body sensor networks - universiteit twenteessay.utwente.nl/65921/1/karuppiahramachandran_ma_ewi.pdfi...

Master’s thesis on

Characterization of Communication Mechanismsfor Implantable Body Sensor Networks

Focusing on Physical layer and Medium Access Control Sub-layer

Vignesh Raja Karuppiah Ramachandran.

In partial fulfillment of the requriements for the degree of

Master of Sciencein

Embedded Systems

Faculty of Electrical Engineering, Mathematics and Computer Science (EEMCS)Pervasive systems Research group.

Thesis Committee:

Prof.dr.ir. Paul Havinga Chair, Pervasive systemsDr. Nirvana Meratnia Associate Professor, Pervasive systemsIr. Bert Molenkamp Lecturer, Computer Architecture for

Embedded Systems

Enschede, August 2014

Abstract

The use of wireless sensor networks (WSN) in health-care has been rapidly increased in thelast few years. Miniaturization of sensor nodes, in-body data communication, bio-compatibilityare the main outcomes of the on-going extensive research on nano-technology, wireless commu-nication and bio-medical engineering, respectively. Sensing of various life-critical physiologicalsignals such as heart-rate, blood-pressure, blood-glucose level, is made possible with miniaturizedimplantable wireless bio-sensors. Challenging requirements of WSN in health-care applicationshave resulted in the advent of a specific type of WSN called Implantable body sensor network(IBSN), in which the sensor nodes are implanted either subcutaneously or by an invasive surgeryinto the patient’s body. These sensor nodes continuously monitor the physiological signals whichis necessary for the patients with life-threatening diseases such as epilepsy. Life-critical implant-able medical devices (IMD) such as pace-makers and neural stimulators, are also connected to theIBSN. A closed control loop of medical devices is envisioned through a network of IMD and bio-sensors using IBSN, in which the IMDs are programmed for different types of therapies throughwireless channel, based on the real-time response of the patient by continuously monitoring themedical symptoms with the implantable bio-sensors.

IBSN is less researched compared to the body sensor network (BSN) where the wireless com-munication between sensor nodes takes place on the surface of the body. Extensive research oncharacterizing the communication mechanisms for IBSN is needed to standardize a reliable RFcommunication within the human body. In order to create a reliable and energy-efficient sensornetwork, two main layers of the Open System Interconnection (OSI) network model is required.The physical layer which is aimed at unification of the hardware requirements in a network toenable the successful transmission of data. The medium access control (MAC) sub-layer thatis aimed at controlling the access to the wireless channel, which directly affects the networkperformance and energy efficiency of the nodes. This master thesis focuses on characterizing thephysical layer and MAC sub-layer with different configurations of IBSN by means of softwaresimulation and hardware experimentation. Two main state-of-the-art mechanisms are focusednamely Medical Implant Communication Service (MICS) band specifications in the physicallayer and wake-up radio integration in the MAC sub-layer which enables globally standardizedcommunication strategies for IBSN and ultra-low power communication with reliable networkperformance respectively. These mechanisms are studied and characterized for different IBSNscenarios with a bio-medical implant in this thesis work. .

As a result, an optimum configuration of the physical layer and MAC sub-layer for the IBSNis found. The added-value of wake-up radio in MAC layer and the effect of MICS band config-urations in physical layer and MAC layer are identified. The evaluation results will also indicatethe potential drawbacks in the existing configurations at physical and MAC layers of BSN, andidentify why the existing BSN mechanisms cannot be used for IBSN scenarios. Possible solutionsto overcome these drawbacks are suggested for the future research work.

Keywords. Implantable body sensor network, Wake-up radio, Medium access control

iii

Acknowledgement

It was around August 2012 when I started the Master’s education in Embedded systems atthe University of Twente. I am always fascinated about wireless sensor networks, which is a partof embedded systems and is also widely applied in different industries for different scenarios. Inthis regard, I was fortunate to get introduced to Prof. Paul Havinga, by Ir. Andrea SanchezRamirez, during one of my courses called ”Energy efficient embedded systems”. Based on mywork in EEES course, I was given an opportunity to work as a student assistant in an EU-FP7project ”WiBRATE”, which involves application of WSN in industrial vibration monitoring. Icontinued working in the project for my intern-ship and also was given an opportunity to extendthe work as a part-time job during my thesis. It was during this tenure as a student assistant Iwas supervised and guided by Dr. Niravana Meratnia. With highly constructive comments andprogressive meetings together with Dr. Paul and Dr. Nirvana, I successfully proposed my ideaof research about in-body sensor networks and continued with this thesis work. Although theassistantship work was not related to my thesis, I was given an opportunity to explore my ownresearch interests. With their support, I was able to publish two international articles duringmy master study and of course a trip to Singapore for presentation. I was able to successfullycomplete this thesis overcoming all the difficulties. Even though this master thesis is just anexploration and characterization of existing wireless communication mechanisms, I hope withfurther research I can materialize the closed loop architecture for medical devices. I thank myprofessors for giving me another opportunity to continue with my research towards a PhD degree.

I must thank Dr. Niels Moseley for his continuous and valuable technical support and Dr.Berend Jan van der Zwaag, for his constructive feedback at my writing skills. I thank Ir. Kylezhang who developed the medical implants which were also used for the final hardware char-acterization in this thesis. I thank Ir. Saeid Yazdani for all his support in debugging codes attimes. Apart from being thanked, I must acknowledge them for being excellent colleagues forthe last one year. I also thank all the people of PS for making me feel comfortable at the office.

I thank my friends, Alex, Frank, Yoppy, Gebremedhin, Anantha, Hasib, Anand, Nolie,Ramesh, Morshed, for being with me at difficult times. Apologies, for not having a wholelist of names. I thank all of my friends who supported me either directly or indirectly during mymaster studies. Without all their support, staying far away from home and focusing on studieswould not have been possible

I thank my family and relatives for supporting me financially and emotionally throughout themaster education. And a special thanks to my amma, thambees,and logapa family for allowingme to stay abroad for years to complete my education. In-spite of all the difficult times, I thankyou all for letting me cherish my dreams.

Last but not the least, I thank God almighty for this wonderful life.

v

Contents

Contents vii

List of Figures xi

List of Tables xiii

1 Introduction 11.1 Implantable medical devices and sensors in health-care . . . . . . . . . . . . . . . 2

1.1.1 Implantable bio-sensors for monitoring physiological and contextual signalsin a human body . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.1.2 Drawback of current medical devices in health-care . . . . . . . . . . . . . 61.1.3 Challenges of IBSN in a closed loop architecture . . . . . . . . . . . . . . 7

1.2 Context of Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71.2.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

1.3 Research question . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81.3.1 Research approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

1.4 Outline of thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2 Requirements of PHY and MAC layers for IBSN 112.1 Communication mechanisms for IBSN . . . . . . . . . . . . . . . . . . . . . . . . 12

2.1.1 MAC protocols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122.2 Architectural framework for closed loop medical devices using IBSN . . . . . . . 142.3 Low power design and Power Scavenging . . . . . . . . . . . . . . . . . . . . . . . 14

2.3.1 Wake-up radio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142.3.2 Power Scavenging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2.4 Application scenario and requirements for MAC protocols in IBSN . . . . . . . . 18

3 Survey of MAC protocols with and without wake-up radio for implantablesensor network 213.1 Features of MAC protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3.1.1 Attributes of MAC as proposed by IEEE 802.15 TG-6 . . . . . . . . . . . 223.2 Requirements of MAC protocol for IBSN with different medical devices . . . . . 223.3 Taxonomy of MAC protocols for IBSN . . . . . . . . . . . . . . . . . . . . . . . . 233.4 Access mechanisms without wake-up radio . . . . . . . . . . . . . . . . . . . . . . 25

3.4.1 TDMA based MAC protocols for IBSN . . . . . . . . . . . . . . . . . . . 263.4.2 CSMA based MAC protocols for IBSN . . . . . . . . . . . . . . . . . . . 263.4.3 Hybrid MAC protocols for IBSN . . . . . . . . . . . . . . . . . . . . . . . 283.4.4 Other Access Mechanisms for IBSN (FDMA, UWB, ALOHA) . . . . . . . 30

Contents vii

CONTENTS

3.5 Access mechanisms with wake-up radio . . . . . . . . . . . . . . . . . . . . . . . . 343.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363.7 Conclusions of Literature survey . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

4 Analysis of wake-up radio based MAC protocols 414.0.1 Radio Triggered sensor MAC . . . . . . . . . . . . . . . . . . . . . . . . . 414.0.2 OnDemand MAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424.0.3 SCM MAC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

4.1 Simulations and performance evaluation . . . . . . . . . . . . . . . . . . . . . . . 444.1.1 Simulation results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

4.2 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

5 Characterisation of PHY layer of an implanatable sensor node 515.1 Description of test environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

5.1.1 Animal flesh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515.1.2 Different location in muscular tissue for an implant . . . . . . . . . . . . 52

5.2 CC430 based implant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545.2.1 Pseudo-implementation of wake-up radio and antenna matching circuit . . 55

5.3 Physical layer configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 565.3.1 Parameters of physical layer configuration . . . . . . . . . . . . . . . . . . 565.3.2 Medical scenarios for different configurations . . . . . . . . . . . . . . . . 57

5.4 Implementation and experimental setup . . . . . . . . . . . . . . . . . . . . . . . 585.4.1 Implant location in flesh for different scenarios . . . . . . . . . . . . . . . 585.4.2 Collective evaluation of PHY parameters . . . . . . . . . . . . . . . . . . 585.4.3 Set of physical parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 605.4.4 Set of network parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

5.5 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635.5.1 Results from evaluating the set of physical parameters . . . . . . . . . . 635.5.2 Results from evaluating the set of network parameters . . . . . . . . . . . 67

5.6 Conclusion from characterization of implantable sensor node . . . . . . . . . . . . 705.7 Optimum parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

6 Performance evaluation of wake-up feature based CSMA/CA protocol 716.1 CSMA/CA without wake-up radio . . . . . . . . . . . . . . . . . . . . . . . . . . 716.2 CSMA/CA with wake-up radio . . . . . . . . . . . . . . . . . . . . . . . . . . . . 736.3 Performance analysis of CSMA/CA protocol with and without wake-up radio . . 76

6.3.1 Network setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 766.3.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

6.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

7 Conclusions 817.1 Answer to the research question . . . . . . . . . . . . . . . . . . . . . . . . . . . . 827.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

Bibliography 85

Appendix 91

A Animal flesh in SC1 91

viii Contents

CONTENTS

B Animal flesh in SC2 92

Contents ix

List of Figures

1.1 Implantable cardiac pacemaker . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Implantable neural stimulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.3 Implantable drug-delivery system (on left is the USB stick for size-comparison) . 31.4 Implantable glucose sensor[11] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.5 Wireless, Battery-less, implantable MEMS sensor for ECG measurement [10] . . 41.6 Intra-cranial pressure sensor[28] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.7 Ingestible endoscopic pill size-compared with one cent coin [32] . . . . . . . . . . 41.8 Electro-CardioGram of Human [16] . . . . . . . . . . . . . . . . . . . . . . . . . . 51.9 Thoracic pressure signal and Photo-PlethysmoGraph of a Human [18] . . . . . . 51.10 SpO2 signal with different contextual measurements . . . . . . . . . . . . . . . . 61.11 Representation of physiological signal (ECG) and contextual signal (respiration

impedance sensor) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.1 Different communication strategies of BSN . . . . . . . . . . . . . . . . . . . . . . 112.2 Block diagram of sensor nodes with wake-up radio . . . . . . . . . . . . . . . . . 152.3 Transmitter implementation of 2.4 GHz wake-up radio . . . . . . . . . . . . . . . 162.4 Recevier implementation of 2.4 GHz wake-up radio . . . . . . . . . . . . . . . . . 17

4.1 RTM scheme. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424.2 On-Demand MAC scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434.3 SCM-MAC scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444.4 Effect of IPAT on power consumption . . . . . . . . . . . . . . . . . . . . . . . . 464.5 Effect of IPAT on End-to-End delay . . . . . . . . . . . . . . . . . . . . . . . . . 474.6 Effect of IPAT on packet delivery ratio . . . . . . . . . . . . . . . . . . . . . . . . 484.7 Effect of IPAT on duty cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

5.1 A dissected part of pig flesh used for testing the sensor node. . . . . . . . . . . . 525.2 A cross section of skin showing different tissue layers including skin. . . . . . . . 525.3 A biological safety cabinet used for carrying out the experiments with meat. . . . 535.4 Block diagram of radio chip used in the evaluation. . . . . . . . . . . . . . . . . . 545.5 Custom made CC430 based implant. . . . . . . . . . . . . . . . . . . . . . . . . . 555.6 CC430 based implant enclosed in a paraffin coating. . . . . . . . . . . . . . . . . 565.7 Flowchart for PHY evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605.8 RSSI vs Distance at 0 degree antenna orientation . . . . . . . . . . . . . . . . . . 645.9 RSSI vs Distance at 90 degree antenna orientation . . . . . . . . . . . . . . . . . 645.10 RSSI vs Distance at 180 degree antenna orientation . . . . . . . . . . . . . . . . . 645.11 RSSI vs Frequency at a fixed tx power, distance, orientation . . . . . . . . . . . . 64

List of Figures xi

LIST OF FIGURES

5.12 Received signal strength information (RSSI) for in-body to in-body communica-tion SC1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

5.13 RSSI vs Distance at 0 degree antenna orientation . . . . . . . . . . . . . . . . . . 655.14 RSSI vs Distance at 90 degree antenna orientation . . . . . . . . . . . . . . . . . 655.15 RSSI vs Distance at 180 degree antenna orientation . . . . . . . . . . . . . . . . . 655.16 RSSI vs Frequency at a fixed tx power, distance, orientation . . . . . . . . . . . . 655.17 Received signal strength information (RSSI) for in-body to on-body communica-

tion SC2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655.18 RSSI vs Distance at 0 degree antenna orientation . . . . . . . . . . . . . . . . . . 665.19 RSSI vs Distance at 90 degree antenna orientation . . . . . . . . . . . . . . . . . 665.20 RSSI vs Distance at 180 degree antenna orientation . . . . . . . . . . . . . . . . . 665.21 RSSI vs Frequency at a fixed tx power, distance, orientation . . . . . . . . . . . . 665.22 Received signal strength information (RSSI) for on-body to on-body communica-

tion SC3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 665.23 Tx Rate vs Packet delivery ratio evaluated with fixed packet length. Measured in

SC1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 685.24 Tx Rate vs Packet delivery ratio evaluated with fixed packet length. Measured in

SC2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 685.25 Tx Rate vs Packet delivery ratio evaluated with fixed packet length. Measured in

SC3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

6.1 Packet format of the CSMA/CA protocol. . . . . . . . . . . . . . . . . . . . . . . 726.2 Flowchart for CSMA/CA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 736.3 Wake-up feature of CC430 (ref: CC430 datasheet[45]) . . . . . . . . . . . . . . . 746.4 Flowchart for CSMA/CA with wake-up feature . . . . . . . . . . . . . . . . . . . 756.5 The network topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 766.6 Inter packet arrival time vs Duty cycle. Comparison between CSMA/CA with

and without Wake up radio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 776.7 Inter packet arrival time vs Packet delivery ratio. Comparison between CSMA/CA

with and without Wake up radio . . . . . . . . . . . . . . . . . . . . . . . . . . . 786.8 Inter packet arrival time vs End to End delay. Comparison between CSMA/CA

with and without Wake up radio . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

A.1 Scenario 1, in-body - in-body communication Distance = 16cm . . . . . . . . . . 91A.2 Scenario 1, in-body - in-body communication, Distance = 6cm . . . . . . . . . . 91

B.1 Scenario 2, in-body 1 (to) in-body2 communication,Distance= 20 cm . . . . . . . 92B.2 Scenario 2, in-body 1 (to) in-body2 communication,Distance= 160 cm . . . . . . 92

xii List of Figures

List of Tables

2.1 Requirements for a MAC protocol in IBSN. . . . . . . . . . . . . . . . . . . . . . 18

3.1 Features of MAC protocols as suggested by IEEE 802.15 TG 6 [17] . . . . . . . . 223.2 Requirements of data communication in implantable medical devices used to com-

municated to the base station controller . . . . . . . . . . . . . . . . . . . . . . . 233.3 Taxonomy of MAC protcols based on Wakeup radio for IBSN . . . . . . . . . . . 243.4 TDMA-MAC protocols for IBSN . . . . . . . . . . . . . . . . . . . . . . . . . . . 273.5 CSMA-MAC protocols for IBSN . . . . . . . . . . . . . . . . . . . . . . . . . . . 293.6 hybrid-MAC protocols for IBSN . . . . . . . . . . . . . . . . . . . . . . . . . . . 313.7 FDMA, UWB, ALOHA based access mechanisms . . . . . . . . . . . . . . . . . . 333.8 MAC protocols with wakeup-radio for IBSN . . . . . . . . . . . . . . . . . . . . 353.9 Comparison of MAC protocols in terms of network parameters . . . . . . . . . . 39

4.1 Simulation parameter values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454.2 Strengths and weaknesses of selected MAC protocols . . . . . . . . . . . . . . . . 49

5.1 Evaluation of hardware with two different sets of physical and network parameters. 595.2 Set of physical parameters for SC1. . . . . . . . . . . . . . . . . . . . . . . . . . . 615.3 Set of physical parameters for SC2. . . . . . . . . . . . . . . . . . . . . . . . . . . 615.4 Set of physical parameters for SC3. . . . . . . . . . . . . . . . . . . . . . . . . . . 625.5 Set of network parameters. Repeated for SC1, SC2, SC3 . . . . . . . . . . . . . 635.6 Optimum values derived from validating the set of physical parameters . . . . . . 675.7 Optimum parameters from hardware evaluation . . . . . . . . . . . . . . . . . . . 695.8 Optimum parameters from hardware evaluation . . . . . . . . . . . . . . . . . . 70

List of Tables xiii

Abbreviations

BSN Body sensor networksCSMA Carrier sense multiple accessIBSN Implantable body sensor networkTDMA Time division multiple accessFDMA Frequency division multiple accessMICS Medical implant communication serviceWBAN Wireless Body Area NetworkIEEE Institute for electrical and electronics engineersTG Task groupMAC Medium access controlPHY Physical layerOSI Open systems interconnectQoS Quality of ServiceWSN Wireless sensor networksIMD Implantable medical deviceWuR Wake-up radioWoR Wake-on timerECG Electro cardiogramEEG ElectroencephalogramEMG ElectromyogramPPG Photo-plethysmographRF Radio frequencyRTS Ready to sendCTS Clear to sendIPAT Packet inter-arrival timePDR Packer delivery ratioCPU Central processing unitRAM Random access memorykB Kilo bytesKBPS Kilobytes per secondMHz Mega hertzTx TransmissionRx ReceptionSC ScenarioUART Universal asynchronous receiver transmitterRSSI Received signal strength informationCCA Clear channel assessmentISM Industrial Scientific and medical band

List of Tables xv

Chapter 1

Introduction

A disease is an abnormal condition that affects the body of an organism. The human-kind isprone to different kind of diseases that are chronic and sometimes fatal. These diseases requirecontinuous monitoring and constant therapies for reducing the negative impact or to prevent anyfatal incidents. Most of the elderly patients are easily affected by diseases due to the process ofageing. It is common that the young patients also suffer due to unexpected chronic diseases whichdirectly affects the quality of living. Thankfully, the advent of advanced healthcare systems helpsthese elderly and young patients who suffer from diseases which in the previous decades werenot possible to be medically treated. In the last decades, technology had helped to find relieffrom certain diseases in neurological disorders [1] [2], cardio-vascular problems[3], chronic pain,and also efficient remedies for syndromes such as Parkinson’s syndrome [4]. The most commondiseases, which are found to be treatable with medical devices, are:Cardio-vascular problem

The Heart is an organ which pumps blood to different parts of the body. The pumpingcycle has four distinct operation which is regulated by a muscular node that provides electricalimpulse to the heart. On failure of this muscular node, due to various reasons, the pumping actionbecomes out of phase. A medical device called pace-maker is implanted near to the heart, whichacts as the node providing electrical impulses [3]. A Physician programs the pace-maker to pacethe heart in a rhythmic fashion. The physiological signals such as Electro Cardiogram(ECG),blood pressure are measured and analyzed before programming the pace of the heart.Neurological disorders

The human body is composed of nerves, which acts as medium of communication between thebrain and other parts of the body. Any dysfunction in the neural system will lead to neurologicaldisorders. Epilepsy is one of the most common neurological disorder which results in seizuresin brain which causes random bodily behaviour. Symptoms such as rapid movement of limbscalled myoclonic jerks. This is the most common symptom of generalized stroke caused due toepilepsy[1].Hormonal deficiency

The cases where the naturally secreting hormone of the body fails, causing severe diseases suchas diabetes. To overcome this situation, artificial hormone are injected into the body throughdrug-delivery devices. These device inject drugs resembling the hormone regaining the normalfunctioning of the body. Programming of these devices as per the requirement of patient is doneby physicians.

1

CHAPTER 1. INTRODUCTION

Figure 1.1: Implantable cardiac pacemaker Figure 1.2: Implantable neural stimulator

1.1 Implantable medical devices and sensors in health-care

The medical devices referred to in this study are the devices that are implanted inside thehuman body, which simulates the natural function of any organ. For example, pace-maker is amedical device that simulates the function of sinus-articular node which is a muscular node onthe heart that provides periodic electrical impulses for a proper functioning of the heart. Devicesalso include nerual stimulators, drug delivery systems which are placed as a replacement for thenatural body organs. These devices are typical electronic system, which comprises of a energysource, a small micro-controller, and a communication module for external communication. Someof the devices also contains a sensing part to sense vital bio-signals of the human body whichincludes heart-rate, blood-pressure, core-temperature of the human body, blood glucose level,blood and tissue oxygen level.

Many medical devices have been found to be useful in treating diseases without knowing theexact mechanism behind the cure. However, most of the devices and its mechanisms are beingpublished in the last few years. For example, the medical reason for curing depression with deepbrain stimulator is unknown when it was first used in a patient[5]. Nevertheless, devices withknown mechanism of cure are established in the past. Some of such devices are pace-maker, neuralstimulator, drug-delivery systems, retinal implant, cochlear implant, semi-functional prostheticlimbs.

2 1.1. Implantable medical devices and sensors in health-care


Figure 1.3: Implantable drug-delivery system(on left is the USB stick for size-comparison)

Pace-makerThe pace-maker is a device which is used to electrically stimulate the heart for normal oper-

ation. As shown in 1.1, it has two leads delivering the electrical stimulus. The device itself isimplanted near the heart and programmed externally using a telemetry link, usually a magneticlink. Due to the progressing standardization of radio communication inside of human body,latest devices have radio frequency communication link between the implanted device and theexternal programmer[6]. The pace of electrical stimulus is programmed as per the requirementof the patient. A patient have to visit the doctor frequently in order to reprogram the device.Neural stimulator

The neural stimulator is an electronic device which provide electrical impulses to the brainand neural systems to regain the functionality the neural system. The stimulation can be con-structive inducing the activity, or destructive by reducing the activity. The concept is same asthat of a pace maker, where two leads delivering the current is connected to a controlling deviceusually implanted near the collar-cuff below the shoulders[7]. The programming of the stimulatoris carried out as per the symptoms and requirements of the patient. Robust mechanism of actionin using neural stimulation for many diseases is yet to be known, however a cure was possible inmost of the neural-disordered patients. A commercially available neural stimulator is shown inFig. 1.2. Drug delivery device

The drug delivery systems are devices which store small quantities of refillable drug capsules,and releases the drug either to either blood stream or to the nerve bundles. In patients with dia-betes, insulin pump in placed near to the vertebral column and the insulin is inject periodicallyas per the requirement of the patients as shown in Fig. 1.3. The period of the drug release iscontrolled by the physician based on patient’s medical history and symptoms[8]. Chronic pain,spinal cord injuries, hormonal deficiency are few applications of drug delivery systems. Thereare literature [8] [5] [3] which witness the introduction of sensing circuitry in the device itself,which can monitor as well as provide the therapy. However the symptoms are not local, but haveto be precisely monitored all over the body due to dynamic nature of the human body.

1.1.1 Implantable bio-sensors for monitoring physiological and contex-tual signals in a human body

Successful research on biological, chemical, electrical and mechanical sensor technologies haveled to a wide range of wearable and implantable sensors suitable for continuous monitoring. In

1.1. Implantable medical devices and sensors in health-care 3


addition to sensor sensitivity, several factors have to be considered in the design of a pervasivebiosensor[9], such as reliability, ease of use, selectivity, sensor packaging, biocompatibility, andpower consumption. Bio-sensors are often affected by noise due to bio-fouling, motion artefact,and interference. For example, ECG (Electrocardiogram) sensors are highly sensitive to motionartefact, which can hinder its ubiquitous use. To improve the sensor reliability, multi-sensor orsensor array approaches are commonly used. Sensor fusion techniques can then be applied to fuseinformation from these sensors. For example, source recovery can be employed to fuse the inform-ation from multiple sensors and infer the intrinsic signal characteristics [10][11]. Although by theintroduction of additional sensors can improve the overall system performance, increasing thenumber of sensors can potentially increase the complexity of the system and affect its practicaldeployment. To circumvent this problem, minimum number of sensors should be used for differ-ent application scenarios. In fact, selecting only relevant features or sensors not only simplifiesthe system set-up but also improves the classification accuracy [16]. In practice, feature selectiontechniques can be employed to identify relevant sensors and their optimum location. The figures1.4 - 1.7 show different implantable sensor enabled with a wireless communication. Based on thetypes of commercially available sensors, potential use of them in monitoring physiological andcontextual signals are mentioned in the following section.

