route data to the sink with guaranteed collision free access to the medium. Energy-Balanced Rate Assignment and Routing (EBRAR) protocol is presented in [18]. It is an energy efficient routing protocol in which routing is based on the residual energy of nodes. As a result, instead of one fixed path, data is intelligently sent through different routes by equally distributing the load among the nodes. Authors in [19] focus on increasing the network lifetime by relaying and cooperation techniques. First, the relay nodes perform relaying of traffic only so that, more energy is available for communication purposes. Furthermore, the relays cooperate in forwarding the data from nodes to the sink. Authors in [20] suggest a scheme in which nodes are grouped into a number of clusters. There is a CH in each cluster which is responsible for collecting the data from nodes. CH aggregates the received data and sends it to the sink. M. R. Senouci et al. [21] analyze different sensor network routing protocols and propose a new technique for increased network lifetime. Experiments show that their protocol can extend the network lifetime and can be very effective. Authors in [22] propose clustering algorithm for WSNs named as Fast and Flexible Unsupervised Clustering Algorithm (FFUCA). It gives low complexiy along with optimal energy consumption. In [23], authors give the techniques for transmitting the vital signs to the cloud. They propose energy efficient routing and data security mechanisms. In [24], authors propose Markov decision process model to 7

24 study the charging and discharging of sensor s battery. They also study the properties of optimal transmit policies. Authors in [25] propose an energy efficient routing protocol for WSNs. They reduce the transmission power of nodes to save energy. They also form a virtual back bone of high energy nodes to transmit data efficiently. Authors in [26] analyze the WBANs channels and bit error performances. They attach the receiving antenna to the back side and evaluate its performance. S. Ivanov et al. [27] use cooperation between WBANs and environmental sensors to efficiently transmit data to the distant gateway. Their suggested technique gives improved results in terms of packet loss, power consumption and delay. Authors in [28] present a method to elect controlled nodes which inform any abnormal behaviour to the CH. It saves energy and gives a better dynamic approach. Authors in [29] give routing algorithm based on global optimization cost function. Simulation results show that the protocol gives improved results relative to previous techniques. In [30], authors present evidence-based sensor coverage model. It is close to reality and can be extended to tackle the issues related to deployment of nodes. Simulations show that their model performs better than traditional models. Authors in [31 33] use clustering schemes to efficiently use the energy of nodes in WSNs. Nodes send their data to the CH which further routes it to the sink. Different challenges in body area networks are discussed in [34]. Energy efficiency is a major challenge which is a big hindrance in widespread use of WBANs. Other challenges are interference and reliability. Authors in [35] decrease the inter-ban interference by using cooperative scheduling. It results in increased throughput. Their proposed technique gives better packet reception rate than other schemes. In [36], authors propose Random Incomplete Coloring (RIC) in WBANs to overcome the interference. It increases the throughput and reduces the energy consumption. Simulations show that RIC efficiently reduces the interference in WBANs. Different issues in WBANs like, energy efficiency and packet delivery are discussed in [37]. Authors present some techniques to overcome the issues related to the successful delivery of packets to the sink. Authors in [38] use WBAN to monitor different physiological parameters of human body. They deploy nodes on the human body which send the real time data to the sink. The data received from nodes is used to check the physical condition of the body. Authors in [39] discuss IntraBody Communication (IBC) for WBANs. In this scheme, body is used as a communication medium. Signal travels in the body from transmitter to receiver. The advantage of this scheme is that it provides increased data security. In [40], authors propose a method to efficiently use the energy of heterogeneous sensor nodes to increase the network lifetime. Their proposed algorithm considers 8

25 the heterogeneity of nodes and requirement of the application. Their suggested scheme saves energy of nodes. Authors in [41] use relay nodes to transmit data from sensors placed onthe human body. It saves the energy of nodes and increases their lifetime. Simulations show that their scheme gives improved lifetime and bit error rate. Authors in [42] present a method to monitor the position of nodes on the human body. They continuously monitor the position of limbs of the human body. Authors in[43] present collision avoidance protocol for reliable data delivery in WBAN. They also propose security mechanism to restrict illegal access to the network. L. Yao et al. [44] present a secure mechanism for transmitting the vital parameters to the control unit. They give ECG-signal based secure communication. It gives security and confidentiality of data. Authors in [45] propose a method to find a fault in the nodes in WBANs. In some situations, nodes become inactive and cannot monitor the vital signs correctly. The suggested scheme finds the inactive node and informs about any kind of abnormality in WBAN. In [46], authors propose a thermal-aware protocol for routing the data of nodes. The path having minimum distance is selected. Alternative paths are selected in case of hotspots; nodes which are heated due to increased energy dissipation. Authors in [47] propose a new Media Access Control (MAC) protocol for reliable data delivery in WBANs. They propose a channel access mechanism for increased throughput. In [48], authors propose security mechanism for WBANs. They also design a microcontroller to reduce the energy consumption. Experimental results show that their scheme works better. Authors in [49] present interference avoidance scheme for reliable data delivery. It is based on Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) and Time Division Multiple Access (TDMA). It increases the throughput in WBANs. Authors in[50] present fair data collection scheme for WSNs. As nodes are located at different distances from the sink, so fair data distribution leads to extended network lifetime. In a WBAN, nodes are located close to each other and within the communication range of each other. Therefore, efficient MAC layer protocols are employed to avoid collision. Y. Zhang et al. [51] propose a priority-guaranteed MAC protocol for WABNs. In this protocol, control channels are separated from data channels. Priority-specific control channels are used for life-critical applications. Authors also present wakeup trigger mode to facilitate priority traffic. Authors in [52] propose PLA-MAC for priority-based traffic in WBANs. The sensed data is bifurcated according to their Quality-of-Service (QoS) (i.e., delay, reliability and throughput) requirements and is assigned priorities. These priorities determine the transmission schedule of packets. The superframe structure also 9

26 varies according to the amount of data causing minimum energy consumption. Due to continuous movements of the human body, the link between the sink and the node may not be connected all the time. The link breakages result in loss of data. P. Ferrand et al. [53] describe a cooperative transmission scheme in WBANs to overcome the disconnected links. They use a multi-hop scheme to ensure good connectivity. In their proposed work, some sensors are elected to support the nodes having bad links. This way, data is efficiently routed to the sink. Authors in [54] propose an obesity control framework using WBANs. They propose software and hardware architectures for obesity control. In their proposed framework, sensors are placed on the human body. These sensors monitor different vital signs and compare them with the predefined thresholds. If the sensed value exceeds the threshold, the information is sent to a smart phone or a personal computer to allow taking the appropriate action to prevent body harm. In [55], authors place wearable sensors on the human body and study the link behaviour in dynamic conditions. They record the link quality, packet delivery and Received Signal Strength Indicator (RSSI) values in real-time. They also describe the packet delivery and energy efficiency obtained by using dynamic routing and adaptive transmission power schemes, respectively. Authors in [56] estimate the lifetime of Health Monitoring Network (HMN) using probabilistic analysis. It is important to estimate the lifetime of the network to replace/recharge the batteries of nodes to continuously monitor the required parameters. In [57], authors use wireless accelerometer sensor to determine the link performance and lost packets for different runners and for different sensor locations. They conclude that sensors placed on the wrist give best results. In the following sections, we discuss some routing protocols in detail. 2.1 Mobility-supporting Adaptive Threshold-based Thermal-aware Energyefficient Multi-hop ProTocol (M-ATTEMPT) for WBANs In M-ATTEMPT [58], the high data rate nodes are placed near the sink on the human body. Whereas, the nodes having low data rate are placed away from sink. The nodes near the sink have more energy than the other nodes in M- ATTEMPT. The protocol operation is categorized into different phases. In the initialization phase, all nodes broadcast hello messages. This hello packet contains the information about neighbors and distance from sink in terms of hop-counts. In the routing phase, routes with minimum hops are selected for data transmission from nodes to the sink. In case of critical data, the nodes send their data directly 10

