DEVELOPING AN ENERGY-EFFICIENT WIRELESS SENSOR NETWORK FOR

IMPROVED OPERATION ON CARDIOVASCULAR PROBLEMS IN

DEVELOPMENT ECONOMY

BY

AJAYI EBENEZER ADENIYI

FUOYE/Ph.D/EEE/2020/001

DEPARTMENT OF ELECTRICAL ELECTRONICS ENGINEERING,

FACULTY OF ENGINEERING,
FEDERAL UNIVERSITY, OYE-EKITI

DECEMBER, 2021

1
 DEVELOPING AN ENERGY-EFFICIENT WIRELESS SENSOR NETWORK FOR
IMPROVED OPERATION ON CARDIOVASCULAR PROBLEMS IN
DEVELOPMENT ECONOMY

BY

AJAYI EBENEZER ADENIYI

FUOYE/Ph.D/EEE/2020/001

B.ENG (ELECTRICAL/ELECTRONICS ENGINEERING)

UNIVERSITY OF PORT HARCOURT
M.ENG (ELECTRONIC/COMMUNICATION ENGINEERING)
MICHAEL OKPARA UNIVERSITY OF AGRICULTURE, UMUDIKE
PROGRESS REPORT SUBMITTED TO
DEPARTMENT OF ELECTRICAL ELECTRONICS ENGINEERING,
FACULTY OF ENGINEERING,
FEDERAL UNIVERSITY, OYE-EKITI

SUPERVISOR:
PROF. A.O. AKINSANMI

CO-SUPERVISORS:
DR. I.B. OLUWAFEMI
DR. A.S.OLUWOLE

MARCH, 2023

2
 TABLE OF CONTENTS

CHAPTER ONE
INTRODUCTION

1.1 Background to the Study 1

1.2 Statement of the problems 6
1.3 Aim and Objectives of the study 7
1.4 Justification 8
CHAPTER TWO

LITERATURE REVIEW

2.1 Wireless Sensor Networks Technology 9

2.2 Potential impact and benefits of WSN 12
2.3 Applications of Wireless Sensor Networks 16
2.3.1 Military Applications 16
2.3.2 Environmental Applications 17
2.3.3 Health or Medical sensor network Application 17
2.3.4 Home Applications 19
2.3.5 Other Commercial Applications 19
2.4 Energy Consumption Challenges in Wireless Sensor Networks 19
2.4.1 Throughput 20
2.4.2 Energy Efficient Design 20
2.4.3 Energy saving in interference environment 21
2.4.4 Message Delivery 21
2.4.5 Network Lifetime 21
2.4.6 Limited Energy 22
2.4.7 Scalability 22
2.4.8 Data aggregation 22
2.5 Energy Consumption of Sensor Nodes 23
2.5.1 Power Consumption 23
2.5.2 Energy Consumption during transmission 24

3
2.6 Energy Efficiency 25
2.7 Wireless Biomedical Sensor Networks 25
2.8 The comparison of technologies used in WSN in Medical Applications 27
2.9 Cardiovascular Applications (Telemedicine In Stroke) 30
2.9.1 The burden of the disease 30
2.9.2 The rural challenge 31
2.9.3 Telemedicine/Telestroke 33
2.9.4 Current and potential uses of telestroke 33
2.9.5 Telestroke requirements 34
2.9.5.1 Technology options for telestroke 34
2.10 Barriers to the implementation of telestroke models 36
2.10.1 Technical advances 36
2.11 Different Protocol Layer Approaches Towards Energy Efficiency 38
2.11.1 Energy Efficiency in Device Level 38
2.11.2 Energy Efficiency in BSs 40
2.11.3 Energy Efficiency in Aerial Networks 42
2.11.4 Energy Efficiency in Small Cell Network 44
2.12 Energy Efficiency at Device Level 46
2.13 Energy Efficiency at Access Level 48
2.14 Energy-efficient radio resource management in WSN 54
2.15 Fundamentals in Markov Decision Process 56
2.16 Handover 59
2.17 Related Works and Identified Knowledge Gaps 99
2.18 Overview of Massive Mimo Technology 62

2.19 MIMO (Multiple Input Multiple Output) 65

2.20 Benefits 68
2.21.1 Time Division Duplex
69
2.21.2 Linear Processing 70
2.21.3 Favourable Propagation 70

4
2.21.4 Array Size 70
2.21.5 Scalability 71
2.22 Working Principle of Massive MIMO 71
2.23 Channel Estimation 74
2.23.1 Uplink Data Transmission 75
2.23.2 Downlink Data Transmission 75
2.24 Knowledge Gap 78

CHAPTER 3
MATERIALS AND METHODS
3.2 Research Method 79
3.2.1 Method Framework 79
3.2.2 Design Method 79
3.3 Materials Requirement for the project 80
3.3.1 Base station module 80
3.3.2 Power Amplifiers 80
3.3.3 Antenna Interface: 81
3.3.4 Power supply unit 81
3.3.5 User equipment module (UE) 81
3.3.6 Set of sensor nodes 81
3.3.7 Medical Server (MS) 81
3.3.8 An electrocardiogram (ECG) 81
3.3.9 Computation and Software 82
3.2.2 Mathematical model of the Massive MIMO system 82
3.3 Channel State Information (CSI) 83
3.4 MIMO Channel Model with m Antenna at TX and n Antennas at RX 83
3.5 Mathematical Formulation of Channel Map By Vector Mapping 84
3.5.1 Single Antenna Channel Map 84
3.5.2 MIMO Channel Map 85
3.6 Implementation of Massive MIMO on MATLAB/Simulink 85

5
3.6.1 Massive MIMO using Hybrid Beamforming 85
3.6.2 Channel State Information 86
3.7 Design of the MIMO-OFDM LTE 86
3.7.1 MIMO System model 87
3.7.1.1 Convolution Coding 87
3.7.1.2 Interleaving 88
3.7.1.3 Orthogonal Frequency Division Multiplexing (OFDM) 88
3.7.1.4 Space Time diversity coding blocks 88
3.8 Real time cardiovascular (CDV) detection model using
ECG and QSR algorithm 89

3.8.1 ECG Signal Source 89

3.8.2 Modeling the ECG signal processing and filtering 90
3.8.3 Real-Time QRS Detection of CDV Signal 90
3.5 Knowledge Gap 91
3.7 Expected Conttribution to the Body of Knowledge 91

REFERENCES

6
 LIST OF TABLES

Table 2.1:Comparing traditional MIMO and Massive MIMO 68

Table 2.2:Summary of Identified Gaps 82

Table 2.3:Research Timelines 84

7
 LIST OF FIGURES

Figure 2.1: A general scheme of wireless sensor 10

Figure 2.2(a). The typical architecture of a sensor node used

in Wireless Sensor Networks (WSNs) 12

Figure 2.2(b) The typical architecture of a WSN 13

Figure 2.3: A medical sensor network application 18

Fig. 2.4: Architecture example of a healthcare system based on a WSN 27

Figure 2.5: Cellular network power consumption 41

Figure 2.6: Approaches for BS energy 42

Figure 2.7: Aerial network supporting terrestrial cellular coverage 43

Figure 2.8: A typical heterogeneous network with macro, micro,

pico and femto cells 45

Figure 2.9 States, Action and Reward of a MDP 58

Figure 2.10: Horizontal handover. (a) Hard handover; (b) Soft handover 59

Figure. 2.11 Understanding of SISO, SIMO, MISO and MIMO: 62

Figure 2.12 Multiuser MIMO System 64

Figure 2.13 Massive MIMO Technology 65

Figure 2.14 MIMO - Multiple Input Multiple Output 66

Figure 2.15: Downlink of a multi-cell massive MIMO system 66

Figure 2.16: Uplink and Downlink Operations of a massive MIMO system 72

Figure 2.17: TDD protocol of Massive MIMO transmission 74

Figure 3.1: Methodological Framework 79

Figure 3.2: WSN for cardiovascular operation 79

Figure 3.3: Spatial representation of uplink transmission matrix 80

8
Figure 3.4: Spatial representation of downlink transmission matrix 80

Figure 3.5: Typical MIMO model with M antennas at the TX and N

antennas at the RX front 81

Figure 3.6: MIMO from channel perspective 82

Figure 3.7: Schematics for channel sounding model 83

Figure 3.8: Block diagram representing MIMO OFDM model 83

Figure 3.9: Simulink implementation of OFDM system 84

Figure 3.10: STDC encoder implementation on simulink 84

Figure 3.11: STDC decoder implementation on Simulink 85

Figure 3.12: MIMO OFDM model in Matlab/Simulink 85

Figure 3.13: Normalized ECG waveform 86

Figure 3.14: ECG signal processing unit 86

Figure 3.15: Overall CDV detection system using ECG signal processing 87

Figure 3.16: Simulink implementation of ECG signal source 88

Figure 3.17: Design Resultant Simulink Block 89

9
 CHAPTER 1

INTRODUCTION

1.1 Background to the Study

A Wireless Sensor Network (WSN) is a collection of spatially disseminated sensor nodes, which

are interconnected by using wireless communication (Akyildiz et al., 2020;Kandris et al., 2020).

According to (Ye et al., undated), wireless sensor networking is evolving equipment that has an

extensive variety of prospective applications including environment monitoring, smart spaces,

medical systems and robotic exploration. They upheld that such a network normally consists of a

large number of distributed nodes that organize themselves into a multi-hop wireless network

and that each node has one or more sensors, embedded processors and low-power radios, and is

normally battery operated. Typically, these nodes coordinate to perform a common task. Vujic

(2015) added that, over the last decade Wireless Sensor Networks(WSNs) have been

successfully applied in manyengineering fields such as: structural healthmonitoring, industrial

applications, environmentalmonitoring, traffic controls, health applications,etc.