Figure 1.4: Implantable glucose sensor[11] Figure 1.5: Wireless, Battery-less, implantableMEMS sensor for ECG measurement [10]

Figure 1.6: Intra-cranial pressure sensor[28]

Figure 1.7: Ingestible endoscopic pillsize-compared with one cent coin [32]



Physiological signals of Human bodyPhysiological signals of the human body reflect the functioning of the human body and symp-

toms of certain diseases can be reflected in the physiological signals. Vital organs of human bodysuch as heart, brain, lungs etc., have unique pattern of electrical signals which can be recordedusing the bio-sensors. Apart from vital organs of human body, unique pattern of signals canbe sensed from non-vital organs, such as muscles and skin. Some of the physiological signals ofhuman body are shown in the figures 1.8 and 1.9

Figure 1.8: Electro-CardioGram of Human [16] Figure 1.9: Thoracic pressure signal andPhoto-PlethysmoGraph of a Human [18]

Contextual signals of Human bodyThe contextual signals of human body such as chemo-sensory responses in terms of sweat,

hormones and enzymes such as adrenalin can be monitored for diagnosing diseases [3]. For ex-ample, a prediction of heart failure can be done if the information from contextual sensor such asanxiety through sweat and adrenalin sensor is available. Monitoring these contextual values willalso help in rehabilitation of patients both physically and mentally. It is interesting to note thatthe mental diseases such as Treatment resistant depression depend on contextual responses of apatient in the therapy. For example, McKeown et al., in [12] presented a way to monitor thehealth of a patients by monitoring the perception of laughter of a person at different social places.Deep brain stimulation therapies can be made efficient if the contextual signals of human bodyis monitored continuously along with physiological responses. Similarly, motion based activitysensors, respiration sensor and hormonal sensor implants can be involved in forming a closed loopmedical systems. Agarwal et al., in [13] have presented a novel method of using different con-textual sensors such as sweat sensor, heat loss sensor, number of walking steps using pedometer,skin temperature, vertical acceleration, which are then fed into a Statistical vector machine. Asa result Agarwal et al, were able to find the health condition of the patient efficiently. Also,Adolph et al., in [14] have performed experimental trials with chemo-sensors and measured theanxiety of a person. However, in order to use sophisticated implantable sensor network, variousaspects of the network should be defined by rigorous research defining dependency on the sensornetwork and reliability of the medical devices.

1.1. Implantable medical devices and sensors in health-care 5


Figure 1.10: SpO2 signal with different contex-tual measurements

Figure 1.11: Representation of physiologicalsignal (ECG) and contextual signal (respiration

impedance sensor)

1.1.2 Drawback of current medical devices in health-care

Medical environment for critical care patients often involve many medical devices. Physiolo-gical sensors such as pulse rate monitor, oxygen sensor, provide vital information about thefitness of the patient. Medical devices, for example, drug-delivery device, provide therapies fordiseases. That is, they stabilize the fitness of the patient, for example, by infusing medication.Medical systems, considered together with the patient and caregivers, represent an importantclass of cyber-physical systems. Patient safety is the primary concern in such systems, yet reas-oning about patient safety is very difficult because of insufficient understanding of the dynamicsof human body response to treatments. Physicians and care-givers manually program the med-ical devices by monitoring the dynamic response of human body measured by the physiologicalsensors.Human errors, another important source of patient safety problems, are also difficult toreason about in the framework of conventional medical system development. Traditionally, care-givers perform the role of the controller in such a system. This means that the caregiver needs tocontinuously monitor all sensor devices and apply an appropriate treatment. The large numberof sensor data from different patients and appropriate control of a specific patients, makes the jobof the caregiver very difficult. Two simultaneous emergency situation may divert the care-giver’sattention, making him or her miss an important event. As a result, patient safety may suffer andsome times the results may be fatal. Multiple such occurrences are documented in the clinicalliterature[1].Possible solution to overcome human errors and improve patient safety One ofthe possible solution to overcome these drawbacks is making the operation of medical deviceautonomous. The medical device should be able to perform the therapies as prescribed by thephysician for different symptoms of a particular disease. The symptoms are continuously mon-itored by the sensors which is given as an input to a closed-loop control system which actuatesthe medical devices to provide the therapy for a specific symptom of a disease. This mechanismof closed loop operation is possible if three types of technologies combine. Monitoring the symp-toms, which require bio-sensors either invasive or non-invasive to the human body. Processingthe sensor-data which is acquired by the bio-sensors and data communication to the implanted



medical devices and sensors. Design of bio-sensors and processing the data is an ongoing re-search, where a substantial progress has already been made. IBSN is a suitable candidate toform a network of implantable medical devices and sensors, integrating the autonomous opera-tion with a closed loop operation. Focus on the wireless data communication to and from theimplantable medical devices and sensors is needed.

1.1.3 Challenges of IBSN in a closed loop architecture

The need of the closed loop medical devices can be fulfilled by carefully orienting the featuresof IBSN towards requirements of advanced medical devices. IBSN is highly inter-disciplinaryfield of research, where communication mechanism depends on the properties of the human bodyalong with other factors related to human body such as movement, noisy RF environment etc.In order to establish a wireless communication in IBSN to operate in a closed loop control, thefollowing technical challenges have to be addressed.

• establishing reliable link in different positions of the body

• sensor nodes operated at ultra-low power consumption

• ability to operate in link-failures

• achieving zero-latency in emergency conditions

• providing QoS requirement for medical device inter-communication

1.2 Context of Research

IBSN and BSN are suitable candidates to continuously monitor the vital signs of humanbody and control the medical devices based on the physiological response from the patient. Asmentioned earlier,the characteristics of IBSN is not known compared to BSN. Hence a thoroughcharacterization and evaluation of IBSN is required. The context of this research is to characterizethe PHY and MAC layer in an IBSN.

1.2.1 Motivation

The communication mechanism in an in-body environment is not completely standardized.Many physical layer parameters and MAC layer parameters are unknown for a given hardware.In order to understand the effect of communication mechanism, an evaluation of physical layerand data link layer is necessary. Moreover, the behaviour of real hardware inside the body ofliving organism depends on various parameters of wireless communication such as transmittedfrequency, power of transmission, modulation format, and orientation of antenna. Knowledgeof these parameters and effect of hardware design of the implant is an important starting pointfor the IBSN. Having known the physical layer parameters, for a given hardware design it isimportant to evaluate the data link layer. MAC sub-layer is important layer of the networkstack of a sensor network.

The MAC protocols are one of the important aspects of the communication strategies toreduce power consumption, increase reliability and throughput. It is also known from literature(refer Chapter 3), that the wakeup-radio can be incorporated in the MAC protocols in order toprovide a high reliability by reducing the collision and reduce power consumption by overcom-ing the problem of over-hearing and idle-listening. The benefits of wake-up radio based MACprotocols in BSN is enormous, but the effect of the same in different physical medium such as

1.2. Context of Research 7


in IBSN is still unknown. Hence, characterizing the existing MAC protocols with and withoutwake-up radio in an IBSN scenario is required. A thorough characterization of PHY and MAClayers is required to design the infrastructure for IBSN.

1.3 Research question

Wireless communication to the implanted sensor node is a challenging task to achieve dueto inter-dependency of power consumption, network performance and dynamic nature of humanbody. Hence, research is needed in the direction of characterizing the communication strategiesfor IBSN to establish a robust wireless link within the human body. This research will be acarried out by characterizing the RF communication mechanisms inside the human body.. Thefollowing question will be answered as a result of this research

Can a wake-up radio integrated with MAC protocol, meet the QoS requirements and powerconstraints of an IBSN while operating inside the human body ?

1.3.1 Research approach

The objectives of the research is focused on the IBSN to address the reliability, accuracy andpower efficiency. The main aim of the IBSN is to increase the reliability of the medical devicesand reduce the power consumption. Four steps are carried out to characterize the physical layerand MAC layer of the IBSN,

• Investigate the existing MAC protocols and the use of wake-up radio in MAC protocols.

• Verify the findings of existing MAC protocols in an IBSN scenario using software simula-tions.

• Characterize the physical layer parameters depending on different medical scenarios ofimplantable sensor networks using a implantable sensor node implanted inside the animaltissue.

• Evaluate the network performance by choosing optimum physical layer parameters for theimplantable sensor network.

1.4 Outline of thesis

This thesis is organized as follows,

• Chapter 1 presents the general need for Implantable sensor network and its applications.

• Chapter 2 describes the general requirements for an implantable sensor network.

• Chapter 3 presents an elaborate survey of existing MAC protocols with and without wake-up radio and their applicability in IBSN.

• Chapter 4 describes a detailed analysis of different kinds of access mechanisms in a MACprotocol with and without wake-up radio in an IBSN scenario by software simulation.

• Chapter 5 presents a detailed characterization of the PHY layer in an implantable sensornode and provides optimum parameters for evaluating the network performance.

8 1.3. Research question


• Chapter 6 analysis the advantages and disadvantages of the wake-up radio with carriersense multiple access mechanism in a network of sensor nodes implanted inside the animaltissue

• Chapter 7 concludes the results from characterization and evaluation of PHY and MAClayers of implantable sensor node and some of the open research questions are presented.

1.4. Outline of thesis 9

Chapter 2

Requirements of PHY and MAClayers for IBSN

In the last few decades, wireless body sensor network have been established in three differentscenarios namely, Off-body communication, On-body communication and In-body communica-tion. Off-body communication is the communication from the base station to the transceiveron human side. On-body communication is the communication with on-body sensor nodes andwearable system. In-body communication is the communication between invasive or implantabledevices and external base station. Monitoring in-body functions and the ability to communicatewith an implanted therapeutic device, such as a pacemaker, are essential for its best use. Outof the three scenarios, most of the research was mainly on the first and second scenarios rulingout in-body sensor network due to its complex nature[15][16]. Technologies enabling the firsttwo scenarios were already existing, which resulted in establishment of Body Area Networks andPersonal Area Networks. However, the in-body sensor networks demands critical requirementssince the sensor nodes are placed inside the body.

Figure 2.1: Different communication strategiesof BSN

11

CHAPTER 2. REQUIREMENTS OF PHY AND MAC LAYERS FOR IBSN

2.1 Communication mechanisms for IBSN

IBSN sensors are either implanted into patients or worn by the patients, low power radiohas to be used to minimize the radiation. In addition, as sensitive physiological data is beingtransmitted in a IBSN, reliable and secure wireless links are essential. Due to the growing demandof the medical devices, ITU have allocated 402-405 MHz under the citizen band radio services.This frequency have to be shared with the satellite communication. MICS is one of five CitizensBand Radio Services. The others are the Citizens Band Radio Service at 27 MHz, the WirelessMedical Telemetry Service (WMTS) at 216-217 MHz, the Low Power Radio Service (LPRS) at216-217 MHz, and the Family Radio Service (FRS) at 460 MHz.The 402-405 MHz frequencieshave propagation characteristics conducive to the transmission of radio signals within the humanbody. In addition, equipment designed to operate in the 402-405 MHz band can fully satisfythe requirements of the MICS with respect to size, power, antenna performance, and receiverdesign. Further, the use of the 402-405 MHz band for the MICS is compatible with internationalfrequency allocations. Finally, the use of the 402-405 MHz frequency band for the MICS doesnot pose a significant risk of interference to other radio operations in that band. MICS systemsconsist of the transmitters connected to medical implant devices, and programming, monitoringand control equipment. International regulation of the frequency band for life-critical medicaldevices will benefit the user, the wireless medical industry, and regulators. It will also impactpositively on cost-saving, quality, reliability and delivery of healthcare. MICS is accepted globallyfor the use of medical devices and the work-group for standardization is set.

Why not ISM band for medical devices?The ISM band also has the same properties as MICS band in terms of electrical design.

However, due to the threat of interference from other users which can significantly increase thenoise floor and cause unwanted and unpredicted behaviour to the medical devices operating in thesame band. The interference can even occur from BMW’s Comfort Access and MB’s Keyless Gowhich uses the same 433 MHz ISM band for security operations in auto-mobiles. Also, patientstravel internationally and often carry and use medical devices across national borders, thus globalharmonization of rules and standards is essential. The frequency bands, service status (level ofprotection), technical standards and certification requirements are not efficiently controlled andpredicted in the ISM band. The introduction of IBSN and the use of dedicated spectrum for somemedical applications (e.g. telehealth) need to be standardized and harmonized internationallywhich is not possible in the case of ISM band. Thus use of ISM band in the medical devices isnot a suitable candidate for the medical devices which are implanted inside the body.

2.1.1 MAC protocols

The development of an affordable IBSN induces a number of issues and challenges such asinteroperability, scalability, Quality of Service (QoS), and energy efficient communication. Thereare various low-power techniques to ensure energy efficient communication in a wireless sensornetwork such as fixed duty cycling technique in SMAC[17] and wake-up slots in TDMA[18].However, they are not energy efficient in case of a heterogeneous IBSN. Unlike SMAC, thetraffic characteristics in a IBSN vary from periodic to non-periodic and vice versa[18]. Theconcept of fixed duty cycling technique gives limited answer when it comes to the heterogeneousbehaviour of autonomous sensor nodes in a IBSN. The dynamic nature of these nodes does noturge synchronized periodic wakeup periods. Some nodes, e.g., electrocardiogram (ECG), maysend data at 1/hour rate to the coordinator, while other may send data twice in a week. Thesenodes should also have the capabilities to sense and transmit emergency information. The data is

12 2.1. Communication mechanisms for IBSN


classified into three categories, i.e., Normal Traffic, On-Demand Traffic, and Emergency Traffic.

The IBSN MAC protocol is required to accommodate the entire traffic classification in a soph-isticated manner. Most of the well-known low-power MAC protocols such as IEEE 802.15.4[17], SMAC, TMAC, and WiseMAC cannot accommodate these diverse traffic requirements[18].They give limited answers in terms of energy efficiency and reliability. Furthermore, they cannothandle both medical and non-medical IBSN applications. Medical data usually needs high pri-ority and reliability than non-medical data. Time critical event needs highest reliability. IEEE802.15.4 [17] Guaranteed Time Slot (GTS) can be utilized to handle time critical events but theyexpire in case of low traffic[21].

The IEEE 802.15.6 aims to provide low-power in-body and on-body wireless communicationstandard for medical and non-medical applications. The standardization committee has sugges-ted four options to design MAC and PHY layer for a IBSN:

• To define MAC and PHY standard for on-body communication to serve immediate marketneeds. A slight modification to the existing MAC standard with an alternative PHY layerfor in-body communication is also suggested.

• To define MAC and PHY only for on-body communication to serve immediate marketneeds

• To define MAC and PHY only for in-body communication

• To define MAC and PHY for in-body and on-body communication simultaneously regard-less of their effects on the availability of specification

Due to the demanding market needs and the option to serve immediately, more focus wasgiven to the on-body sensor networks, rather than in-body sensor networks. However, it isimportant to note that use of in-body sensor networks is essential in life-critical situations wherethe implanted sensor has to be used for continuous monitoring of physiological signals withoutany external interferences.

Design challenges of In-Body MACThe most challenging task in developing a low-power IBSN-MAC protocol is to accommodate

in-body sensor nodes in an energy-efficient way. In-body nodes are implanted under humanskin and have critical power requirements. They are totally different than on-body nodes interms of power efficiency and data transmission rate (10kbps for medical and up to 10Mbps fornon-medical applications). Moreover, they need to send emergency data in less than 1 secondto the coordinator. This is a hot issue in the design and implementation of an in-body MAC.The nodes are required to be self-triggered when exceeds a predefine threshold for emergencysituation. Critical data requires low latency and high reliability [19]. The solution is to adjustinitial back-off windows for critical and non-critical traffics [22]. Non-critical traffic nodes musthave larger initial back-off window than critical traffic nodes. The smaller initial back-off windowfor the critical nodes results in lower latency.

According to Zhen et all, the use of CSMA/CA for in-body communication does not providereliable solution [20]. The main reason is that the path loss inside human body due to musculartissues is much higher than the path loss in free space. The in-body nodes cannot performClear Channel Assessment (CCA) in a favourable way. Alternatively, Ullah et all proposed a

2.1. Communication mechanisms for IBSN 13


TDMA-Based solution to accommodate the entire traffic classification of in-body nodes. Thecommunication is based on pre-defined wake-up patterns, stored in a Pattern-Based wake-uptable.

2.2 Architectural framework for closed loop medical devicesusing IBSN

The sensor nodes for measuring the physiological data is available off the shelf. However,for a closed loop operation it is important that the location of sensor nodes and the topology ofconnection between this sensor nodes, location of central control unit and communication to theexternal control unit have to be optimized for reliable operation. This has to be defied under anarchitectural framework of body sensor networks which defines its operation.

An architectural framework of communication strategies in the IBSN will define,

• Connection topology of sensor network

• Communication mechanism that is used under the topology

• Data handling in different scenarios

• Number of nodes and the duty cycle of the sensor node

• Communication to the top level protocols in OSI layer

The foreseen framework consists of different technologies fused together. Off-the-shelf sensorsand chip level electronics for processing, radio communication, power scavenging, control andmedical devices will be used. Communication strategies of the on-body sensor network and off-body sensor network are well established [30] [32]. Different approaches to improve the reliabilityfactor of on-body sensor network have been published [20] [24].

2.3 Low power design and Power Scavenging

Power source is one of the key elements for IBSN. It often dominates the size and lifetimeof the sensor nodes. Thus far, battery remains the main source of energy for sensor nodes.There are different ways of using the power source efficiently. One of the ways is to reducethe power wastage in wireless radio. Wakeup radio is a key technology that is emerging as asolution to reduce the power consumption of the sensor nodes. There are also other ways suchas on-node processing of data, there by reducing the amount of data transmitted. Since thepower consumed by the processors now a days is much lesser than the power consumed by thewireless radio. Energy harvesting is also a key researched area, where the power required for theoperation of a sensor node can be harvested from various sources. The following sections discussin brief about the low power design and power harvesting techniques involved in IBSN.

2.3.1 Wake-up radio

The wireless communication is the most power hungry part of a sensor network apart fromprocessing and sensing. The power consumption of the wireless radio is controlled and reducedby software mechanisms such as duty-cycling using MAC protocols as discussed in the previous

14 2.2. Architectural framework for closed loop medical devices using IBSN


section. In the last decade, an interesting concept of wake-up radio is thoroughly being investig-ated. A wake-up radio is a transceiver radio which consumes very small amount of energy in fewnW (nano-watts), than the main transceiver radio which operates usually in orders of µwatts(micro-watts). The main features of wake-up radio to consider it as a suitable candidate for theIBSN are

• Low power consumption.

• Operable only in smaller range of network.

• Reliable performance in smaller network size.

• Ability to operate outside MICS band.

• Less complex hardware required.

The main purpose of the wake-up radio is to reduce the power wastage of the main radio. Thepower wastage of the main radio is due to the idle-listening, over-hearing, data collision andstate-switching (on state to off state and vice-versa). The wake-up radio is used to turn-on themain radio only for useful data communication. By doing so, the power consumed by the mainradio for idle listening is eliminated, along with the over-hearing problems preventing the datacollision to occur. Unwanted state switching is eliminated, since the main radio is turned on onlyfor useful communication. This process will not only increase the reliability of the low powerwireless communication, but also increase the energy efficiency. The wake-up radio operated in adifferent frequency band than the normal radio is used to send wake-up signals to the node, whichon positive verification will turn on the main radio. The data communication is then initiatedand completed using the main radio, reducing the total amount of time that the main radio isturned-on. The main radio is used only for useful data communication instead of idle-listeningand waiting for a slot to communicate. A normal schematic of a wake-up radio [21] is shown inthe Fig. 2.2

Figure 2.2: Block diagram of sensor nodes with wake-up radio

The wake-up radio can be implemented using a wake up transmitter circuit [22] as shown in the2.3 and wake-up receiver circuit [23] as shown in Fig. 2.4. It is certain that the wake-up radiowill occupy more space on the hardware of the implant. However due to the miniaturization of

2.3. Low power design and Power Scavenging 15


electronic component, the implementation of wake-up radio can be done in the chip level. Onesuch example is the Microsemi ZL70120 [23] which is a implant module comprising of both mainradio and wake-up radio in a package of 80mm× 80mm area [22]. The Fig. 2.4 and 2.3 showsthe block diagram of the chip-level implementation of wake-up radio. The CC2550 from ChipConLtd., is implemented in ZL70120[24] as a die along with the main radio on the same module. Themodule from Microsemi also comprises of the matching circuit for the antenna in MICS bandand 2.4 Ghz band. Thus, a wake-up radio implementation is feasible for the medical devices,meeting the requirement of the size constraints and the power reduction is highly possible bycareful implementation of software. An analysis of MAC protocols using the wake-up radio ispresented in chapter 3.

Practicality of MAC protocols with WUR in IBSN

Wake-up radio typically introduces additional hardware, leading to energy overhead, perform-ance overhead and chip-area overhead. With the advancement in nano-electronics and advancedchip-fabrication technologies, it is possible to have two radio transceivers operating at differentfrequency bands in a single chip package. Also, as shown in [25] a wake-up radio with OOKmodulation can operate in the nanowatt range. The radio hardware should also have very lowtransition time from off-state to on-state, coping with the state transition request from the MACprotocol. The practicality of MAC protocols with WUR in IBSN can be stated as the capab-ility to communicate using dual-band radio transceivers implanted inside a human body witha bio-compatible package, meeting the requirements of different health-care services. For thehardware to be practical for the implementation in implants [26][27][28], a power-efficient yetperformance-oriented control is required over the wireless medium. The power consumption andperformance of MAC protocols with WUR are not traded off as it is in the case of normal MACprotocols. So clearly, the MAC layer has to be modified for a practical use of hardware available.

Figure 2.3: Transmitter implementation of 2.4 GHz wake-up radio

16 2.3. Low power design and Power Scavenging


Figure 2.4: Recevier implementation of 2.4 GHz wake-up radio

Operation of wake-up radioThe basic operation of wake-up radio takes place in four steps as shown in fig 2.2. There can

be different software to control the wake-up radio, however the steps below explain the simplemethodology implemented in the Microsemi implant module.Step 1 : Start up base stationStep 2 : Send 2.4 GHz wake-up messageStep 3 : Implant node (IMD) receives 2.4 GHz messageStep 4 : Implant node (IMD) send wake-up response in 400 MHz using main radio.Each steps use different access mechanisms, duty cycling and power level. The efficiency of thehardware depends on the software implemented on the module. In chapter 3, different softwaremethodology for wake-up radio is discussed. In a nutshell, wake-up radio is a suitable candidateto increase energy efficiency of the sensor nodes by eliminating the conventional energy issuesfaced in the main radio.

2.3.2 Power Scavenging

In parallel to power reduction, perpetual energy supply with power scavenging can prolong thelifetime of the sensor and enable long term monitoring of the patient. A number of power scaven-ging sources have currently been proposed, which include motion, vibration, air flow, temperaturedifference, ambient electromagnetic fields, light and infra-red radiation. For instance, Mitchesonet al. developed a vibration based generator designed for wearable/implantable devices, whichis capable of delivering 2uJ/cycle [29]. Similar vibration based thin film piezoelectric energyscavenging systems was proposed by Reilly et al. [28]. A thermal micro-power generator hasbeen developed by IMEC, which can convert thermal energy to 4uW power at 5’C temperaturedifference on the thermopile [29]

2.3. Low power design and Power Scavenging 17


2.4 Application scenario and requirements for MAC pro-tocols in IBSN

In order to characterize and evaluate the PHY and MAC layers of IBSN, an applicationscenario with implantable medical devices and sensors is explained. Consider an epileptic patientimplanted with a deep-brain stimulator which stimulates the brain with electrical impulses inthe event of seizure occurrences. It is possible to detect the occurrence of a seizure beforehandwith symptoms that can be measured with implantable sensors such as an EEG sensor, bloodflow sensor, and external sensors such as inertial accelerometers. To predict the onset of aseizure, the data from these sensors have to be processed in real time at a base station, andcontrol signals have to be sent to the implanted stimulator to suppress the seizure. The wholeprocess including the processing and communication has to be in real time in order to predictand suppress the seizure on time. The wireless communication between these sensor nodes andmedical devices should be highly reliable with negligible delay in order to handle the situationflawlessly. Also, the occurrence of the symptoms is completely random, which means sensingshould be done continuously, and occurrence of an event should be predicted locally by the sensornode. Processing the signal locally is out of the scope of this paper, but wireless communicationshould be efficient not only in terms of performance but also in terms of energy consumption toensure a long-time operation. As an indication, the battery life time of implantable deep-brainstimulators is typically two to three years [30]. Few off-the-shelf stimulators are equipped witha rechargeable battery that can be recharged via an inductive link [30].

Having short-range communication such as a magnetic-induction system will not serve thepurpose of communication with the base station, hence a radio frequency (RF) link with acoverage of at most ten meters and at least two meters is required. To ensure reliable RF com-munication, the underlying MAC protocol is crucial in terms of reliability and energy efficiency.In order to standardize in-body and on-body communication, IEEE 805.14 task group 6 was set.

Network parameter Requirement

Frequency of operation 402-405 MhzBandwidth 3Mhz

No. of Channels 10, each channel is 300 Khz bandwidthPower of operation 25 µW isotropic radiation power

Interference AcceptedTopology Star, P2P

Network size 20 nodes max.Duty cycle 0.1% for non emergency data communication

Latency upto 60 msThroughput upto 100 Kbps

Table 2.1: Requirements for a MAC protocol in IBSN.

This task group has defined the physical layer properties of body sensor networks. Four typesof communication links are foreseen: in-body to in-body communication (Scenario 1 (SC1)),in-body to on-body communication (SC2), on-body to on-body communication (SC3), and on-body to external nodes communication (SC4). Only the first two channel models are considered,where in-body communication is involved, focusing only on the IBSN scenario. In this analysisthe characteristics of the proposed Physical (PHY) and MAC layer constraints set by the task

18 2.4. Application scenario and requirements for MAC protocols in IBSN


group are considered. RF links do not propagate well inside the human body which has to betaken into account at the MAC layer for packet-loss and fading. For in-body communication,a dedicated RF band has been allocated called the Medical Implant Communication Service(MICS) band operating at 402-405 MHz, containing ten channels of 300 KHz bandwidth eachwith effective isotropically radiated power (EIRP) of 25 µW [31]. It is observed that, inside thehuman body, the MICS band can propagate with less loss and fading than the other frequencybands. The recommended network topology is a star network with a central network controller.However, in a complex application scenario as explained earlier, the need of peer-to-peer (P2P)communication is optimal if the processing can be done locally on the sensor nodes. Scalabilityis not an issue, since the number of nodes is typically less than fifteen [28] [32] [33]. The datarate of the network can vary for different sensor nodes depending on the type of sensed data.However, in our application scenario, there is no need for high-bandwidth data transmissionsuch as video or audio. For the scenario explained above, a data rate of 20 Kbps is sufficient forreliable transmission of data from stimulator communication, blood flow sensor, inertial sensor,and EEG sensor. The requirements are set based on the recommendations from task group 6and to meet the applications similar to the scenario explained in this section. Table 2.1 lists therequirements of IBSN.