27 to the sink. If two routes are available then route with minimum hop counts is selected. The low data rate nodes send their data to the nearest high data rate nodes which send the aggregated data to the sink. In M-ATTEMPT, single-hop and multi-hop communication is utilized to enhance the network lifetime. After the route selection, TDMA slots are assigned to the nodes. All the nodes transmit data in their scheduled time slots. In M-ATTEMPT, nodes are categorized into different levels according to their data rates as parent nodes, first-level child nodes and second-level child nodes. During themovement ofthehumanbody, ifachildnodemoves away fromitsparent node, it can associate to another nearest parent node to save energy. Due to excessive energy consumption, a node may get heated which is known as hot-spot. In this case, alternative paths are selected until the node returns to its original normal state. However, nodes deplete their energy quickly resulting in shorter stability period and lack of critical data transmission from some nodes. We propose new protocols to overcome the deficiencies in M-ATTEMPT. We compare the proposed protocols with M-ATTEMPT and discuss different performance metrics in detail in chapters 3 and Stable Increased-throughput Multi-hop Protocol for Link Efficiency (SIMPLE) in WBANs In SIMPLE [59], eight nodes are placed at different positions on the human body with sink at the waist. The working of SIMPLE protocol is divided into different phases. In the initial phase, sink broadcasts a short information packet to inform the nodes about its position on the human body. Each node broadcasts a packet which contains the node ID, its residual energy value and its location. In the next phase, a forwarder node is selected which routes the data of other nodes, thus saving their energy. The forwarder is selected based upon its distance from sink and its residual energy status. The node having minimum distance from sink and having maximum residual energy value is selected as a forwarder. All the corresponding nodes send data to the forwarder node which aggregates the received data and routes it to the sink. Furthermore, the nodes having critical data send their data directly to the sink to observe minimum delay. In the scheduling phase, forwarder node assigns TDMA based time slots to its children nodes. All the nodes transmit data in their scheduled time slots to avoid any collision and loss of data. In this way, data is efficiently routed from nodes to sink. However, routing load is not uniformly distributed among all the nodes in SIMPLE 11

28 protocol. The placement of sink is also an important parameter as it greatly affects the throughput. In addition, the human comfort level must also be taken into account when deciding the position of sink. We propose new routing protocols to overcome the above mentioned drawbacks. Therefore, we compare the proposed protocol with SIMPLE and discuss different performance parameters in detail in chapters 3 and On Increasing Network Lifetime (OINL) in body area sensor networks using global routing with energy consumption balancing In OINL, global routing based on Dijkstras algorithm is used to enhance network lifetime in WBANs. A link-cost function is also proposed for enhancing the network lifetime. In OINL, link-cost information is periodically gathered at the Access Point (AP) in the form of channel attenuation. All the routing calculations are performed at the AP as it has more energy than nodes. The channel attenuation for the selected link between nodes j and k is given as: α j,k = RSSI P tx (2.1) Where, RSSI denotes the received signal strength at node k and P tx is the transmitted power. The energy of nodes used thus far is calculated using eq E j i = Ej i 1 + RSSI T α j,k (2.2) Here, j denotes the node ID and i is the current round. E j i is the accumulated energy of node j at round i and α j,k is the attenuation of the selected link. RSSI T is the predefined target RSSI level. The link cost Cj,k i is computed as: C i j,k = RSSI T α j,k ( 1+ Ek i i 2 E min ) (2.3) The link-cost function is derived by dividing the accumulated energy of node i with the minimum energy across all nodes, Ei min. The ratio is then raised to the power of M 0, which reflects the effect of balanced energy consumption. The nodes send the sensed data through relay nodes having minimum link-cost. In this way, nodes consume energy in a balanced way which enhances the network lifetime. However, thedrawback ofthisscheme isthatitburdensthenodesnearthesink(or 12

29 AP). The network lifetime can further be improved by using direct transmission of nodes near the sink. We propose a new technique which gives improved performance than OINL with reduced computational overhead as discussed in chapter 5. 13

30 Chapter 3 FEEL: Forwarding Data Energy Efficiently with Load Balancing in Wireless Body Area Networks 14

31 3.1 Motivation WBANs monitor human health with limited energy resources. In these network, different routing schemes are used to route data towards sink which further sends data to the medical server or other monitoring station. M-ATTEMPT uses multihop communication for normal data delivery to sink. Nodes communicate directly to the sink for routing critical data. However, they deplete their energy quickly resulting in shorter stability period and lack of critical data from some nodes. SIMPLE uses a cost function for forwarder node selection which prolongs the stability period. However, load is not uniformly distributed among all the nodes. The placement of sink is also an important parameter as it greatly affects the throughput. In addition, the human comfort level must also be taken into account when deciding the position of sink. SIMPLE and M-ATTEMPT protocols are discussed in chapter 2 in detail. Sink 8 Node Figure 3.1: Deployment of nodes on the human body in FEEL The radio model used for calculating the energy consumption of nodes in discussed in the next section in detail. 15

32 3.2 Radio Model There are different radio models in the literature. We use first order radio model given in [60]. The equations for first order radio model are given as: E TX (k,d) = E TXelect (k)+ε amp (k,d) (3.1) E TX (k,d) = E TXelect.k +ε amp.k.d 2 (3.2) E RX (k,d) = E RXelect (k) = E RXelect.k (3.3) Where, E TX istheenergyconsumed intransmission processande RX istheenergy consumed by the receiver. E TXelect and E RXelect are the energies required to run the electronic circuit of transmitter and receiver respectively. ε amp is the energy required by the amplifier circuit, k is the packet size whereas d is the distance between transmitter and receiver. In WBANs, the communication medium is human body which contributes attenuation to the radio signals. Therefore a path loss coefficient parameter n is included in the radio model. Equation for the transmitter energy consumption is: E TX (k,d) = E TXelect.k +ε amp.k.d n (3.4) The energy parameters depend upon the hardware of the system. We consider two transceivers, Nordic nrf 2401A and Chipcon CC2420, which are used frequently in WBAN technology. The energy parameters for these transceivers are shown in table 3.1. Table 3.1: Energy Parameters of Transceivers Parameter nrf 2401A CC2420 Units DC current (TX) ma DC current (RX) ma Min. supply voltage V E TXelect nj/bit E RXelect nj/bit ε amp nj/bit/m n 3.3 FEEL: Proposed Protocol In this section, we discuss a novel routing protocol for WBANs. Uniform energy consumption of nodes is important for long term health monitoring in WBANs. 16

33 We propose FEEL, a new routing protocol with improved stability period and throughput. The following subsections give detail of the proposed protocol Deployment of Nodes In FEEL, we deploy eight homogeneous nodes on the human body. Node 8 is ECG and node 7 is glucose level sensor. These two nodes send their data directly to the sink. We use two different topologies for the placement of sink on the human body. In the first case sink is placed on the chest while in the second case it is placed on the wrist. We place the sink on the chest and wrist to study the performance of the proposed protocol. We study the impact of sinks placement on energy consumption of nodes. Fig. 3.1 shows the placement of nodes and sinks on the human body. It also shows the distances of nodes from sinks Start-up Phase In the initial phase sink broadcasts a HELLO message containing following three types of information. Location of sink. Location of neighbours. Information about possible routes to the sink. The nodes receive this HELLO packet and update their routing table. They also send information about their IDs and residual energy status to the sink. Fig. 3.2 shows the contents of HELLO message. Figure 3.2: Contents of HELLO message in FEEL 17

34 3.3.3 Selection of Forwarder Node In this section, we present the selection criteria of forwarder node. In order to save energy and balance the energy consumption of the network, FEEL selects a new forwarder in each round. As sink knows the residual energy of all nodes, it broadcasts the ID of the node having maximum residual energy to make it the the forwarder node. Forwarder node = Node max(r.e) (3.5) Where R.E is the residual energy of a node. Residual energy is calculated by subtracting the consumed energy from initial energy. Energy residual = Energy initial Energy consumed (3.6) The node having maximum residual energy is selected as a forwarder node. All the neighboring nodes send their data to the forwarder node. The forwarder node aggregates the received data and routs it to the sink. In the next round, again a new forwarder node is selected based upon the residual energy. In this way, forwarder node rotates uniformly and all the nodes get a chance to become a forwarder. Therefore, energy is consumed more uniformly as compared to SIMPLE and M-ATTEMPT resulting in increased stability period and throughput Scheduling Phase In this phase, forwarder node assigns Time Division Multiple Access (TDMA) based time slots to its children nodes. All nodes send their data to the forwarder node in their allocated time slots. Proper scheduling of nodes minimizes their energy consumption Data Transmission Phase All other nodes except ECG and glucose level measuring nodes send their data to the forwarder. The forwarder node aggregates the received data and routs it to the sink. Nodes measuring ECG and glucose level communicate directly to the sink as they have critical data. If a node possesses energy less than a threshold (γ), it communicates directly to the sink. In addition, it does not further take part in the selection of forwarder. This is done to save the data aggregation energy of nodes. If a node has shorter distance to the sink than forwarder node, it routs its data directly to the sink. 18