Botezatu et al. (2013) postulated that, it is already known fact that the world population is

ageing. This happens because of scientific and technological developments which increase the

life expectancy at birth (Mann, 2004) by almost 20 years (from 46.5 years in 1950-1995 to 66.0

years in 2000-2005) and second because fertility dropped almost to half, from 5.0 to 2.7 children

per women (UN, Department of Economic and Social Affairs, Population Division). According

to them, by 2050 the aged population will increase by 70 percent in developed regions and by

400 percent in less developed regions. The Economist Intelligence Unit, “Healthcare strategies

for an ageing society’’ maintained that, one of the healthcare strategies for ageing society is the

use of technology to provide care to ageing population in their own environment at home or at

10
elderly care institutions. This is possible through the use of telehealth solutions such as remote

patient monitoring or telemonitoring. Wireless monitoring devices represent an attractive option

for telehealth solutions because they do not restrict the movement of the patient (i.e. lightweight

and small form factor)(Botezatu, 2013). Wireless monitoring devices are an ideal option for e-

health solutions because of their usage convenience. Vujić (2015) maintained that, wireless

network is any type of computer network which is not connected by cables. It is a method by

which homes, telecommunications networks and business installations avoid the costly process

of introducing cables into a building,or as a connection between various equipment locations.

Wireless telecommunications networks are generally implemented and administered by using a

transmission system called radio waves. AM radio, FM radio, satellite radio, satellite TV,

satellite Internet access and broadcast TV are, in fact, wireless networks. Hence, the usage of

wireless technology is very convenient (Stankovic, 2006). WSNs consist of spatially distributed

autonomous sensors designed to monitor physical parameters or environmental conditions, such

as temperature, strain, pressure, vibration, sound, motion, pollutions, etc. Consequently, the

sensors cooperatively pass their data through the network to a main location. The base station

may communicate with the user or task manager node via Internet or Satellite. A wireless sensor,

also known as a mote (re-MOTE), smart dust, smart sensor or sensor node within the network

performs the function of sensing, data processing and wireless data transmission. It is powered

by an individual power source which often consists of a battery with a limited energy budget

(Vujić, 2015).

According to Lalitha et al. (2017), the use of WMSN in health care appliance is a potentially

very useful. Recent advances in wireless networks and electronics have led to the emergence of

Wireless Sensor networks (WSNs). Healthcare applications are measured as promising fields for

11
wireless sensor networks, where patients can be examined using wireless medical sensor

networks (WMSNs) .Wireless sensor networks have standard considerable attention from both

the academic and industry communities for many years, since these networks are the vital

component for realizing next generation networking and computing. More so, with ageing of the

people, existing medical property cannot assure future healthcare demands of seniors and

patients. Resources are limited and it is impossible for most patients to meet the expense of long-

term hospital stays due to economic limitations, work, and other reasons, even though their

health status must be examined in a real-time or short periodic time mode. As a result, wireless

monitoring medical systems will become part of mobile healthcare centers with real-time

monitoring in the future. Kollipara (2010) noted that, Wireless sensor networks consist of small,

cheap, low-powered embedded de-vices called motes or nodes. A set of nodes communicate

wirelessly with each other in a network to perform a particular task. Each node consists of a

microcontroller, a radio for wireless communication and sensors for sensing the environment. To

establish wireless communication between the nodes in the network MAC protocols are used.

Wireless sensor networks are usually used for monitoring the environment or surrounding

conditions. According to the requirement of the application each node in the network has

different types of sensors embedded on them such as temperature, humidity, light motion etc.

Vujić (2015) studied the wireless sensor networks applications in aircraft structural health

monitoring with the aim to monitor structures using embedded or attached non-destructive

evaluation sensors and to utilize the data in order to assess the state of the structure. Often

structures equipped with various types of sensors are compared to human nervous system. This

means that SHM is the imitation of the human nervous system. SHM is a new and improved way

to make a non-destructive evaluation with a minimum of manual intervention. It includes all

12
monitoring aspects which are related to damages, loadsand conditions, which have a direct

influence on the structure.Most important mechanisms which the present work sought to handle

are the energy efficiency required to proficiently sustain the network for adequate operation.

Wireless sensor network (WSN) applications involve deployment of battery powered nodes

which are active for a considerable time and they usually do not have any control by humans

after deployment. As the battery capacity of a node is limited, this draws the attention for energy

management of a node in the network. This research deals with the application of WSNs for

Cardioevascular health monitoring with the intention of modeling the energy-efficient wireless

sensor network to improve its operation for cardiovascular glitches in Nigeria.

Cardiovascular diseases (CVD) are the major cause of mortality globally. According to world

health organization (WHO) analytics nearly 32% of adult deaths all over the world are due to

cardiovascular diseases which are caused by disorders of the heart and blood vessels (Lancet,

2015).These include various heart related diseases including coronary heart disease (heart

attacks), rheumatic heart disease, raised blood pressure (hypertension), cerebrovascular disease

(stroke), peripheral artery disease, congenital heart disease and heart failure (Gogate and Bakal,

2018). These types of cardiovascular diseases need continuous monitoring of certain body

parameters which need long hospital stays. In the hospitals patients are monitored continuously

by hospital staff using various instruments like bedside monitors. These instruments are bulky

and immobile and thus keep patients stick to the bed. Their wired connections are very

uncomfortable to patients and medical staff also. Due to mounting hospital costs and shortage of

qualified healthcare professionals it is difficult to continuously monitor the essential body

parameters of the patients suffering from CVD (Gogate and Bakal, 2018).

13
According to Mukadas and Misbau (2009), cardiovascular diseases are the leading cause of

mortality and morbidity in developed countries and they are emerging as prominent public health

problems in developing countries. Boon et al. (2002) defined cardiovascular disease as any

disorder, abnormality or failure to function well, relating to the heart and blood vessels or the

circulation. Concern about increasing rates of death and disability clue to cardiovascular diseases

in non-western countries is often met with skepticism; whether they constitute a serious public

health problem is unknown (Reddy, 2004). He further stated that with justifiable alarm about the

spread of Human Immunodeficiency Virus and AIDS, and with old foes such as malaria and

tuberculosis still posing formidable challenges in many developing countries, it is understandable

that epidemics of cardiovascular disease have insidiously established themselves without

attracting global attention or local action.

World Health Organization report (2002) revealed that 80% of deaths from cardiovascular

diseases and 87% of related disability currently occur in low and middle income countries. In

this report, it was further emphasized that cardiovascular diseases have become the leading cause

of death in many developing countries and will soon attain that status in several others

(WHO.2002). The high burden of mortality from cardiovascular causes in developing countries

which is estimated at 9 million in 1990 and expected to increase to 19 million by 2020, is only

partially explained by their large populations(Murray and Lopez, 1996) . Reddy (2002) reported

that the rate of death due to cardiovascular disease among persons 15 to 59 years of age is 3 to 8

times as high in Tanzania and Nigeria as in England and Wales. It was concluded in this report

that death and disability occurring in midlife have disastrous consequences for families who lose

wage earners, and the resulting loss of productivity adversely affects national development

14
Reddy (2002). Oviasu was of the opinion that almost all unexpected deaths of medical origin in

Nigeria are due to cardiovascular causes (Oviasu, 1982).

A study conducted in South Western Nigeria by Adedoyin and Adesoye (2005) revealed that, of

the 1004 cardiovascular disease patients recorded between 1997 and 2001, those with heart

failure had the highest occurrence 384(35%) while congenital heart disease has the Slowest

occurrence 9.0 (0.8%). In that study hypertension accounted for 353(32%), stroke 192(17.4%),

cardiomyopathies 84(7.6%) and others 62(5.6%)8. The study further revealed that men

671(60.8%) were found to have higher incidence of cardiovascular disorders than women

633(39.2%) (Adedoyin R. A., Adesoye). Shakaib et al (2005) confirmed the existence of ethnic

differences in hypertension control and in cardiovascular and renal outcomes, which they

attributed to factors such as biological, cultural, social, healthcare provider and healthcare system

factors like insurance and access to care and medication.Although sharp shifts in demographic

pattern and life style have resulted from urbanization and industrialization, the globalization that

constituted the tailwind of the 20th century propelled developing countries into the world wide

epidemic of cardiovascular diseases (Mukadas and Misbau, 2009). They maintained that, out of

the 24 million people expected to die of cardiovascular diseases in 2020, about 9.3 million will

be between 30 and 69 years of age, most of them will be in the developing countries.

1.2 Statement of the problems

Wireless sensor network is an energy constraint network with the requirement of deploying

battery to power remotely positioned nodes for a given network. Since the battery capacity of

the nodes is always limited and usually, it is difficult to replace them when deployed, there is

therefore, the need for energy management of the nodes in a network. If the designed protocol is

energy efficient it will enhance the life time of a wireless sensor network especially as it

15
concerns the enhancement of Telemedicine operation in developing Nations. Research has

shown that Africa is home to less than 1 billion people, and is a major contributor to the global

burden of Cardiovascular Disease (CVD). In 2019 alone, an estimated 1 million death were

attributable to Cardiovascular Disease in Sub-Saharan African which constituted 5.5.% of all

global CVD related deaths and 11.3% of all deaths in Africa

The burden of CVD is increasing rapidly in African, most importantly hypertension, stroke,

Cardiomyopathies and coronary heart disease. Therefore the increasing in incidence cases of

CVD across Southern Nigerian call for concerted efforts in dealing with energy requirement

issues, so as to have an efficient Energy wireless sensor network with the objectives of balancing

the increasing in routing load throughput of the network, bandwidth expansion, reduction in Bit

Error Rate (BER) and to increase the lifetime of their network.

1.3 Aim and Objectives of the study

The aim of the research work is to develop an energy-efficient wireless sensor network for

improved operation of telemedicine in the developing Economy, a case study of Cardiovascular

Diseases. The following are specific objectives:

i. To evaluate different energy efficiency techniques for wireless sensor Network.

ii. To investigate the distribution and generate useful data of different cardiovascular

problems and cases within the study area.

iii. To characterize the network in order to maintain various key performance Indicators

(KPIs) for energy efficiency.

iv. Application of Massive Multiple Input Multiple Output (MIMO) Algorithm to

solve the formulated optimization problem.

16
 v. To formulate a thorough optimization problem for the proposed energy saving model.

vi. To validate the obtained results by comparing with existing models.

1.4 Justification

An excellent healthcare strategies for ageing society is the use of technology to provide care to

ageing population in their own environment at home or at elderly care institutions. Also as a

way of restricting the spread of COVID 19 pandemic, infected patients can be attended to with

the aid of wireless network sensor mechanism. This is possible through the use of telehealth

solutions such as remote patient monitoring or telemonitoring. Wireless sensor network is an

energy constraint network. The designed protocol is energy efficient that will enhance the life

time of a wireless sensor network especially as it concerns the enhancement of Telemedicine

operation in developing Nations

17
 CHAPTER 2

LITERATURE REVIEW

2.1 Wireless Sensor Networks Technology

Wireless network refers to any type of computer network which is not connected by cables. It is

a method by which homes, telecommunications networks and business installations avoid the

costly process of introducing cables into a building, or as a connection between various

equipment locations. Wireless telecommunications networks are generally implemented and

administered by using a transmission system called radio waves. AM radio, FM radio, satellite

radio, satellite TV, satellite Internet access and broadcast TV are, in fact, wireless networks.