2.4. Application scenario and requirements for MAC protocols in IBSN 19

Chapter 3

Survey of MAC protocols withand without wake-up radio forimplantable sensor network

In an IBSN, sensor nodes have limited resources such as energy, size of the components(including the sensor, processor and radio), and range of communication. Despite the limitedresources available, IBSN applications impose strict requirements for the wireless network interms of communication reliability, delay, throughput, energy-efficiency and in some applicationseven Quality of Service(QoS). MAC protocols in wireless networks, aim for minimum delay,maximum throughput and an increased network life-time by controlling the main sources ofenergy waste such as idle-listening, collisions, over-hearing, and packet lost. There are a numberof MAC protocols available for wireless sensor networks, even some are focussed on the IBSNapplications. The following section presents a survey on the existing MAC protocols that areoptimized for IBSN applications and attempts to find the potential problems in MAC protocolsthat needs to be solved by further research.

3.1 Features of MAC protocol

The IBSN is a special type of WSN, which varies from WSN in various features such asscalability, reliability, latency and energy-efficiency. As explained in section 2.1, IBSN has threetypes of communications namely, In-body communication, On-body communication, and Off-body communication. This work focusses on MAC protocols that are available for In-body andOn-body sensor networks. The fundamental task of MAC protocol is to avoid collision of datapackets and to prevent simultaneous transmissions while preserving maximum throughput, min-imum latency, communication reliability and maximum energy-efficiency [34]. QoS is also animportant factor of good MAC protocol. In medical applications a latency of 125 ms of is onlyallowed, whereas in consumer electronics latency can be less than 250ms [17].Other importantfeatures include adaptability to changes in network topology, maximum achievable throughputin different network scenarios, least jitter in heterogeneous traffic, efficient bandwidth utilizationwith high payload, safety and security. The following table presents the expected values fordifferent features of IBSN as per the IEEE 802.15.6 [17].As a summary, a good IBSN-MAC should have energy-efficiency, reliability even in heterogen-

eous traffic, safety and security in addition to QoS. [34]

21

CHAPTER 3. SURVEY OF MAC PROTOCOLS WITH AND WITHOUT WAKE-UP RADIOFOR IMPLANTABLE SENSOR NETWORK

Feature of good IBSN MAC Acceptable value for Implanted medical devices

Throughput upto 200 KBPS for medical devicesupto 4Mbps for non-medical devices

Latency upto 100ms in life critical implantsupto 2 seconds in monitoring medical devices

Bandwidth 300KHz MICS band100MHz in 2.4GHz ISM band1.74 MHz in 433 MHz ISM band

Duty cycling less than 0.1 % in MICS band medical devicesno restriction if Listen before talk is incorporated

Interference mitigation CRC, FEC, frequency agility are recommended forsafety purposes.

Table 3.1: Features of MAC protocols as suggested by IEEE 802.15 TG 6 [17]

3.1.1 Attributes of MAC as proposed by IEEE 802.15 TG-6

Wireless Body Area Network (WBAN) has attracted many researchers in academia andindustry, because of its great potential to revolutionize the technology for healthcare. Due to itsgrowing requirements, a task group have been set to standardize in-body, on-body and off-bodycommunication [35]. The purpose of the task group is to define new physical and Medium AccessControl (MAC) layers optimized for low power in-body/on-body nodes (not limited to humans)to serve a variety of medical and non medical applications. This section will briefly explain theattributes of MAC layer set for in-body and on-body by the task-group and in compliance withMICS band regulations.

IEEE 802.15.6 specificationThe IEEE 802.15 task-group 6 [17] suggests that the nodes should be organized into one-hop

or two-hop star network. In the case of single-hop star network a single co-ordinator controls theentire operation of the network whereas in the case of two hop star network a relay-capable nodemay be used to exchange data between hub and the destination-node. The entire physical channel(in time-axis) is divided into super-frame structures. Each super-frame is usually bounded by abeacon period of equal length. In MICS band regulation, the transmission of beacons boundingthe super-frame is prohibited. For such non-beacon modes, where beacons are not used, thesuper-frame boundaries are defined by polling frames.

3.2 Requirements of MAC protocol for IBSN with differ-ent medical devices

The features of MAC protocol specified by the IEEE 8025.15 TG-6 are generalized for the in-body and on-body BSN, however the features do not exactly suit the requirements of closed-looparchitecture. The focus of the TG-6 specifications of MAC protocol aim at general monitor-ing of vital signals, periodic transfer of data to the base station through inter-networking, and

22 3.2. Requirements of MAC protocol for IBSN with different medical devices


emergency event handling in case of life-critical events. The closed-loop functioning of medicaldevices does not only require these features but also additional features such as ability to in-telligent medium access for emergency events, reliable and low latency communication betweennodes and medical devices, ability to respond quickly within the a specific time-frame for a givenmedical devices. It is important to know the requirements of the wireless communication in caseof closed loop access. The requirements are derived for different medical devices based on theliterature and medical case histories.

Network parameterRequirement of implantable medical devices

Pace-maker Neural Stimulators Drug-delivery systems Retinal implants

• Throughput upto 100 KBPS upto 100 KBPS upto 150 KBPS upto 150 KBPS• Latency upto 10 ms upto 30ms upto 60 ms upto 20ms• Payload 40 KBPS 60 KBPS 30 KBPS 80 KBPS• Duty cycling (MICS band) 0.1% 0.1% 0.25% 0.1%

Table 3.2: Requirements of data communication in implantable medical devices used tocommunicated to the base station controller

3.3 Taxonomy of MAC protocols for IBSN

The requirements of the existing medical devices are studied in the previous section. In orderto meet the needs of the medical devices and introduce closed loop mechanism, a sophisticatedMAC protocol for reliable wireless communication is needed. IBSN is a relatively new field whencompared to the BAN, BSN. The closed loop medical devices fall in the category of the IBSN.IBSN being relatively new, is not completely a different wireless domain. Many characteristicsof the BSN and WSN also applies to IBSN. In order to understand the existing literature witha focus of IBSN, good classification of the state-of-the-art MAC protocols is needed. Table 3.3shows the classification of existing MAC protocols based on wake-up radio, for the IBSN.

From the literature it is clear that use of Wake-up radio will eliminate the power wastageof the main radio. This study is focused on the MAC protocols with low energy consumption.Moreover, impact of Wakeup radio in MAC protocols is also the focus of this study. Hence, it iswise to classify the existing MAC protocols that are developed with and without wake-up radio.Not many literature is available on IBSN MAC protocols. Hence, features of existing MACprotocols are extracted and presented, such that these features will match the requirements ofIBSN.

Time Division Multiple Access (TDMA), Carrier Sense Multiple Access with Collision Avoid-ance (CSMA/CA), Frequency Division Multiple Access (FDMA), Slotted ALOHA are the mostlyused multiple access schemes in body sensor networks. Each of the access mechanism has itsown advantages and drawbacks. The following sections will explain about each specific accessmethods, and other researches that have been done in different mechanisms.

3.3. Taxonomy of MAC protocols for IBSN 23


Implantable

body

sensornetw

ork

Without

Wake-U

pR

adio

Tim

eD

ivisionM

ultipleA

ccess

Global

SynchronousT

DM

ASchedule

basedT

DM

A

Carrier

SenseM

ultipleA

ccess

Collision

avoidanceA

cknowledgm

entbased

Hybrid

access

TD

MA

+C

SMA

TD

MA

+F

DM

A

CSM

A\CA

+U

WB

Other

accessm

echanism

FD

MA

SlottedA

LO

HA

With

Wake-U

pR

adio

Tim

eD

ivisionM

ultipleA

ccessC

arrierSense

Multiple

Access

Hybrid

access

TD

MA

+A

LO

HA

TD

MA

+C

SMA

Other

accessm

echanism

FD

MA

SlottedA

LO

HA

Tab

le3.3

:T

axon

om

yof

MA

Cp

rotco

lsb

ased

on

Wakeu

pra

dio

forIB

SN

24 3.3. Taxonomy of MAC protocols for IBSN


3.4 Access mechanisms without wake-up radio

The access mechanisms are broadly divided into two categories.

i Scheduled access and its variants (connection-oriented and contention free)

ii Random access mechanism (CSMA/CA or Slotted Aloha)

The type of access made by the wireless radio is the key classifier. Random access mechanismaccesses the wireless medium randomly with or without schedule. A back-off in access is madefor a specific duration if the medium is found to be accessed by another node. Scheduled accessis done with a specific time schedule and the time duration can be fixed by a network coordinatoror can be determined locally by the sensor node. Time synchronization between the nodes isrequired for the case of scheduled access mechanism. Unscheduled access is a hybrid mechanismwhich accesses the medium randomly with specific duration to access. The access mechanismcan differ for different application purpose. Each access mechanism has its own advantage andits disadvantages. The basic principle of each mechanism is explained below.

• Scheduled AccessThe scheduled access mechanism of the physical medium is carried out by synchronizing allthe nodes in the network to a specific time frame, and allocating a scheduled time slot foreach sensor node. The advantage of the method is collision-free data transmission due tothe time-schedule, which reduces the power consumption and increase the throughput of anetwork. However, the time schedule needs a accurate synchronization of time between thenodes is necessary. A deviation in the synchronization will cause the network to operateless-efficiently or sometimes even fail to operate. TDMA is an example of scheduled accessmechanism, which is contention free access methodology. TDMA is selected to be the bestcandidate for the IBSN because of the following reasons.

1. The size of the IBSN is small compared to other WSN applications

2. The star topology is preferred for IBSN where a central network controller (CNC) isalways present outside the body in close proximity to the network. This enables easiertime synchronization

3. Collision avoidance is easier with less power consumption

There have been a lot of research done in TDMA MAC protocols that are focused on theIBSN . Some of the TDMA-MAC protocols which best fits the requirements of in-body andon-body IBSN is presented in table 3.4 and explained in section 3.4.1.

• Random AccessThe random access of the physical medium is carried out by Carrier Sense Multiple Ac-cess/Collision Avoidance (CSMA/CA). In contrast to the IEEE 802.3 standard which usedcollision detection, IEEE 802.15 (wireless medium) standard uses collision avoidance inorder to save some power. The collision avoidance is performed by setting a back-offcounter to a random integer number uniformly distributed over the interval [1, CW ] whereCW ∈ (CWmin, CWmax). Although scheduled access mechanism are selected to be thesuitable candidate for IBSN , there are few researches been done on Random access mech-anism. Some of the random access MAC protocols developed for the BSN and WSN arepresented in table 3.5 and explained in section 3.4.2.

3.4. Access mechanisms without wake-up radio 25


3.4.1 TDMA based MAC protocols for IBSN

Marinkovic et. al. [36] present an energy-efficient, low duty cycle MAC protocol basedon TDMA. The MAC protocol enables access to the physical layer for a hierarchical topologyconsisting of nodes communicating with master nodes, which in turn communicate with the mon-itoring station. The hierarchy removes the need for sensors to expend power by transmitting tothe monitoring station. Also, the use of TDMA ensures collision-free transfer and minimizationof idle-listening. The protocol is implemented using the Analog Devices 70XMBZ2 platformwith ADF7020 RF transceivers. Measurements reveal that the protocol is energy-efficient forstreaming and short burst data communications. A novel TDMA-based protocol for BSNs,called H-MAC,is proposed in [44]. This protocol improves energy-efficiency by using the heart-beat rhythm to perform TDMA synchronization, avoiding energy consumption associated withtransmitting time synchronization beacons. Power efficiency is also accomplished in H-MAC asaTDMA-based protocol assigns time slots to each biosensor to guarantee collision-free trans-mission. Simulations show that H-MAC prolongs the network life of sensors dramatically. In[37] a TDMA-based MAC protocol called BodyMAC is proposed. Three types of bandwidthallocation schemes are devised to cope with different types of data communications, such asperiodic data sensing and important event allocation. In conjunction with bandwidth allocation,a sleep mode mechanism is introduced, which turns off a node’s radio during beacon, uplinkand downlink periods, as much as possible. Simulations results show superior performance ofBodyMAC compared to that of IEEE 802.15.4 MAC. Timmons et. al. [46] introduce an ad-aptive TDMA-based MAC protocol called MedMAC. MedMAC incorporates a novel adaptiveTDMA synchronization mechanism in which only a multi-superframe beacon has to be listenedto by the nodes. An optional contention period is also available for low-grade data, emergencyoperation and network initialization procedures. Simulations show that MedMAC consumes lesspower than IEEE 802.15.4 for two classes of medical applications. In [48] a power efficient MACprotocol is proposed for WBANs. This work presented a traffic based wakeup mechanism thatutilizes the three categories of traffic patterns of the body sensor nodes, namely: normal traffic,on-demand traffic and emergency traffic. The wakeup patterns of all body sensor nodes areorganized into a table called traffic-based wakeup table. The table is maintained and modifiedby a network coordinator according to the application requirements. Based on the body sensornode’s wakeup patterns, the network coordinator can also calculate its own wakeup pattern.During normal traffic, both the body sensor nodes and the network coordinator send data basedon the traffic–based wakeup table. Authors propose a new MAC protocol based upon staticnature of BAN [38]. TDMA approach is used for streaming large amount of data. Static natureand TDMA approach are being utilized efficiently to maximize network life. In target topologya Master Node (MN) collects data from on body nodes and communicates with a MonitoringStation (MS). Received data is being analysed by MS while the on-body network coordinationand synchronization is being performed by MN. Time slots S1 to Sn are allocated to sensor nodeswhile time slots RS1 to RS2 are reserved which are being assigned when requested. Number ofthese extra time slots depends upon targeted packet drop, packet error rate and number of sensornodes.

3.4.2 CSMA based MAC protocols for IBSN

The diverse nature of in-body BNs together with the electrical properties of the human bodyinfluences the development process of a power-efficient MAC for in- body sensor networks. Thedata rate of implanted BNs varies, ranging from few kbps in pacemaker to several Mbps incapsular endoscope. In the in-body sensor network, critical traffic requires low latency and highreliability than non-critical traffic. One of the solutions is to adjust initial back-off windows in

26 3.4. Access mechanisms without wake-up radio


Nam

eof

the

pro

tocol

Desc

rip

tion

Sp

ecia

lfe

atu

reP

ote

nti

al

dra

wb

ack

Refe

ren

ce

LD

-TD

MA

Low

du

tycy

cle

TD

MA

2.0

4m

Wat

3V

DC

usi

ng

CO

TS

tran

scei

ver.

Pow

erco

nsu

mp

tion

isle

ast

com

par

edto

oth

erp

ro-

toco

ls.

Hig

hla

ten

cyin

the

even

tof

ap

ack

etfa

ilu

re.

Re-

qu

ires

acc

ura

tesy

nch

ron

isat

ion

Mari

nko

vic

etal

.[3

6]

HD

MA

C-T

DM

AH

eart

bea

trh

yth

msy

nch

ron

ized

TD

MA

Incr

ease

dn

etw

ork

life

tim

eby

15%

-300

%m

ore

than

oth

ersi

mil

ar

BS

NM

AC

Su

ffer

sfr

omse

vere

sin

gle

-poi

nt

of

fail

ure

pro

ble

m.

No

accu

rate

hea

rtrh

yth

mis

mea

sure

dal

l-ov

erth

eb

od

y,h

ence

use

of

net

work

coor

din

ator

isn

eces

-sa

rytr

adin

goff

wit

hen

ergy

-effi

cien

cyan

db

an

d-

wid

theffi

cien

cy.

Li

etal

.[3

9]

CF

-MA

CC

onte

nti

on-f

ree

MA

Cp

roto

col

Sel

f-st

abil

izin

gan

dd

oes

not

requ

ire

aglo

bal

tim

ere

fere

nce

.T

he

pro

toco

lw

ill

au

to-s

tab

iliz

efo

ran

yn

etw

ork

chan

ge

Can

not

han

dle

coll

isio

neff

ecti

vely

,sp

ecia

lly

wh

ena

new

nod

ejo

ins

the

net

work

.P

erfo

rman

ceis

seve

rely

aff

ecte

don

the

even

tof

chan

ge

inn

etw

ork

top

olo

gy

Bu

sch

etal.

[38]

SS

D-T

DM

AS

elf-

Sta

bil

izin

gD

eter

min

isti

cT

DM

AE

ner

gyeffi

cien

tp

erfo

rman

ce.

Sel

fst

ab

iliz

ing

inca

seof

dyn

amic

dat

ava

riati

on

s.C

an

sup

port

chan

ges

inn

etw

ork

top

olog

y.N

ovel

two

laye

rap

-p

roach

for

data

-lin

kcr

eati

on.

Any

slig

ht

vio

lati

onin

the

ass

um

pti

ons

mad

efo

rth

ep

roto

col

wil

ld

evia

teth

ep

erfo

rman

ced

rast

ic-

ally

.C

lust

erti

me

syn

chro

niz

atio

nis

nee

ded

,d

ir-

ectl

yp

rop

orti

onin

gto

the

per

form

an

ce.

Som

eof

the

assu

mp

tion

sm

ade

can

not

be

met

inre

al-w

orl

dim

ple

men

tati

on

Aru

mu

gam

etal.

cite

Aru

mu

gam

200

5

DQ

BA

N-M

AC

Dis

trib

ute

dQ

ueu

ing

Bod

yA

rea

Net

wor

kM

AC

Hig

hQ

oSsu

pp

ort

wit

hli

mit

edp

roto

col

over

-h

ead

.L

ess

com

pu

tati

on

alco

mp

lexit

yan

dea

syim

ple

men

tati

on

.N

ovel

inte

grat

ion

offu

zzy

rule

sch

edu

lin

gal

ong

wit

hT

DM

Ab

ased

app

roach

ren

der

sa

per

form

an

ceor

iente

dcr

oss-

laye

rop

tim

-iz

edM

AC

Glo

bal

tim

esy

nch

ron

izati

on

isa

lim

itin

gfa

ctor.

Pow

erhu

ngry

du

eto

exte

nd

edop

erati

on

of

cross

-la

yer

opti

miz

atio

n.

Fu

zzy

logi

cw

ill

bec

ome

ab

ur-

den

for

the

senso

rn

od

esin

case

ofdyn

amic

dat

a-

load

vari

atio

ns.

Ota

let

al.

[40]

HE

H-M

AC

Hu

man

En

ergy

Har

vest

ing

MA

CP

rovid

esp

riori

tyd

iffer

enti

atio

nto

the

sen

sor

nod

esan

dfl

exib

ilit

yto

the

net

work

.H

ighly

adap

t-iv

eto

envir

onm

enta

lch

an

ges

.E

ner

gy

har

vest

ing

rate

s,n

etw

ork

size

and

pac

ket

inte

r-arr

ival

tim

esar

ed

yn

am

ical

lyad

apte

dw

ith

inth

ep

roto

col

Th

rou

gh

pu

t,an

dot

her

QoS

par

amet

ers

are

not

anal

yze

dan

dp

rese

nte

d.

Su

ffer

sfr

omglo

bal

tim

esy

nch

ron

izat

ion

issu

es,

fail

ing

ofw

hic

hse

ver

lyh

ind

ers

the

net

wor

kp

aram

eter

s.

Ibar

raet

al.

[41]

PB

-TD

MA

Pre

amb

le-B

ase

dT

DM

AH

eter

ogen

eou

ssu

pp

ort

for

dyn

amic

data

.C

an

pro

vid

ere

alti

me

gu

ara

nte

e.V

ery

low

ener

gy

con

-su

mp

tion

,ye

tle

ssla

ten

cyan

dhig

hth

rou

ghp

ut

isp

rovid

ed.

QoS

dep

end

son

the

pre

am

ble

and

tim

esy

nch

ron

-iz

atio

n.

Ull

ah

etal.

[42]

Bod

yM

AC

En

ergy

effici

ent

TD

MA

base

dB

SN

MA

CB

od

yM

AC

use

sfl

exib

lean

deffi

cien

tb

and

-w

idth

all

oca

tion

sch

emes

wit

hd

yn

amic

slee

pm

od

e.S

up

port

sd

yn

amic

app

lica

tion

sin

IBS

N.

Bet

ter

per

form

ance

inte

rms

of

the

end

-to

-en

dp

acke

td

elay

and

ener

gysa

vin

g

No

imp

lem

enta

tion

isd

one.

Res

ult

sare

bas

edon

soft

ware

sim

ula

tion

.H

igh

lyacc

ura

tegl

obal

syn

-ch

ron

izati

on

isre

qu

ired

.

Fan

get

al.

[37]

Tab

le3.

4:

TD

MA

-MA

Cp

roto

cols

for

IBS

N



a traditional CSMA/CA for critical and non-critical traffic. Due to high path loss inside thehuman body, the use of CSMA/CA does not provide reliable solution in multi-piconets [37].For a threshold of -85dBm and -95dBm, the on-body BNs cannot see the activity of in-bodyBNs when they are away at 3 meters distance from the body surface. However, within 3 metersor less distance, the CCA works correctly in the same piconet. The in-body MAC should alsoconsider the thermal influence caused by the electromagnetic wave exposure and circuit heat.Nagamine et al. discussed the thermal influence of the BNs using different MAC protocols [38].CA-MAC is a novel approach of using a threshold value for deciding whether the packets aretransmitted or not,based on the distance to the sink node. Energy efficient implementation forsmall scale dynamic network topology is achieved with CA-MAC. Latency is reduced by a noveladaptive algorithm based on the context of the packets. Simulation of CA-MAC by the authorsin [43] shows a reliable QoS parameters, required for the IBSN. Authors in [44] have presentedULP-MAC which increased network lifetime by 15%-300% more than other similar CSMA basedIBSN MAC.

Authors in [45] present B-MAC which renders typical properties of a IBSN such as simpleimplementation on hardware, highly predictable performance parameters, and tolerance to net-work changes. B-MAC also guarantees a reliable data packet delivery ratio of 98.5%. However,B-MAC is optimized only for star topology. B-MAC also suffers from hardware constraints suchas memory and computational extravaganza. In [46] Huq et al. presents MEB-MAC which fo-cuses on the channel access delay reduction for medical emergency traffic with high reliability.No energy efficiency is concerned in implementation and analysis. It has adverse effect on newnode insertion and mobility of network. An energy efficient MAC protocol called O-MAC ispresented in [47]. O-MAC has achieved increased energy efficiency by novel receiver schedulingmethods such as Staggered On and Pseudo-randomized Staggered On. Theoretical analysis andpractical implementation reveals that the protocol is 70% energy efficient than B-MAC, S-MACand T-MAC. CSMA based MAC protocol presented in [48], called DISSense. A good analysis ofMAC problems in different typologies and the benefits of cross layer optimization are addressedin the work presented. Features such as data delivery ratio, latency, duty cycling and adapt-ability are better than other similar protocols. DISSense can achieve good QoS in small scalenetworks and proportionately increase with network size. However, the performance of DISSenseis traded off with energy consumption. No clear analysis of energy-efficiency is carried out.

3.4.3 Hybrid MAC protocols for IBSN

CSMA and TDMA are the most common techniques used in a sensor network MAC. CSMA/CAis used for contention-based protocols. Majority of the CSMA/CA base MAC uses RTS andCTS packets before data communication. This causes lots of packet overheads.A TDMA-basedapproach has many advantages over other similar techniques such as CSMA/CA and FDMA.Authors of [12] have provided a comparison between TDMA and CSMA/CA as shown in Table 3.Otal and Alonso proposed an energy-saving MAC protocol, DQBAN (Distributed Queuing BodyArea Network) for WBAN in [47] as an alternative to the 802.15.4 MAC protocol which suffersfrom low scalability, low reliability and limited QoS in real-time environments. The proposedDQBAN is a combination of a cross-layer fuzzy-logic scheduler and energy-aware radio-activationpolicies. The queuing of access packets and data packets is determined by fuzzy-logic rules, whichpermit body sensors to find out ‘how favorable’ or ‘how critical’ their situation is in a given time-frame. The fuzzy-logic scheduling algorithm is shown to optimize QoS and energy-consumptionby considering cross-layer parameters such as residual battery lifetime, physical layer quality andsystem wait time. The authors tested their proposed protocol on two scenarios: a homogeneous



Nam

eof

the

pro

tocol

Desc

rip

tion

Sp

ecia

lfe

atu

reP

ote

nti

al

dra

wb

ack

Refe

ren

ce

CA

-MA

CC

onte

xt

Ad

apti

ve

MA

CP

roto

col

CA

-MA

Cis

an

ovel

ap

pro

ach

ofu

sin

ga

thre

shold

valu

efo

rd

ecid

ing

wh

eth

erth

ep

ack

ets

are

tran

s-m

itte

dor

not

,bas

edon

the

dis

tan

ceto

the

sin

kn

od

e.E

ner

gy

effici

ent

imp

lem

enta

tion

for

small

scale

dyn

am

icn

etw

ork

top

olo

gyis

ach

ieved

wit

hC

A-M

AC

.L

ate

ncy

isre

du

ced

by

an

ovel

ad

apti

vealg

ori

thm

bas

edon

the

conte

xt

of

the

pack

ets.

Com

pu

tati

on

al

com

ple

xit

yis

hig

her

wh

ich

isa

thre

at

tosm

all

erre

sou

rce

con

stra

int

nod

esan

dlo

ng

term

net

work

op

erati

on.

Eva

luati

on

of

the

pro

toco

lis

lim

ited

wit

hth

eore

tica

ld

ata

an

did

eal

ass

um

pti

ons.

Kim

etal.

[43]

PN

P-M

AC

Pre

emp

tive

slot

allo

cati

onan

dN

on-P

reem

pti

vetr

ansm

issi

onM

AC

Su

pp

ort

sva

rious

typ

esof

traffi

cs:

conti

nu

ous

stre

am

ing,p

erio

dic

data

,ti

me-

crit

ical

emer

gen

cyal

arm

,as

wel

las

non

-p

erio

dic

data

.H

igh

lyre

liab

leQ

oS

sup

port

.N

ovel

com

bin

ati

on

of

conte

nti

on

-fre

ean

dco

nte

nti

onacc

ess

mec

han

-is

ms.

Su

ffer

sfr

omse

vere

reso

urc

eex

hau

stio

n.

En

ergy

con

sum

pti

on

isn

ot

con

sid

ered

as

acr

iter

iafo

rd

esig

n.