35 3.4 Energy Consumption Analysis In this section, we develop equations for single-hop and multi-hop communications. Energy consumed for single-hop communication is: E SH = E TX (3.7) E TX is the transmission energy as given by: E TX = k (E elect +ε amp ) d 2 (3.8) Where, E elect is the energy consumed by electronic circuit. Now, energy consumed during multi-hop communication is given by: E MH = k[m (E TX )+(m 1) (E RX +E da )] (3.9) Here, E RX is the reception energy and m is the number of nodes. 3.5 Simulation Results and Analysis In order to verify the performance of FEEL protocol, simulations are performed in MATLAB. We study the performance of the proposed protocol in comparison with SIMPLE and M-ATTEMPT. The initial energy of all nodes is same i.e. 0.5 J. In simulation, we ignore the sensing energy consumed by the nodes. Simulations are performed five times and average results are plotted. Table 3.2 shows the values of different parameters used in simulation. We evaluate different performance metrics of the proposed protocol. Introduction to some of the metrics is given below. A. Network Lifetime It is the total time till the death of last node. It represents time for which the network operates. In WBANs, a protocol is required to offer maximum network lifetime. B. Stability Period It is the time before the death of the first node. It is an important parameter in WBANs. C. Throughput Throughput is the number of packets successfully received at sink. 19

36 D. Residual Energy It is the difference of initial energy and consumed energy. E. Path Loss It is the difference between transmitted power and received power. It is represented in decibel (db). Table 3.2: Simulation Parameters Parameter Value Units E RXelect 36.1 nj/bit E TXelect 16.7 nj/bit ε amp 1.97 nj/bit/m 2 E da 5 nj/bit d o 0.1 m γ 0.1 J Packet size (k) 4000 bits Frequency (f) 2.4 GHz Initial energy (E o ) 0.5 J Network Lifetime Figs. 3.3 and 3.4 show the stability period and network lifetime of FEEL protocol. Our protocol selects the forwarder node on the basis of residual energy of nodes. So, energy is consumed in a balanced way. As a result, stability period of FEEL protocolisincreased. InSIMPLE, thenodescloser tothesinkhave morechance to become forwarder node. So energy is consumed in an imbalanced way, decreasing the stability period. FEEL has stability period of about 5428 rounds and network lifetime of 7486 rounds in the first case. In the second case, the stability period is increased to 5635 rounds. It is due to the fact that sink is closer to most of the nodes in this case. As a result less distance between nodes and sink causes less energy consumption of nodes. So, the stability period is increased Throughput It shows the number of packets successfully received at sink. WBANs require maximum data reception at the sink with minimum packets dropped. We use Random Uniformed Model [61] for packet drop calculation. The status of communication link can be good or bad depending upon the probability. We suppose the probability of link status to be good is 0.7. FEEL protocol achieves higher throughput than M-ATTEMPT and SIMPLE as shown in figs. 3.5 and 3.6. Throughput depends upon the number of nodes which are alive. More nodes send more packets 20

37 Number of dead nodes % 81% FEEL SIMPLE M ATTEMPT Rounds Figure 3.3: Comparison of stability period and network lifetime for case 1 so throughput increases. As the stability period of M-ATTEMPT and SIMPLE is less, so less number of nodes send packets resulting in less throughput. Whereas, the FEEL protocol has longer stability period, so more nodes send packets resulting in increased throughput. Throughput of the FEEL protocol is even higher in second case due to increased stability period Residual Energy The residual energy of the network is shown in figs. 3.7 and 3.8. The FEEL protocol uses multi-hop communication for data transmission to the sink. All nodes except 7 and 8, transmit their data to the forwarder node which routs it to the sink. The forwarder node is selected at the start of each round. The selection of new forwarder in each round saves energy. In FEEL protocol a new forwarder node is selected in each round, removing the burden of data transmission from a single node. In M-ATTEMPT and SIMPLE, nodes die early due to heavy traffic load and non-uniform load distribution Path Loss Path loss shows the difference in the transmitted and received power represented in decibels (dbs). The posture of human body affects the signal. As a result path loss shows different behaviour during the movement of human body. There are 21

38 Number of dead nodes % 78% FEEL SIMPLE M ATTEMPT Rounds Figure 3.4: Comparison of stability period and network lifetime for case 2 different models used to estimate the path loss. It is a function of distance and frequency as expressed in [62] and shown as: PL(f,d) = PL o +10.n.log 10 ( d d o )+Xσ (3.10) Where, PL o is path loss at reference distance d o and n is path loss exponent. The distance between transmitter and receiver is d, X is a gaussian random variable and σ is the standard deviation. Path loss at reference distance d o is given as: Where, λ is the wavelength of electromagnetic waves. ( ) 2 4.π.do PL o = 10.log 10 (3.11) λ Figs. 3.9 and 3.10 show the path loss in each round. In simulation, we use a fixed frequency of 2.4 GHz from ISM band. We use path loss coefficient of 3.38 and standard deviation of 4.1. FEEL has lower path loss as shown in the figs. 3.9 and In the proposed protocol, path loss decreases after 4000 rounds. It is due to the fact that some nodes die after 4000 rounds. So less number of nodes have lower path loss. FEEL protocol has lower path loss than M-ATTEMPT. The improvement (%) provided by the FEEL protocol to M-ATTEMPT and SIM- PLE is shown in tables 3.3 and

39 Packets received at sink 3.5 x FEEL SIMPLE M ATTEMPT Rounds Figure 3.5: Comparison of network throughput (aggregated) for case 1 Table 3.3: Improvement in Percentage for case 1 Parameter Improvement (%) Improvement (%) in M-ATTEMPT in SIMPLE Stability period Network lifetime Throughput 72 7 Average residual energy Average path loss Table 3.4: Improvement in Percentage for case 2 Parameter Improvement (%) Improvement (%) in M-ATTEMPT in SIMPLE Stability period Network lifetime Throughput Average residual energy Average path loss

40 Packets received at sink 3.5 x FEEL SIMPLE M ATTEMPT Rounds Figure 3.6: Comparison of network throughput (aggregated) for case FEEL SIMPLE M ATTEMPT Residual Energy (J) Rounds Figure 3.7: Comparison of residual energy for case 1 24

41 4 3.5 FEEL SIMPLE M ATTEMPT Residual Energy (J) Rounds Figure 3.8: Comparison of residual energy for case FEEL SIMPLE M ATTEMPT Path Loss (db) Rounds Figure 3.9: Comparison of path loss for case 1 25

42 FEEL SIMPLE M ATTEMPT Path Loss (db) Rounds Figure 3.10: Comparison of path loss for case 2 26

43 Chapter 4 REEC: Reliable Energy Efficient Critical data routing in Wireless Body Area Networks 27

44 4.1 Motivation The routing schemes in WBANs use different data transmission mechanisms like, single-hop, multi-hop, minimum-hop, etc. In single-hop routing scheme, distant nodes die faster than the nodes nearer to the sink. On the other hand, in multihop and minimum-hop routing schemes, the nearer nodes die earlier as they have more data to route than the distant nodes. M-ATTEMPT uses multi-hop scheme for routing data from sensor nodes to sink. It is a thermal aware routing protocol which selects a new route after a hotspot detection. However, the hotspot detection causes more energy consumption. SIMPLE overcomes the deficiencies in M-ATTEMPT. It selects a new forwarder in each round that receives and aggregates the data of other nodes and routes it to the sink. However, this protocol burdens the single forwarder node by routing all the data through it. In SIMPLE, nodes send all the data (normal and critical) which is unnecessary in most of the scenarios in WBANs. Therefore, we present REEC which sends only critical data and avoids the transmission of redundant data. SIMPLE and M-ATTEMPT protocols are discussed in chapter 2 in detail. The radio model used for calculating the energy consumption of nodes in discussed in the next section in detail. 4.2 Radio Model There are different radio models in the literature. We use first order radio model given in [52]. The equations for first order radio model are given below: TX (κ,l) = TXelect (κ)+ε amp (κ,l) (4.1) TX (κ,l) = TXelect.κ+ε amp.κ.l 2 (4.2) RX (κ,l) = RXelect (κ) = RXelect.κ (4.3) Where TX is the energy consumed in transmission process. RX is the energy consumed by the receiver. TXelect and RXelect are the energies required to run the electronic circuit of transmitter and receiver, respectively. ε amp is the energy required by the amplifier circuit whereas κ is the packet size. The distance between transmitter and receiver is represented by l. In WBANs, the communication medium is human body which introduces attenuation to the radio signals. Therefore a path loss coefficient parameter n is included 28