Hence, the usage of wireless technology is very convenient (Stankovic, 2006). WSNs consist of

spatially distributed autonomous sensors designed to monitor physical parameters or

environmental conditions, such as temperature, strain, pressure, vibration, sound, motion,

pollutions, etc. Consequently, the sensors cooperatively pass their data through the network to a

main location. The base station may communicate with the user or task manager node via

Internet or Satellite. A wireless sensor, also known as a mote (re-MOTE), smart dust, smart

sensor or sensor node within the network performs the function of sensing, data processing and

wireless data transmission. It is powered by an individual power source which often consists of a

battery with a limited energy budget. The general scheme of WSNs is presented on Figure 2.1.

18
Figure 2.1: A general scheme of wireless sensor networks (Source:Vujić, 2015)

Wireless sensor technology is playing a vital role in many of the commercialized industrial

automation processes and various other real life applications (Basha et al., 2008; Ellaboudy.,

2012; Zhao2011 and Paavola, 2007). It is particularly suitable for harsh environment applications

where deploying of other network infrastructure is difficult and/or almost impossible such as in

battlefield, in hazardous chemical plant, and in high thermal environment. It is not uncommon to

see that most of the crucial surveillance and security applications also rely on sensor based

applications. Sensors which are tiny in size and cheap in cost have the capabilities to be deployed

in a range of applications as explained in (Maraiya, et al., 2011; Erol-Kantarci and Mouftah,

2011; Lu and Gungor, 2009; Buckley et al., 2006) . Essentially all sensor networks comprise

some forms of sensing mechanism to collect data from an intended physical environment either

by a time driven approach or by event triggering approach. By these approaches a sensor will

convey the sensed data to a destination or sink (multiple destinations/sinks are also possible) via

some kinds of routing algorithm such as Minimum Cost Forwarding Algorithm (MCFA),

Directed Diffusion Routing Protocol (DDRP), or one of the clusterbased routing protocols. Being

very small in size, sensor nodes are built with limited computational capacity, small storage

memory, and finite battery power capacity (Yang, 2014). The structure of a typical WSN node

(Al-Karaki and Kamal, 2004) consists of four main components: a sensing element normally

19
used for sensing a physically measureable parameter; an Analogto- Digital Converter (ADC),

used for converting analog signals to some digital formats; a processing unit, providing

simple/basic data processing and computation capabilities; and a power unit, responsible for

sensor node’s operation life span. It is a known fact that WSN is a resource constrained network

in which energy efficiency is always the main issue since the operation of WSN depends heavily

on the life span of the sensor nodes’ battery (Kaur and Singh, 2002). The most energy consuming

operation in WSN is the data packet routing activity.The characteristics of theWSN are different

from the conventional networks (Akyildiz et al., 2002; Chu et al., 2002). These unique

characteristics are often taken into account for addressing the issues and challenges related to

network coverage, runtime topologiesmanagement, node distribution, node administration, node

mobility energy efficiency/consumption, network deployment, application areas/environment,

and so forth (Singh et al., 2010; kalpana and Bhuwaneswari, 2011; Cao et al., 2007). Nodes in a

WSN are generally energy, computation, and memory constrained. Consequently, there is a need

for research and development into low-computation resourceaware algorithms for WSNs,

targeting at small, highly resource constrained embedded sensor nodes. Energy consumption is

of prime importance in WSNs and thus some algorithms (Singh and Mittal, 2013; Manda et al.,

2013; Abdulla et al., 2012; Bahi, 2014; Ebrahimi and Assi, 2014; Chang and Ju, 2014; Bhadoria

and Chandra, 2012) and hardware were designed with energy efficiency or energy awareness as

a central focal point of interest. Enhancing energy efficiency of WSN with respect to the

communication routing protocol is the primary concern of this research. We propose a new

routing protocol entitled “Position Responsive Routing Protocol (PRRP)” and compare its

performance with the well-known LEACH and CELRP protocols. The simulation results show a

20
significant improvement over the aforementioned protocols in terms of energy efficiency and the

overall performance of the WSN.

2.2 Potential impact and benefits of WSN

A Wireless Sensor Network (WSN) is a group of spatially dispersed sensor nodes, which

areinterconnected by using wireless communication (Akyildiz et al., 2002). As seen in Figure 1,

a sensor node, also called mote, is an electronic device which consists of a processor along with

a storage unit, a transceiver module, a single sensor or multiple sensors, along with an analog-to-

digital converter (ADC), and a power source, which normally is a battery. It may optionally

include a positioning unit and/or a mobilization unit. A sensor node uses its sensor(s) in order to

measure the fluctuation of current conditions in its adjacent environment. These measurements

are converted, via the ADC unit, into relative electric signals which are processed via the node’s

processor. Via its transceiver, the node can wirelessly transmit the data produced by its processor

to other nodes or/and to a selected sink point, referred to as the Base Station. As illustrated in

Figure 2.2, the Base Station, by using the data transmitted to itself, is able to both perform

supervisory control over the WSN it belongs to and transmit the related information to human

users or/and other networks (Yick , et al., 2008).

Figure 2.2 (a).The typical architecture of a sensor node used in Wireless Sensor
Networks(WSNs) (Source: Kandriset al., 2020).

21
Figure 2.2 (b).The typical architecture of a WSN (Source: Kandriset al., 2020).

The collaborative use of a sufficient quantity of such sensor nodes enables a WSN to perform

simultaneous data acquirement of ambient information at several points of interest positioned

over wide areas. The inexpensive production of sensor nodes of this kind, which despite their

relatively small size, have exceptionally advanced sensing, processing, and communication

abilities has become feasible due to continuous technological advances. For this reason, although

WSNs were initially used mainly for military purposes, nowadays they support an ever-growing

range of applications of different types (Arampatzis et al., 2005).

According to Freitas and Azevedo (2016), the elderly population are growing and generally

getting older. Life expectancy continues to increase with the new advances in healthcare. Today,

over 850 million people worldwide suffer from chronic diseases and spend up to 85% of their

savings in healthcare plans. It is always better to prevent a disease than to treat it, so that

individual monitoring is required as a periodic activity. The growing population in developed

countries leads governments to increase their healthcare budgets, and also presents new

challenges to health systems, especially with older people living either on their own or in old age

homes. (Minaie et al., 2013)

22
Due to the increase of research in the area of wireless sensor networks (WSN), these have

brought new opportunity in medical devices. The WSN are seen as one of the most important

advances in technology of this century and hence its importance in areas as diverse as health,

psychology, fire prevention, security and even in the military. The advantage of this technology

is the ability to track, monitor, study, understand and act on a particular phenomenon or events

(Akyildiz et al., 2002).

Traditionally, healthcare monitoring is performed on a periodic check basis where patients are

constantly updated on their symptoms; the physician checks and makes a”diagnosis, then when

possible monitors the patient’s progress during treatment. In most cases, health monitoring is

done by wireless network infrastructures. But the coverage of these network infrastructures has

limitations from bandwidth. These limitations in continued health surveillance services, it is not

always possible to send emergency signals from patients to healthcare workers. With WSN,

patients can get continuous health monitoring using wireless ad hoc networks which can transmit

vital signs over shorts distances. In most systems, the health data of multiple patients may be

resent using the wireless multi-jump routing scheme for a base station (Aminian and Naji, 2013).

At present telemedicine systems can establish news protocols such as IEEE standards 802.11,

802.15 and 802.16, which are characterized by the distribution networks for medical information

providing life-saving services (Lozano et al., 2011). Wireless sensors can be placed on patients

in a hospital or homecare setting to gather physiological signals. The correct WSN design

depends on accurate traffic models, the selection of the correct model is essential for the correct

management of network traffic, network congestion, interference between nodes and the energy

expended by each node. Currently, there are no traffic models that represent medical WSN

applications (Messier and Finvers, 2007).

23
With ageing of the people, existing medical property cannot assure future healthcare demands of

seniors and patients. Resources are limited and it is impossible for most patients to meet the

expense of long-term hospital stays due to economic limitations, work, and other reasons, even

though their health status must be examined in a real-time or short periodic time mode. As a

result, wireless monitoring medical systems will become part of mobile healthcare centers with

real-time monitoring in the future.

“WSNs are collected of individual embedded systems that are capable of” Interacting with their

environment through various sensors, Processing information locally and Communicating this

information wirelessly with their neighbors.

A sensor node (embedded system) usually consists of three components which are

i. Wireless modules or motes – key components of the network which consists of a

Microcontroller, transceiver, power source, memory unit, and may contain few

sensors. Examples: Mica2, Cricket, MicaZ, Iris, Telos, SunSPOT, and Imote2.

ii. A sensor board which is mounted on the mote and is embedded with multiple types

of sensors. Examples: MTS300/400 and MDA100/300.

iii. A programming board (gateway board) – provides multiple interfaces including

Ethernet, WiFi, USB, or serial ports for connecting different motes to an enterprise or

industrial network or locally to a PC/laptop. These boards are used to program the

motes or gather data from them. Example: M1B510, M1B520, and M1B600.

24
2.3 Applications of Wireless Sensor Networks

According to (Kapoor, 2013) Sensor networks may comprise of different types of sensors:

seismic, low magnetic rate, thermal, visual, infrared, and radar. With such a large variety in the

types of sensors, several kinds of ambient conditions that include the following may be

monitored (Estrin et al., 1999)

• Temperature

• Humidity

• Movement of Vehicles

• Pressure

• Soil Makeup

• Noise Levels

• Lightning Conditions

• Speed, and direction of movement of an object

. Stress level on attached objects

• Presence, or absence of certain kinds of objects

Various applications of sensor networks have been categorized and are discussed in the

following subsections (Akyildiz et al., 2002)

2.3.1 Military Applications

Sensor networks have been used in the military for the purpose of monitoringfriendly forces,

equipment and ammunition, battlefield surveillance, reconnaissance of opposing forces and

terrain, targeting, battle damage assessment; and nuclear, biological and chemical (NBC) attack

detection and reconnaissance (Akyildiz et al., 2002)

25
2.3.2 Environmental Applications

One of the most important applications of wireless sensor networks in the domainof

environmental applications is forest fire detection. Sensors may be deployed densely in a forest

region and the exact origin of fire may be relayed before the fire becomes uncontrollable.