QoS

will

be

trad

edoff

wit

hen

ergy

effi-

cien

cyan

dd

yn

am

icn

etw

ork

top

olo

gy.

Yoon

etal.

[49]

UL

P-M

AC

An

Ult

ra-l

ow-p

ower

Med

ium

Ac-

cess

Con

trol

Pro

toco

lfo

rB

od

yS

enso

rN

etw

ork

Acr

oss

laye

rd

esig

nst

rate

gy

isad

op

ted

.N

etw

ork

coor

din

ator

an

dth

ese

nso

rsin

tera

ctto

ach

ieve

ef-

fici

ent

pow

erm

an

agem

ent.

Var

iab

lesu

per

-fra

me

stru

ctu

reis

ad

ap

ted

.IB

SN

coord

inato

rca

nm

ake

dyn

amic

ad

just

men

tb

ase

don

the

feed

back

toac

hie

veb

ette

rp

erfo

rman

cein

ener

gy

effici

ency

and

late

ncy

.

Op

tim

ized

for

star

top

olo

gy.

Su

ffer

sfr

om

hard

-w

are

con

stra

ints

such

as

mem

ory

an

dre

al-

tim

egu

ara

nte

e.S

imu

lati

on

isca

rrie

dou

tw

ith

idea

ln

etw

ork

con

dit

ion

s.

Li

etal.

[39]

B-M

AC

Ber

kele

y-M

AC

Ver

sati

leL

owP

ower

MA

Cp

roto

col

BM

AC

ren

der

sp

rop

erti

esof

IBS

Nsu

chas

sim

ple

imp

lem

enta

tion

onh

ard

ware

,pre

dic

tab

lep

er-

form

an

cep

ara

met

ers,

and

tole

ran

ceto

net

work

chan

ges

.H

igh

lyre

liab

led

ata

pack

etd

eliv

erof

98.5

%

Ver

yw

ell

suit

edfo

rst

ar

top

olo

gy

net

wor

ks.

Inca

seof

chan

ge

inn

etw

ork

top

olo

gy

the

pro

toco

lh

ind

ers

per

form

an

ce.

En

ergy

effici

ency

can

on

lyb

eex

pec

ted

wh

enin

terf

ace

dw

ith

diff

eren

tse

rvic

esre

sult

ing

incr

oss

-lay

erop

tim

izati

on

.

Pola

stre

etal.

[45]

DIS

Sen

seA

nad

apti

ve,

Ult

ralo

w-p

ower

MA

Cpro

toco

lC

ross

laye

rop

tim

izati

onis

sues

are

con

sid

ered

.F

eatu

res

such

as

dat

ad

eliv

ery

rati

o,

late

ncy

,d

uty

cycl

ing

an

dad

apta

bil

ity

are

bet

ter

than

oth

ersi

mil

arp

roto

cols

.C

anach

ieve

good

QoS

insm

all

scale

net

work

s.

Per

form

an

ceis

trad

edoff

wit

hen

ergy

con

sum

p-

tion

.N

ocl

ear

an

aly

sis

of

ener

gy-e

ffici

ency

isca

rrie

dou

t.D

esig

ned

for

the

pu

rpose

of

larg

esc

ale

and

cove

rage

net

work

s.

Cole

santi

etal.

[48]

ME

B-M

AC

Med

ical

Em

ergen

cyB

od

y(M

EB

)M

AC

ME

B-M

AC

focu

ses

on

the

chan

nel

acce

ssd

elay

red

uct

ion

for

med

ical

emer

gen

cytr

affi

cw

ith

hig

hre

liab

ilit

y.

Imp

lem

enta

tion

isd

on

ein

real-

worl

dsc

enari

os.

How

ever

,n

oen

ergy

effici

ency

isco

nce

rned

.It

has

ad

vers

eeff

ect

on

new

nod

ein

sert

ion

an

dm

ob

ilit

yof

net

work

Hu

qet

al.

[46]

O-M

AC

Oh

ioS

tate

Un

iver

sity

,O

hio

-M

AC

Incr

ease

den

ergy

effici

ency

by

nov

elre

ceiv

ersc

hed

uli

ng

met

hod

ssu

chas

Sta

gger

edO

nan

dP

seu

do-r

and

om

ized

Sta

gger

edO

n.

Th

eore

tica

lan

aly

sis

an

dpra

ctic

al

imp

lem

enta

tion

revea

lsth

atth

ep

roto

col

is70

%en

ergy

effici

ent

than

B-M

AC

,S

-MA

Can

dT

-MA

C.

Qos

isn

ot

con

sid

ered

,p

ara

met

ers

such

as

late

ncy

an

dth

rou

gh

pu

tare

not

evalu

ate

d.

Cao

etal.

[47]

Tab

le3.

5:

CS

MA

-MA

Cp

roto

cols

for

IBS

N



scenario of a body sensor network with 5 – 35 homogeneous ECG wireless sensors and a het-erogeneous scenario of a body sensor network with wireless ECG sensors and four other sensorsfor clinical doctor PDA, respiratory rate, blood pressure and endoscope imaging. The authorsshowed that the DQBAN protocol has higher reliability, while fulfilling certain battery limits andlatency demands and displays energy-saving behaviour compared to most MAC implementations.

3.4.4 Other Access Mechanisms for IBSN (FDMA, UWB, ALOHA)

CSMA and TDMA are the most common techniques used in a sensor network MAC. CSMA/CAis used for contention-based protocols. Majority of the CSMA/CA base MAC uses RTS and CTSpackets before data communication. This causes lots of packet overheads. Use of wakeup radiocan minimize the extra power consumption by the RTS-CTS packet exchange which is doneby the main radio. TDMA is popular in synchronous MAC. Authors of [19-21] have proposedTDMA base MAC protocols for BAN. A TDMA-based scheme combined with wakeup radio canbe used to design a power efficient MAC. A TDMA-based approach has many advantages overother similar techniques such as CSMA/CA and FDMA. Pure ALOHA is the first random accesstechnique introduced and it is so simple that its implementation is straight forward. It belongsto the family of contention-based protocols, which do not guarantee the successful transmissionin advance. In this whenever a packet is generated, it is transmitted immediately without anyfurther delay. Successful reception of a packet depends only whether it is collided or not withother packets. In case of collision, the collided packets are not received properly. At the end ofpacket transmission each user knows either its transmission successful or not. If collision occurs,user schedules its re-transmission to a random time. The randomness is to ensure that samepacket do not collide repeatedly. Each packet is belongs to a separate user due to the fact thatpopulation is large. In ALOHA technique node checks for the availability of data packets to betransmitted. If they are available then node transmits them otherwise process ends.

Slotted ALOHA is a variant of Pure ALOHA with channel is divided into slots. Restriction isimposed on users to start transmission on slot boundaries only. Whenever packets collide, theyoverlap completely instead of partially. So only a fraction of slots in which packet is collided isscheduled for re-transmission. It almost doubles the efficiency of Slotted ALOHA as comparedto Pure ALOHA. Successful transmission depends on the condition that, only one packet istransmitted in each frame. If no packet is transmitted in a slot, then slot is idle. Slotted Aloharequires synchronization between nodes which lead to its disadvantage.

FDMA is a basic technology in analog Advanced Mobile Phone Service (AMPS), most widely-installed cellular phone system installed in North America. With FDMA, each channel can beassigned to only one user at a time. Each node share medium simultaneously though transmitsat single frequency. FDMA is used with both analog and digital signals[56]. It requires high-performing filters in radio hardware, in contrast to TDMA and CSMA. As each node is separatedby its frequency, minimization of interference between nodes is done by sharp filters. In FDMAa full frame of frequency band is available for communication, In FDMA a continuous flow ofdata is used, which improves efficiency of sending data.

In [54], the authors proposed the use of a UWB transmitter for energy-efficient operation ofWBANs. Due to the high interference generated by the human body and its environment, onesolution to develop low power-output transceivers for radios in the sensor nodes is to optimizethe air interface of the network. By creating architectures that exploit the features of robust



Nam

eof

the

pro

tocol

Desc

rip

tion

Sp

ecia

lfe

atu

reP

ote

nti

al

dra

wb

ack

Refe

ren

ce

X-M

AC

Sh

ort

Pre

amb

leM

AC

Pro

toco

lfo

rD

uty

-Cycl

edW

irel

ess

Sen

sor

Net

wor

ks

Low

pow

erco

mm

un

icat

ion

isd

eplo

yed

by

ast

rob

edp

ream

ble

app

roac

hth

attr

ansm

its

ase

ries

ofsh

ort

pre

amb

lep

ack

ets

toth

eta

rget

rece

iver

.T

run

cati

onth

ep

ream

ble

by

the

targ

etre

ceiv

ersa

ves

ener

gyat

bot

hth

etr

ansm

itte

ran

dre

ceiv

eran

din

trod

uce

low

erla

ten

cy.

Nea

r-op

tim

alsl

eep

and

list

enp

erio

ds

are

dem

onst

rate

d.X

-MA

Cou

t-p

erfo

rms

trad

itio

nal

Low

-pow

erli

sten

ing.

Hig

hla

ten

cyin

the

even

tof

ap

ack

etfa

ilu

re.

Re-

qu

ires

acc

ura

tesy

nch

ron

isat

ion

Bu

ettn

eret

al.

[50]

S-M

AC

Sen

sor

MA

CG

ood

ener

gyco

nse

rvin

gp

rop

erti

esw

ith

anab

ilit

yto

mak

etr

ade-

offs

bet

wee

nen

ergy

and

late

ncy

acco

rdin

gto

traffi

cco

ndit

ion

s.T

he

pro

toco

lh

asb

een

imp

lem

ente

deffi

cien

tly

inh

ard

war

eat

real

-worl

dsc

enar

ios.

Sca

lab

ilit

yis

sues

are

not

add

ress

ed.

Net

work

top

olog

yis

con

sid

ered

con

stan

tw

ith

con

stant

nu

mb

erof

nod

es.

Ye

etal.

[18]

V-M

AC

Vir

tual

MA

CV

MA

Cis

emb

edd

edin

Bod

yQ

oSto

mak

eit

rad

io-a

gnos

tic,

soth

atit

can

contr

olan

dsc

hed

ule

wir

eles

sre

sou

rces

wit

hou

tkn

owle

dge

ofth

eim

ple

men

tati

ond

etai

lsof

the

un

der

lyin

gM

AC

pro

toco

l.B

od

yQ

oSad

opts

anas

ym

met

ric

ar-

chit

ectu

re,

inw

hic

hm

ost

pro

cess

ing

isd

one

atth

ere

sou

rcef

ul

aggr

egat

orw

hil

ele

ssp

roce

ssin

gis

don

eat

the

reso

urc

eli

mit

edse

nso

rn

od

es.

En

ergy

effici

ency

isn

otco

nsi

der

edat

all

.E

valu

-at

ion

ofQ

oSp

aram

eter

sis

give

nm

ore

imp

ort

an

ceth

anth

atof

the

ener

gyco

nce

rns.

Zh

ou

etal.

[51]

R-M

AC

Res

erva

tion

Med

ium

Acc

ess

Con

-tr

olP

roto

col

Avo

idan

ceof

over

hea

rin

g,fr

equ

ent

com

mu

tati

onb

etw

een

slee

pan

dw

ake

up

mod

es,

and

dat

aco

llis

ion

sar

ego

od

resu

lts

ofth

isn

ovel

app

roac

h.

R-

MA

Cp

roto

col

also

adju

sts

the

du

rati

onof

the

slee

pan

dac

tive

per

iod

sac

cord

ing

toth

etr

affic

load

inor

der

toav

oid

dat

aco

llis

ion

s.

Not

very

ener

gyeffi

cien

tin

low

data

rate

ap

pli

ca-

tion

.A

imed

ath

igh

dat

ara

teap

pli

cati

on

inla

rge

scal

en

etw

orks

Yes

sad

etal.

[52]

UB

-MA

CU

rgen

cy-b

ased

MA

CP

roto

col

Cri

tica

ln

od

esp

acket

tran

smis

sion

sar

ep

rior

itiz

edov

ern

oncr

itic

aln

od

espac

ket

tran

smis

sion

s.T

he

pro

pos

edp

roto

col

ison

lyev

alu

ate

dm

ath

em-

atic

ally

.N

etw

ork

may

fail

for

diff

eren

tn

etw

ork

top

olog

yan

dnu

mb

erof

nod

esin

an

etw

ork

isli

mit

ed

Ali

etal.

[53]

EE

E-M

AC

En

ergy

Effi

cien

tE

lect

ion

bas

edM

AC

Pro

toco

lA

lgor

ith

mis

good

atp

rese

rvin

gn

etw

ork

top

ology

and

con

nec

tivit

yw

hil

ein

trod

uci

ng

orre

du

cin

gex

tra

nod

es.

Sm

all

erra

teof

dev

iati

onin

ener

gyco

nsu

mp

tion

inh

igh

erd

ata

load

con

dit

ion

s.E

ner

gyeffi

cien

cyis

good

com

par

edto

S-M

AC

and

B-M

AC

Th

ep

roto

col

isn

otan

alyse

dfo

rQ

oS

para

met

ers.

Itis

stat

edth

atQ

oSm

ayh

ind

erth

een

ergy

effi-

cien

cyfo

rsm

alle

rn

etw

orks

Ud

ayaku

mar

etal.

[54]

FE

-MA

CF

orw

ard

ing

Ele

ctio

n-b

ased

MA

Cp

roto

col

Hig

hn

etw

ork

life

tim

ew

ith

ener

gyeffi

cien

cyan

dlo

adb

alan

ce.

Rou

tin

gca

pab

ilit

yof

the

net

wor

kla

yer

isal

soem

bed

ded

inth

ep

roto

col.

Hig

hly

scal

able

and

ener

gy-e

ffici

ent

wit

hm

ore

nu

mb

erof

nod

es

Res

ourc

eu

tili

zati

onis

exh

aust

ive.

Req

uir

esa

rela

tive

lyla

rge

mem

ory

and

hig

hco

mp

uta

tion

al

pow

er.

Qia

ng

etal.

[55]

Tab

le3.

6:

hyb

rid-M

AC

pro

toco

lsfo

rIB

SN



nodes and energy-constrained nodes, low latency and simple network topologies, the authorsshow that an UWB based architecture is advantageous over narrowband radio communication.The use of UWB has been further explored in [55]. Here the authors described the use of UWBfor an in-body WBAN application of capsule endoscopy. Within the 402-405 MHz frequencyband, allowed by the FCC for in-body communication systems, UWB communication has shownto be most effective for integrating in-body and on-body medical sensors into a single system.With the help of radio channel simulation results, the authors presented the link capacity, signalpower spectral density and interference mitigation. Finally a WBAN coordinator acts as aninterface between the network of sensors and a medical server. The current state of this work isdocumented under the MELODY Project [56, 57].



Nam

eof

the

pro

tocol

Desc

rip

tion

Sp

ecia

lfe

atu

reP

ote

nti

al

dra

wb

ack

Refe

ren

ce

Coop

era

tive-M

AC

Low

duty

cycl

eT

DM

AS

uit

able

for

hig

hly

mob

ile

nod

es.

Nov

elco

m-

bin

atio

nof

TD

MA

wit

hF

DM

Ad

eals

wit

hth

ein

terf

eren

cean

dco

llis

ion

cau

sed

by

the

mob

ile

clu

ster

.T

he

coll

isio

ns

bro

ugh

tby

the

mob

ile

clu

ster

are

avoid

edth

rou

ghd

iffer

ent

freq

uen

cies

use

din

WB

AN

Com

ple

xh

ard

war

eis

requ

ired

.R

esou

rce

uti

liza

-ti

on

isex

hau

stiv

ere

nd

erin

gle

ssp

ower

effici

ency

Rom

an

etal

.[5

7]

Hyb

rid

-MA

CH

yb

rid

(TD

MA

+F

DM

A)

MA

CP

roto

col

Red

uce

din

terf

eren

cein

the

inte

rcl

ust

eran

din

tra

clu

ster

com

mu

nic

atio

nu

sin

gn

ovel

com

bin

ati

on

ofF

DM

Aan

dT

DM

Ate

chn

iqu

es.

Ach

ieve

sle

ssen

ergy

con

sum

pti

on

.F

ulfi

lls

the

ban

dw

idth

requ

irem

ent

of

each

nod

ein

the

sen

sor

net

work

.H

ere

aft

erb

an

dw

idth

div

isio

nea

chn

od

eget

sch

ann

elw

hos

eb

and

wid

this

more

than

the

requ

irem

ent.

Imp

lem

enta

tion

isea

sy.

Les

sre

liab

le,

suff

ers

from

hig

hp

acke

td

rop

for

hig

her

dat

alo

ad

scen

ari

os.

Mu

kh

erje

eet

al.

[58]

Hy-M

AC

Hyb

rid

TD

MA

/FD

MA

MA

CP

roto

col

An

ovel

ap

pro

ach

wh

ich

sch

edule

sth

en

etw

ork

nod

esin

aw

ayth

at

elim

inat

esco

llis

ion

san

dp

rovid

essm

all

bou

nd

eden

d-t

o-en

dd

elay

an

dh

igh

thro

ugh

pu

t.It

takes

ad

vanta

geof

mu

ltip

lefr

equ

enci

esav

aila

ble

inst

ate-

of-t

he-

art

sen

sor

nod

eh

ard

ware

pla

tfor

ms

such

as

MIC

AZ

,T

E-

LO

San

dF

ireF

ly.

Ou

t-of-

band

syn

chro

niz

ati

on

iseff

ecti

ve,

ren

der

ing

TD

MA

mec

han

ism

effici

entl

y

Can

not

be

imp

lem

ente

din

con

serv

ati

vera

dio

ban

dsu

chas

MIC

Sw

her

eth

enu

mb

erof

chan

nel

avail

ab

leis

hig

hly

lim

ited

.N

ot

effici

ent

inte

rms

of

ener

gy

Sala

jegh

ehet

al.

[59]

HU

A-M

AC

Hyb

rid

Un

ified

-Slo

tA

cces

sM

AC

Pro

toco

lT

he

spec

ial

des

ign

edm

ini-

slot

met

hod

incr

ease

sth

eco

nte

nti

on

effici

ency

.Con

tenti

on

-fre

ed

ata

traffi

csc

hem

ew

asad

op

ted

togu

ara

nte

eth

eQ

oS

.A

lloca

tion

of

slots

isad

apti

veto

the

traffi

clo

ad

.In

crea

sed

scala

bil

ity

an

dro

bu

stn

ess

for

aB

AN

.

Su

ffer

sfr

om

seve

reli

mit

atio

ns

from

state

-of-

the-

art

har

dw

are.

Rea

lw

orl

dim

ple

men

tati

on

was

car-

ried

out

wit

hid

eal

assu

mp

tion

sof

net

wor

kp

ara-

met

ers.

.E

ner

gyeffi

cien

cyis

lagg

ing

Li

etal

.[3

9]

YN

U-M

AC

YN

UJap

an,

Ult

ra-W

ideB

and

MA

Cpro

pos

alP

roto

col

con

sid

ers

SA

Ror

ther

mal

infl

uen

ceto

hu

man

bod

yby

swit

chin

gcl

ust

erm

ech

an

ism

.P

osi

tion

ing

or

loca

liza

tion

of

BA

Nn

od

esis

hig

hly

poss

ible

Diff

ernt

sup

ple

men

tary

tech

olog

ies

yet

tob

ean

a-

lyze

d.

Imp

lem

enta

tion

isn

otp

ossi

ble

wit

hC

OT

Sh

ard

war

e

En

da

etal

.[6

0]

FM

-UW

BM

AC

CS

EM

Sw

itze

rlan

d,

Fre

qu

ency

Mod

ula

tion

-U

ltra

Wid

eBan

dM

AC

pro

pos

al

Low

ener

gy

at

the

tran

smit

ter

an

dn

als

osa

ves

ener

gy

atth

ed

esti

nat

ion

nod

eas

itd

oes

not

hav

eto

list

ento

aco

mp

lete

wake

-up

pre

am

ble

.S

uff

ers

less

from

over

hea

rin

g.R

edu

ced

chan

nel

usa

ge

an

dth

ereb

yco

llis

ion

s.Im

pro

ved

reli

ab

ilit

yan

dre

du

ced

late

ncy

Extr

eme

requ

irem

ent

for

hard

ware

com

pare

dto

oth

erm

ech

an

ism

s.N

oop

tim

al

physi

cal

laye

rd

esig

nis

pro

pos

ed

Far

sero

tuet

al.

[61]

Tab

le3.

7:F

DM

A,

UW

B,

AL

OH

Ab

ase

dacc

ess

mec

hanis

ms



3.5 Access mechanisms with wake-up radio

Use of wakeup radio can minimize the extra power consumption by the RTS-CTS packetexchange which is done by the main radio. TDMA is popular in power efficiency and are appliedin small scale networks such as IBSN . A TDMA-based scheme combined with wakeup radio canbe used to design a power efficient MAC. The inter-arrival parameters are re-configurable valuesfor each BN. For example, in case of a patient, a doctor/nurse or in-charge person can set thepacket inter-arrival time for temperature monitor (node BN-008) to be 6 h or 21,600 s. This willcause the BNC to send a wakeup radio signal to the particular BN after the specified intervalsand complete the data communication. The BN, between two consecutive wakeup periods canswitch off its main radio and go to sleep state to save power. It does not have to contend for datacommunication. The inter-arrival time can be reset and reconfigured as the need arises. A nodelike heart rate monitor can have very low inter-arrival time (e.g., 100 ms) for a particular amountof time such as exercise or operations. It can be reset again to match other desired level of datacommunication. We have assumed that every node in the network uses wakeup radio with BNCmanaging a wakeup scheduling table. CSMA and TDMA are the most common techniques usedin a sensor network MAC. CSMA/CA is used for contention-based protocols. Majority of theCSMA/CA base MAC uses RTS and CTS packets before data communication. This causes lotsof packet overheads. Use of wakeup radio can minimize the extra power consumption by theRTS-CTS packet exchange which is done by the main radio. TDMA is popular in synchronousMACs. Authors of [39-41] have proposed TDMA based MAC protocols for BAN with wake-upradio. A TDMA-based scheme combined with wakeup radio can be used to design a powerefficient MAC. A TDMA-based approach has many advantages over other similar techniquessuch as CSMA/CA and FDMA. Authors of [62] have provided a comparison between TDMAand CSMA/CA. An Ultra Low Power and Traffic adaptive protocol designed for WBANs isdiscussed in [63]. They used a traffic adaptive mechanism to accommodate on-demand andemergency traffic through wake-up radio. Authors of [64] have proposed a MAC which supportsdependability and QoS guarantee for the most important life-critical message and majority realtime traffic. The protocol can be used with different physical layers UWB, MICS, WMTS, HBC.A dynamic network size from greater than 6 nodes to lesser than 100 nodes per network can beachieved. A Improved quality-of-service addressing throughput, access latency, priority. Highscalability is realized. Star, cluster-tree and the peer-to-peer, are supported in the MAC protocolproposed in [65]. Table 3.8 comprehends the MAC protocols based on the special feature andpotential drawback.A wakeup mechanism based on traffic intensity at each node is used forcommunication in case of normal traffic. Each node maintains the wakeup schedule in a table forevery node in the network which is constructed based on traffics at the particular BN. Wakeupinterval is calculated from inter-arrival of packets for a BN. Use of wakeup table by BNC savesa significant amount of power as all BNs in the network remain in the sleep state (i.e., switch offthe main transceiver) until it is woken up by the BNC.

From the best of the authors knowledge, if a protocol support dynamic rates and multi-hopscheduling, they often focus on always-on radios, which is not feasible in WBANs. Wake-upradios are good alternative for always on radios, however there is no much work being done inthis perspective of the WBAN.

Ad-hoc mesh networks can also be suited for closed-loop In-body sensor networks. Basedon the different characteristic of the ad-hoc networks, Authors in [74] and [75] found out thatapart from all the different topologies, ad-hoc strategy can be applied for body-sensor network.However, the general characteristics of the Ad-hoc network is the power consumption, assume

34 3.5. Access mechanisms with wake-up radio


Nam

eof

the

pro

tocol

Desc

rip

tion

Sp

ecia

lfe

atu

reP

ote

nti

al

dra

wb

ack

Refe

ren

ce

NIC

T-M

AC

NIC

TJap

an,

Pro

pos

alfo

rM

AC

usi

ng

wak

e-up

radio

Supp

ort

for

pro

vid

ing

dep

endab

ilit

yan

dQ

oS

guara

nte

efo

rth

em

ost

imp

ort

ant

life

-cri

tica

lm

essa

ge

and

majo

rity

real

tim

etr

affic.

Can

be

use

din

diff

eren

tphysi

cal

layer

sU

WB

,M

ICS,

WM

TS,

HB

C.

Adynam

icnet

work

size

from

grea

ter

than

6nodes

tole

sser

than

100

nodes

per

net

work

can

be

ach

ieved

.

Pow

erco

nsu

mpti

on

hig

her

due

tohig

hp

erfo

rm-

ance

.A

LO

HA

isuse

din

stea

dofC

SM

A/C

Aw

hic

hm

ayre

sult

inabru

pt

per

form

ance

dev

iati

onin

larg

enet

work

size

Zhen

etal.

[64]

IME

C-M

AC

IME

CN

arro

wban

dM

AC

pro

-p

osal

Impro

ved

quality

-of-

serv

ice

addre

ssin

gth

rough-

put,

acc

ess

late

ncy

,pri

ori

ty.

Hig

hsc

alab

ilit

yis

realize

d.

Sta

r,cl

ust

er-t

ree

and

the

pee

r-to

-pee

r,ar

esu

pp

ort

ed.

The

des

ign

ispro

ne

toco

llis

on

and

Low

reso

urc

eeffi

cien

cy.

Ener

gyco

nsu

mpti

on

isve

ryhig

hco

m-

par

edto

other

sim

ilar

pro

toco

ls.

Dolm

ans

etal

.[6

5]

Mil

ler-

MA

CA

MA

CP

roto

col

toR

educe

Sen

sor

Net

wor

kE

ner

gyC

on-

sum

pti

onU

sing

aW

akeu

pR

adio

The

pro

toco

lis

wel

lsu

ited

for

mult

iple

hop

and

mult

iple

flow

scen

ari

os,

outp

erfo

rmin

gsi

mil

arpro

-to

cols

inte

rms

of

ener

gyand

late

ncy

.