45 in the radio model. Equation for the transmitter energy consumption is: TX (κ,l) = TXelect.κ+ε amp.κ.l n (4.4) Energy parameters depend upon the hardware of the system. We consider two transceivers Nordic nrf 2401A and Chipcon CC2420 that are used frequently in WBANs. The energy parameters for these transceivers are enlisted in table 4.1. Table 4.1: Energy Parameters of Transceivers Parameter nrf 2401A CC2420 Units DC current (TX) ma DC current (RX) ma Supply voltage (min.) V TXelect nj/bit RXelect nj/bit ε amp nj/bit/m n 4.3 REEC: Proposed Protocol In this section, we describe the proposed routing protocol. One of the major challenges in WBANs is to increase the network lifetime for continuous monitoring of patients. REEC consumes energy efficiently that leads to increased network lifetime. The detail is given in the following subsections Deployment of Nodes In the proposed protocol, we deploy eight sensors on the human body. All nodes are homogeneous i.e. having equal initial energy. In REEC, we place the sink at the centre of human body. We choose abdomen for the placement of sink as it is less mobile (as compared to limbs) and has same distance from head and foot. The sink is placed on the abdomen of the human body as shown in fig The information from sink is sent to the physician via internet for inspection and diagnosis. It is also sent to the ambulance service office for immediate help in case of emergency. Medical server stores the patients data for future purposes. The whole scenario of WBAN is shown in fig

46 Information ECG sensor 1 2 Laptop Physician Pulse rate sensor Sink Medical server EMG and motion sensors PDA Assessment and treatment Ambulance Figure 4.1: Deployment of nodes on the human body in REEC Start-up Phase In this phase, sink broadcasts a short information packet which contains the location of sink on the human body. Each node receives this packet and stores the location of the sink. Afterwards, each node broadcasts a packet which contains the ID of node, its location and residual energy status. In this way, all nodes are updated with the location of neighbouring nodes and the sink Forwarders Selection Phase In this section, we present the selection criteria of the forwarder nodes. The complete set of nodes A is given by: A = {1,2,3,4,5,6,7,8} (4.5) 30

47 In order to consume the energy efficiently, REEC uses cost function ξ to select new forwarders in each round. The ξ is calculated as: ξ(i) = ( ) l(i) R(i) i A (4.6) Here, l is the distance between the node and sink and R is the residual energy of node. The node having minimum value of ξ is selected as forwarder. We consider two sets of nodes as: α = {1,2,3,4} (4.7) β = {5,6,7,8} (4.8) α A (4.9) β A (4.10) α β = ø (4.11) A = α β (4.12) In REEC, two forwarders are selected in each round, one from α and second from β. The Ψ α is selected from α and Ψ β is selected from β. The total number of nodes is ℵ. Ψ α = ℵ min(ξ(i)) i α (4.13) Ψ β = ℵ min(ξ(i)) i β (4.14) The node having minimum value of ξ is selected as a forwarder node. Sink broadcasts the IDs of Ψ α and Ψ β after calculating ξ. The nodes from α send their data to Ψ α whereas nodes fromβ send their data to Ψ β. The forwarder nodes aggregate the data of all the nodes and routeit to the sink. Inthe next round, againtwo new forwarder nodes are selected based upon ξ. In this way, forwarder nodes rotate and all the nodes get a chance to become a forwarder. Therefore, energy is consumed more efficiently than in SIMPLE and M-ATTEMPT resulting in increased lifetime and throughput. In REEC, the routing load is shared between the two forwarders which results in efficient energy consumption of nodes. The forwarders are selected dynamically which results in fair load distribution. They collect the data from distant nodes and save their energy. Furthermore, the two forwarders are located in the upper and lower parts of the human body and collect data from their corresponding nodes as shown in fig

48 4.3.4 Scheduling Phase In this phase, forwarder nodes assign Time Division Multiple Access (TDMA) based time slots to their corresponding nodes. The nodes send their data to the forwarders Ψ α or Ψ β in their allocated time slots. Proper scheduling of nodes minimizes their energy consumption. It also avoids collision to achieve better network throughput Data Transmission Phase The initial energy o of all nodes is 0.5 J. The nodes send only critical data. The forwarder nodes aggregate the received data and route it to the sink. If a node possesses energy less than a threshold τ, it communicates directly to the sink. In addition, it does not further take part in the selection of forwarder. This is done to avoid energy consumption in data aggregation. If a node has shorter distance to the sink than forwarder, it routes its data directly to the sink. The nodes from α send their data to Ψ α and nodes from β send their data to Ψ β. The flowchart of the proposed protocol is shown in fig Energy Consumption Analysis In this section, we develop equations for single-hop and multi-hop communications. Energy consumed for single-hop communication is: SH = TX (4.15) Here, TX is the transmission energy as given by: TX = κ ( elect +ε amp ) l 2 (4.16) The energy consumed during multi-hop communication is given by: MH = κ[ℵ ( TX )+(ℵ 1) ( RX + da )] (4.17) Where, da is the data aggregation energy and ℵ is the number of nodes in the network. 32

49 Start Scan Body No No If critical value Yes If node from No If critical value Yes Yes No If energy(j )> Yes Yes If energy(j )> Yes Yes No No No If 0<energy(J) If dis_sink dis_ If dis_sink dis_ If 0<energy(J) Yes Yes No No Yes Yes Send data to Send data to Send data to the sink End dis_sink: Distance of node from sink dis_ : Distance of node from dis : Distance of node from Figure 4.2: Flowchart of REEC 4.5 Experiments and Discussions In order to verify the performance of REEC, simulations are performed five times and average results are plotted. Table 4.2 presents the simulation parameters. We ignore the sensing energy consumed by the nodes in simulation. We assume that the probability of critical data is 70%. In the simulation of REEC, we set the value of τ as 20% of o. We study the performance of the proposed protocol in comparison with SIMPLE and M-ATTEMPT. Different performance metrics of REEC are evaluated and are discussed in the following subsections. 33

50 Table 4.2: Simulation Parameters Parameter Value Units TXelect 36.1 nj/bit TXelect 16.7 nj/bit ε amp 1.97 nj/bit/m n da 5 nj/bit l o 0.1 m κ 4000 bits ν 2.4 GHz o 0.5 J Stability Period and Network Lifetime The network lifetime of the proposed protocol is shown in fig Our protocol selects two forwarders in each round which aggregate the data of other nodes and route it to the sink. The proposed protocol has 25% and 159% improved stability period than SIMPLE and M-ATTEMPT, respectively. It shows that energy of all the nodes is consumed uniformly. Due to efficient energy usage, the proposed protocol also achieves the high network lifetime of about rounds REEC SIMPLE M ATTEMPT No. of dead nodes Rounds Figure 4.3: Comparison of stability period and network lifetime in REEC, SIMPLE and M-ATTEMPT 34

51 4.5.2 Throughput Throughput is the number of packets successfully received at sink. In WBANs, routing protocols are needed which give high network throughput for reliable monitoring of the patients, elderly peopole, etc. REEC consumes energy efficiently resulting in longer network lifetime. The nodes are alive for longer time and send more packets that leads to increased throughput. We use Random Uniform Model [53] for packet drop calculation. The status of communication link can be good or bad depending upon the probability. We suppose the probability of link status to be good is 0.7. The proposed protocol gives better throughput than SIMPLE and M-ATTEMPT as shown in fig x REEC SIMPLE M ATTEMPT Packets received at sink Rounds Figure 4.4: Comparison of network throughput (aggregated) in REEC, SIMPLE and M-ATTEMPT Residual Energy The residual energy of the network is shown in fig The forwarder nodes Ψ α and Ψ β receive the data of their corresponding nodes and route it to the sink. As nodes send critical data to the nearest forwarder node, so less energy is consumed and they stay alive for longer time. In REEC, the energy of nodes depletes slowly as shown in fig