Wireless sensor networks also have numerous applications in the field of agriculture. They may

be used to monitor conditions that may affect crops and livestock. They may be used for

detection of levels of various chemicals in soil. Movement of birds, animals, and insects may be

monitored with the help of sensor networks as well. Sensor networks may also be used for flood

detection (Bonnet et al., 2000; Kapoor, 2013).

2.3.3 Health or Medical sensor network Application

Sensor networks have numerous applications in the area of health monitoring. Sensor networks

may be used for telemonitoring of physiological data (Ogawa et al., 1998) and can also be

deployed to detect the behaviour of elderly people (Celler 1994). With the help of sensors,

doctors may be able to identify predefined symptoms earlier (Nam et al., 1998; Ko et al., 2010).

Also, due to remote monitoring via a wireless sensor network, the patients have an option to stay

at home rather than at a treatment centre (Bauer et al., 2000). Sensors may also be used for

monitoring of patients and tracking of doctors inside a hospital. Sensors are also finding use in

drug administration in hospitals for minimizing the prescribing of wrong medicines to patients

(Sibbald, 2001).

26
Sensor networks have been applying in various aspects of medical care. By equipping patients

with tiny, wearable vital sign sensors, physiological status of patients can be obtained easily. In

emergency or disaster scenario, sensor networks can be used to track healthcare personnel and

patient status as well as location continuously in real-time mode. Figure 2.3 illustrates a medical

sensor network application.

Figure 2.3: A medical sensor network application (source: Lalitha et al., 2017)

Cellular systems (2.5G, 3G and beyond 3G) have the potential to greatly improve telemedicine

services by extending the range of healthcare system, improve the flexibility and heterogeneous

network with an end-to-end telemedicine framework. The system consists of a cellular network

platform, which gathers the information from wearable sensors, monitoring devices and server

platform, which receives, stores, processes collected patients’ vital data and forwards them to the

existing information systems.

27
2.3.4 Home Applications

With the advancement of sensor technology, a number of applications have been proposed for

home. It is proposed that sensors and actuators may be buried in home devices like refrigerator,

and microwave oven etc. The sensors embedded in these devices can interact with each other and

also with the outside world via the internet or satellite.The owner of the devices will then be able

to operate these devices remotely (Petriu et al., 2000). Another scenario of creating a smart

environment is described in (Herring and Kaplan, 2000). This work describes a scenario of

building a smart information environment center building in a campus at the beginning of a

semester.

2.3.5 Other Commercial Applications

Some other commercial applications of sensors include environmental control of office buildings

(Rabaey et al., 2000), interactive museums, detecting car thefts, inventory control, and vehicle

tracking and detection (Shih et al., 2001).

2.4 Energy Consumption Challenges in Wireless Sensor Networks

The behaviour of WSNs is greatly affected by the deployment environment in wireless

communication is unpredictable in different environment. Sensors are usually deployed densely

in wireless sensor network. Due to this dense environment, it can suffer significant interference

which greatly impairs network performance (Varun and Gangwar 2019).

In wireless sensor networks energy consumption is one of the important issues not only because

of battery operated sensor nodes but also due to its significant impact on the idea of green

computing. In wireless sensor network, Clustering approach plays important role. Clustering

28
approach increases network life time, improved bandwidth utilization and also reduces wasteful

energy consumption thereby reducing overhead. However clustering approach has certain

limitations as follows:

2.4.1 Throughput

To ensure the stability in the wireless sensor networks certain level of throughput is required to

fulfill the quality of services to the end user of network. In WSN interfering problems occurs

when the channel sharing for higher data transmission. Besides other issues of WSN effects of

interferences is also big important during simultaneous transmission of data in order to enhance

the WSN capacity. In such scenario high throughput and low delay is difficult to achieve.

Throughput of wireless sensor networks is affected by various factors such as collision

avoidance, control overhead, channel utilization and latency. Throughput maximization is a

fundamental problem in WSNs. Therefore, we need an efficient coordination between

throughput and power consumption (Cached et al., 2013).

2.4.2 Energy Efficient Design

In WSNs, energy consumption and prolonging life time of the network are two critical issues.

WSNs nodes are low powered battery device, replacement of battery or recharge of battery is

very difficult task in hostile environment. The components of senor node consume a large

amount of energy either in active mode or idle mode. Therefore there is a need of power

management scheme to save the energy in idle mode by switching off the components that are

not participated for a particular instance of time (Aanchal et al., 2013).

29
2.4.3 Energy saving in interference environment

The behaviour of WSNs is greatly affected by the deployment environment in wireless

communication is unpredictable in different environment. Sensors are usually deployed densely

in wireless sensor network. Due to this dense environment, it can suffer significant interference

which greatly impairs network performance. Therefore, to discover different technique for

reducing power consumption in the presence of interference and shadowing environments are

also very important (Zhong and Zhang, 2012)

2.4.4 Message Delivery

The role of sensor nodes in WSN is sensing environment and delivering data to the base station.

Since there are various sensing nodes in WSN to pass the data to the destination node, fairness is

an important issues for researcher. Multi hop routing get worse the packet loss in WSNs, node

near the destination have higher packet delivery. If sensor node has packet to send, it must be

able to deliver the data at destination node. Poor data delivery performance may degrade

performance of data transport and expand energy consumption. Therefore, the delivery ratio

should be high (Zhong and Zhang, 2012).

2.4.5 Network Lifetime

Effective clustering approach reduces energy level in intra-cluster and inter-cluster

communication due to this increases network lifetime. The energy consumption in wireless

sensor networks is still challenge in industrial and research field (Junchao et al., 2009).

30
2.4.6 Limited Energy

Nodes of wireless sensor networks have low powered battery and very small in size so that

sensor nodes has limited energy storage for operating in network. So there is a need of an

efficient approach for utilization of this limited energy. A proper clustering scheme can reduced

overall energy consumed in the network (Wang and Li, 2011).

2.4.7 Scalability

In WSN most of the sensor nodes deployed due to infrastructure less property. The node of

sensor network has limited coverage range. For such scenarios a capable routing protocols are

needed for handling a vast amount of sensor nodes. WSN consists of collection of large number

of small nodes; it is not easy to preserve the global information of network for each node in

sensor network (Manjeshwar and Agrawal, 2000).

2.4.8 Data aggregation

Data aggregation is a technique for eliminating the redundant data transmission in WSNs. Data

aggregation is the fundamental procedures for saving the energy. Data aggregation is technique

to gather and aggregate data so that network lifetime is enhanced. Most of times each sensor

nodes duplicates sensed data to its sink node called base station lead redundancy at base station

((Yuet al., 2012; Zang et al., 2014).

31
2.5 Energy Consumption of Sensor Nodes

2.5.1 Power Consumption

The main consumers of energy are the controller,the radio front ends,to some degree the

memory,and depending on the type, the sensors.To use such a battery to power a node even only

a single day,the node must not consume continuously more than 1/(24.60.60) Ws/s≈11.5μw.No

current controller,let alone an entire node,is able to work at such low-power levels.One important

contribution to reduce power consumption of these components comes from chip-level and lower

technologies:Designing low-power chips is the best starting point for an energy-efficient sensor

node.Introducing and using multiple states of operation with reduced energy consumption in

return for reduced functionality is the core technique for energy-efficient wireless sensor node.

At time t1,the decision whether or not a component is to be put into sleep mode should be taken

to reduce power consumption from Pactive to Psleep.If it remains active and the next event occurs at

time tevent,then a total energy of Eactive=Pactive(tevent-t1) has be spent uselessly idling.Putting the

component into sleep mode,on the other hand,requires a time τ down until sleep mode has been

reached:as a simplification,assume that the average power consumption during this phase is

(Pactive+ Psleep)/2.Then,Psleep is consumed until tevent.In total,τdown(Pactive+ Psleep)/2+( tevent-t1- τdown)Psleep

energy is required in sleep mode as opposed to (tevent-t1) Pactive when remaining active.The

energy saving is thus

Esaved=((tevent-t1) Pactive- τdown(Pactive+ Psleep)/2+( tevent-t1- τdown)Psleep) (2.1)

Once the event to be processed occurs,howeverman additional overhead of

Eoverhead = τup ((Pactive+ Psleep)/2 (2.2)

32
is incurred to come back to operational state before the event can be processed,again making a

simplifying assumption about average power consumption during makeup.This energy is indeed

an overhead since no useful activity can be undertaken during this time.Clearly,switching to a

sleep mode is only beneficial if E overhead< Esaved,or,equivalently,if the time to the next event is

sufficiently large:

(tevent-t1)>1/2 (τdown+(Pactive+ Psleep) τup (2.3)

Pactive+ Psleep

2.5.2 Energy Consumption during transmission

The energy consumed by a transmitter is due to two sources: one part is due to RF signal

generation, which mostly depends on chosen modulation and target distance and hence on the

transmission power Ptx, that is, the power radiated by the antenna. A second part is due to

electronic components necessary for frequency synthesis, frequency conversion, filters, and so

on. The transmitted power is generated by the amplifier of a transmitter, ITS own power

consumption Pamp depends on its architecture, but for most of them, their consumed power

depends on the power they are to generate. A more realistic model assumes that a certain

constant power level is always required irrespective of radiated power, plus a proportional offset:

Pamp = αamp + βamp Ptx (2.4)

Where αamp and βamp are constants depending on process technology and amplifier

architecture.