Addit

ional

har

dw

are

isre

quir

edto

allo

wse

nder

sto

forc

ere

ceiv

ers

tow

ake

-up

when

asp

ecifi

ednum

ber

ofpac

ket

sar

ebuff

ered

.E

xtr

ahar

dw

are

consu

med

extr

aen

ergy

,w

hic

his

not

analy

zed

inth

edes

ign.

Mille

ret

al.

[66]

RT

WA

C-M

AC

Rad

ioT

rigg

ered

Wak

e-up

wit

hA

ddre

ssin

gC

apab

ilit

ies

MA

CR

TW

AC

reduce

sid

lelist

enin

gand

also

suppre

ssunnec

essa

ryra

dio

wake

-ups

due

toth

ead

dre

ssin

gin

form

ati

onin

cluded

inth

ew

ake

-up

sign

al.

Ver

yle

ssp

ower

consu

mpti

on

and

low

late

ncy

than

other

MA

Cpro

toco

ls(S

MA

CT

MA

Cand

BM

AC

)

No

expli

cit

anal

ysi

sof

QoS

wit

hre

spec

tto

the

ener

gyco

nsu

mpti

on.

Furt

her

wor

kis

nee

ded

toin

tegra

tew

ith

diff

eren

tM

AC

pro

toco

lsfo

rdat

aco

mm

unic

atio

nusi

ng

mai

nra

dio

.

Ansa

riet

al.

[67]

PE

-MA

CP

ower

effici

ent

MA

Cusi

ng

Wak

e-up

radio

PE

-MA

Cis

aT

DM

Abas

edM

AC

usi

ng

wak

eup

radio

that

can

save

am

ore

than

50%

ofen

ergy

use

din

CSM

A\C

Aw

hile

still

hav

ing

alo

wdel

ayin

data

transf

er

Hig

her

net

wor

ksi

zeis

not

consi

der

ed,

rath

eronly

two

nodes

wer

euse

dto

eval

uate

the

pro

toco

l.Q

oSand

other

net

work

par

am

eter

ssu

chas

inte

rfer

ence

,del

ayar

eid

eally

assu

med

inth

eev

aluati

on.

Am

een

etal

.[6

8]

UL

PA

-MA

CU

ltra

Low

Pow

erA

synch

ronou

sM

AC

Pro

toco

lusi

ng

Wak

e-U

pR

adio

This

appro

ach

can

impro

veup

to82%

QoS

and

53%

ener

gysa

vin

gw

hen

consi

der

ing

wit

hT

ICE

R[6

9]

pro

toco

lfo

rw

irel

ess

com

munic

atio

ns.

Suff

ers

from

seve

redata

and

wake

-up

bea

con

col-

lisi

ons

ina

hig

htr

affic

net

work

,w

hic

hre

duce

sth

eav

erag

edata

rece

ived

rate

by

5.3

9%

.

Le

etal

.[7

0]

Wu

RM

AC

Wak

e-up

radio

MA

CB

yel

imin

atin

glo

wp

ower

pollin

gfo

rdet

ecti

ng

channel

act

ivit

y,th

ism

ethod

pro

vid

esm

ore

ener

gy-e

ffici

ent

solu

tions

than

B-M

AC

and

S-

MA

C.

Ver

ylo

wp

er-h

op

late

ncy

and

aver

age

pow

erco

nsu

mpti

on

Mat

hem

atic

al

analy

sis

isdone

rath

erth

an

real

wor

ldim

ple

men

tati

on.

Rea

lw

orld

diffi

cult

ies

are

bypass

edw

ith

assu

mpti

ons

Ker

sten

etal

.[7

1]

T-M

AC

Adap

tive

Ener

gy-E

ffici

ent

MA

CP

roto

col

Han

dle

slo

ad

vari

atio

ns

inti

me

and

loca

tion

by

adapti

veduty

cycl

ein

anov

elw

ay.

Red

uce

sth

eam

ount

of

ener

gy

wast

edon

idle

list

enin

g

Thro

ugh

put

istr

aded

off

wit

hen

ergy

effici

ency

.E

xp

erim

ente

din

stat

icand

non

-mob

ile

net

wor

ks

VanD

amet

al.

[72]

TB

CD

-TD

MT

ime-

Bas

edC

oded

Dat

a-T

ime

Div

isio

nM

ult

iple

xin

g28

0ti

mes

hig

her

thro

ughput

than

Zig

Bee

pro

-to

col.

Sim

ple

modula

tion

tech

niq

ues

,re

quir

esve

ryle

sshard

ware

com

ple

xit

y

Idea

lca

seof

only

one

single

tran

smit

ted

data

bit

per

round

isuse

d.

No

real

-wor

ldim

ple

men

tati

onis

dis

cuss

edsu

chas

effec

tof

envir

onm

ent

noi

seeff

ect.

For

ouza

ndeh

etal.

[73]

Tab

le3.

8:M

AC

pro

toco

lsw

ith

wake

up

-rad

iofo

rIB

SN

3.5. Access mechanisms with wake-up radio 35


power supply in the range of a laptop or ever wall-power socket. Main characteristic of thead-hoc network is the power consumption due to always on radio. In the case of the body sensornetworks, the power is limited entity and radio cannot be turned on always.By carefully looking at the literature it is found that the wake-up radios are coming up withlower power consumption and can suit the need of always on listening strategies. The recentresearches were mostly on single-hop network with star-topology, consisting of one master nodeand different slave nodes. The main reason to select this type of network topology is to reducethe computational burden on the implanted nodes. The idea of having mesh network in tree-structure or ad-hoc fashion is not well researched due to the reason that computational effortof implant nodes are minimal. However, there are very few researches being done with the aimof having tree based network traversal for the in-body sensor networks. In [48], the authorshave presented an intelligent algorithm which can be used for implanted sensor nodes where thescalability is not a big issue.Combination of wake-up radio and mesh topology should be suited for the IBSN, meeting therequirements such as energy-efficiency, reliability, and Qos.

3.6 Discussion

The state-of-the art MAC protocols are analysed focusing on IBSN and BSN. Different pro-tocols have unique features exactly suited for the IBSN applications, and at the same timedrawbacks which renders the MAC protocols not completely useful. In order to exhibit thiscontradiction in selecting a best suited protocol, Table 3.9 presents an overall comparison of allthe protocols studied. In general for the purpose of the IBSN, certain features are considered fora reliable performance. It is important to select characteristic features that are most importantfor the operation of Implantable sensor nodes.

The features that are selected for this comparison are based on the requirements from Sec-tion 2.4 and common added values of the MAC protocols studied:

• Low latency

• Hardware complexity

– Frequency of operation

– Bandwidth utilization

– Effective radiated power

• Energy efficiency

– Duty cycle

– Energy aware operation of main radio and wake-up radio.

• Software overhead

– Headers for link establishment

– Cross-layer features

– Higher payload data per frame

• Reliability

36 3.6. Discussion


– Packet delivery ratio

– Retransmission mechanism and redundancy in data retransmission.

• Effective throughput

– Data rate

– Higher payload

• Quality of Service

More than 40 protocols have been considered in order to broadly evaluate their use in IBSN.A + indicates that the specific feature is an advantage for IBSN and a − indicates that thefeature is a disadvantage for IBSN. Table 3.9 is a classification of different protocols based onnetwork features such as mentioned in the list.The choice analysing the MAC protocols basedon the specific network parameters is explained below.

Energy efficiency is the overall consumption of energy of a sensor node in establishing a linkfor data transfer. A MAC protocol determines the energy efficiency by controlling the accessto the wireless medium. A efficient MAC protocol should have minimum access to the wirelessmedium and at the same time not compromising the performance of the underlying network. In acomplex IBSN, the longevity of the network is decided by the power available. Hence, analysingthe MAC protocol based on the energy efficiency and identifying the major design flaw in existingdesign is important. Reliability is the successful data transfer from a Tx node to a Rx node.Reliable MAC protocols typically incur more overhead than unreliable protocols, and as a result,are slower and less scalable. This is not an issue for star-topology based unicast communicationin IBSN. Analysis of reliability is to ensure that all the emergency data can be communicated tothe base station, while not compromising the energy efficiency. In a network Software overheadis the non-useful data which is used to successfully establish a link between Tx and Rx. In orderto understand the how the reliability is affected with software overhead, each protocol have beenevaluated with the software overhead incurred in each protocol. Quality of Service parametersconsidered in this analysis include error rate, bandwidth, throughput and transmission delay. InIBSN, the MAC protocol may not provide all the QoS features, but knowing the dependencyof the QoS with energy efficiency and reliability is important. Latency is the amount of timerequired to establish a link before any useful data transfer. In case of IBSN, latency of upto 60ms can be tolerated [3]. However, increase in latency directly affects the reliability and networkfailure if a time critical application is considered. The main reasons for the latency in a networkcan be due to the path loss and RF interferences. But latency due to overheads can also be areason in some of the MAC protocols as discussed in section 3.4.1. Thus, considering the IBSNscenario it is important that the latency needs to be evaluated for a given access mechanism.Hardware complexity is hardware components that are required for establishing a link in wirelessmedium, for example, a extra transceiver is required in case of wake-up radio mechanisms apartfrom the main radio transceiver.All the MAC protocol mentioned in the table 3.9 are focussed on the IBSN application scenario,while none of them is explicitly developed for IBSN applications. Although some of the protocolsare developed for general WSN application, it is still considered since, the features of the protocolalso meet the requirement of the IBSN application. Table 3.9 is a qualitative analysis of theprotocols based on the requirements of IBSN application.

3.6. Discussion 37


MAC protocol

Features

Energy-efficiency

Reliability Software-overhead

Effective-throughput

QoS Low-latency

Hardware complexity

CF-MAC[38]Y-2005

+ + + ++ + + + + −− + TDMA w/o WuR

SSD-TDMA [76]Y-2005

+ + + −− − + + + −− + + + TDMA w/o WuR

DQBAN MAC[40]Y-2009

−− − + + + + − − TDMA w/o WuR

HEH-MAC [41]Y-2007

−− + −− ++ −− + + + TDMA w/o WuR

BodyMAC [37]Y-2009

+ + + + −− ++ −− −− TDMA w/o WuR

UB-MAC [53]Y-2010

−− ++ − + + + − + + + TDMA+CSMA w/o WuR

X-MAC[50]Y-2006

+ + + −− ++ −− + + + + TDMA+CSMA w/o WuR

V-MAC [51]Y-2008

+ −− + + + −−− − ++ TDMA+CSMA w/o WuR

R-MAC [52]Y-2007

+ + + − − + + + − + + + TDMA+CSMA w/o WuR

PNP-MAC [49]Y-2010

−−− + −−− + −− + + + TDMA+CSMA w/o WuR

O-MAC [47]Y-2006

+ + + −− − + + + − + + + TDMA+CSMA w/o WuR

MEB-MAC [46]Y-2012

+ + + + − + + + − ++ TDMA+CSMA w/o WuR

EEE-MAC [54]Y-2013

−− + + + − + + + −− + + + TDMA+CSMA w/o WuR

FE-MAC [55]Y-2007

+ + + −− − + + + − + + + TDMA+CSMA w/o WuR

P-MAC [77]Y-2013

+ −− − + + + − + + + TDMA+CSMA w/o WuR

CA-MAC [43]Y-2009

−−− ++ −− ++ −−− + + + TDMA+CSMA w/o WuR

ULP-MAC [44]Y-2005

+ + + + −−− ++ −−− + + + TDMA+CSMA w/o WuR

BMAC [45]Y-2007

+ + + + −−− ++ − + + + TDMA+CSMA w/o WuR

BSN-MAC [39]Y-2010

+ + + −− ++ + − + + + TDMA+CSMA w/o WuR

ULPD-MAC [78]Y-2008

−− + − + + + − −−− TDMA+CSMA w/o WuR

DISSense [48]Y-2007

+ + + + −−− + + + −− ++ TDMA+CSMA w/o WuR

S-MAC[18]Y-2002

−−− + −−− ++ − + + + TDMA+CSMA w/o WuR

38 3.6. Discussion


MAC protocol

Features

Energy-efficiency

Reliability Software-overhead

Effective-throughput

QoS Low-latency

Hardware complexity

Cooperative - MAC [9]Y-2008

++ + −− ++ −− ++ TDMA+FDMA w/o WuR

Hybrid-MAC [58]Y-2014

−−− + + + −−− + + + −− + + + TDMA+FDMA w/o WuR

HyMAC [59]Y-2012

+ + −−− ++ + ++ TDMA+FDMA w/o WuR

HUA-MAC[39]Y-2010

+ + + ++ −−− ++ −− + + + Hybrid ALOHA w/o WuR

YNU-MAC [60]Y-2009

− + −− ++ ++ −− CSMA + UWB w/o WuR

FM-UWB MAC [61]Y-2009

− ++ −− ++ −− + + + CSMA + UWB w/o WuR

NICT-MAC [64]Y-2014

+ + + + −− ++ −− + + + Slotted ALOHA + WuR

IMEC-MAC [65]Y-2009

+ + + + −− − −− ++ ALOHA + TDMA + WuR

Miller-MAC [66]Y-2005

+ + + + −− −−− −− + + + TDMA+CSMA+ WuR

RTWAC[67]Y-2009

+ + −− + + ++ TDMA+ CSMA + WuR

PE-MAC [68]Y-2011

+ + −− −− −− −−− TDMA+ CSMA + WuR

ULPA-MAC [70]Y-2013

++ − −− + + + −− − CSMA + WuR

WuR MAC [71]Y-2006

+ + + ++ −− ++ − + CSMA + WuR

T-MAC[72]Y-2003

− ++ −− ++ −− + CSMA + WuR

TBCD-TDM [73]Y-2009

+ + + + −−− + −− + + + TDMA+WuR

Table 3.9: Comparison of MAC protocols in terms of network parameters

3.6. Discussion 39


3.7 Conclusions of Literature survey

The main focus of this survey was on MAC protocols with and without wake-up radio thatmeet the requirements of network parameters in IBSN such as topology, performance in termsof reliability, latency, duty cycle and the hardware and software complexity. In literature thereare numerous MAC protocols available. Different optimization techniques such as cross-layeroptimization, use of wake-up radio, hybrid access mechanism benefiting from features of differentmechanisms were studied and classified based on their advantages and disadvantages. MACprotocols that were not actually developed for the purpose of IBSN were also investigated, sincethe features of those protocols were very much oriented towards the requirements of IBSN suchas latency less than 60ms and very low duty cycles in order of 0.1% while providing reliabilityfor data transmission. More than 40 MAC protocols were analyzed. As a result it is found thatHybrid TDMA + CSMA access mechanism incorporated with Wake-up radio is best suited forthe IBSN application. Moreover, only few research literature focusing the IBSN is available withthe hybrid TDMA+CSMA specification. It is evident that further research have to be done todesign an optimal access mechanism for IBSN.

From survey, the added-value of wake-up radio in medium access is known. However, differentaccess mechanisms has its own advantage over the other. For instance TDMA is excellent inenergy efficiency when compared to CSMA/CA, but the required software overhead for timesynchronization in the entire network. In order to understand the effect of access mechanismswith wake-up radio, an analysis of wake-up radio based MAC protocols with different accessmechanism have to be done. Out of the protocols analyzed, RTM-MAC which is CSMA/CAbased, OnDemand MAC which is TDMA based, and SCM-MAC which involves hybrid accessare best performing in comparison in wake-up radio based MAC protocols. Analyzing theseprotocols in terms of design and performance can be done to find the best suited wake-up basedaccess mechanism for IBSN.

40 3.7. Conclusions of Literature survey

Chapter 4

Analysis of wake-up radio basedMAC protocols

In this chapter, three MAC protocols selected from the survey (Chapter 3) are analyzedbased on their design and the strengths and weaknesses in terms of software and hardwarecomplexity and network performance are identified for IBSN scenario. RTM-MAC [79] which iswake-up radio (WUR) based MAC protocol with CSMA/CA protocol (contention-based accessmechanism), OnDemand MAC [81] which is WUR based TDMA protocol (contention-free accessmechanism), SCM-MAC [80] which is WUR based hybrid access protocol (CSMA/CA+TDMA)are chosen and analysed with a software simulation of the protocols using MATLAB. In SCM-MAC, an interesting wake-up radio architecture which uses a single radio for emulating both themain radio and the wake-up radio is mentioned [80], reducing the hardware overheads withoutaffecting the network performance. All protocols are analyzed based on a single-hop star topology,where all the sensor nodes compete in the medium to transfer data to the base station. Peer-to-peer (P2P) communication can also be made possible with the selected protocol [80]. However,the network topology is defined so that it meets the basic requirements of IBSN as stated insection 2.4.

4.0.1 Radio Triggered sensor MAC

Radio Triggered sensor MAC (RTM) [79], has good potential for application in BSN asclaimed by its authors. The wake-up feature is adapted using additional hardware which consistsof a passive wake-up radio (WUR) sensor with a bandpass filter implementation. The main radioand the micro-controller are triggered by the passive wake-up sensor upon an intended wake-upcall (WUC). A sender uses the CSMA/CA contention mechanism for sending the wake-up callto other nodes. Upon gaining access to the medium, the sender sends out a wake-up preamblethrough the medium, which energizes the passive WUR and triggers the main radio of the nodes.The communication is carried out using ReadyToSend/ClearToSend (RTS/CTS) as shown inFig. 4.1. The RTS packet has information about the destination, which prevents other radiosto overhear the data communication. The node to which the RTS is intended will send a CTSand complete the communication using an acknowledgement. Collision avoidance is incorporatedusing a Network Allocation Vector, which stores the time period for which the communicationof the other node will take place. This information is available from data packets. By this, othernodes will know how long they should wait before they can access the medium. Back-off periodand number of retries can be set easily in the protocol. In order to prevent energy wastage

41

CHAPTER 4. ANALYSIS OF WAKE-UP RADIO BASED MAC PROTOCOLS

Figure 4.1: RTM scheme.

through overhearing, the nodes are put to sleep once they receive a RTS or CTS not intendedfor them.

Strengths: RTM does not require a complex wake-up radio, but uses simple hardware toemulate the same effect. It is CSMA based, which means there are no timing beacon and errorsdue to clock drifts. It can also be used in P2P topology, however energy efficiency and hardwarecomplexity will be traded off.

Weaknesses: Reliability is an important factor for IBSN, which is not guaranteed with theCSMA/CA approach for emergency data. This MAC protocol did not discuss any GuaranteedTime Slot (GTS) thus fails to have high reliability in implantable sensor networks. Although itis claimed that energy wastage is minimized by the CSMA/CA-based mechanism it still faces alot of energy wastage in unnecessary wake-ups when compared with a TDMA-based approach.

4.0.2 OnDemand MAC

OnDemand MAC is a contention-free protocol, which uses wake-up radio to trigger the mainradio for data communication [81]. The on-demand mechanism is designed for a single-hop star-topology network. A wake-up schedule table is used for normal sensing operation where thecentral controller sends a wake-up signal to the node and reads the data from the implantednode. A node can wake up the controller by sending a wake-up signal for emergency medicalevents. Thus, the nodes in the network communicate on demand. The data communication iscarried out by the main radio upon waking up in a TDMA-based access mechanism as shownin Fig. 4.2. The controller node of the network is responsible for slot allocation and channelallocation, in order to reduce the burden on the sensor nodes. Data communication starts onlyafter the acknowledgement from the receiver and synchronization beacon from the controller.The data communication is then acknowledged, and the main radio is put to sleep.

Strengths: OnDemand MAC is suitable for delay-sensitive scenarios such as data transfer inthe event of an emergency. Collision is minimized and as a result, retransmissions are greatlyreduced thus saving power.

Weaknesses: A synchronization beacon is sent by the controller during normal sensing periodsand also during emergency periods. Any minimal drift in the clock can occur, since high-precisionclocking cannot be done in small nodes. This scenario renders very low reliability and requires a

42


Figure 4.2: On-Demand MAC scheme

large number of beacons to increase the reliability of the network. A higher number of beaconscauses a lower effective throughput and more bandwidth wastage. The performance of the on-demand MAC is compared with the CSMA approaches in section 4.1.1

4.0.3 SCM MAC

SCM MAC takes advantage of the Sub Carrier Modulation Wake-up Radio (SCM WuR) [80].A very low operation power of 8µW is achieved for coverage distances of up till 30 meters. TheSCM-MAC makes use of the simple OOK modulation for the wake-up procedure, encoding thenode ID. The node ID will only wake up the main radio if the wake-up call (WUC) is actuallyintended for it. The advantage of the node ID-based WUC will reduce the unwanted listeningto data packets as it is in the case of RTM-MAC. The main radio uses the Clear ChannelAssessment (CCA) based approach for data communication as shown in Fig. 4.3. Comparisonwith standard MAC without wake-up radio, such as B-MAC and IEEE802.15.4, has shown thatSCM-MAC performs better in terms of better latency and high energy-efficiency [80]. This workis an example of state-of-the art wake-up radio-based transceivers which can out-perform normalMAC protocols for use in IBSN.

Strengths: A novel radio architecture is used which meets the requirements of MAC protocolsfor IBSN. Low power consumption and larger coverage range is supported. Excellent performancein terms of low latency and high packet delivery ratio in moderate inter-packet arrival time.

Weaknesses: No provision for the emergency data handling is provided. The emergency datapacket has to compete with normal packets to reach the controller. The approach to handlecollisions is pretty simple, which may introduce problems during higher data-rates when nodescompete to send even non-emergency data.

43


Figure 4.3: SCM-MAC scheme

4.1 Simulations and performance evaluation

Simulations were performed in MATLAB by adapting the 802.15 channel mode for in-bodycommunications [31]. The choice of using MATLAB over a network simulator is due to theavailability of channel models and physical layer models for in-body communication scenarios.Discrete event simulation is repeated using packet inter-arrival time with a fixed data rate. Be-cause from literature a common network architecture was found to be a single-hop star topologycontaining 11 nodes [31], this topology is considered in the simulations. Simulation was repeatedfor different values of packet inter-arrival time, while observing parameters such as delay, powerconsumption, packet drop ratio and duty cycle. A perform evaluation is made based on theseparameters in order to compare the effect of MAC protocols in in-body scenarios and these para-meters will help us to understand the reliability and power consumption trade-off in differentnetwork conditions. Inter-packet arrival time is a commonly used parameter to evaluate networkperformance, as it is directly affected by changing network conditions. The physical layer para-meters are modified based on the recommendation of the IEEE 802.15 task group 6 proposalon the MICS band [31] and summarize them in table 4.1. The same network architecture isused for all simulation runs. Parameters of different protocols are derived from the respectivesimulation models as explained in the referred literature. As an optimum for the chosen MACprotocols, a packet size of 50 bytes was chosen. CSMA-based protocols were implemented witha random back-off time. Other settings of the wake-up mechanism, such as wake-up interval,power consumption, and modulation techniques were chosen as proposed for each protocol in thereferred literature.

4.1.1 Simulation results

Discrete event simulation in MATLAB has resulted in comparison of delay, packet deliveryratio, power consumption and duty cycle of three different varieties of MAC protocols withwake-up radio as shown in figures 4.4 to 4.7.

Power consumption Power consumption is the total power consumed by the radio commu-nication including the wake-up call and data communication. Evaluating the power consumption

44 4.1. Simulations and performance evaluation


Table 4.1: Simulation parameter values

Parameter ValueChannel bandwidth 300 KHzReception current (main) 7 mATransmission current (main) 15 mAIdle current (main) 3 mASleep current 3 µASlot time 10 msData rate 250 KbpsReception current (WUR) 8 µATransmission current (WUR) 1 mAData rate (WUR) 1 KbpsPathloss coefficient a 1.92Pathloss coefficient b 39.85Pathloss parameter σN 6.59 dBPolarization parameter Xc 0.45Pathloss parameter d 10 m

based on IPAT will show how much the power can be traded off during emergency events, andhow long a node could perform in different networks scenario. Upon evaluating it is observedthat the power consumption of CSMA/CA based RTM-MAC is the highest for both low andhigh IPAT values. This is the effect of the unwanted listening of control packets by the nodesand unnecessary waking up of nodes for preambles. Also, the CSMA-based approach for datacommunication keeps the radio always switched on until the CTS is sent when data is completelytransferred. The power consumption of TDMA-based on-demand MAC is the lowest, which ismainly due to the reduced number of control packets and intelligent control of synchronizationby the On-Demand protocol. However, SCM-MAC is still in the acceptable region for moderateIPAT, when compared to the on-demand MAC. The lower power consumption of the SCM-MACis due to the special chip that has been developed to operate in ultra-low power consumptionon top of straightforward implementation of Clear Channel Assesment (CCA) based main radio.The main radio is completely switched off in the SCM-MAC until the intended wake-up call isdecoded by the ultra-low power WUR. SCM-MAC is also interesting since the authors claimthat it can operate with a coverage area of up to 30 meters at 8µA [80]. From this analysis,it is evident that simple passive hardware cannot be completely power efficient in the case ofimplants. Although the TDMA-based approach has an intelligent synchronization mechanism,in real hardware the clock drift will be influencing the power consumption of the sensor node.It can be clearly seen from the simulation results, that a wake-up radio which can have an lowpower decoding option alongside the hybrid CSMA and guaranteed time slot is the best choicefor power-efficient implantable sensor node communication.

Delay The delay in the sensor network is defined as the difference in time taken for a set of datathat is sent to reach at the destination. Varying IPAT in different protocols will have influenceon the delay due to the effect of preambles and control beacons sent alongside the data. In Fig.4.5, three different protocols are compared in terms of delay. It is shown that the CSMA-basedRTM has the highest delay with high IPAT. The delay is larger in RTM because at higher IPAT,the carrier sense overheads such as RTS, CTS and the wait time of passive wake-up for the data

4.1. Simulations and performance evaluation 45


0 5 10 15 20 25 30 35 40 45 500

10

20

30

40

50

60

Packet Inter−arrival Time (seconds)

Pow

er c

onsu

mpt

ion

(mW

)

SCM−MACOn−Demand MACRTM−MAC

Figure 4.4: Effect of IPAT on power consumption

communication to complete. In the case of on-demand MAC, delay is almost constant for thedata communication, because the only major delay is by wake-up and transmission of packets.The synchronization beacon is encoded in the packets which is also a reason for constant delayeven though it reduces the effective throughput of the transmitted data. SCM-MAC on the otherhand, still performs better than RTM, due to the node-id encoding in the wake-up call and alsoless unnecessary overhead such as passive activation like RTM. From the simulation analysisof the delay, it is evident that TDMA-based protocols are excellent in performance. The slotallocation for the channel access for each node is the crucial reason for constant delay. However,similar results can be achieved with CSMA-based protocols using GTS mechanisms. Hybridprotocols can be adapted with the wake-up feature to make the delay performance better.