52 4 3.5 REEC SIMPLE M ATTEMPT Residual energy (J) Rounds Figure 4.5: Comparison of residual energy in REEC, SIMPLE and M-ATTEMPT Path Loss Path loss is the difference between the transmitted and received power represented in decibels (dbs). The posture of the human body affects the electromagnetic signals. As a result, path loss shows different behaviour along different body parts. There are different models used to estimate the path loss. Path loss is a function of distance and frequency as shown below: Γ(ν,l) = Γ o +10.n.log 10 ( l l o )+Xσ (4.18) Where, Γ o is path loss at reference distance l o and n is path loss exponent. The distance between transmitter and receiver is l and ν is the frequency. X is a gaussian random variable and σ is the standard deviation [63]. Path loss at reference distance l o can be expressed as: Here, λ is wavelength of electromagnetic waves. ( ) 2 4.π.lo Γ o = 10.log 10 (4.19) λ Fig. 4.6 shows the path loss in each round for the proposed protocol. In simulation, we use a fixed ν of 2.4 GHz from ISM band. We use the values of n and σ as 3.38 and 4.1, respectively. The improvement in percentage provided by the proposed protocol to M-ATTEMPT 36

53 REEC SIMPLE M ATTEMPT Path Loss (db) Rounds Figure 4.6: Comparison of path loss in REEC, SIMPLE and M-ATTEMPT and SIMPLE is shown in table 4.3. Table 4.3: Improvement in Percentage Parameter Improvement (%) Improvement (%) in M-ATTEMPT in SIMPLE Stability period Network lifetime Throughput Average residual energy Average path loss

54 Chapter 5 BEC: A Novel Routing Protocol for Balanced Energy Consumption in Wireless Body Area Networks 38

55 5.1 Motivation In WBANs, balanced energy consumption of nodes helps to monitor the vital signs of the human body for increased time period. OINL has increased network lifetime due to the balanced energy consumption of nodes. It collects link-cost periodically at the sink, where all routing decisions are performed. In OINL, nodes send the data via routes that have minimum cost. The cost function of OINL is given as: C i j,k = RSSI T α j,k ) E k M 1+( i Ei min (5.1) 2 Where, RSSI T is the target RSSI value required to achieve reliable communication and α j,k is the channel attenuation for the link between j and k. E k i is the accumulated energy of node i at round k and Ei min is the minimum accumulated energy across all nodes. In eq. 5.1, M 0 which shows the effect of imbalanced energy consumption. Eq. 5.1 transforms to conventional cost function when M = 0, which is the power required to transverse a link regardless of the accumulated energy of nodes. OINL protocol is discussed in chapter 2 in detail. However, one deficiency of OINL is that it results in increased energy consumption of nodes in data reception and aggregation. As data is routed through shortest path, so intermediate nodes may be involved in data reception and aggregation. It results in increased energy consumption of nodes near the sink. Table 5.1: Energy Parameters of Transceivers Parameter nrf 2401A CC2420 Units DC current (TX) ma DC current (RX) ma Supply voltage (min.) V E TXelect nj/bit E RXelect nj/bit ε amp nj/bit/m n 5.2 Analysis of Energy Consumption In WBANs, nodes consume different amount of energy in single-hop and multi-hop communications. Energy consumption in a single-hop communication is given as: E sh = E TX (5.2) 39

56 Where, E TX is the transmission energy which is calculated as: E TX = (ε amp +E elect ) s d 2 (5.3) Where, ε amp is the energy consumed by the amplifier and E elect is the energy consumed by the electronic circuit. The packet size is denoted by s and d shows the distance between the node and the sink. On the other hand, energy consumption in multi-hop communication is given as: [ E mh = s n E TX +(E DA +E RX ) (n 1) ] n (5.4) In eq. 5.4, n is the number of hops and E DA is the energy consumed in data aggregation. E RX is the energy consumed in data reception and we assume that E TX = E RX. 5.3 BEC: The Proposed Protocol In this section, we discuss the proposed routing protocol. The detail is given in the following subsections Radio Model A number of radio models are proposed in the literature. We use first order radio model [64] given as: E TX (s,d) = E TXelect (s)+ε amp (s,d) (5.5) E TX (s,d) = E TXelect.s+ε amp.s.d 2 (5.6) E RX (s,d) = E RXelect (s) = E RXelect.s (5.7) In WBANs, the human body contributes attenuation to the radio signals. Therefore, a path loss coefficient parameter n is included in the radio model. The expression for the energy consumption is given as: E TX (s,d) = E TXelect.s+ε amp.s.d n (5.8) Different types of sensors are available for the monitoring of physiological parameters of human body in WBANs. Table 5.1 shows the energy parameters of two 40

57 transceivers which are widely used in WBAN technology Placement of Nodes In BEC, eight nodes are placed on the human body. All nodes have equal initial energy (i.e. nodesarehomogeneous). The sink isplacedonthechest ofthehuman body as shown in fig Table 5.2 shows the distances between the nodes and Node Sink OINL BEC Figure 5.1: Placement of nodes on the human body and mechanism for path selection in OINL and BEC the sink. Table 5.2: Distances of nodes from the sink Node Distance (m)

58 5.3.3 Start-up Phase In this phase, the sink broadcasts a HELLO packet to all the nodes. Each node receives this packet and stores the location of the sink. Then each node broadcasts a packet which contains the ID of a node, its location and the value of the residual energy. In this way, all nodes are updated with the location of neighbouring nodes, position of the sink and possible routes to the sink. Fig. 5.2 depicts the format of the HELLO packet. t Position ation ation Figure 5.2: Format of the HELLO packet in BEC Routing Phase In this phase, nodes select their path to the sink. The nodes closer to the sink send their data directly to the sink. However, the nodes far away fromthe sink use intermediate (relay) nodes to route the data. The mechanism of the path selection for the proposed protocol is shown in fig OINL selects the path with minimum attenuation and more nodes are involved (see fig. 5.1) which results in more energy consumption in the form of data reception and aggregation, so nodes die quickly. On the other hand, the proposed protocol selects a path with suitable number of intermediate nodes and successfully routes the data to the sink. This way, less energy is consumed and nodes stay alive for a long time. The cost function used in the proposed routing scheme is given as: C(i) = 1 R.E(i) (5.9) Where, C(i) is the cost of node i. In eq. 5.9, R.E(i) represents the residual energy of node i. In BEC, a node having minimum cost is selected as a relay node. In this way, balanced energy consumption results in increased network lifetime. There is a trade off between having only critical (emergency) data for long term and normal (continuous) data for small time period. As critical patients need 42

59 immediate medical treatment, therefore, critical data is routed without any hindrance. In BEC, we implement reactive routing when the energy of nodes decreases below a threshold (τ) Scheduling Phase In this phase, the sink assigns Time Division Multiple Access (TDMA) based time slots to all the nodes. All the nodes use the same frequency band and transmit their data in different time slots. The nodes send their data in their scheduled time slots to avoid any collision Data Transmission Phase The initial energy of all nodes is the same (i.e. E o = 0.5 J). The nodes sense the vital parameters of the human body and send data to the sink continuously. However, after the nodes are left with energy less than τ, the proposed protocol uses reactive routing. Therefore, human vital parameters are monitored for long term. 5.4 Experiments and Discussions In order to verify the performance of the proposed protocol, simulations are performed five times and average results are plotted. Table 5.3 shows the simulation parameters. We ignore the sensing energy consumed by the nodes in our simulation. Furthermore, we assume that 30% of the data is critical. We study the performance of the proposed protocol in comparison with OINL. The following subsections contain the detail of different performance parameters. Table 5.3: Simulation Parameters Parameter Value Units E RXelect 36.1 nj/bit E TXelect 16.7 nj/bit ε amp 1.97 nj/bit/m 2 E DA 5 nj/bit τ 0.1 J d o 0.1 m s 4000 bits f 2.4 GHz E o 0.5 J 43