The efficiency of the power amplifier ηpa = Ptx/Pamp =1Mw/174Mw+5.0.1mW≈0.55%

2.6 Energy Efficiency

33
Energy is a precious resource in wireless sensor networks and that energy efficiency should

therefore make an evident optimization goal.It is clear that with an arbitrary amount of energy,

most of the QOS metrics defined above can be increased.The most commonly considered aspects

are:

 Energy per correctly receive bit

 Energy per reported event

 Delay/energy trade offs

 Network Lifetime

 Network half-life

 Time to partition

 Time to first node death

 Time to loss of coverage

 Time to failure of first event notification

2.7 Wireless Biomedical Sensor Networks

Advances in technology have allowed the integration and miniaturization of sensors, embedded

microcontrollers and radio interfaces in a single microcontroller; WSN and micro fabrication

have allowed a new generation of WSN suitable for many applications. One of the most exciting

and important areas of application are in health supervision. WBSN refers to the wireless body

area network (WBAN) and it uses a wireless architecture that consists of a number of body

sensor units (BSUs) combined with a single body central unit (BCU). This network is made up

of wearable computing devices which are in development. The WBAN is an on-body sensor

34
having little power to include wireless devices for easy isolated control. In real time, the situation

of several patients is being monitored constantly in this type of architecture. So it can monitor

physiological parameters. However, this type of application has faced several challenges when

designed. For example, communications between the sensors need to be reliable and free of

interference and it should also provide flexibility to the users. The growth of WBAN must

include the advance of diagnostic tools of the medical monitoring system (López, 2011). A

variety of physiological sensors that monitor vital signs, environmental sensors (temperature,

humidity and light), and a location sensor may be integrated in a WWBAN.

The WWBAN is constituted by inexpensive, lightweight and miniature sensors, and it may allow

a long-term, discrete and ambulatory health monitoring with instantaneous feedback to the user

about the current state of health in real-time and the user's medical records. Such a system can be

used to supervise the rehabilitation by computers in multiple conditions, and even early detection

of medical conditions. For example, intelligent heart monitors can alert patients about their

medical conditions or provide information to a specialized service in the event of catastrophic

events. When integrated in a broader telemedicine system with the patient’s medical records,

WWBAN shows a revolution in medical research by sending all collected data. The large

amount of collected physiological data will allow quantitative analysis of various conditions and

patterns(Liu, 2015). Recent advances in miniaturization of intelligent biosensors opened new

opportunities for the continuous monitoring of patients (Neves et al., 2011). Unobtrusive tiny

wearable sensors will collect large amounts of data automatically, reducing the cost and

inconvenience of regular visits to the physician (Neves et al., 2011)

35
2.8 The comparison of technologies used in WSN in Medical Applications

In this area the WSN has a wide application and can be used in almost all applications, and all

technologies used in WSN can be applied to health, including sensors and positioning systems

such as GPS (Lozano et al., 2011). For example, the wireless data communication is a

bidirectional radio frequency communication with ad-hoc routing, which allows each patient's

node to send the data to a base station, even if they are not within its direct radio range (Walker,

et al., 2007). As shown in the figure 2.4.

Fig. 2.4: Architecture example of a healthcare system based on a WSN (Walker, 2007).

Aminian and Naji (2013) designed a prototype of a ubiquitous health system for hospitals, which

is the concept of ubiquitous, placing wireless sensors unobtrusively in a person's body to form a

wireless network that can communicate the state of health of the patient with the base station

connected to the PC monitor. However, the relay nodes in the middle do not require use of a high

frequency band and can transmit the data over a short range frequency module (RFM). But it is

important to note that, there are medical bands such as MICS (Medical Implant Communication

36
Service), 10 meters of 402-405 MHz frequency band coverage that can be used for body area

network applications because of their low power transmissions (Aminian and Naji, 2013).

Tolentino et al. (2011) presented an architecture of a health system ubiquitous for continuous

monitoring of patients in their natural physiological conditions or elderly patients with chronic

diseases, the biggest difference it is designed to monitor the elderly who live in remote areas or

in small nursing home without enough health technical support, instead of monitoring patients in

a large hospital environment. This WSN ad hoc is used integrated with existing medical

practices and technologies in real time remote monitoring to provide medication and a patient

status monitoring system assisted by incorporated the wireless sensor. The system transfers the

data wirelessly to a base station connected to a server on an ad hoc network using IEEE 802.15.4

or LR-WPANs (Gaddam et al., 2011). Sheltami and his colleagues developed a telemetry project

with the definition of alerts priorities depending on the emergency. In this WSN, profiles of

patients are updated with the information processed in the central database. The central computer

is responsible for sending e-mails and/or messages in case of emergencies (Tolentino et al,.

2011; Sheltami et al., 2006). Mbakop and his team developed a monitoring of patients in real-

time systems, which is made up by a system of two nodes where vital signs are collected and

transmitted wirelessly to a base station and then the data can be stored and presented on a

continuous base station. The great innovation in this design is the correct operation, but the

biggest challenge is its’ actual use in the future of the movements of the patient to produce

energy for sensors.Sheltami et al. (2006) developed and implemented a SMART, it is based on

Selective Activity Monitoring (SAM) and it integrates various intelligent sensors which

communicate via the standard radio protocols. The system depends on a number of selected set

of wireless intelligent sensors and a controller. This controller receives and processes inputs from

37
the sensors. The selected sensors are installed as monitor in televisions, reading lamps and beds,

in different locations in a house (Chung, 2009).

Walker and his team showed the need to create a single platform, but robust enough to support a

number of complex applications. The fundamental technical contributions of the proposal and

the improvements platform include: 1) the application of JADE (Java Agent-based Development

framework) for WSN environment, 2) the abstraction of sensors and types of events, 3) a user

interface for application development, and 4) Architecture used to monitor the application states.

They monitored patient's behaviour and health, when using a health professional system or a

relative in question would be able to plan the application to monitor the patient's routine, and

define actions based on detection of significant events with sensors in the elderly’s house. (Khan,

et al., 2009).

The WearIT@work project was established by the European Commission as an integrated

project to investigate "Wearable Computing" as a technology dealing with integrated computer

systems in clothing (wearIT@work). One possible application of this project is the rapid

availability of patient’s medical information at any time; this could mean an interesting reduction

in costs in medical examinations, also the power to perform medical evaluations on a daily basis

and according to the circumstances of patients and in extreme cases, it could save patient’s lives.

(Lozano et al., 2011).

Alarm-Net presents preventive healthcare an adaptable solution for continuous exams over

wireless sensors for intelligent medical care, creating a historical record and preserving the

privacy of the patient. These sensors can detect small changes but key signals that humans

cannot see and alert to like heart rate, oxygen levels in the blood and the circadian rhythm which

38
may indicate changes in health requirements (López, 2011). Alarm-Net accomplishes these goals

through the unification and aggregation of heterogeneous devices in a common architecture

(Leroux, 2013).[CodeBlue (developed by Harvard University) is a wireless infrastructure

intended to provide a common protocol and software structure in a disaster scenario response,

allowing monitoring and wireless monitoring of patients and rescuers. The CodeBlue is a self-

organized platform that is easy to connect because of its ad hoc architecture and it integrates

different nodes of wireless device sensors.

The system integrates low-power wireless sensors, and it offers services for the establishment of

credentials, handoff, location tracking, and network’s filtering and aggregating of produced data

by the sensor. The simple interface allows emergency medical technicians to request data from a

group of patients. The CodeBlue is designed to pass through a wide range of density networks

and operates in a range of wireless devices, from the limited resources to the PDA and more

powerful PCs. The CodeBlue has several types of sensors (oximetry, ECG and motion sensor)

and it is used together with the ZigBee trading platforms, Mica2, MICAz and Telos. Researchers

consider that such platforms have a good response in research settings but have many failings in

actual scenarios due to the dimensions of the modules and batteries; a support platform has also

been developed for lighter sensors to be used in accident monitoring modules in a non-invasive

way(Neves et al., 2008)

2.9 Cardiovascular Applications (Telemedicine in Stroke)

2.9.1 The burden of the disease

Stroke is a high frequency neurological disorder and the most common cause of complex

disability in adults (Adamson et al. 2004). It is the second and third most common cause of

39
mortality in the developing and developed worlds respectively (Lyons & Rudd 2007). In the

United States, 780,000 people per year experience either a new or a recurrent stroke. In 2004,

stroke mortality was estimated to be in excess of 150,000 and the prevalence of stroke in people

over the age of 20 was 5.8 million in 2005. About 30% of stroke survivors are permanently

disabled and about 20% require institutional placement at three months (Rosamond et al. 2008).

Stroke is an example of a largely preventable disease that presents acutely, with a short time

window for amelioration. It is associated with a high mortality rate, a significant risk for residual

disability, and has a large impact on society, the patients and their families (Demaerschalk et al.

2010). Stroke recurrence can lead to a stepwise decline into dependency, resulting in a financial

burden on society. According to Rosamond et al (2008), the indirect and direct costs of stroke in

the United States were calculated at $65.5 billion in 2008. The majority of strokes are due to

cerebral infarction (87%) and as such are amenable to a variety of pre-stroke risk factor

modification strategies, as well as thrombolysis or intravascular clot retrieval strategies during

the acute phase (Hachinski 2002; Yusuf 2002; Rosamond et al. 2008). The remaining subtype of

stroke, i.e. intracerebral haemorrhage, is largely preventable by pre-stroke blood pressure control

(Hachinski et al. 2010).

2.9.2 The rural challenge

Globally, the majority of strokes occur in rural areas where there is often a lack of stroke

services. In these areas, stroke care is often fragmented and does not adhere to recommended

guidelines (Hess et al. 2005; Joubert et al. 2008). This, together with the world-wide focus on

provision of health services, the geographical barriers that are associated with a general

attenuation of access to healthcare resources and the paucity of stroke experts results in an

40
inequitable distribution of resources, which frequently limits access to evidence–based care (Park

& Schwamm 2008). Service delivery is frequently variable even within developed countries, but

even more so in developing countries. In rural Australia, for example, over 90% of hospitals

have 24-hour access to CT scanning, but residents have to travel, on average, about 100km. A

study carried out in Montana and Northern Wyoming by Okon et al (2006) revealed that only

39% of hospitals had 24-hour CT capabilities. A study in China by Liu et al (2007) showed the

use of CT in rural areas to be “low” compared to urban areas, but exact data are not available. In

India, there are no reliable data on rural Indian CT services. In a study of stroke services in 21

rural hospitals in Idaho by Gebhardt et al., (2006), 77.8% reported patient delays and 66.7%

reported transport delays. There were equipment delays in 22.2% and ancillary service delays in

61.1%. Only 33.3% of hospitals were involved in quality improvement programmes,

thrombolytic therapy was available for stroke in only 55.6% and no hospital had a designated

stroke team. In Scotland, although it was revealed that the admission rate for symptomatic

carotid disease was significantly higher in deprived rural populations, less carotid

endarterectomies were performed in the rural compared to the urban areas. An assessment by

Read et al (2005) of the differences in stroke care practices between regional and metropolitan

hospitals in Australia showed that rural patients were less likely than their urban counterparts to

receive CT of the head within 24 hours of admission, swallowing assessment, echocardiography,

carotid imaging, lipid or glucose estimations or services from allied health professionals.