Packet delivery ratio The packet delivery ratio (PDR) is the ratio of the total number ofpackets received at the receiver to the total number of packets generated at the sender. A 100%PDR means there is no loss of packets. PDR is an important metric to analyze the networkperformance with different physical layer parameters with a specific topology. In case of thein-body sensor network, inter-packet arrival time can vary due to the highly varying physicalproperties of the human body. The PDR is evaluated for different IPAT to find the reliabilityof the different MAC protocols with wake-up radio. As a result, it is shown that all protocolsperform with high PDR at lower IPAT. However, as the IPAT increases the PDR of the protocolsdecreases. Out of the three MAC protocols chosen, SCM has a better delivery ratio, which is dueto the reduced overheads and hence packets can be transferred in less time when compared to theTDMA-based approach. On-demand MAC has the highest PDR because the effective payload



0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 50

0.5

1

1.5

2

2.5

3

3.5

4

4.5


End

−E

nd D

elay

(se

cond

s)


Figure 4.5: Effect of IPAT on End-to-End delay

is reduced with longer preambles at higher IPAT. It is good to notice that in on-demand MAChas a provision to send emergency traffic in higher priority. RTM has a better performance thanon-demand MAC for similar reasons as SCM. However, the passive wake-up overheads and lackof device-specific wake-up caused RTM to have more packet loss at higher IPAT.

Duty cycle Duty cycle is the proportional duration for which the main radio is turned onin useful data communication. Variation in IPAT will cause the duty cycle to be higher, whichmeans radio has to be turned on for much longer time, if the time between the arrival of packets atthe receiver is longer. In RTM the CSMA-based access mechanism will have to sense the networkrandomly after every back-off time, which causes the radio to be turned on unnecessarily. In thecase of on-demand MAC, the duty cycle is almost the same for different IPAT because of thefact the main radio is turned on only in pre-assigned time slots.

4.1. Simulations and performance evaluation 47


5 10 15 20 25 30 35 40 45 500

10

20

30

40

50

60

70

80

90

100


Pac

ket d

eliv

ery

ratio

(%

)


Figure 4.6: Effect of IPAT on packet delivery ratio

0 5 10 15 20 25 30 35 40 45 5030

40

50

60

70

80

90

100


Dut

y cy

cle

(%)


Figure 4.7: Effect of IPAT on duty cycle



4.2 Discussion

Different access mechanisms can be used in the MAC protocols with WUR. There is alwaysa trade-off between performance in terms of Quality of Service (QoS) and energy efficiency.Simulations were performed to understand the effect of QoS and power consumption in a complexphysical layer with unique conductivity and path loss effects. It is seen from the simulations that ahybrid access mechanism with contention-free access during emergency situations and contention-based access during the normal sensing process is the optimal choice for MAC protocols with awake-up based access mechanism. Even though it is claimed that TDMA is a good choice forin-body communication in terms of energy efficiency, the effect of clock drift will have seriouscomplications. However, energy efficiency of the CSMA approach as implemented in SCM-MAC is similar to the TDMA approach as implemented in On-Demand MAC. The wake-upradio feature in SCM clearly removes the conventional problems of CSMA approaches such asidle-listening, over-hearing and packet-collision by implementing node-id encoded wake-up calls.Thus from the simulations it can be concluded that a wake-up based MAC protocol can performwithout trading off the energy-efficiency by adapting the advantages of both TDMA and CSMAapproaches.

Table 4.2: Strengths and weaknesses of selected MAC protocols

MAC Strengths Weaknesses

RTM [79] simple WUR hardware; CSMA based,hence no clock drifts errors; suitable forP2P

reliability not guaranteed with CSMA/CA approachwithout GTS for emergency data; much energyis wasted in unnecessary wake-ups compared withTDMA

OnDemand[81]

suitable for delay-sensitive scenarios(e.g. emergency data transfer); min-imized collision resulting in reducedretransmissions and lower power con-sumption

synchronization beacons sent in both normal andemergency periods, lack of high-precision clockingin implanted nodes may lead to clock drift, hencevery low reliability or large number of beacons, lead-ing to lower effective throughput and more wastedbandwidth

SCM [80] novel IBSN radio; low power consump-tion; larger coverage; low latency; highpacket delivery ratio

no separate emergency data handling; simple col-lision handling may be problematic at higher datarates

4.3 Conclusion

In this chapter three different MAC protocols with wake-up radio are analysed based on theirperformance and access mechanisms. Simulations were performed to understand the effect ofreliability and power consumption in a complex physical layer with unique conductivity and pathloss effects. Strengths and weaknesses of the analyzed WUR based MAC protocol is organized intable 4.2. It is seen from the simulation results that a hybrid access mechanism with contention-free access during emergency situations and contention-based access during the normal sensingprocess is the optimal choice for MAC protocols with a WUR. Even though it is claimed thatTDMA is a good choice for in-body communication in terms of energy efficiency, the effect ofclock drift will have serious complications. However, energy efficiency of the CSMA approach asimplemented in SCM-MAC is similar to the TDMA approach. The WUR feature in SCM clearlyremoves the conventional problems of CSMA approaches such as idle-listening, over-hearing and

4.2. Discussion 49


packet-collision by implementing node-id encoded WUCs. Thus from the simulations it can beconcluded that a wake-up based MAC protocol can perform reliably and energy-efficiently byadapting the advantages of both TDMA and CSMA approaches. The best choice of wake-uparchitecture is to use a dedicated WUR which can operate with several µW.

50 4.3. Conclusion

Chapter 5

Characterisation of PHY layer ofan implanatable sensor node

This chapter focuses on the description of test environment, and characterization and evalu-ation of physical layer configurations of the sensor node in different IBSN scenarios. The sensornode will be placed subcutaneously inside the animal flesh. Performance of radio inside the tissueis evaluated with different physical layer configurations such as transmission power and packetlength. The results of this evaluation will be used for selecting an optimum set of physical layerconfiguration for wireless communication of implants. This optimum set of parameters will beused later to evaluate the performance of medium access control protocol in chapter 6

5.1 Description of test environment

The test environment for the implants is the piece of animal tissue extracted from the muscularpart of the animal. In most cases, the implants are placed in the dermis layer of the skin, asshown in Fig. 5.2. Evaluating the effect of bones in the flesh is not necessary since the implantsare placed in the layers of flesh and not deep inside the tissues near the bones. From literature themost resemblance to the human flesh is found with the flesh of pigs. The muscular organizationcan also be simulated using artificial recipe of chemicals and silica mixture. However, due tothe simplicity and as a quicker and easier starting point, a piece of flesh is dissected and used inthese experiments. Fig. 5.1 shows the part of meat dissected from thigh of the pig.

5.1.1 Animal flesh

Animal flesh is made of different types of tissues. A tissue is a collection of similar cells fromthe same origin that together carry out a specific function. There are four types of tissue presentin animals and also humans i.e., Connective tissue, Muscular tissue, Nervous tissue, epithelialtissue are the four types of tissues. In general the medical implants are implanted based on thepurpose of the sensor or device. For example, to monitor the activity of nervous system in thelimbs, a sensor is located close to the nervous tissue in the flesh. A detailed cross section of howthe animal flesh is composed is shown in Fig. 5.2. It is important to note that different layers ofskin has different conductivity(µ) and permittivity (ε) .

51

CHAPTER 5. CHARACTERISATION OF PHY LAYER OF AN IMPLANATABLE SENSORNODE

Figure 5.1: A dissected part of pig flesh used for testing the sensor node.

Figure 5.2: A cross section of skin showing different tissue layers including skin.

5.1.2 Different location in muscular tissue for an implant

The implantable sensor node is an existing practise in the last decade. Research is still beingcarried out to find an optimum location of the implants. The challenging issue of the implantlocation is that the RF characteristics (ε, µ) vary randomly in different parts of the body. Placinga implant just under the skin (epidermis layer in Fig. 5.2) has a different effect than havingan implant subcutaneously i.e, under the Dermis layer(Fig. 5.2). However, depending on thepurpose of the implant, the location is decided. Currently, magnetic readers are used to readthe data from implants which is not in a location in favour of the RF communication. Thereare different types of implants such as sensors, which sense the vital signs of life such as in-bodytemperature, blood glucose level, tissue oxygen level, heart rate, and intra-venous blood pressure(pressure of blood flow inside the veins), and medical devices such as pace-maker, drug deliverydevices, and stimulators (refer Chapter 1). These implants are commonly placed subcutaneouslyunder the skin. To sense the symptoms of the vital organs, some of the sensors can be placedclosed to the vital organs. In such cases the location in the body is chosen such that the implantstays close to the subcutaneous layer of the skin, so that the implant can be reached easily incase of any negative response from patient’s immune system.

52 5.1. Description of test environment


Implant location in the flesh

Although different types of implantable medical devices exists, the most common place of theimplant is selected. From the literature, more than 80 % of the medical implant surgeries aredone subcutaneously and the implant is placed in the hypo-dermis layer of the flesh [73][74][12].The subcutaneous layer is chosen because the control of foreign body1 rejection is done with easeat this layer of skin. More deeper the implant, the rejection of the foreign body will be difficultto diagnose. Also, the second important reason is to keep implant close to the skin for easierdata communication, and diagnosis of the device after the implantation.

Thus for the evaluation of physical layer parameters, the subcutaneous layer is chosen as theimplant location. All the experiments carried out, will have the sensor node implanted in thesubcutaneous layer of the animal flesh.

Dimension of flesh is selected in order to accommodate the implant and also should supportthe evaluation scenarios as mentioned in the section 5.3.2. The flesh used for evaluation, asshown in Fig. 5.1, is of the dimension 22 cm in length, 20 cm in width and 11 cm in depth. Thenumbers are not chosen but the given dimension depends on the characteristics of the animal.

Test bed for evaluating the physical layer communication with in-body sensor nodes

As the flesh of pig can be highly contaminating with microbial organisms, proper precautionshave to be taken before any experiment is carried out. A biological safety cabinet is used for car-rying out the experiments. The biological safety cabinet (Fig. 5.3)will prevent any contaminationand can be cleaned with acetone after the experiments.

Figure 5.3: A biological safety cabinet used for carrying out the experiments with meat.

1sensor node is a foreign body inside the animal or human. As a consequence the immune system of theanimal or human will try to expel the sensor node from the body by creating wounds and isolation of cells overthe implanted region

5.1. Description of test environment 53


5.2 CC430 based implant

The design of hardware should be done based on two important criteria. One is the ultra lowpower radio operation and the second is the compactness of the hardware. There are differentradio chips available in the market which can comply to the requirements of the medical implant.The chip used in this evaluation is CC430 from Texas Instruments. The main features whichmake CC430 suitable for evaluation are:

• Ultra low power radio operation - 160µA/MHz

• Processing capability

– 16 bit CPU, with 50ns instruction cycle time

– memory - 32 kB of flash & 4 kB of RAM

• Radio features

– ability to operate in MICS band (402-405 MHz)

– clear channel assessment

– RSSI measurement

– configurable access to PHY layer

– configurable MAC layer

The chip is also suitable for operating simpler on-node signal processing at very low energy ex-penditure. However, the signal processing is not an immediate scope of the thesis. For evaluatingthe wireless communication parameters, CC430 is an optimum choice. A development board isused to first test the software developed and a custom made PCB is used for evaluation of thewireless performance inside the meat.

Figure 5.4: Block diagram of radio chip used in the evaluation.

54 5.2. CC430 based implant


5.2.1 Pseudo-implementation of wake-up radio and antenna matchingcircuit

The cc430 based implant does not have a separate wake-up radio. The wake-up feature canbe simulated via the software and chip has a support for wake-up feature using a local timer.This feature is used to evaluate the performance of wake-up based MAC protocols. The timer isset for a predefined and a definite period by the master node. The wake-up timer interval is setbased on the choice of eliminating congestion in the network when two nodes try to transmit atthe same time. The wake-up information is sent in the data packet which can be dynamicallyadjusted by the master node after a transmission is completed. However, the lack of real wake-upradio converges to the fact that, the energy measurement cannot be made since no separate hard-ware for wake-up radio is present on chip. However, the network performance can be measuredassuming the timer is accurate.

The chip also lacks antenna matching circuit which will contribute for losses in the meatconstantly. Effective radiation will be destructed due to the mismatch of antenna impedancewhen the sensor node is implanted inside the meat. The purpose of antenna matching circuit isto match the impedance of the antenna to that of the surrounding medium. As a work-around,the radio is set to operate in the frequency for which the antenna is designed. The loss due to themismatch in antenna impedance is considered to be constant and is subtracted from the RSSImeasurement. The loss in signal strength is calculated by measuring difference between the RSSIin air and RSSI in the meat for the same transmission power and distance. This procedure willhelp to reduce the effect of antenna matching circuit.

The initial testing of the chip and the corresponding software were performed in the de-

Figure 5.5: Custom made CC430 based implant.

velopment board provided by the TI. Once the software is validated on the chip, the customdeveloped board is used. The board is developed by Ir. Kyle Zhang at Pervasive system researchgroup. The implant is shown in Fig. 5.5, which comprises of CC430, a ceramic antenna matchedto operate at 433 MHz, and an accelerometer. The implant has a UART and a SPI connectionthrough which the software is loaded and the output will be debugged. The implant is wrappedin a paraffin wax paper and later by bubble wrapping paper. The later case is considered toinvestigate the effect of air surrounding the antenna inside the meat.

5.2. CC430 based implant 55


Figure 5.6: CC430 based implant enclosed in a paraffin coating.

5.3 Physical layer configurations

The impact of flesh in the radio communication is not completely standardized. Differentdesign of hardware may influence the meat differently. First step is to understand the effectof flesh on the designed hardware. This understanding should be normalized and an optimumshould be chosen for the further experiments for the performance measurement of the network.

5.3.1 Parameters of physical layer configuration

The hardware designed supports different configurations for the PHY and MAC layer. Theconfigurations of physical layer should be verified with real world physical characteristics suchas distance, antenna orientation, etc. To understand the different configuration for the PHYlayer, different parameters should be studied and the effect of the parameters in the networkperformance is needed. This is explained as follows,

• Transmission (Tx) Power : The power at which the power is delivered to the antenna.It is measured in the units of dbm. If the antenna is exactly matched with the outputimpedance of the radio chip, the power delivered at the antenna should ideally match thetransmission power. Varying Tx power will result in the variation in the signal receivedat the receiver end. Lower the RSSI, higher the bit error rate and lesser the number ofsuccessful transmission. In order to save power at both receiver and the transmitter, anoptimum level of Tx power should be chosen.

• Tx Rate : The rate at which the data is sent. The unit is bytes per second. Higherthe transmission rate, quicker the transmission and hence power will be saved. However,if the receiver misses any information, the re-transmission has to be done, which again

56 5.3. Physical layer configurations


contradicts to the power consumption. Also, the network throughput is increased withhigher data rates. However, the same failure rate effect the effective throughput of thenetwork.

• Frequency of Tx : Higher the frequency, lesser the propagation inside the human body.As the human body is conductive, RF properties in air donot follow inside the body. Thein-body communication as mentioned earlier is standardized to operate in MICS band.However, it can be useful to investigate if the network performance is increased with adifferent frequency inside the meat.

• Distance : The distance between Tx and Rx is not software configurable, however, thevarying the distance will change the physical layer configuration. In order to cope up withthe changes in distances, the maximum limit of the distance is found out with differentconfigurations of the PHY.

• Orientation : The orientation of antenna inside the meat. The orientation is again not asoftware defined PHY parameter, but it is taken into consideration for evaluating the PHYof hardware. Different scenarios of communication will be explained in the section 5.3, outof which is the communication of nodes placed in different location inside the body. Forthis configuration, the antenna orientation can be a possible factor to decide on the locationof the implants. Moreover, other scenarios such as inbody to outbody communication willalso benefit from the antenna orientation.

All these options are possible to configure though software with the given implant. The distanceand orientation are taken care in the implantation of the sensor node in the sub-cutaneous layerof the given meat.

5.3.2 Medical scenarios for different configurations

The IEEE 802.15 Task group 6 have defined four main configurations for in-body communic-ation [31] :

• Scenario (SC) 1 : In-body to in-body communication

• SC2 : In-body to on-body communication

• SC3 : On-body to on-body communication

• SC4 : On-body to external nodes communication

PHY layer parameters for all the four configurations will be evaluated. Different medical scenariosfor these configurations are foreseen as follows,

SC1 : The in-body to in-body communication is necessary when an implant medical deviceneeds to communicate to the implanted sensor. Consider a patient who has a drug-delivery deviceand pace maker implanted. The drug delivery device will deliver the insulin for the regulatingthe blood glucose level. The effect of blood glucose level has impact on the heart rate. Lowerthe glucose level, lower the heart rate is the medical symptom [3]. In such a situation the patientwill have sensors implanted for measuring the glucose level and hear rate. In the event of an highheart rate, the pacemaker should regulate the rhythm of the heart. To confirm that the heartrate is changing because of the glucose level, the pace maker should communicate to the glucosesensor and also send command to the drug delivery implant, while regulating the pace of the

5.3. Physical layer configurations 57


heart by itself. All the data communication in this case is inside the human body, provided thepacemaker acts as the controller. The data communication should be reliable and also quickerto save the life of the patient.

SC2: Consider the same scenario explained above. If the controller is not the pace-maker thenan external controller outside the body must communicate to the implants for the same actionto take place. However, for tele-medicine application where the sensor data and operation of theimplanted medical devices are to be logged externally to a server. In such cases, logging of theperiodic data to an external device is mandatory.

SC3: The on-body sensor nodes are the nodes which monitor the physiological and physicalactivities non-invasively. These sensor nodes must communicate to the controller or the datalogger which is also placed external to the body. This communication is not critical and can takeplace with less priority than the in-body sensor nodes. The effect of shadowing and fading dueto human body is considered in this situation i.e, placing a node in the chest of a human (ECG)and another node in the leg (activity sensor)of human will severely face the effects of shadowingfrom human body. Validating the PHY layer parameters for this scenario is also considered inthis thesis.

SC4 : The communication between external controller which is attached to the surface of thebody to the external internet server placed in the vicinity of the human. The telemedicine ap-plication requires this communication path for remote monitoring of the patient and healthcare.This communication model need not be considered for the evaluation of PHY since it is out ofscope of the thesis.

5.4 Implementation and experimental setup

Implementation is done with different sets of parameters. Each group will be evaluated forthe three scenario as discussed in section 5.3. The setup with the flesh and medical implant isshown in the appendix.

5.4.1 Implant location in flesh for different scenarios

The scenarios explained in section 5.3.2 is exhibited as shown in the figures in Appendix A.The sensor encapsulated in a paraffin coat is placed in the subcutaneous layer. Power source isplaced outside the flesh. The effect of metal and electrical interference to RF link is reduced byplacing all the power source outside the flesh. The UART connection is made in order to sendand receive the RSSI values. Once the physical layer validation is set and the optimum is chosen,the network parameters will be logged locally on the flash memory. The flash is read out oncethe timespan for respective experiment is completed.

5.4.2 Collective evaluation of PHY parameters

The evaluation parameters of PHY are interdependent. In order to find an optimum, theirinter-dependency should be studied in detail. The choice of evaluation parameters are donebased on the linearity based statistical approach. The main aim of the choice of parameters isto find and verify the linear dependencies between each parameter. Once the parameters from

58 5.4. Implementation and experimental setup


Set of parameters Evaluating parameters

Set of physical parameters Transmission power (dbm)Transmission distance (cm)Antenna orientation (degrees)

Set of network parameters Operating frequency (Hz)Data rate (Kbps)Packet length (bytes)Packet delivery ratio (%)

Table 5.1: Evaluation of hardware with two different sets of physical and network parameters.

set of physical parameters are evaluated, an optimum setting is chosen from the results, forthe evaluation of set of network parameters. Out of the results, the optimum is chosen for theevaluation of network performance in Chapter6. In order to simplify the measurement pro-cess, a UART interpreter is developed. A command with different PHY parameters is sent andthe corresponding results is read out using the same USRT interpreter. The implementation ofUART interpreter is shown in appendix. The radio has to be reset for every change in the PHYparameters. The resetting of radio will put the radio in sleep mode. Thus for every change, thenode has to initialize before sending any data. A flowchart for the evaluation of physical layerparameter is shown in Fig. 5.4.2.The evaluation is carried out in groups to find the optimum parameters chronologically. A setof basic parameters including transmission power and distance with antenna orientation is car-ried out. Secondly, the set focusing on the transmission rate, frequency of operation and packetlength is carried out. Modulation is not an important consideration as the chosen parameterswill not greatly affect from modulation format.

Optimum definition The optimum value is defined as the value which is neither the bestcase nor the worst case for the given evaluation. By doing so, the further results are not influ-enced by the previous results. Also, in reality the optimum value will make will be chosen as aclose resemblance to the real-world operation.

5.4. Implementation and experimental setup 59


initializeCC430

RADIO UART

Applyradio

settingswith PHYparameters

Read theRSSI valueto UART-

buffer

Clear radiosettings

Check ifUART in-terpreter

has avalid

command

stop

yes

no

Figure 5.7: Flowchart for PHY evaluation

5.4.3 Set of physical parameters

The first set of parameters to be evaluated in the in-body scenario is the transmission powerand the received signal strength for different transmission distance, orientation of the antennasand power of transmission. As explained previously, the nodes are evaluated at different scenariosSC1-SC3. The application may vary for different configurations, but the need of the evaluatingthe physical layer parameters is required to design the upper layers of the OSI model. Table 5.2shows the different configurations of first set of experiments for evaluation with scenario 1. Thismodel is defined as the communication between the node within the body. The nodes are placedin the distance of maximum 16 cms, which is large enough for accommodating two differentsensors within the torso. Perhaps, most of the application such as drug delivery devices, pace-maker, and neural stimulators are placed in the torso along with sensors such as blood pressure,glucose, heart rate and movement. Further more, evaluating this Scenario will help to study theeffect of flesh in the radio communication and the design dependency for upper layer protocols.

Scenario 2 Consecutively, Table 5.3 shows different parameter settings for the Scenario 2. InSC2 the sensor node is placed inside the body and communicates to the sensor node outside thebody. The maximum distance chosen for this evaluation is 160 cms, which covers the body of an



Antenna orientation Evaluating parameters

0◦ Transmission power (dbm) - 5,0,-10,-15,-30Transmission distance (cm) - 0,2,4,8,16



Table 5.2: Set of physical parameters for SC1.

average human being. In this situation, a sensor may be placed in a part of the body such as thethigh or hip. In more than 80 % of the medical cases, the sensor is not placed in a highly mobilehuman part such as leg. This is because the noise in the measured signal level gets distorted andthe behavior of the sensor values contradicts to the normal measurement. However, even if thesensor node is placed in mobile part of the human, the values chosen will be within the range ofthe average human body.






Scenario 3 Scenario 3 is evaluated with larger distances, where the nodes are placed outsidethe body, and communicate with each other. In this case, the application scenario would be tohave two different controllers such as drug-delivery device and central data logger. The medicalsensor data could be logged to a device placed externally to the human body. It can also bethe case that the medical device need to communicate to the logger for a intelligent decision. Insuch situation the device may be in need to communicate to the device worn outside the body.The maximum distance evaluated in this case is 300 cms. Another interesting evaluation wouldbe to find the impact of the flesh if placed in the surface. In this case, the node is place on theskin of the animal flesh and the same experiments are repeated.

5.4.4 Set of network parameters

The set of network parameters used for the evaluation are data rate, packet length, frequencyof operation. The transmission power, distance and antenna orientation are chosen from the

5.4. Implementation and experimental setup 61



0◦ Transmission power (dbm) - 5,0,-10,-15,-30Transmission distance (cm) -0,25,50,75,100,150,225,300




previous evaluation. From this set of network parameters, the values are varied and the packetdelivery ratio is measured. This will help in finding the optimum settings for measurement ofnetwork performance. This chronological evaluation will help to find the network performancein a reliable setting of the PHY parameters. The values of different parameters and its corres-ponding values for characterizing are shown in table 5.5. Upon the selection of parameters, animplementation is done in the implant using UART interpreter. A command is sent from the PC,and based on the input given through UART. Different results are read back from the interpreterthrough UART and stored in the PC. These data are later analyzed in PC with MATLAB. Theresults from data analysis are discussed in section 5.5.



Packet length Evaluating parameters

5bytes Data rate(Kbps) - 2,20,40,60,80,100,150Packet delivery ratio PDR (%)

20bytes Data rate(Kbps) - 2,20,40,60,80,100,150PDR (%)





Table 5.5: Set of network parameters. Repeated for SC1, SC2, SC3

5.5 Results and discussion

The data collected from the chip is loaded in MATLAB and organized based on differentsetting as explained in the previous section. Data is analyzed using the functions shown inappendix B.

5.5.1 Results from evaluating the set of physical parameters

Validation results of the set of physical parameters are presented in the Fig. 5.12 - 5.17.The set of results are plotted with distance in x asis and RSSI in y axis. Different data setsof transmitting power are used for plotting. In order to compare the results meaningfully,different antenna orientation are chosen as the criteria for analysis. In Fig. 5.12, the antennaorientation is chosen as a classifier. The results are presented based on the Scenario in thefollowing subsections.SC1 In Scenario 1, where the nodes are placed inside the flesh, the attenuation of signal is veryhigh even in the smaller distances. The plot show in Fig. 5.12, at 0 degree which represents theantenna are placed closed to each other, even at 2 cm separation, a transmission of 5 dbm isreceived as -20 dbm. This indicates almost 25 % of the signal is attenuated even at such lowerdistances. Increasing the distances indicate that the RSSI is decreasing linearly. This boils downto the fact that ” as the distance increases, the received signal strength decreases”. However,it is interesting to note that the decrease in RSSI is not completely linear at higher distances.The relationship between RSSI and distance become logarithmic after 12 cms inside the flesh.