60 5.4.1 Stability Period and Network Lifetime The stability period is the time from the start of the network till the death of the first node. On the other hand, network lifetime shows the time from the start of the network till the death of the last node. The proposed routing scheme selects relay nodes on the basis of cost value. The node having the minimum cost value is selected as a relay node for data transmission. Therefore, nodes exhibit a uniform energy consumption which increases the network lifetime. Our proposed protocol sends data to the sink by consuming less energy. BEC has 49% improved network lifetime than OINL. It shows that the energy of all the nodes is efficiently consumed. Due to efficient energy usage, the proposed protocol achieves increased network lifetime. Fig. 5.3 shows the comparison of the stability period and the network lifetime. It is evident that the BEC achieves improved stability period and network lifetime. Figure 5.3: Comparison of stability period and network lifetime in BEC and OINL Network Throughput The throughput is the number of packets successfully received at the sink per unit time. The proposed protocol consumes energy efficiently resulting in longer network lifetime. The nodes are alive for longer time and send more packets that leads to increased throughput. In this work, we use a random uniformed model [64] for packet drop calculation. The status of the communication link can be good or bad depending upon the probability. We assume the probability of 0.7 for 44

61 the link status to be good. BEC offers increased throughput than OINL as shown in fig The throughput of the proposed protocol decreases after 5245 rounds. It is due to the fact that nodes are left with energy less than τ and only critical data is routed. Therefore, BEC has increased the network lifetime at the cost of lower throughput after 5245 rounds (see figs. 5.3 and 5.4). Packets received at sink 4 x OINL BEC Rounds Figure 5.4: Comparison of network throughput in BEC and OINL Residual Energy The residual energy of the network in the proposed routing scheme is shown in fig The intermediate nodes receive the data of their corresponding nodes and route it to the sink. As nodes send critical data to the nearest forwarding nodes, so less energy is consumed and they can stay alive for longer time. Fig. 5.5 shows that initially OINL and BEC have the same residual energy. However, after 5245 rounds the proposed scheme offers better residual energy curve than OINL due to reactive routing strategy Path Loss Path loss is the difference between the transmitted and received power represented in decibels (dbs). The posture of the human body affects the electromagnetic signals. As a result, the path loss shows different behaviours along different body 45

62 2 1.8 OINL BEC Residual energy (J) Rounds Figure 5.5: Comparison of residual energy in BEC and OINL parts. There aredifferent models used toestimate thepathloss which isafunction of distance and frequency as: PL = PL o +10.n.log 10 ( d d o )+σ s (5.10) Where, PL o is the path loss at reference distance d o and n is the path loss exponent. The distance between the transmitter and the receiver is d and σ s is the standard deviation [64]. The path loss at reference distance d o can be expressed as: ( ) 2 4.π.do PL o = 10.log 10 (5.11) λ Here, λ is the wavelength of the electromagnetic waves. In our simulation, we use a fixed frequency (f) of 2.4 GHz from Industrial, Scientific and Medical (ISM) radio band. We use the values of n and σ s as 3.38 and 4.1, respectively. Fig. 5.6 shows the path loss in each round for OINL and BEC. We observe that after 5245 rounds the path loss exhibits continuous fluctuations. These fluctuations are due to reactive routing in which data is not sent if it is not critical (i.e. normal data). In this way, there is no path loss in some rounds and the path loss curve goes to zero (see fig. 5.6). The improvement in the percentage provided by 46

63 OINL BEC 400 Path loss (db) Rounds Figure 5.6: Comparison of path loss in BEC and OINL BEC as compared to OINL is shown in table 5.4. Table 5.4: Improvement in Percentage Parameter Improvement (%) in OINL Stability period Network lifetime 49 Network throughput 0.6 Average residual energy Average path loss

64 Chapter 6 Mobility Modeling for Wireless Body Area Networks 48

65 6.1 Motivation Mobility models have a big impact on the accuracy of simulations in WBANs. Although a number of mobility models for ad-hoc networks are proposed in existing literature, they are not suitable for WBAN because of its limited area and small communication range. The models in WABNs use certain mobility models (like RPGM, etc.) for moving the logical center of the group and the individual nodes. It is not necessary that all the nodes in WBANs follow the logical center. The model in[65] does not specifically implement different postures of human body. Postures are of great importance in WBANs as the network topology may entirely change due to their changes. 6.2 Mobility Modeling In WBANs, nodes are deployed on the human body to monitor different physiological parameters like, blood pressure, temperature, heart beat level, etc. These nodes send their sensed data to the sink placed on the chest of the human body. The distances between nodes and sink are constant in static position. However, as the human body is mobile in reality, so, the distance between node and sink changes. Mobility models of WSNs are not suitable for WBANs due to limited area and small communication range in the later. Furthermore, they do not consider different postures of the human body. In this work, we consider different postures and propose a method to calculate the distances between nodes and sink when the human body is in motion. We devise a mechanism consisting of two phases; (i) Posture selection phase and (ii) Nodes movement phase. In the posture selection phase, a posture of the human body is selected like, standing, sitting, laying, walking, and running. The probability of posture change can be determined from real human mobility traces. However, we take probabilities of different postures from [65] as shown in fig Markov chain in the figure shows the probability of posture change from one state to another after a fixed time interval defined by the user. After posture change, the new position of nodes is selected in the second phase. We assume that sink is placed on the chest of human body and all positions of nodes are measured relative to it. The following sections discuss the different postures of human body in detail. 49

66 LAY 0.2 SIT 0.2 STAND 0.4 WALK 0.7 RUN Figure 6.1: Markov model for posture pattern selection Standing In this position, the distances between nodes and sink are constant as body is in static position Sitting In this posture, we assume that the human body is sitting on a chair. In this position, there is little movement of trunk of the human body. Most of the time, the human arms and legs exhibit motion in three dimensions. We calculate the positions of nodes placed on arms and legs. As nodes placed on arms show similar behaviour, so, we calculate the position of a single node placed on elbow. Similarly, we calculate the position of node placed on knee. The node e is placed on the elbow while node k is placed on the knee. The normal position of e in sitting position is given as: P e = P(ρ e,θ e,φ e ). (6.1) Where, ρ e is the radial distance, θ e is polar angle and φ e is azimuthal angle of e from sink. During movement of the human arm, the maximum and minimum distances between e and sink in sitting position are ρ emax and ρ emin, respectively. So, the difference between these distances is: d e = ρ emax ρ emin. (6.2) We form a sphere at a distance of ρemax+ρ emin from sink as shown in fig This 2 sphere has a radius of de. Now, during movement, the node e will always lie in 2 50

67 this sphere. The new position of node e is calculated using the following equation. ρ e (t) = ρ e (t 1)+(η e rand(1) ζ e ). (6.3) Where, η e is given as: [ 1 0] if ρ e (t) = ρ emax η e = [0 1] if ρ e (t) = ρ emin (6.4) [ 1 1] if ρ emin < ρ e (t) < ρ emax Fromeq. 6.3, it is clear that the new position of a node depends upon the previous position. A random number is added to the current location to find new location. If the new position of node goes out of bound then η e will decrement the distance between node andcenter of the sphere. Onthe other hand, if the distance between node and sink approaches ρ emin, then η e will be positive and it increases the distance (see eq. 6.4). In eq. 6.3, ζ e is the step size which can be adjusted according to the application. Its value is always greater than zero. As the main concern in WBANs is the distance between nodes and sink, so, we will not calculate other parameters like, θ e and φ e. It should be kept in mind that these values will also change according to eq Now, we discuss the movement of node k placed on the knee of the human body. The normal position of k in sitting position is given as: P k = P(ρ k,θ k,φ k ). (6.5) Here, ρ k is the normal distance of k from sink. θ k and φ k represent the polar and azimuthal angles, respectively. During movement of human body, the maximum and minimum distances between k and sink are denoted by ρ kmax and ρ kmin, respectively. d k = ρ kmax ρ kmin. (6.6) We form a sphere at a distance of ρ kmax+ρ kmin 2 from sink as shown in fig This sphere has radius of d k 2. The node k always lie in this sphere during movement. The new position of k is calculated using the following equation. ρ k (t) = ρ k (t 1)+(η k rand(1) ζ k ). (6.7) 51