Furthermore, no rural hospital in New South Wales had a stroke-specific clinical nurse compared

with 21 stroke nurse case manager positions in metropolitan New South Wales, and only one

third of the rural hospitals had access to a neurologist.

41
2.9.3 Telemedicine/Telestroke

Telemedicine has been defined as the delivery of healthcare services to the underserved,

employing telecommunication (Misra et al. 2005). A more extensive definition is “the process by

which electronic, visual and audio communications (including the telephone) are used to provide

diagnostic and consultation support to practitioners at distant sites, assist in or directly deliver

medical care to patients at distant sites, and enhance the skills and knowledge of distant medical

care providers (Deshpande et al. 2008). “Telestroke” refers to the application of telemedicine to

stroke care. It is a new application of existing technology in the care of stroke patients (Sato &

Ohta 1993). Demonstration projects have proven the feasibility of telestroke (Goldstein &

Rothwell 2007) and suggested its potential to facilitate access to specialist stroke expertise in

hospitals without access to specialist clinicians. Importantly, use of this technology may promote

implementation of best-practice management of vascular risk factors in the stroke survivor after

discharge (Bouffard 1997; Susman 1997; Park & Schwamm 2008). The main drivers have been

technological advances, such as the digitisation and compression of data permitting the rapid

transfer of images (Levine & Gorman 1999).

2.9.4 Current and potential uses of telestroke

Telemedicine, as a distant communication tool, was first attempted in radiology 50 years ago

(Jutras 1959) and subsequently in psychiatry (Wittson et al. 1961). Since 1999 there has been a

gradual increase in telemedicine programmes and, more recently, a growing interest in its use in

stroke, mainly in facilitating thrombolysis, (Wang 2003; Audebert 2006; Park & Schwamm

2008), but also in establishing diagnoses and guiding treatment options (Wiborg & Widder

2003). Telestroke has the potential to improve the care of the stroke patient in the acute phase,

the subacute phase, the rehabilitation phase and in the long term for the prevention of recurrence

42
of cerebrovascular or cardiovascular events. Attention at all these levels can substantially lower

the net cost of the condition to society by reduction of lost productivity, nursing home costs and

rehabilitation (Hachinski et al. 2010). So, although in the past telemedicine has mainly focused

on the area of thrombolysis in the acute stage of the disease, it has now been acknowledged to

have the potential to also bring substantial benefits to the remaining stages of the stroke victim’s

journey.

2.9.5Technology options for telestroke

A range of generic and personalised technology options have been used in different studies,

including telephone, facsimile, email, videoconferencing, and internet-based communication.

i. Use of telephone

Simple telephone contact has been shown to be useful in different situations, such as giving

practical problem-solving advice to caregivers. Moreover, the contact was useful in reducing

stress in a study by Grant et al (2002). Telephone interviews have been shown to be reliable in

the application of a variety of assessment and measuring tools, such as the Stroke Impact Scale

(SIS) (Kwon et al. 2006), and proved reliable for evaluation of function, disability and cognitive

function in community outpatients (Meschia et al. 2004; Merino et al. 2005). Telephone

administration of the Patient Health Questionnaire (PHQ9) has been validated in stroke patients

(Lee et al. 2007). This instrument has demonstrated reliability as a screening tool for Post Stroke

Depression (Williams et al. 2005). There is a need to determine the best modality for achieving

the requisite goal in stroke survivors. These goals may be risk factor modification, patient and

career education, detection and management of post-stroke depression, career support, strategy

implementation, rehabilitation or simple surveillance of health service usage. Different

43
modalities may be appropriate for different goals in that simple telephone calls may suffice for

patient and career education or support (Grant et al. 2002; Buckley et al. 2004), whereas

videoconferencing would be required for a telerehabilitation programme (Hersh et al. 2006), or a

psychiatric intervention for depression (Doze et al. 1999).

ii. Use of Videoconferencing

Videoconferencing enables the patient, the caregivers, and the local and distant physiciansto

interact visually and audibly. However, minimum specifications need to be determinedon the

technical aspects including the degree of resolution of the computer screens andbandwidth to

transmit videoconferencing and images. Some studies have also indicated thatsound quality

needs to be improved (Audebert et al. 2005; Meyer et al. 2008). In order to address these

technical issues, the following requirements have been proposed: a remote-control camera with

zoom, tilt and rotation functions at the distant or “spoke” hospital, and large monitors with high

resolution are needed in both the “hub” and “spoke” hospitals. To maintain sufficient visual

quality (25–30 frames per second), a bandwidth of at least 300 kilobits per second is required

(Audebert 2006). A controlled trial, allocating hospital-based or mobile teleconsulting and

evaluating technical parameters, acceptability,and the impact on immediate clinical decisions

showed that critical treatment decisions can be made on the basis of laptop-based telestroke

consultations using the available European mobile network technology. However, although the

technical quality was sufficient to make relevant immediate clinical decisions, the quality of the

video examination was considered inferior to hospital-based consultations and there were critical

comments regarding the lack of a video stream on the spoke side (Audebert et al. 2008).

iii. Innovative Technologies

44
The literature reveals an interest in experimenting with new information and communication

technologies from videoconferencing, Internet web-cams, to virtual reality haptic workbenches.

Indeed, technological advances and developments in communication methods have led to new

ventures within the field of telemedicine, and more specifically, telestroke. For example,

CovotemTM Video Solutions is a telestroke tool that has been developed for emergency stroke

treatment and secondary prevention. It comprises a software platform, HD videoconference and a

high quality sound system. CovotemTM enables the neurologist to remotely manipulate a

motorised camera in order to dynamically visualise the patient. A graphical editor enables

medical imaging to be shared in real-time in DICOM format so that the type of stroke can be

diagnosed and thrombolysis administered if necessary. Patient data can also be shared via an

electronic file.This is just one example of the way in which information technology has taken

giant strides in developing effective and efficient tools for delivering health services to widely

dispersed populations.

2.10 Barriers to the implementation of telestroke models

2.10.1 Technical advances

One of the major barriers to the success of telestroke is that, despite the numerous advantages of

recent technological advances, technical problems arise with telemedicine technology, including

non-connecting or malfunctioning devices. This problem must be eradicated since it can lead to

distrust by users and low levels of satisfaction, which can be further aggravated in cases of lack

of interoperability due to fears of rapid obsolescence and wasted capital. In terms of

telemedicine-directed stroke care, there have been 3 different methods used for interaction: (1)

telephone service; (2) HQ-VTC with an on-call stroke team using an Internetbased wireless or

high-speed landline connection; and (3) a combination of telephone and video methods. Each of

45
these methods has strengths and weaknesses, and several trials are seeking to determine whether

videoconferencing is superior to exclusively telephonic interaction (Schwamm et al. 2009).

Despite innovative technological solutions making medical imaging available for simultaneous

viewing by spoke hospital personnel and the telestroke consultant, which reduces the number of

individuals required to make definitive recommendations regarding thrombolytic therapy or

other time-critical interventions, many rural areas (as well as some urban and suburban areas) do

not have access to consistent low-latency, high-speed bandwidth sufficient to support reliable,

high quality video transmission and reception over open, standards-compliant networks.

The presence of essential infrastructure (telephone lines, wireless broadband) must be assessed

for hospitals participating in the exchange of telemedicine data as part of an SSCM

implementation. Effective information technology systems and supporting infrastructure need to

be put in place to initiate a telemedicine programme for stroke treatment. For example, the mode

of data transmission must provide adequate bandwidth to transmit large amounts of data quickly,

accurately, and securely. The fact that the spectrum of IT options is expanding rapidly, whilst

costs are decreasing, is perhaps more important. However, deployment of these options is not

uniformly available across all geographical and demographic users. Fibre-optic cable is not yet

as ubiquitous and high bandwidth mobile phone networks (such as G3) do not have the same

coverage as GSM (Webb and Williams, 2006). More importantly, money is often spent on

technical equipment but not sufficiently on the personal resources needed in telemedicine

services.

2.11 Different Protocol Layer Approaches towards Energy Efficiency

According to Islam (2017)Wireless networks are bounded by limited resources like: bandwidth,

power, time, complexity, battery life, energy, capacity etc. Therefore efficient utilization of

46
resources play very important role in designing efficient networks. In fact, the main objective of

an efficient network design is to optimize the system performance in terms of:

∙ Energy consumption minimization

∙ QoS provision

∙ Mobility management

∙ Network access delay

∙ Security awareness.

There are many cases where a trade-off exists between some of the above mentioned criteria,for

example energy-QoS trade-off, energy-delay etc. Energy efficiency of a wireless network can be

reached over different protocol layers since different protocols exploit the source of energy

consumption in different ways, such as by power amplifiers, mixers, processors, registers, filters

etc. Many pioneering works have proposed many energy efficient approaches for wireless

communication considering both active mode and idle mode. Following a hybrid protocol

architecture based on the Internet and the IEEE 802 architectures, we can list some major energy

saving algorithms located at different protocol layers.

2.11.1 Energy Efficiency in Device Level

In recent years, advanced signal processing techniques and wireless radio devices have boosted

the implementation of many applications of wireless communication. Thepromising ubiquitous

computing system has been envisioned by the technologicaldevelopments in digital signal

processing, microelectronics, wireless communicationand networking, sensing material and

Wireless Sensor Network (WSN). Low powerwireless communications based sensing and

computing devices are the main part of theWSN technology. These devices are called sensor

nodes which are usually powered by the finite energy of a non-rechargeable battery. WSN is a

47
quickly growing technologythat has attracted well-deserved attention of the academic and

industrial researchersand in the global business market. The advanced hardware technologies

allow more signal processing functionality to be integrated into a single chip. Coin sized fully

functional wireless node are becoming very demanding and popular among industries and

researchers, in which all the application interfaces, such as, a radio frequency (RF) transceiver,

analog-to-digital (A/D) and digital-to-analog (D/A) converters, baseband processors etc. are

integrated within it. Such wireless nodes are usually power by small batteries. The replacement

of these batteries is very difficult and expensive, even for some cases not possible. Therefore,

energy consumption minimization is a very crucial design aspect for a WSN. Most of the

pioneering work (Fedrizzi et al., 2014) related to energy-constrained communication has focused

on minimizing the transmission energy of the network. The emphasis on minimizing

transmission energy is reasonable in the traditional wireless link where the transmission distance

is large (100 m), therefore the transmission energy is dominant in the total energy consumption.