5.5. Results and discussion 63


2 4 6 8 10 12 14 16−110

−100

−90

−80

−70

−60

−50

−40

−30

−20

−10

Distance cm

RS

SI d

bm

Distance vs RSSI (Orientation = 0 degree)

Tx 5dbmTx 0dbmTx −10dbmTx −15dbmTx −30dbm

Figure 5.8: RSSI vs Distance at 0 degreeantenna orientation

2 4 6 8 10 12 14 16−110

−100

−90

−80

−70

−60

−50

−40

−30

−20

Distance cm

RS

SI d

bm




2 4 6 8 10 12 14 16−120

−110

−100

−90

−80

−70

−60

−50

−40

−30

−20

Distance cm

RS

SI d

bm




390 400 410 420 430 440 450 460 470−110

−105

−100

−95

−90

−85

−80

−75

Frequency MHz

RS

SI d

bm

Freqeuncy vs RSSI

Figure 5.11: RSSI vs Frequency at a fixed txpower, distance, orientation

Figure 5.12: Received signal strength information (RSSI) for in-body to in-body communicationSC1

SC2 In Scenario 2, where the nodes are placed inside the flesh and outside the flesh, theattenuation of signal is lesser than the SC1. The plot show in Fig. 5.17, at 0 degree whichrepresents the antenna are placed closed to each other, even at 2 cm separation, a transmissionof 5 dbm is received as -10 dbm. This indicates 20 % of the signal is attenuated even atlower distances. Increasing the distances indicated that the RSSI is decreasing linearly.However, as obtained in the case of SC1, the decrease in RSSI is not completely linear athigher distances. The relationship between RSSI and distance become logarithmic evenafter 10 cms inside the flesh when the node is communicating to the node outside the flesh.

64 5.5. Results and discussion


2 4 6 8 10 12 14 16−110

−100

−90

−80

−70

−60

−50

−40

−30

−20

Distance cm

RS

SI d

bm




2 4 6 8 10 12 14 16−110

−100

−90

−80

−70

−60

−50

−40

−30

−20

−10

Distance cm

RS

SI d

bm




2 4 6 8 10 12 14 16−120

−110

−100

−90

−80

−70

−60

−50

−40

−30

−20

Distance cm

RS

SI d

bm




390 400 410 420 430 440 450 460 470−110

−105

−100

−95

−90

−85

−80

−75

Frequency MHz

RS

SI d

bm

Freqeuncy vs RSSI


Figure 5.17: Received signal strength information (RSSI) for in-body to on-body communicationSC2

SC3 The Scenario 1 where the nodes are placed outside the flesh, the attenuation is similar tothat of the communication in free space. The plot show in Fig. 5.22, at all the orientations, theattenuation follows the same pattern. However, it is interesting to note that the decrease inRSSI is still linear at higher distances. There is no logarithmic dependency at higher distances.The effect of orientation is not a major contribution in the RSSI information in free space.The antenna in this case is faced towards the air, and not sandwiched between PCB and skin.



2 4 6 8 10 12 14 16−120

−110

−100

−90

−80

−70

−60

−50

−40

−30

−20

Distance cm

RS

SI d

bm




2 4 6 8 10 12 14 16−120

−110

−100

−90

−80

−70

−60

−50

−40

−30

−20

Distance cm

RS

SI d

bm




2 4 6 8 10 12 14 16−120

−110

−100

−90

−80

−70

−60

−50

−40

−30

−20

Distance cm

RS

SI d

bm




390 400 410 420 430 440 450 460 470−110

−105

−100

−95

−90

−85

−80

−75

Frequency MHz

RS

SI d

bm

Freqeuncy vs RSSI


Figure 5.22: Received signal strength information (RSSI) for on-body to on-body communicationSC3

In general all the cases indicate that the RSSI is linearly decreasing with distance, which can beformulated as

RSSIα

(k

Distance

)· (µ · Txpower)

where, k & µ are constants. In order to match the standards of MICS, the transmission powershould not be higher than 25µW. By referring to the datasheet of antenna and CC430 it is foundthat at a transmitting power of -10dbm the EIRP of the node will not exceed 25µW . However,no hardware measurement is made in order to verify the EIRP. Also, lowering the power furtherbelow -10dbm will not be able to communicate at distances greater than 5cms. A distance of16 cms is chosen in order to match the requirements of pace-maker and drug-delivery devicestandards, where the distance between the sensor and the device cannot be larger than 16 cms.Larger distance will reduce the sensitivity of the sensor, hence the distance can be ranged from



3-15 cms. The antenna orientation is chosen to be 0 degree, because at higher distances, it isfound that the node cannot have a higher RSSI values. Having lower RSSI values may result inthe higher error rates in data transmission, which will not be reliable for in-body communication.In order to evaluate the set 2 parameters of the PHY layer, the optimum parameters are set based

Parameter Value

Transmission power -10dbmTransmission distance SC1 - 16 cms

SC2 - 40cmsSC3 - 100

Frequency of operation 433MHzAntenna orientation 0 ◦

Table 5.6: Optimum values derived from validating the set of physical parameters

on the analysis. The optimum is chosen where the node is able to receive valid signal strengthmeeting the MICS standards.

5.5.2 Results from evaluating the set of network parameters

The set of physical parameters such as transmission power, distance and antenna orientation arefixed based on evaluation presented in Section 5.5.1 and listed in Table 5.6. The set of networkparameters parameters are validated in this section. The results will be able to define the PHYparameters for the network performance. As setup section 5.4.3, the same scenarios are used toevaluate the set of network parameters. Data rate is swept from the minimum of 2 Kbps till150 Kbps, and the corresponding paket delivery ratio is measured. The packet delivery ratiowill indicate how many packets succeed in reaching the receiver. Upon varying the transmissionrate, the effect of the physical medium on data communication can be fixed. Different packetsize will help in identifying the number of retransmission in case of higher packet lengths. NoMAC is used, hence until the packet is received at the receiver, the retransmission will be carriedout. Upon exceeding the maximum number of retransmissions, the node will ignore the packet.However, in this experiment, as soon a packet fails, it is considered as a dropped packet. Resultsfor different Scenarios are discussed in the following subsections. SC1 Transmission power,distance and orientation are listed in table 5.6. The packet length is fixed for each iteration ofthe measurement. In Scenario 1, as the packet length increases, even in the higher data rates,the packet delivery ratio is decreased. This indicates that the higher packet length, will make thenode to communicate for a longer time. However, the failure of the packet can occur if the frameis corrupted due to very low RSSI. It is also conversely found that the higher data rates will notincrease the packet delivery ratio, in case of higher packet length. For a fixed packet length, thePDR increases as the data rate increases. An important observation is that the PDR is almostconstant at higher data rates for a fixed packet length. SC2In Scenario 2, as the packet lengthincreases, even in the higher data rates, the packet delivery ratio is decreased. This behaviouris similar to that of the Scenario 1. The data rate is directly proportional to the packet deliveryratio, however it changes with that of the packet length. The packet length decreases the PDRin lower data rates and PDR increases as the data rate is increases. This indicates that thehigher packet length, higher data rates is required to maintain the reliability of the network.The performance can be related to that of the SC1, since the Scenario is partially inside the



0 50 100 15055

60

65

70

75

80

85

90

95

100

Tx rate Kbps

Pac

ket d

eliv

ery

ratio

%

Tx rate vs PDR (fixed packet length)

packet legnth (PL) 5 bytesPL 20 bytesPL 40 bytesPL 60 bytesPL 80 bytesPL 100 bytes

Figure 5.23: Tx Rate vs Packet delivery ratio evaluated with fixed packet length. Measured inSC1

flesh.

0 50 100 15060

65

70

75

80

85

90

95

100

Tx rate Kbps

Pac

ket d

eliv

ery

ratio

%




SC3 In Scenario 3, the behaviour of the network totally differs than that of the implanted nodes.



The implications are similar to that of the data communication in the free space. It also importantto notice that the distance in the SC3 validation is higher than that of the previous Scenarios.The dependency of the data rate and packet delivery ratio still is linearly proportionate. Forhigher packet length, the packet delivery ratio is lower in lower data rates. The analysis of the

0 50 100 15080

82

84

86

88

90

92

94

96

98

100

Tx rate Kbps

Pac

ket d

eliv

ery

ratio

%




experimental results were studied and the results are explained. The selection of parameters forthe next set of experiments are fixed as shown in table 5.7. The table indicates the values of thedata rate and the packet length. The choice of the value is based on the results from the previousexperiments. Small packet length with twice the data rate will have better performance in allthe Scenarios. Thus in order to verify the network performance, the most suited parameters arechosen which is twice the packet length. In this scenario, the node can also send packets muchfaster, which will also reduce the congestion in the network and increase the reliability. Higherpacket length resulting in lower packet delivery ratio is not a suitable candidate for the in-bodysensor network.

Parameter Value

Packet length 20 bytesData rate 40 Kbps

Table 5.7: Optimum parameters from hardware evaluation



5.6 Conclusion from characterization of implantablesensor node

Different Scenarios were chosen for evaluation of PHY parameters. Each validation is performedbased on the requirements of the network for different application scenarios. Upon selection ofvalidation parameters, the experiment is carried out such that each parameter is chosen andvalidated robustly. To obtain a data point, each experiment is performed redundantly for 10times and the average is chosen to reduce the human error. Upon verifying the results, it seemsthat the effect of flesh in radio communication is not best performed as in the case of the freespace propagation of radio. To compensate for the losses induced by the animal tissue, thesettings of PHY layer can be adjusted. In order to characterize the given hardware, for a givenanimal tissue, it is necessary that various experiments are carried out with different parametersas explained in the previous sections. These results can be used also in future experiments whenvalidating the network performance. However, it is wise to note that different animal tissueswill have different electromagnetic properties. Thus, if the hardware is changed or if a differentpart of animal tissue is used, it is necessary to repeat the characterization. In such cases, thesame approach as explained in this section can be carried out. From the results, the optimumparameters are chosen as shown in section 5.8.

5.7 Optimum parameters

The optimum parameters for the implementation of MAC layer protocols is chosen based on thecharacterization results of CC430 based implant, as listed in table 5.8. This setting of the PHYwill be used here after in the network evaluation.

Parameter Value

Transmission power -10dbmTransmission distance SC1 - 16 cms

SC2 - 40cmsSC3 - 100

Frequency of operation 433MHzAntenna orientation 0 ◦

Packet length 20 bytesData rate 40 Kbps

Table 5.8: Optimum parameters from hardware evaluation

70 5.6. Conclusion from characterization of implantable sensor node

Chapter 6

Performance evaluation ofwake-up feature basedCSMA/CA protocol

This chapter discusses about the effect of wake-up radio in Carrier sense multiple access withcollision avoidance (CSMA/CA) protocol in the sensor nodes and evaluation of the networkperformance using the optimum parameters for physical layer listed in Chapter 5. The effectof wake-up feature in the CSMA/CA protocol is also evaluated. The main aim of this chapteris to verify the software simulation results of Chapter 4 in implantable sensor node and to findthe effect of wake-up radio in the chosen CSMA/CA protocol. In order to know evaluate thenetwork performance with sensor node and in a implanted environment, a CSMA/CA basedMAC is implemented in the sensor node. The results from characterising the physical layerof the sensor nodes are used in order to confirm the functioning of the radio communicationas per the IEEE standards when implanted inside the animal flesh. The results from softwaresimulation has proved that the wake-up radio will decrease the power consumption and increasethe performance of the network parameters listed in Chapter 2. To evaluate the performanceof the wake-up radio based MAC protocol in IBSN scenario, experimentation with network ofsensor nodes implanted inside the animal tissue and evaluation of network parameters is required.The CSMA/CA access mechanism with wake-up radio had shown interesting results in chapter 4where, the impact of wake-up radio had increase the network performance by 20% and behavedsimilar to that of TDMA based access mechanism without any need of time-synchronisation overthe network. Hence in this chapter, CSMA/CA access mechanism is chosen and evaluated withimplanted sensor nodes.

6.1 CSMA/CA without wake-up radio

In CSMA/CA MAC protocol without wake-up feature, as soon as a node receives a packet thatis to be sent, it checks to be sure the channel is clear (no other node is transmitting at thetime). If the channel is clear, then the packet is sent. If the channel is not clear, the node waitsfor a randomly chosen period of time, and then checks again to see if the channel is clear. Thisperiod of time is called the backoff factor, and is counted down by a back-off counter. If thechannel is clear when the back-off counter reaches zero, the node transmits the packet. If thechannel is not clear when the backoff counter reaches zero, the back-off factor is set again, and

71

CHAPTER 6. PERFORMANCE EVALUATION OF WAKE-UP FEATURE BASEDCSMA/CA PROTOCOL

the process is repeated. The protocol should keep the number of collisions to a minimum, evenunder the highest possible load. To this end, the range of the random delay, or the contentionwindow (CW), is set to vary with the load. In the case of a collision, the delay is doubledprogressively: 15, 31, 63,...1023, until a successful transmission occurs and the delay is resetto the minimal value. If a node has a packet to send in slotted CSMA/CA, then the threeparameters are initialized and the boundary of the next backoff period is located. The MAClayer delays the assessment of the channel for a random number of backoff periods in the range0 to 2(BO) . After the delay on next backoff boundary, the node will start transmitting, if twoCCAs, the packet transmission, and any acknowledgement can be completed before the end ofthe CAP within the current window. If it cannot be completed, then it will start evaluationat the start of next window. If the channel is assessed to be busy, then the MAC incrementsback-off period by one and also ensures that back-off be less than or equal to maximum backoff.CW would be reset to 2 and if the value of backoffs, the CSMA/CA shall start another randomdelay. Otherwise, the CSMA/CA terminates with a Channel Access failure status.

Packet format The packets are formatted in accordance with the IEEE 802.15.6 standards.Each packet have a preamble, synchronization bits ( only in WuR implementation), length of thepacket information, dest address and source address, device info as guided by the manufacturersand the user data. Size of each section is shown in the Fig. 6.1. The purpose of preamble isnot to time synchronize the data, but to inform the receiver that it is the start of the frame.Synchronization bits is not present in the CSMA/CA implementation and is zero padded duringthe transmission. In WuR implementation the time at which the node should be waken-up ispresented in this part of the frame. Destination and receiver address is not necessary in thiscase, however as the size of the network increases, it will be helpful to prevent idle listening fromother nodes. Port information and the device information is send along with the packet in orderto satisfy the requirements of the hardware provided. The payload data is set to 20 bytes in thiscase as an optimum from the previous experiments. There is no provision of error checking inthis stage, as the network size is smaller and simplest topology is used.

Figure 6.1: Packet format of the CSMA/CA protocol.

Flowchart for CSMA/CA without wake-up radio implementation The CSMA/CA im-plementation is carried out as shown in the Fig. 6.2. The back-off timer is set based on theformula, BT = 2BE−1 where BE is the exponentially increasing number.

72 6.1. CSMA/CA without wake-up radio


start

Set back-off timeto zero

Assemblethe frame

PerformCCA

Is thechannelidle ??

Transmitapplication

data

Wait forrandombackofftime

stop

no

yes

Figure 6.2: Flowchart for CSMA/CA

f o r ( i = 1 ; i <= MAXNODES; i++) {i f (m. nbr [ i ] . weight != −1) {

node weight = moddif f (m. nbr [ i ] . weight , m. nbr [ source ] . weight ) ;weight [ node ] [ i ] = node weight ; // master node has the h i ghe s t p r i o r i t ymake pkt ( pkt ) ;pkt−>type = type ;pkt−>s i z e = 0 ;pkt−>dest = i ;i f (CSMADEBUG)send pkt ( pkt ) ;

}

Code 6.1: CSMA/CA pseudo-code

6.2 CSMA/CA with wake-up radio

The Wake-On-Radio (WOR) feature of the CC430 is a suggested method for conserving powerin wireless systems in which the radio sleeps all the time but periodically wakes up from SLEEPmode and listens for incoming packets. The wake-up radio is emulated using the WOR function

6.2. CSMA/CA with wake-up radio 73


for every node in the network. WOR is a timer, which is available in CC430. The value can beset manually and the radio wakes-up as set by the user. Upon waking up the radio by internaltimer, clear channel assessment(CCA) of the channel is performed. If the channel is free, theradio is sends the data to the base station which is always active. The information about thewake-up period is stored in the central base station. If the base station wants to communicateto the node, the communication to the node is carried out only during the active period of thenode after performing CCA. If the base station fails to communicate in the active period itwaits for the next active period of the specific node. From the node, the emergency data is sentimmediately with CSMA/CA access mechanism. The emergency data simulated using the timerA (different from WOR timer) interrupt in a specific node. In real medical scenario any one ofthe node in a network will sense critical data [3]. Hence, one of the node in the network is set tointerrupt by random timing interval using timer A interrupt. In normal mode, the sensor nodesare triggered by the base station at their active period(0.1% duty cycle) and request the data.The CSMA/CA is implemented as explained in section 6.1. The operationg of Wake-up featureincluded CSMA/CA access mechanism is shown in Fig. 6.4.

WOR functioning The WOR feature on the CC430 family devices provides two events, Event0 and Event 1, which can be leveraged to wake up and stabilize the radio core oscillator, and tochange the radio to RX mode. Another programmable parameter that the WOR feature uses isthe RX timeout, which determines the period during which the radio stays in RX mode. If apacket is received before the period reaches the RX timeout value, the CC430 can process thereceived packet and return to SLEEP mode. On the other hand, if no packet is received duringthe RX active period, the radio resumes the SLEEP state after the RX timeout.The applicationhas the flexibility to specify the wake up interval (tEvent0) as well as the RX active period (tRXtimeout) within each interval. RF systems that require power optimization could lengthen thewake up interval or decrease the RX active duty cycle. Vice versa, more responsive systemswith higher packet reception rates could be obtained by decreasing time duration of Event0 orincreasing time duration of RX timeout values. Operation of the CC430 wake-up feature is shownin Fig. 6.3

Figure 6.3: Wake-up feature of CC430 (ref: CC430 datasheet[45])

74 6.2. CSMA/CA with wake-up radio


start

Set back-off timeto zero

Assemblethe frame

Is thenode

awake ?

Wait tillthe nodeis awake

PerformCCA

Is thechannelidle ??

Transmitapplication

data

Wait forrandombackofftime

stop

yes

no

no

yes

Figure 6.4: Flowchart for CSMA/CA with wake-up feature

6.2. CSMA/CA with wake-up radio 75


6.3 Performance analysis of CSMA/CA protocol with andwithout wake-up radio

Experiments were repeated for different values of packet inter-arrival time, while observing para-meters such as delay, power consumption, packet drop ratio and duty cycle. A performanceevaluation is made based on these parameters in order to compare the effect of CSMA/CA withand without wake-up radio feature in in-body scenarios and these parameters will help us tounderstand the reliability trade-off in different channel model conditions. Inter-packet arrivaltime is a commonly used parameter to evaluate network performance, as it is directly affected bychanging network conditions. The data is recorded locally in the flash memory of the device. Thedata stored is of 30 minutes long. The choice of 30 minutes is because of the smaller networksize and simpler topology. The network gets saturated in 30 minutes of duration. Moreover,the battery life is shorter, which will not enable all the experiments to be carried out in singleexperimental setup. Changing the setup for each experimental run is not advisable as it mayalso influence the measurement and this deviation is measurement is assumed to be constant.Hence, no change in set-up for different experimental run is made recording the data for 30minutes. Information from software simulation also conveys that the network topology plays animportant role in selection the time duration of evaluation. Later the data from the flash is readout through a UART communication, and is analyzed on a PC using MATLAB.

6.3.1 Network setup

In order to evaluate the performance of communication between the sensor nodes and betweenthe controller and the sensor nodes, a network has to be setup. Different scenarios as explainedin the Chapter 5, is used to evaluate the network. The network is setup in a star-topology witha central coordinating node and four client nodes. As shown in Fig. 6.5, the implementation isdone with a central coordinator and four clients.

Figure 6.5: The network topology

6.3.2 Results and discussion

The data obtained are analyzed in MATLAB. The results are presented in the following subsec-tions.Duty cycle Duty cycle is the proportional duration for which the main radio is turned on in use-ful data communication. Variation in inter-packet arrival time (IPAT) will cause the duty cycleto be higher, which means radio has to be turned on for much longer time, if the time betweenthe arrival of packets at the receiver is longer. In the CSMA/CA-based access mechanism thenode have to sense the medium randomly after every back-off time, which causes the radio to beturned on unnecessarily. In case of the WuR mac, the duty cycle is lesser because the node willhave the information about the data at which it has to send. Ensuring that the node will turn

76 6.3. Performance analysis of CSMA/CA protocol with and without wake-up radio


on the radio only when the node sends a data to the controller.Power consumption The power consumption can be predicted from duty cycle and the valuesof current and voltage consumption from the data sheet of the radio. The CC430 has a CC1101radio chip embedded on it. The current consumption for sleep mode is 200nA and 15 mA inactive mode [30]. The radio is operated with a battery power supply of 3.3 V. Hence if the radiois active for 1 second 49.5 mW (power(W ) = current(A) ∗ voltage(V )) is consumed and whilesleeping 0.06 mW is consumed. This power calculation is to show that the longer the sleep mode,shorter the duty cycle and hence lesser the power consumption1. From the Fig. 6.6, the dutycycle of WUR based CSMA/CA is much lower than the CSMA/CA, which means that in WURbased CSMA/CA the radio is in longer sleep duration than the CSMA/CA itself. Hence thepower consumption of the sensor node with WUR based CSMA/CA access mechanism will belesser than CSMA/CA based access mechanism.

0 5 10 15 20 25 30 35 40 45 500

10

20

30

40

50

60

70

80

90

100


Duty

cycle

(%

)

CSMA/CA without WuR

CSMA/CA with WuR

Figure 6.6: Inter packet arrival time vs Duty cycle. Comparison between CSMA/CA with andwithout Wake up radio

Packet delivery ratio The packet delivery ratio (PDR) is the ratio of the total number ofpackets received at the receiver to the total number of packets generated at the sender. A 100%

1In reality, the power consumption of the sensor node also depends on the processor power consumption whichwill be added to the power consumption of the embedded radio.

6.3. Performance analysis of CSMA/CA protocol with and without wake-up radio 77


5 10 15 20 25 30 35 40 45 500

10

20

30

40

50

60

70

80

90

100


Packet deliv

ery

ratio (

%)

CSMA/CA without WuR

CSMA/CA with WuR

Figure 6.7: Inter packet arrival time vs Packet delivery ratio. Comparison between CSMA/CAwith and without Wake up radio

PDR means there is no loss of packets. PDR is an important metric to analyze the networkperformance with different physical layer parameters with a specific topology. In case of thein-body sensor network, inter-packet arrival time can vary due to the highly varying physicalproperties of the human body. The PDR is evaluated for different IPAT to find the reliabilityof the different CSMA/CA protocols with wake-up radio. As a result, it is shown that all thetwo implementations perform with high PDR at lower IPAT. However, as the IPAT increasesthe PDR of the protocols decreases. Out of the two implementations chosen, CSMA/CA withWuR has a better delivery ratio, which is due to the reduced overheads and hence packets canbe transferred in less time similar the case of TDMA-based approach. The wake-up featureimplemented is similar to that of the TDMA based approach except for the fact that there is noreal hardware based time synchronization in the network.

Delay The delay in the sensor network is defined as the difference in time taken for a set of datathat is sent from the source to reach at the destination. Varying IPAT in different protocols willhave influence on the delay due to the effect of preambles and control beacons sent alongsidethe data. In Fig. 4.5, two different implementations of CSMA/CA with and without WuR arecompared in terms of delay. It is shown that the CSMA/CA without WuR has the highestdelay with higher IPAT. The delay is larger in CSMA/CA without WuR because at higher

78 6.3. Performance analysis of CSMA/CA protocol with and without wake-up radio


0 1 2 3 4 5 60

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5


End−

End D

ela

y (

seconds)

CSMA/CA without WuR

CSMA/CA with WuR

Figure 6.8: Inter packet arrival time vs End to End delay. Comparison between CSMA/CA withand without Wake up radio

IPAT the carrier sense overheads such as CCA, and back-off time for the data communicationto complete adds to the effective data communication. In the case of WuR CSMA, delay isalmost constant for the data communication, because the only major delay is by wake-up timerdata and transmission of packets. The synchronization bits are encoded in the packets whichis also a reason for constant delay, however it does not reduce the effective throughput of thetransmitted data. From the experimental analysis of the delay, it is known that WuR-basedprotocols has less delay even when the real hardware is not present. The slot allocation based onthe predefined timer value for the channel access for each node is the crucial reason for delay incase of WuR. In case of higher IPAT, the delay increases linearly in both the cases of CSMA/CAimplementation, which is due to the fact that the number of nodes and topology remains thesame whereas the inter-packet arrival time is increased. However, similar results can be achievedwith CSMA-based protocols without WuR using guaranteed time slot (GTS) mechanisms whichrequires time synchronization in the network, adding extra overheads.

6.3. Performance analysis of CSMA/CA protocol with and without wake-up radio 79


6.4 Conclusion

Despite the fact that there is no dedicated for wake-up radio,we simulated the wake-up radiousing the WOR feature of the chip. The WuR implementation is more close to the TDMA basedapproach but with no time synchronization required. The time is synchronized with informationfrom the data packets from base station, no separate preamble is required for synchronisation.The analysis of CSMA/CA with WuR feature performs better than CSMA/CA without WuRdue to unique reasons such as prevention of idle listening, and over hearing using WUR feature.The power measurement is not made since no dedicated wake-up radio is used. However, thepower consumed can be predicted using the duty cycle analysis. From the results the CSMA/CAwithout wake-up feature has higher duty cycle than the CSMA/CA with the wake-up feature.Higher duty cycle shows that long period of time the radio is switched on, hence higher powerconsumption. As explained in section 6.3.2, CSMA/CA with WUR has lower power consumptionthan CSMA/CA without WUR.As a conclusion, the use WuR feature increases the network performance in terms of low latency,shorter overheads, guaranteed packet delivery ratio. The results obtained in section 6.3.2 verifiesthe software simulation performed earlier except for the power consumption. The performance ofthe network increases in terms of packet delivery ratio and lower latency,by incorporating WuRto the CSMA/CA implementation. From the duty cycle measurement, it can also be claimedthat the power consumed will also be lesser when real wake-up radio is implemented, as observedin the software simulation of Chapter 4 the power consumed by the external wake-up radio ismuch lower than that of the main radio.