68 The value of η k is calculated using: [ 1 0] if ρ k (t) = ρ kmax η k = [0 1] if ρ k (t) = ρ kmin (6.8) [ 1 1] if ρ kmin < ρ k (t) < ρ kmax The new position of k is calculated using eq. 6.7 where η k is a random number which is calculated using eq ζ k represents the step size and is adjusted according to the application. Its value is always greater than zero. Sphere of Radius dk/2 Sphere of Radius de/2 Figure 6.2: Human body in sitting position Walking During walking, the arms and legs of human show repetitive and similar movement patterns. When the left arm moves forward, the right leg also moves in the forward direction. Similarly, right arm and left leg are synchronized. This defined trajectory helps to efficiently model the mobility of human body. When the body moves from static position, the new position of sink is given as: P s = P(ρ s,θ s,φ s ). (6.9) 52

69 Where, ρ s is calculated as: ρ s (t) = ρ o +tu. (6.10) Where, u denotes the speed of the human and t is the time after which we are calculating new position. ρ o denotes the initial position of sink. The normal position of node e is given as: P e = P(ρ e,θ e,φ e ). (6.11) Let us denote the distances between sink and node in forward and backward positions by ρ front e and ρ back e, respectively. We assume their magnitudes are same. and ρ back e as shown in fig The node e will So, we form a curve between ρ front e move along this curve and its position at any time t is calculated as: ρ e (t) = ρ e (t 1)+η e d e. (6.12) Where, d e is calculated as: d e = ρ front e ρ e = ρ back e ρ e. (6.13) The value of η e changes with time as shown in fig Its value ranges from 0 to 1. Now, we see the movement of nodes placed on legs. The normal position of node k is given as: P k = P(ρ k,θ k,φ k ). (6.14) During walking, the legs move in the forward and backward directions. We denote thedistancesbetweensinkandnodek inforwardandbackwarddirectionsbyρ front k and ρ back k, respectively. We assume that these two distances are same and form a curve between them as shown in fig The moving node k always lies on this curve. The new position of k is calculated as: ρ k (t) = ρ k (t 1)+η k d k. (6.15) In the above equation, d k is calculated as: d k = ρ front k The value of η k changes with time as shown in fig ρ k = ρ back k ρ k. (6.16) 53

70 Figure 6.3: Human body in walking position Running In the running position, there are repetitive movements of certain limbs of the body such as arms and legs, similar to walking position. The arms and legs undergo continuous movements in forward and backward directions. We find the position of nodes placed on arms and legs of the human body during running. In the running position, sink also changes its position in each time interval. The new position of sink at time t is given as: P s = P(ρ s,θ s,φ s ). (6.17) Where, ρ s is calculated as: ρ s (t) = ρ o +tu. (6.18) Where, ρ o is the initial position of sink, u is its speed and t the time after which its new position is calculated. The normal position of node e in running position is given as: P e = P(ρ e,θ e,φ e ). (6.19) During running, the node e moves in the forward and backward direction continuously. Let ρ front e and ρ back e denote the distances of sink from node in forward and backward directions. We assume that both of these distances are same. So, 54

71 we form a curve centered at ρ e having length of ρ front e ρ e in each direction (i.e. forward and backward) as shown in fig During running, the nodes always lie on this curve. The position of e at any time instant is calculated as: ρ e (t) = ρ e (t 1)+η e d e. (6.20) In the above equation, d e is calculated as: d e = ρ front e The value of η e changes with time as shown in fig ρ e = ρ back e ρ e. (6.21) Now, wediscussthemovement ofnodek placedonrightknee. Thenormalposition of node k is given as: P k = P(ρ k,θ k,φ k ). (6.22) Let ρ front k and ρ back k denote the distances between sink and node in forward and backward directions. We assume that these two distances are equal. We form a curve centered at η k having length of ρ front k ρ k in each direction (i.e. forward and backward) as shown in fig The node always moves along this curve during motion. Its position at any time t is calculated as: ρ k (t) = ρ k (t 1)+η k d k. (6.23) In the above equation, d k is calculated as: d k = ρ front k The value of η k changes with time as shown in fig ρ k = ρ back k ρ k. (6.24) Laying In thelaying position, nodes placed onthe trunk of the human bodyareminimally mobile. Ontheother hand, nodesplaced onarms andlegs aremobile. The normal position of node e placed on the elbow is given as: P e = P(ρ e,θ e,φ e ). (6.25) Let us denote the minimum and maximum distances between sink and node e by ρ emin and ρ emax, respectively. So, we make a sphere at distance of ρemax+ρ emin 2 from sink and having radius of ρemax ρ emin 2 as shown in fig Now, the position of 55

72 Figure 6.4: Human body in running position node is calculated using following equation: ρ e (t) = ρ e (t 1)+(η e rand(1) ζ e ). (6.26) Where, η e is given as: [ 1 0] if ρ e (t) = ρ emax η e = [0 1] if ρ e (t) = ρ emin (6.27) [ 1 1] if ρ emin < ρ e (t) < ρ emax In eq. 6.26, ζ e is the step size and its value is always greater than zero. Now, we calculate the new position of node k in laying position. The normal position of node k in standing position is given as: P k = P(ρ k,θ k,φ k ). (6.28) During laying, the nodes placed on legs show random mobility. Let the maximum and minimum distances between sink and node k are denoted by ρ kmax and ρ kmin respectively. We form a sphere at distance of ρ kmax+ρ kmin from sink and having 2 radius of ρ kmax ρ kmin, as shown in fig The new position of node is calculated 2 56

73 as: ρ k (t) = ρ k (t 1)+(η k rand(1) ζ k ). (6.29) Here, the value of η k is calculated as: [ 1 0] if ρ k (t) = ρ kmax η k = [0 1] if ρ k (t) = ρ kmin (6.30) [ 1 1] if ρ kmin < ρ k (t) < ρ kmax In eq. 6.29, ζ k is the step size and its value is always greater than zero. It determines the distance covered in a single time interval. In laying position, larger value of ζ k is selected as nodes move suddenly to larger distances and, after longer time intervals. Sphere radius Sphere radius (ρ kmax -ρ kmin )/2 (ρ emax -ρ emin )/2 Figure 6.5: Human body in laying position 1 time Figure 6.6: Value of η e and η k 57

74 6.3 Impact of Mobility in WBANs During the movement of human body, nodes move to different positions and therefore, distance between sink and node changes. It affects the energy consumption of nodes, propagation delay and path loss of the signal. In the following sections we discuss them in detail Energy Consumption As shown in the mobility model, the distance between nodes and sink changes with the movement of human body. As a result, transmission energy consumption of nodes changes as given in [60]: E TX (k,d) = E TX elec k +ǫ amp k d 2. (6.31) Where, E TX is the transmission energy, E TX elec is the energy required to run the electronic circuit and ǫ amp is the energy required to run the amplifier. k is the packet size and d is the distance between sink and node. It is clear from eq and fig. 6.7 that as the distance between node and sink increases, the energy consumption also increases x 10 4 Energy consumption of nodes (J) Distance between node and sink (m) Figure 6.7: Effect of distance on energy consumption of nodes 58

75 6.3.2 Delay It is time required by a signal to reach from source to destination. Distance between sink and node affects the delay as: delay = d c. (6.32) Where, d is the distance between sink and node and c is the speed of electromagnetic waves. Delay increases with the increase in distance as shown in fig x Delay (s) Distance between node and sink (m) Figure 6.8: Effect of distance on delay Path loss Path loss is the reduction in power density of a wave as it propagates through space. It depends on distance as given in [62]: PL(f,d) = PL o +10nlog 10 ( d d o )+σ s. (6.33) Where, PL o isthepathlossatreferencedistanced o andnisthepathlossexponent whose value varies from 4 to 7 for human body. d is the distance between node and sink (transmitter and receiver) and σ s is the standard deviation. 59