However, the nodes are densely distributed in many recently proposed wireless ad hoc networks

(such as sensor networks), and the average distance between the nodes is usuallybelow 10 m. For

these circumstances, the circuit energy consumption along the signal path becomes comparable

to or even dominates the transmission energy in the total energy consumption. Hence we

summarize the following unsolved challenges in this area:

 The overall energy consumption including both transmission and circuit energy

consumption needs to be considered in order to find the optimal transmission scheme.

 An optimal mechanism is needed to reduce energy consumption under QOS constraint.

On the other hand, rate adaptation (RA) is a popular mechanism to improve the

performance of wireless sensor networks (Li et al., 2012; Douet al., 2011). RA is used to

48
 optimize the various modulation and coding based physical-layer configuration

depending on time varying channel conditions. The traditional goal of RA is to achieve

effective throughput andhigh goodput.

2.11.2 Energy Efficiency in BSs

The dense deployment of base stations (BSs), which is necessary to satisfy the high demand of

traffic, is causing enormous energy consumption with more challengingoperational cost.

Therefore all stockholders of wireless market possess keen interest for making improvement in

energy efficiency at the network level and are putting a large research effort for finding

innovative solutions. Most of the pioneering works have shown that mobile networks have a

strong potential for energy savings. Most of the works done in literature, have emphasized on

reducing energy consumption at the user end, so that the battery life of mobile terminals can be

increased (Xenakis et al., 2012). However, it has been reported in many studies that, state of the

art BSs, also known as eNodeBs in LTE networks, are the major source of energy consumption,

consumingapproximately 60 − 80% of the total energy of a cellular network (Marsan et

al.,2009). Fig.2.5(also shows that BSs are the major source of energy consumption of a wireless

network. This is mainly because of the always-active operation of current systems.

49
Figure 2.5: Cellular network power consumption [source: Islam 2017) .

Thisalways-active mode offers full-time coverage but fails to adapt energy consumption totraffic

load variations. Therefore designing energy efficient BSs has become the mostimportant issue

for any green communication networks. Hence, operators, vendorsand researchers are

collaborating to propose innovative technologies and algorithms to improve energy efficiency in

the BSs. Researchers in many different papers have proposed various distinctive approaches to

reduce energy consumptions in BSs (Wu et al., 2015)which can be summed up in the categories

as shown in Fig. 2.6. The first two approaches from the above list involve architectural changes

as well as the cost of purchasing, replacing, and installing new equipment. These costs also

include the expenditure involved in manpower, transportation as well as associated energy and

direct cost. On contrary, rests of the three approaches that are applicable on the operating

protocols of the system are less expensive and easily implementable as they do not require

changes to current network architecture. In this thesis we limit our research scope to the last

50
challenge area from the above list, where we identify literature gaps and formulate our research

questions accordingly.

Figure 2.6: Approaches for BS energy saving (source: Islam, 2017)

2.11.3 Energy Efficiency in Aerial Networks

Aerial networks have recently become very popular as key enablers for rapid deployable wireless

networks where coverage is provided by on-board radio heads. The quick deployment of the

aerial platforms such as helikites, drones or airships, with respectto terrestrial infrastructure,

make them suitable candidates in tackling a number of different challenges including, increased

coverage in remote areas, better line-of-sight (LoS) conditions and resilience to unexpected

disastrous situations. Facebook Aquila Drone [Zuckerberg, 2014] is a good examples of ongoing

AeBS projects, which propose a novel solutionfor providing internet access from the sky by

using the AeBSs. Aerial networks can also be deployed by the telecom operators in remote areas

as temporary solution of patching coverage gaps [El-Jabu and Steele, 2001]. The Google Loon

(Katikala, 2014) experiment is an ambitious project intended to provide network coverage to

rural and remote areas. A major advantage of the aerial base stations over static terrestrial base

stations is that they can change their positions to serve the dynamic network of users optimally.

An AeBS can be efficiently integrated into terrestrial cellular wireless networks to either serve

51
the ground users directly or relay traffic to the terrestrial network (Guo et al., 2014; Rohde and

Wietfeld, 2012). Fig.2.7provides a good overview of how aerial networks co-exist with terrestrial

cellular infrastructure. Although there has been increased interest in this topic, research is still at

its nascent stage and there are quite a number of challenges that need more research attention:

∙ Energy efficient AeBS design

∙ Optimal altitude for placement of an aerial platform

∙ Aerial channel modeling etc.

Figure 2.7: Aerial network supporting terrestrial cellular coverage (Source: Islam, 2017).

In order to address these challenges, some pioneering work has been found in literature. In the

European Commission project ABSOLUTE [Gomez, et al., 2016], a hybrid satellite- UAV

ground network is developed using AeBSs to address public safety and capacity enhancement

based on LTE communication systems. The main objective of the ABSOLUTE project is to

design and validate an innovative holistic network architecture ensuring dependable

communication services based on the rapid deployment, flexibility, scalability, resilience and

provision of inter-operable broadband services. Although, significant amount of work on design

and implementation of AeBS networks is found in literature, very few of these works focus on

the energy efficient design ofaerial base stations. Energy is a scarce resource for aerial base

52
stations, hence energy-efficient operation of such networks is important given that the entire

network infrastructure, including the battery-operated ground terminals,exhibits requirements to

operate under power-constrained situations. Therefore, the wise management of energy is quite

beneficial for the network lifetime.

2.11.4 Energy Efficiency in Small Cell Network

A wide range of heterogeneous deployments are supported by LTE or LTE-A, that mainly

includes femtocells, picocells, and relay, with aim of extending coverage of the network,

increasing the capacity of the system and reducing transmit power. Fig.2.8depicts an example of

an environment where small cells like micro, pico and femto cells co-exist in the coverage area

of macro cell. Such small cells play a critical role in adopting LTE-A by bringing access network

near to user in a cost-effective way. Femtocells, also known as, Home eNBs, are low-cost, short-

range, user-deployed cellular access points. Femtocells interconnect standard User Equipment

(UE) to the mobile operator network via the broadband access backhaul of the end user. Though

normally few users are supported by femtocells, they have the functionality of regular

Figure 2.8: A typical heterogeneous network with macro, micro, pico and femto cells
source: Arnold et al.,2010, modified].

53
base stations operating in licensed band of the mobile operator. Femtocells substantially enhance

the user-perceived Quality of Service and greatly improve the energy saving potential for the

network nodes at the cost of employing more sophisticated interference and

mobilitymanagement procedures. The requirement of advanced interference and mobility

management has arisen from unplanned femtocell deployment, denser network layout, short

femtocell radius and employment of access control. The unplanned deployment pattern results in

increased Radio-Frequency (RF) interference at the LTE-A network nodes and complicated

mobility management procedure. On the other hand, the denser network layout and the short

femtocell radius increase the number of handovers (HOs) in the system and enlarge the number

of candidate cells, compromising seamless connectivity and increasing the network signaling

load. Additionally, access control may severely degrade Signal to Interference plus Noise Ratio

(SINR) under certain interference scenarios, for instance, when an LTE-A user is not a member

of a Closed Subscriber Group (CSG) femtocell in proximity. Even though femtocell deployment

comprises several technical challenges, but still they significantly reduce the energy expenditure

for both the UEs and the LTE-A network. As mentioned in [Xenakis et al., 2015], the transmit

power of both the mobile terminals and the cellular stations can be reduced by four to eight

orders of magnitude by deploying femtocells. Self-optimization is another feature of femtocell

that leads to further energy savings. Therefore we can say that, even though femtocell

deployment enhances the EE at the access network nodes, but the actual EE gain strongly

depends on: the interferencemanagement and the mobility management algorithm.

2.12 Energy Efficiency at Device Level

Energy is a scarce resource of network nodes, particularly in remote areas or rigorous

environments where recharging is very difficult or for some cases not possible. Energy

54
consumption is dominated by the transmit power required from each node; therefore, reducing

the transmit power could reduce energy consumption. However, many applications, such as

search and rescue or military surveillance, require a network topology capable of withstanding

sudden node/link failures. Furthermore, many algorithms, such as consensus or swarming

algorithms, require highly connected networks for fast convergence to leverage the more

efficient in-network information diffusion. As the transmit power decreases, the number of

network links reduces, consequently decreasing network connectivity. Therefore, there is

significant interest in developingtopology control schemes capable of achieving a good tradeoff

between theobjectives of energy consumption and network connectivity. Various topology

control protocols have been developed for wireless ad hoc networks to obtain energyefficient

topologies, with minimal energy consumption. That is the case of the localized minimum

spanning tree (LMST) [Li, Hou, and Sha, 2005], and some game-based topology control

schemes [Komali, MacKenzie, and Gilles, 2008;[Zhao, et al.,2016]. In addition to energy-

efficiency, some published works [Shang, et al.,2014; {Xu, Yang, and Kwak, 2016] have also

considered the problem of energy balancing in topology design to extend network lifetime.

Topology control approaches designed to minimize other network characteristics (link price,

interference, and others) were proposed in [Nahir, Orda, and Freund, 2014; {Li, et al., 2015]. In

[Tolba, Tolba, and Lorenz, 2016], the mobility strategy for network coverage was controlled,

while efficiently managing energy resources. However, all these works neglected the importance

of network connectivity on the capability to withstand sudden node/link failures. The authors in

[Miyao, et al., 2009], [Nishiyama, et al., 2012] have focused on constructing k-edge connected

topologies by executing LMST k times, which improves the robustness of network connectivity.

In [Li, Cai, and Zhang, 2016], a specific problem of all-to-one topology control for wireless

55
sensor networks was investigated, in which k node-disjoint paths from each node to the sink

were required. Instead of using the conventional connectivity metrics (node/edge connectivity),

the authors employ algebraic connectivity here, a metric that has been shown to adequately

represent the robustness of network connectivity [Moharet al., 1991; {Bertrand and Moonen,

2013]. Some of the pioneering work in this research area has taken the transmission energy into

account, and have proposed several ways to reduce the transmission energy. In [Verdu,

2002{Nuggehalli, Srinivasan, and Rao, 2002] various strategies have beenproposed to minimize

the transmission energy, which are suitable for long range applications.In [Cui, Goldsmith, and

Bahai, 2005] the authors have shown that optimized transmission time can reduce energy

consumption for both Mary quadrature amplitude modulation (MQAM) and M-ary frequency

shift keying modulation (MFSK) techniques. Some of the other works that have presented rate

adaptation for energy efficiency are given in [Fenget al.,2013;Nuggehalli, et al., 2002]. In

[Nuggehalli, et al., 2002] the authors analyzed MIMO based rate adaptation in an 802.11n

wireless network interface card and showed the trade-off between high throughput and energy

efficiency. In [Feng et al., 2013] the authors have considered real world network topologies and

traffic workloads from Abilene and Intel and have developed two power managementschemes to

utilize sleeping mode and rate adaptation for energy efficiency.