80 6.4. Conclusion

Chapter 7

Conclusions

In this work, the communication mechanisms for the in-body sensor networks is characterizedby evaluating the physical layer and MAC sub-layer of the IBSN. A brief introduction about theneed for IBSN in health-care is presented in chapter 1. The main drawbacks of the state-of-the-art medical devices are listed and the possible solution with the closed loop medical devices ispresented. In order to realize the closed loop architecture the need of IBSN is mandatory, hencethe requirements of the IBSN are formed aiming at a network of closed loop medical devices. Anintroduction to IBSN and the current research topics of MICS band in physical layer and wake-up radio integration in the MAC sub-layer was given and their need to materialize the IBSN isdiscussed. Based on this discussion a research question is framed to justify the hypothesis whichstates that wake-up radio integration in the MAC layer will increase the network performanceand reduce the power consumption while operating in the MICS band physical layer. In chapter2, the requirements for the IBSN are presented with the description of appropriate technologiesneeded to fulfill the requirements of IBSN. Wake-up radio operation is explained and the needfor MICS band in medical device communication is discussed. As a result, the requirements forcharacterizing the IBSN are listed.There are numerous MAC protocols which already exists for different applications of WSN, andmore specifically for the body sensor networks. Very few protocols have been proposed so far inliterature for an application in IBSN. However, some of the existing MAC protocols for BSN maysuit the requirements of IBSN as well. In order to understand the existing MAC protocols, aanalytical survey was made in Chapter 3, with more than 30 MAC protocols which has the wake-up feature and that could fulfill the requirements of IBSN. The MAC protocols were surveyedon their design, and network performance based on the selected parameters such as reliability,latency, effective throughput, duty cycle. As a result it is found that the MAC protocols withwake-up radio are better than the MAC protocols without wake-up radio, in terms of energyefficiency, latency, less overhead and reliability. However, it is also found that the wake-up radiowill increase the hardware overheads if a dedicated radio is embedded on the implant. Neverthe-less, practicality of wake-up radio in the current radio chips is studied and found that the addedhardware overhead is not a big problem since differnet implants are packed with dual radio onchip and adds no extra hardware. The minor increase in hardware overhead will be negligiblewhen compared to the benefits in increase performance and reduced power consumption.Out of the surveyed MAC protocols with and without wake-up radio, three protocols with wake-up radio is chosen for software evaluation. The reason is that in literature the performance ofcontention based (CSMA/CA) access mechanism with wake-up radio was better than perform-ance of contention-free (TDMA) access mechanism. In order to study the reasons behind the

81

CHAPTER 7. CONCLUSIONS

performance increase, and to validate the design of the MAC protocols with wake-up radio asoftware simulation of the implantable sensor network is done in MATLAB following the IEEE802.15.6 standard. The results matched with the survey results and main reason for the im-proved performance was the wake-up radio eliminated the conventional problems of CSMA/CAapproach such as over-hearing and idle-listening.The results from software simulation was as alongside to the hypothesis made. To verify thefindings from software analysis, an analysis of physical layer and MAC layer with the implant-able sensor node was made in Chapter 5. The implants were first characterized for the physicalparameters such as distance between transmitter and receptor, transmission power, orientationof antenna. Upon characterizing the optimum parameters, the values are used to characterizethe physical layer of OSI network model such as transmission data rate, packet delivery ratioand packet length. Upon characterizing the optimum values for each parameter are found, whichare then used for the evaluation of MAC sub-layer. The characterization was done in order toknow the effect of flesh as the physical medium of radio communication. The values are foundto be deviating largely for different values of physical layer parameters. This explains why acharacterization of physical layer was necessary for a given implantable sensor node. Upon find-ing the physical layer characteristics, the MAC sub-layer is anaylsed with and without wake-upfeature in chapter 6. The performance of CSMA/CA was found to be largely deviating from theCSMA/CA with wake-up radio. To verify this in real hardware, the CSMA/CA with wake-upfeature was analyzed with the sensor node inside the animal flesh. The three main networkparameters, interpacket arrival time, latency, packet delivery ratio, were analyzed, since theseparameters ensure the reliability of a network and also influences the energy efficiency of thenetwork. The CSMA/CA with wake-up radio was implemented using a wake-on radio timer ofthe chip simulating the effect of dedicated wake-up radio. In all the cases, the wake-up radiowas performing well in terms of the analyzed network parameters. Lower latency due to reducedoverheads and lesser collision, higher packet delivery ratio due to lesser collision of data pack-ets and the absence of unwanted back-off from the medium are the findings of the performanceanalysis CSMA/CA with wake-up feature.

7.1 Answer to the research question

The research question which was raised during the start of this thesis work was

Can a wake-up radio integrated with MAC protocol, meet the QoS requirements and powerconstraints of an IBSN while operating inside the human body ?

.The chapters 3 -6 were focused on the research about wake-up radio integration int he MAClayer and find the effect of the same in software environment and in real hardware environment.We found that a MAC protocol integrated wake-up radio can meet the QoS requirements andpower constraints of an IBSN more appropriate than a MAC protocol without wake-up radio.From our analysis it is known that the least performing access mechanism can be improved usingthe wake-up radio integration and careful designing of the MAC protocols. In all the expectedmedical scenarios of IBSN, the wake-up radio based MAC protocol can serve the purpose betterthan the non-WUR MAC protocols.

82 7.1. Answer to the research question

CHAPTER 7. CONCLUSIONS

7.2 Future work

Different suggestions and ideas were raised during the development of this thesis work. Due tothe time constraints there were not included in this thesis, but they are listed in this section asfuture work,

• The IBSN in different application perspective may require a different topology than thestar topology. For instance P2P was one of the considered topology for medical scenarioSC2.

• TDMA, CSMA/CA kinds of access mechanisms have their unique advantages in IBSN scen-ario. Their advantages in accessing the medium can be combined to form a hybrid accessmechanism integrated with WUR. The resulting protocol will be meeting the requirementsof IBSN nodes.

• In Chapter 6 the wake-up feature is integrated to CSMA/CA using emulated timers. Theresults using dedicated hardware should be more oriented towards the better performance.Radio chips available for the medical implants with dual band radio should be used forimplementation

• The effect of bones was not characterized in this study. A study characaterizing the effectof boned would be needed if radio communication is needed for in-heart pacemakers, wherethe RF link is established across the bones.

• The results from power measurement in Chapter 5was not made across the processor. Thepower measurement with dedicated wake-up radio and across the processor of the sensorneeds to be measured. This information will be useful to design a MAC protocol withdedicated energy awareness feature.

• Different architecture of wake-up radio mechanisms can be used, and the effect of usingdifferent architecture in MAC protocol in terms of performance and energy efficiency canbe characterized.

• The effect of insulating material is not completely characterized. Bio-compatible poly-mers have different conductive properties than normal plastics or paraffins. Effect of theinsulating materials can be studied.

7.2. Future work 83

Bibliography

[1] Teodor Goroszeniuk and David Pang. Peripheral neuromodulation: a review. Current painand headache reports, 18(5):412, May 2014. 1

[2] Joseph F Jasper and Salim M Hayek. Implanted occipital nerve stimulators. Pain physician,11(2):187–200, 2008. 1

[3] Gordon F Tomaselli. Late-Breaking Clinical Trials II Four Seasons Ballroom 1. pages 1–5,2013. 1, 3, 5, 37, 57, 74

[4] Mathieu Favier, Carole Carcenac, Guillaume Drui, Sabrina Boulet, Salah El Mestikawy,and Marc Savasta. High-frequency stimulation of the subthalamic nucleus modifies theexpression of vesicular glutamate transporters in basal ganglia in a rat model of Parkinson’sdisease. BMC neuroscience, 14:152, January 2013. 1

[5] Didier Sagan. RF Integrated Circuits for Medical Applications. Zarlink Semiconductor. 2,3

[6] Pranesh Sthapit and Jae-Young Pyun. Intelligent network synchronization for energy savingin low duty cycle MAC protocols. 2009 IEEE International Symposium on a World ofWireless, Mobile and Multimedia Networks & Workshops, pages 1–6, June 2009. 3

[7] Arib Std. Specific Absorption Rate ( SAR ) Estimation for Cellular Phone. 1998. 3

[8] Technical Report. TR 102 434 - V1.1.1 - Electromagnetic compatibility and Radio spectrumMatters (ERM); Short Range Devices (SRD); Alternative Interference Mitigation Technolo-gies to Listen Before Talk (LBT) for Ultra Low Power Active Medical Implants (ULP-AMI)operating f. 1:1–20, 2005. 3

[9] Joseph Wang. Electrochemical glucose biosensors. Chemical reviews, 108(2):814–825, Feb-ruary 2008. 4, 39

[10] C. Beck, B. Schulz, J. a. Nagel, H. Guth, U. Gengenbach, and G. Bretthauer. Low dutycycle inter-implant communication of the Artificial Accommodation System. 2010 3rd In-ternational Symposium on Applied Sciences in Biomedical and Communication Technologies(ISABEL 2010), pages 1–5, November 2010. xi, 4

[11] Ali Barati, Ali Movaghar, Samira Modiri, and Masoud Sabaei. A Reliable & Energy-Efficient Scheme for Real Time Wireless Sensor Networks Applications. Basic. Appl. Sci.Res, 2(10):10150–10157, 2012. xi, 4

Bibliography 85

BIBLIOGRAPHY

[12] Gary McKeown, William Curran, Denise Kane, Rebecca Mccahon, Harry J. Griffin, CiaranMcLoughlin, and Nadia Bianchi-Berthouze. Human Perception of Laughter from Context-Free Whole Body Motion Dynamic Stimuli. 2013 Humaine Association Conference onAffective Computing and Intelligent Interaction, pages 306–311, September 2013. 5

[13] Sonali Agarwal and G. N. Pandey. SVM based context awareness using body area sensornetwork for pervasive healthcare monitoring. Proceedings of the First International Con-ference on Intelligent Interactive Technologies and Multimedia - IITM ’10, pages 271–278,2010. 5

[14] Dirk Adolph, Lukas Meister, and Bettina M Pause. Context counts! social anxiety modulatesthe processing of fearful faces in the context of chemosensory anxiety signals. Frontiers inhuman neuroscience, 7(June):283, January 2013. 5

[15] David Layerle. A Power Efficient Pulsed MAC Protocol for Body Area Networks by ThesisRelease Permission Form Rochester Institute of Technology. 2011. 11

[16] Salim A Hanna. Regulations and Standards for Wireless Medical Applications. pages 1–5.11

[17] Invited Paper. An Overview of IEEE 802.15.6 Standard. pages 2–7. 13, 21, 22

[18] Wei Ye, John Heidemann, and Deborah Estrin. An Energy-Efficient MAC Protocol forWireless Sensor Networks. 00(c):1567–1576, 2002. 13, 31, 38

[19] Nuria Rodrıguez, Jack Weissberg, and Gerald E Loeb. Communication and Control forReanimating Paralyzed Limbs via a Network of Wireless Micro-Implants. 13

[20] N Javaid, I Israr, M A Khan, A Javaid, S H Bouk, and Z A Khan. Analyzing MediumAccess Techniques in Wireless Body Area Network. 13

[21] Mics R F Telemetry. ZL70102 Medical Implantable RF Transceiver. (May), 2012. 15

[22] Compact Design and Fully Shielded Package. ZL70120 MICS-Band RF Base Station Module( BSM ) Table of Contents. (February), 2013. 15, 16

[23] Implant-grade Quality Assurance, Integrated High-performance Antenna Matching, An-tenna Tuning Circuitry, Compact Design, Fully Shielded Package, Integrated Reference,and Frequency Crystal. ZL70321 MICS-Band RF Standard Implant Module ( SIM ) Tableof Contents. (February), 2013. 15, 16

[24] Short Range Devices, Ultra Low, Power Active, and Medical Implants. EN 301 839-1 -V1.3.1 - Electromagnetic compatibility and Radio spectrum Matters (ERM); Short RangeDevices (SRD); Ultra Low Power Active Medical Implants (ULP-AMI) and Peripherals(ULP-AMI-P) operating in the frequency range 402 MHz to 405 MHz; Part 1: T. 1:1–52,2009. 16

[25] N.E. Roberts and D.D. Wentzloff. A 98nw wake-up radio for wireless body area networks. InRadio Frequency Integrated Circuits Symposium (RFIC), 2012 IEEE, pages 373–376, June2012. 16

[26] Megan C Frost and Mark E Meyerhoff. Implantable chemical sensors for real-time clinicalmonitoring: progress and challenges. Current Opinion in Chemical Biology, 6(5):633–641,2002. 16

86 Bibliography

BIBLIOGRAPHY

[27] Peter D Bradley. Implantable ultralow-power radio chip facilitates in-body communications.RF DESIGN. 16

[28] Aleksandar Milenkovic, Chris Otto, and Emil Jovanov. Wireless sensor networks for personalhealth monitoring: Issues and an implementation. Computer communications, 29(13):2521–2533, 2006. 16, 19

[29] Paul D Mitcheson, Student Member, Tim C Green, Senior Member, Eric M Yeatman, andAndrew S Holmes. Architectures for Vibration-Driven Micropower Generators. 13(3):429–440, 2004. 17

[30] Nerupace. Neural stimulator for treating epilepsy @http://www.neuropace.com/, June 2014.18

[31] IEEE802.15 Task group 6. Channel model for body area network. IEEE 802.15.6 channelmodeling committee. 19, 44, 57

[32] Sai Anand Gopalan and Jong-Tae Park. Energy-efficient mac protocols for wireless bodyarea networks: Survey. In Ultra Modern Telecommunications and Control Systems andWorkshops (ICUMT), 2010 International Congress on, pages 739–744. IEEE, 2010. 19

[33] M. Lont, D. Milosevic, P.G.M. Baltus, A. H M Van Roermund, and G. Dolmans. Analyt-ical models for the wake-up receiver power budget for wireless sensor networks. In GlobalTelecommunications Conference, 2009. GLOBECOM 2009. IEEE, pages 1–6, Nov 2009. 19

[34] Sai Anand Gopalan. Energy-efficient MAC protocols for wireless body area networks: Sur-vey. International Congress on Ultra Modern Telecommunications and Control Systems,pages 739–744, October 2010. 21

[35] Sana Ullah, Manar Mohaisen, and Mohammed A Alnuem. and Security Specifications. 2013,2013. 22

[36] Stevan Jovica Marinkovi, Emanuel Mihai Popovici, Senior Member, Christian Spagnol, As-sociate Member, Stephen Faul, and William Peter Marnane. Energy-Efficient Low DutyCycle MAC Protocol. 13(6):915–925, 2009. 26, 27

[37] Gengfa Fang and Eryk Dutkiewicz. BodyMAC: Energy efficient TDMA-based MAC protocolfor Wireless Body Area Networks. 2009 9th International Symposium on Communicationsand Information Technology, pages 1455–1459, September 2009. 26, 27, 38

[38] Costas Busch, Malik Magdon-ismail, and Fikret Sivrikaya. Contention-Free MAC protocolsfor Wireless Sensor Networks. 26, 27, 38

[39] Changle Li, Jiandong Li, Bin Zhen, Huan-Bang Li, and Ryuji Kohno. Hybrid Unified-Slot Access Protocol for Wireless Body Area Networks. International Journal of WirelessInformation Networks, 17(3-4):150–161, August 2010. 27, 29, 33, 38, 39

[40] Begonya Otal, Luis Alonso, Christos Verikoukis, and Senior Member. Highly ReliableEnergy-Saving MAC for Wireless Body Sensor Networks in Healthcare Systems. 27(4):553–565, 2009. 27, 38

[41] Ernesto Ibarra, Angelos Antonopoulos, Elli Kartsakli, and Christos Verikoukis. HEH-BMAC: Hybrid Polling MAC Protocol for WBANs Operated by Human Energy Harvesting. 27,38

Bibliography 87

BIBLIOGRAPHY

[42] Sana Ullah, Xizhi An, Youngwoo Choi, and Hyungsoo Lee. Project: IEEE P802.15 WorkingGroup for Wireless Personal Area N etworks ( WPANs ). (July):1–16, 2008. 27

[43] Kyung Tae Kim, Won Jun Choi, Myung Jin Whang, and Hee Yong Youn. CA-MAC: ContextAdaptive MAC Protocol for Wireless Sensor Networks. 2009 International Conference onComputational Science and Engineering, pages 344–349, 2009. 28, 29, 38

[44] Huaming Li and Jindong Tan. An Ultra-low-power Medium Access Control Protocol forBody Sensor Network. Conference proceedings : ... Annual International Conference ofthe IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine andBiology Society. Conference, 3:2451–2454, January 2005. 28, 38

[45] Joseph Polastre, Jason Hill, and David Culler. Versatile Low Power Media Access forWireless Sensor Networks Categories and Subject Descriptors. 28, 29, 38

[46] Mohammad a. Huq, Eryk Dutkiewicz, and Rein Vesilo. MEB MAC: Improved channelaccess scheme for medical emergency traffic in WBAN. 2012 International Symposium onCommunications and Information Technologies (ISCIT), pages 371–376, October 2012. 28,29, 38

[47] Hui Cao, Kenneth Parker, and Anish Arora. O-MAC: A Receiver Centric Power Man-agement Protocol. Proceedings of the 2006 IEEE International Conference on NetworkProtocols, pages 311–320, November 2006. 28, 29, 38

[48] Ugo Maria Colesanti and Silvia Santini. DISSense : An Adaptive Ultralow-power Commu-nication Protocol for Wireless Sensor Networks. 28, 29, 36, 38

[49] June S. Yoon, Gahng-Seop Ahn, Seong-Soon Joo, and Myung J. Lee. PNP-MAC: Pree-mptive Slot Allocation and Non-Preemptive Transmission for Providing QoS in Body AreaNetworks. 2010 7th IEEE Consumer Communications and Networking Conference, pages1–5, January 2010. 29, 38

[50] Michael Buettner, GV Yee, Eric Anderson, and Richard Han. X-MAC: a short preambleMAC protocol for duty-cycled wireless sensor networks. . . . Embedded networked sensor . . . ,pages 307–320, 2006. 31, 38

[51] G. Zhou, J. Lu, C.-Y. Wan, M. D. Yarvis, and J. a. Stankovic. BodyQoS: Adaptive andRadio-Agnostic QoS for Body Sensor Networks. 2008 IEEE INFOCOM - The 27th Confer-ence on Computer Communications, pages 565–573, April 2008. 31, 38

[52] Samira Yessad, Farid Nait-Abdesselam, Tarik Taleb, and Brahim Bensaou. R-MAC: Reser-vation Medium Access Control Protocol for Wireless Sensor Networks. 32nd IEEE Confer-ence on Local Computer Networks (LCN 2007), pages 719–724, October 2007. 31, 38

[53] Khaled a. Ali, Jahangir H. Sarker, and Hussein T. Mouftah. Urgency-Based MAC Pro-tocol for Wireless Sensor Body Area Networks. 2010 IEEE International Conference onCommunications Workshops, pages 1–6, May 2010. 31, 38

[54] P Udayakumar, Ranjana Vyas, and O P Vyas. Energy Efficient Election protocol for wire-less sensor networks. 2013 International Conference on Circuits, Power and ComputingTechnologies (ICCPCT), pages 1028–1033, March 2013. 31, 38

[55] Jian Qiang, Gong Zheng-hu, and Gui Chun-mei. A MAC-Routing Integrated Protocol forGeographic Forwarding in Wireless Sensor Networks. Future Generation Communicationand Networking (FGCN 2007), pages 285–290, 2007. 31, 38

88 Bibliography

BIBLIOGRAPHY

[56] A Rahim, N Javaid, M Aslam, Z Rahman, U Qasim, and Z A Khan. A ComprehensiveSurvey of MAC Protocols for Wireless Body Area Networks. 30

[57] Hector Eduardo Roman. A cooperative medium access control protocol for mobile clustersin wireless body area networks. 2013 First International Symposium on Future Informationand Communication Technologies for Ubiquitous HealthCare (Ubi-HealthTech), pages 1–4,July 2013. 33

[58] Sankar Mukherjee, G P Biswas, A Fdma A Frame, and A Cluster A Mac. Recent Advancesin Information Technology. 266:11–19, 2014. 33, 39

[59] Mastooreh Salajegheh, Hamed Soroush, Antonis Kalis, and Athens Information Technology.HYMAC : HYBRID TDMA / FDMA MEDIUM ACCESS CONTROL PROTOCOL FOR.33, 39

[60] Koji Enda, Keisuke Sodeyama, Shun Nagamine, and Hideki Mochizuki. YNU PHY andMAC Design. Number 1. 2009. 33, 39

[61] John Farserotu, S A Csem, and S A Technologies. CSEM FM-UWB Proposal. 2009. 33, 39

[62] Benny Lo, Eric Yeatman, and Louis Atallah. Body Sensor Networks – Acknowledgement.34

[63] Jamshid Abouei and JD Brown. Energy efficiency and reliability in wireless biomedicalimplant systems. . . . in Biomedicine, IEEE . . . , pages 1–12, 2011. 34

[64] Ieee P. Table of contents. The FEBS journal, 281(7):1930, April 2014. 34, 35, 39

[65] Guido Dolmans, Li Huang, and Dusan Milosevic. Project: IEEE P802.15 Working Groupfor Wireless Personal Area N etworks ( WPANs ). 2009. 34, 35, 39

[66] Matthew J Miller, Student Member, Nitin H Vaidya, and Senior Member. A MAC Protocolto Reduce Sensor Network Energy Consumption Using a Wakeup Radio. 4(3):228–242, 2005.35, 39

[67] Junaid Ansari, Dmitry Pankin, and Petri Mahonen. Radio-triggered Wake-ups with Ad-dressing Capabilities for Extremely Low Power Sensor Network Applications. InternationalJournal of Wireless Information Networks, 16(3):118–130, July 2009. 35, 39

[68] Moshaddique Al Ameen, Niamat Ullah, and Kyungsup Kwak. Design and analysis of a MACprotocol for wireless body area network using wakeup radio. 2011 11th International Sym-posium on Communications & Information Technologies (ISCIT), (Iscit):148–153, October2011. 35, 39

[69] R.C. Shah, a. Bonivento, D. Petrovic, E. Lin, J. van Greunen, and J. Rabaey. Joint optimiza-tion of a protocol stack for sensor networks. IEEE MILCOM 2004. Military CommunicationsConference, 2004., 1:480–486, 2004. 35

[70] Trong Nhan Le, Michele Magno, Alain Pegatoquet, Olivier Berder, Olivier Sentieys, andEmanuel Popovici. Ultra low power asynchronous MAC protocol using wake-up radio forenergy neutral WSN. Proceedings of the 1st International Workshop on Energy NeutralSensing Systems - ENSSys ’13, pages 1–6, 2013. 35, 39

[71] Bram Kersten. MAC Layer Energy Concerns in WSNs : Solved ? 4(question 3):1–8. 35, 39

Bibliography 89

BIBLIOGRAPHY

[72] Tijs van Dam and Koen Langendoen. An adaptive energy-efficient MAC protocol for wirelesssensor networks. Proceedings of the first international conference on Embedded networkedsensor systems - SenSys ’03, page 171, 2003. 35, 39

[73] F. Fereydouni Forouzandeh, O. Ait Mohamed, M. Sawan, and F. Awwad. TBCD-TDM:Novel Ultra-Low Energy Protocol for Implantable Wireless Body Sensor Networks. GLOBE-COM 2009 - 2009 IEEE Global Telecommunications Conference, pages 1–6, November 2009.35, 39

[74] Pichitpong Soontornpipit. Effects of radiation and SAR from wireless implanted medicaldevices on the human body. Journal of the Medical Association of Thailand = Chotmaihetthangphaet, 95 Suppl 6:S189—-97, June 2012. 34

[75] David Arney, Miroslav Pajic, Julian M. Goldman, Insup Lee, Rahul Mangharam, and OlegSokolsky. Toward patient safety in closed-loop medical device systems. Proceedings of the 1stACM/IEEE International Conference on Cyber-Physical Systems - ICCPS ’10, (Iccps):139,2010. 34

[76] Mahesh Arumugam and S S Kulkarni. Self-stabilizing deterministic TDMA for sensor net-works. 2005. 38

[77] Nourchene Bradai, Lamia Chaari Fourati, Saadi Boudjit, and Lotfi Kamoun. New priorityMAC protocol for wireless body area networks. Proceedings of the 3rd ACM MobiHocworkshop on Pervasive wireless healthcare - MobileHealth ’13, page 1, 2013. 38

[78] LC Zhong, Rahul Shah, Chunlong Guo, and Jan Rabaey. An ultra-low power and distributedaccess protocol for broadband wireless sensor networks. IEEE Broadband Wireless . . . , 2001.38

[79] P. Sthapit and Jae-Young Pyun. Effects of radio triggered sensor mac protocol over wirelesssensor network. In Computer and Information Technology (CIT), 2011 IEEE 11th Interna-tional Conference on, pages 546–551, Aug 2011. 41, 49

[80] Joaquim Oller, Ilker Demirkol, Jordi Casademont, Josep Paradells, Gerd Ulrich Gamm, andLeonhard Reindl. Wake-up radio as an energy-efficient alternative to conventional wirelesssensor networks mac protocols. In Proceedings of the 16th ACM International Conference onModeling, Analysis & Simulation of Wireless and Mobile Systems, MSWiM ’13, pages173–180, New York, NY, USA, 2013. ACM. 41, 43, 45, 49

[81] Moshaddique Al Ameen, Niamat Ullah, M. Sanaullah Chowdhury, and Kyungsup Kwak. Amac protocol for body area networks using out-of-band radio. In Wireless Conference 2011- Sustainable Wireless Technologies (European Wireless), 11th European, pages 1–6, April2011. 41, 42, 49

90 Bibliography

Appendix A

Animal flesh in SC1

Figure A.1: Scenario 1, in-body - in-bodycommunication Distance = 16cm

Figure A.2: Scenario 1, in-body - in-bodycommunication, Distance = 6cm

91

Appendix B

Animal flesh in SC2

Figure B.1: Scenario 2, in-body 1 (to) in-body2communication,Distance= 20 cm

Figure B.2: Scenario 2, in-body 1 (to) in-body2communication,Distance= 160 cm

92

body sensor networks - universiteit twenteessay.utwente.nl/65921/1/karuppiahramachandran_ma_ewi.pdfi...

Documents