76 PL o is given as: Similarly, it can be written as: ( ) 2 4πd PL o = 10log 10. (6.34) λ ( ) 2 4πdf PL o = 10log 10. (6.35) c Where, f is the frequency, λ isthe wavelength of the propagatingwave andcis the speed of light. We use frequency of 2.4 GHz from ISM band. Path loss increases with the increase in distance as shown in fig It is obvious from the above discussion that distance between nodes and sink Path loss (db) Distance between node and sink (m) Figure 6.9: Effect of distance on path loss affects the energy consumption, delay and path loss. So, in order to find these parameters correctly, we propose and implement the mobility model in our protocol. In this way, it gives more accurate and realistic results. 6.4 Implementation of Mobility Model in the Routing Protocols Wireless Body Area Network (WBAN) consists of nodes placed on the human body to monitor different vital signs like heart rate, glucose level, blood oxygen level, etc. We propose two new routing protocols and discuss their advantages and disadvantages. We discuss their functionality in the following sections in detail. 60

77 6.5 Energy Consumption Analysis Energy consumed in single-hop communication is given as: E SH = E TX. (6.36) Where, E TX is the transmission energy which is calculated as: E TX = k (E TXelect +ε amp ) d 2. (6.37) k is packet size, E TXelect is energy consumed by the electronic circuit, ε amp is the amplification energy and d is the distance between transmitter and receiver. Energy consumed in multi-hop communication is given as: E MH = k (h E TX +(h 1)(E RX +E DA )). (6.38) Where, h is the number of hops, E RX is the energy consumed in receiving the data and E DA is the data aggregation energy. 6.6 Multi-hop Technique In multi-hop routing technique, data is transmitted using neighbouring nodes. Fig shows the placement of nodes on human body. In multi-hop scheme, node 4 sends data to node 1 and node 3 sends data to node 2. Similarly, nodes 7 and 8 send their data to nodes 5 and 6 respectively. The receiving nodes (i.e. nodes 1, 2, 5 and 6) send the aggregated data to sink. If these receiving nodes become dead then the other nodes send their data directly to the sink as shown in fig In this scheme, the far away nodes send data to their neighboring nodes and thus save energy. However, the drawback of this scheme is that nodes near the sink are burdened with heavy load. They consume extra energy in aggregating and receiving the data from other nodes. In this way, they deplete their energy soon, and become dead nodes. 6.7 Data Transmission using Forwarder Nodes In this routing technique, forwarder nodes are selected in each round. These forwarders receive data from their respective group members and forward it to the sink. Fig shows the placement of nodes on the human body. We discuss this protocol in the following sections in detail. 61

78 1 2 Sink Node Figure 6.10: Placement of nodes on the human body Initialization phase In this phase, sink broadcasts a HELLO message containing the following information: Location of sink. Location of neighbors. Information about all possible routes to the sink. All nodes receive this HELLO message and update their routing table Forwarders selection phase In this phase, forwarders are selected to route the data of other nodes. We divide N number of nodes into two sets; A and B, based on their distance from sink, which are given as: N = {1,2,3,4,5,6,7,8} (6.39) 62

79 e e Normal data routing Routing after the death of nodes near the sink Figure 6.11: Network flow tree in multi-hop routing scheme A = {1,2,3,4} (6.40) B = {5,6,7,8} (6.41) In the forwarders selection phase, two forwarder nodes are selected (one from each group) on the basis of cost functions C.F A and C.F B, which are calculated as: C.F A = d(i) R.E(i). i A (6.42) C.F B = d(i) R.E(i). i B (6.43) The node having minimum value of C.F A is selected as a forwarder node from group A. Similarly, the node having minimum value of C.F B is selected as forwarder node from group B. These forwarder nodes collect the data from their respective group members and send it to the sink. 63

80 6.7.3 Scheduling phase In the scheduling phase, forwarders assign Time Division Multiple Access(TDMA) based time slots to their children nodes. All the nodes transmit in their scheduled time slots to avoid collision Data transmission phase In the data transmission phase, nodes transmit data to their respective forwarder nodes in their scheduled time slots. Forwarder nodes receive data from their children nodes, aggregate it and route it to the sink. If a node has less energy than a threshold (τ), it does not take part in forwarders selection and routes its data directly to the sink. This is incorporated to save the data aggregation energy of low energy nodes. If a node has less distance to the sink than forwarder then it routes its data directly to the sink. Fig shows the network tree for forwarder based routing technique. In the initial rounds, nodes send data to their respective forwarders which route it to the sink. However, after some rounds, some nodes may have less energy than others as shown in fig For example, if node 5 has less energy than τ, it sends its data directly to the sink and all other nodes route their data through forwarder nodes. 6.8 Simulation Results and Analysis We simulate the proposed protocols and analyze their results. Table 5.1 shows the simulation parameters and their values. We implement the proposed mobility model in the two routing protocols and assume the values of ρ e and ρ k as 0.15 and 0.20, respectively. We ignore the sensing energy consumed by the nodes in simulation. The initial energy (E o ) of all nodes is 0.5 J. The simulations are run five times and their average results are plotted Network lifetime Network lifetime represents the time from the start of network till the death of last node. On the other hand, the time from start of the network till the death of first node is called stability period. Fig shows the comparison of number of dead nodes and fig shows the comparison of stability period and network lifetime. Forwarders based routing protocol has larger stability period and network lifetime. It is due to the fact that new forwarders are selected in each round and 64

81 rder rder of set A τ tree in initial rounds tree after the death of node 8 Figure 6.12: Network flow tree in forwarder based routing scheme the load is uniformly distributed to all the nodes. On the other hand, in multihop routing protocol, nodes near the sink are heavily burdened and consume more energy in the form of reception and data aggregation energy. As a result, these nodes die quickly. Multi-hop routing protocol has stability period of 1191 rounds and network lifetime of about 2500 rounds. On the other hand, forwarders based routing scheme has stability period of 3913 rounds and network lifetime of 6878 rounds as shown in fig Throughput Throughput shows the number of packets successfully received at sink. A protocol having longer network lifetime sends more packets to the sink and have higher throughput. Fig shows the number of packets sent to the sink in the multihop and forwarders based routing protocols. As forwarders based routing protocol haslongernetworklifetime(see fig. 6.14),so,itsendsmorepacketstothesink. All of the sent packets are not successfully received at sink. We use random uniformed model [64] to calculate the number of dropped and received packets. The status 65

82 Table 6.1: Simulation Parameters Parameter Value Units E RXelect 36.1 nj/bit E TXelect 16.7 nj/bit ε amp 1.97 nj/bit/m 2 E DA 5 nj/bit/signal d o 0.1 m τ 0.2 J k 4000 bits f 2.4 GHz E o 0.5 J 8 7 Multi hop Forwarder based No. of dead nodes Rounds Figure 6.13: Comparison of number of dead nodes in multi-hop and forwarder based routing techniques of the communication link can be good or bad. We assume the probability of 0.7 for link to be good. Figs and 6.17 show the number of packets dropped and successfully received at sink, respectively. It is clear from fig that multihop routing technique continue sending packets to sink till 2500 rounds whereas forwarders based routing technique sends data to the sink till 6878 rounds Residual energy Comparison of residual energy of the multi-hop and forwarders based routing protocols is shown in fig As nodes near the sink consume more energy in multi-hop routing, so, they deplete their energy soon. On the other hand, nodes in 66

83 Figure 6.14: Comparison of stability period and network lifetime in multi-hop and forwarder based routing techniques the forwarder based routing protocol consume less energy and stay alive for longer time. Fig shows the gradual decrease in the residual energy of forwarder based routing protocol. Whereas, in the multi-hop routing protocol the residual energy decreases more quickly Delay Delay is the time required by a signal to reach from source to destination. It varies according to the distance between source and destination as given in eq Fig shows the delay for multi-hop and forwarders based routing protocols. It is clear from the figure that multi-hop routing protocol has less delay as compared to forwarders based routing protocol. It is due to the reason that in multi-hop routing technique, nodes send their data using neighbouring nodes. As these neighbouring nodes are located at a small distance, therefore, less delay is occurred. On the other hand, in forwarders based routing protocol, nodes send their data to a forwarder which can be located at a large distance. As new forwarders are selected in each round, therefore, they may be located far away from other nodes. As a result, large delay will occur due to larger distance between source and destination. The fluctuations in delay are due to the different distances between nodes and sink as the body is mobile. During routine activities, body exhibits different postures and the distances between nodes and sink vary according to that posture. As delay depends on distance, so, it changes in each round. Furthermore, delay starts decreasing after 3913 rounds in forwarders based routing protocol as nodes start dying and therefore, less number of nodes have lower cumulative delay. Similarly, 67