2.13 Energy Efficiency at Access Level

A widely acknowledged fact of the recent age is that, the cellular communication networks will

have greater economic and ecological impact in near future. As a consequence, an innovative

new research discipline has been formed, namely 'green cellularnetworks, which has drawn the

attention of many researchers who are dedicated to reduce the global footprint of cellular

networks. The term green is originally a nickname of dedicated efforts to reducing unnecessary

56
greenhouse gases (such as, CO2) emissions from industries. Another motivation and objective of

'green' approaches, particularly for the mobile operators, is to achieve more commercial benefits,

by minimizing the operating cost related to energy consumption. Hence, it has become necessary

to shift the attention of network designers from pursuing spectral efficiency and optimal capacity

to implement energy efficient strategies. Energy-efficient wireless communication is also

imperative from the users' perspective. According to the 2010 wireless smartphone customer

satisfaction study presented by J. D. Power and Associates (Power et al., 2010), the iPhone

received top marks in every category except for its battery life. The latest report [López, 2011] in

China also rejects the same issue, based on the data in, up to 60% of the users complained that

battery endurance was the greatest hurdle when using 3G services. Without a breakthrough in

battery technology, the battery life of the terminal sets will be the biggest limitation for energy-

hungryapplications (such as, video games, mobile P2P, interactive video, video monitors,

streaming multimedia, mobile TV, 3D services, and video sharing). Therefore withthe explosive

growth of high-data rate applications in wireless networks, EE in wireless communications has

recently drawn increasing attention from the research community. Several international research

projects dedicated to energy-efficient wireless communications are being carried out. There are

various distinctive approaches to reduce energy consumptions in a mobile cellular network.

Approaches found in the pioneering works can be broadly classified into the following five

categories.

∙ Improving energy efficiency of hardware components.

∙ Turning off the network components selectively.

∙ Optimizing energy efficiency of the radio transmission process.

∙ Planning and deploying heterogeneous cells.

57
∙ Adopting renewable energy resources.

Approaches of the first category aim to improve hardware components (such as power amplifier)

with more energy efficient design (Xu et al., 2012). The performance of most components used

in current cellular network architecture is unsatisfactory from the energy efficiency perspective.

For example, the power amplifier consumes the largest amount of energy in a typical cellular

base station (BS), where more than 80% of the input energy is dissipated as heat. Generally, the

useful output power is only around 5% to 20% of the input power (Claussen et al., 2008). Studies

showed that the potentially optimized ratio of output power to input power for power amplifiers

(also known aspower efficiency) could be as high as 70% (Claussen et al., 2008). Accordingly,

substantial amount ofenergy savings can be achieved if more energy efficient components are

adopted in the network. However, the implementation cost for these approaches is high. For

example, a power amplifier module with 35% power efficiency for small cell WCDMA or LTE

BSs (cover at most an area of a radius of 2 km) costs around $75. The cost will be even higher

for larger coverage or higher power efficiency. Therefore, careful consideration in both

operational and economical aspects by network operators is required before decisions on

hardware replacement are made. The secondcategory covers approaches that selectively turn off

some resources in the existing network architecture during non-peak traffic hours. Approaches in

this category generally try to save energy by monitoring the traffic load in the network and then

decide whether to turn off (or switch to sleep mode, also referred as low-power mode or deep

idle mode in some literature), or turn on (or switch to active mode, ready mode or awake mode)

certain elements of the network. Unnecessary energy consumptions, for example, air

conditioning under-loaded BSs, can be avoided by adopting such sleep mode mechanisms. These

approaches generally involve switching certain elements including but not limited to power

58
amplifiers, transceivers, signal processing unit, cooling equipment, the entire BS, or the whole

network back and forth between the sleep mode and the active mode (Marsan et al., 2013). Most

often sleep mode techniques aim to save energy by selectively turning off BSs during off-peak

hours. BSs consume the highest proportion of energy in cellular networks. On the other hand,

dense BSs deployments lead to small coverage area and more random traffic patterns for

individual BS, which make sleep mode operations more desirable. Given the constraint that some

components (e.g., a minimum number of BSs) must alwaysstay on to support the basic operation

of the network, as well as the execution of the switch operation depends on the actuations in

traffic profile, the reported energy saving is not as high as that of component-based approaches.

Also, while it is good to save energy, BS sleeping might negatively impact Quality of Service

(QoS) in the network because of decreasing capacity, unless specific remedial solutions are

adopted concurrently (Soh et al., 2004, (Tabassum, et al.,2014). Nonetheless, because sleep

mode techniques are based on current architecture, they have the advantage of being easier to test

and implement as no replacement of hardware is required and the performance can be evaluated

by computer simulation. The third category focuses on the radio transmission process.

Approaches for this category work on the physical or MAC layer. Advanced techniques

including MIMO technique, cognitive radio transmission, cooperative relaying, channel coding

and resource allocation for signaling have been studied to improve energy efficiency of

telecommunication networks. A variety of approaches have been proposed to efficiently utilize

resources in time, frequency and spatial domains toachieve energy saving. Similar to approaches

based on sleep mode, this type of approaches generally does not require upgrade of hardware

components in the system. However, trade-offs between energy efficiency and other

performance metrics of the network are probably inevitable. Moreover, measuring errors due to

59
complicated uncertainty issues such as interference and noise have not yet been well

investigated. Based on information theory, four fundamental trade−offs related to energy

efficiency on wireless networks have been acknowledged, namely spectrum efficiency energy

efficiency, deployment efficiency−energy efficiency, bandwidth−power and delaypower

efficiency (Chen et al., 2011). The fourth category tackles the energy efficiency issue by

deploying small cells, like micro cells, pico cells and femto cells, in the cellular network

(Chandrasekhar et al.,2008). These smaller cells serve small coverage areas with low energy-

consuming cellular BSs (Xenakis et al., 2013) which usually support plug-and-play feature and

are affordable for user-deployment. Such heterogeneous deployment reduces energy

consumption in the network by shortening the propagation distance between nodes in the

network and utilizing higher frequency bands to support higher data rates, in contrast to

conventional homogeneous macro cell deployment. However, the major drawback of these

approaches is that, these additional small cells add more radio interferences as compared to

conventional homogeneous macro cell networks, which might adversely affect the quality of

service. Furthermore, the deployment of too many small cells may reverse the trend of saving

energy because of extra embodied energy consumed by newly deployed cells as well as because

of the overhead introduced in transmission. Therefore, the quantity and location of the small cells

needs to be carefully optimized in order to reduce total energy consumption. Some research

outcomes have also shown that, integrating heterogeneous network deployment with sleep mode

schemes has proved to be very good approach to achieve significant gain in terms of energy

saving ( Guo and Farrell, 2011; Zheng et al., 2012).The last category of the above mentioned list

includes approaches that adopt renewable energy resources. Renewable resources such as hydro,

wind and solar power stand out for their sustainability and environmental friendliness (Rowlands

60
et al., 2002, Chia et al., 2014) compared to widely used energy resources (such as hydrocarbon

which produces greenhousegases). Some telecom operators have implemented solar power

operated cellular BSs in underdeveloped countries such as Bangladesh and Nigeria, where roads

are in poor and unsafe condition, hence delivering traditional energy resources for off-grid BSs

(such as, diesel) cannot be guaranteed (Chan ,2009; Okoye, 2011). Another popular approach is

the energy harvesting techniques, where available energy is exploited from such renewable

resources to complement existing electric operated infrastructure. This energy harvesting

approach would probably be the long-term environmental solution for the mobile cellular

network industry, especially for the particular areas without mature network infrastructure.

However, it is technically challenging to preserve fault tolerance and data security without any

service interruption while service migrates from the obsolete electric-operated BSs to the new

energy harvesting BSs. Generally speaking, green cellular network is a relatively new area of

research, where the main aim is to make cellular networks greener by reducing total power

consumption through various approaches described above. It was estimated that ICT roughly

accounted for about 10% of global electricity consumption and up to 4% of global carbon

dioxide emissions (around 1 billion tons, approximately equal tothat of aviation industry and one

fourth of emission by cars worldwide) as of early 2013 (Eur, 2013). ICTs share in global carbon

emissions is expected to grow every year, and become double by the year 2020 (Langedem ,

2012). Furthermore, the prevalence of smart phones and tablets accessing cellular network

remarkably contributes to the increasing energy consumption. Smart phones were introduced

around the year of 2000. However, it was the success of mobile operating systems such as iOS,

Android and Windows Phone about a decade later that finally helped them take over traditional

feature phones. Tablet computers became popular almost at the same time, marked by the release

61
of the iPad by Apple Inc. With the help of higher data transmission rate in 3G and 4G (and 5G in

the future) cellular networks, smart phones and tablets enable users to perform much more tasks

than ever before using cellular networks, including, but not limited to, streaming videos,

downloading and reading e-books etc. As a consequence, the number of mobile subscribers are

expected to increase from 4.5billion (in 2012) to 7.6 billion by 2020; and the amount of data

traffic requested by each subscriber are expected to increase from 10 GB (on average) per

subscriber in 2012 to 82 GB per subscriber by 2020. Also, more dynamic and burst mobile data

and video traffic are dominating the mobile voice in cellular networks. All of these factors lead

to significant increase in energy consumption. Manner et al. [2014] showed that, an LTE

network consumes about 60 times more energy as compared to a 2G network in order to provide

the same level of coverage. The pioneering works anticipate that, more BSs, data centers and

other network equipment are required to support the ever-growing mobile traffic. Since BSs

consume more than half of the total energy in a typical cellular network, therefore the increase in

the number of BSs has a significant impact in overall energy consumption. Researchers in [Wu et

al., 2015] have shown that the number of BSs has approximately doubled from 2007 to 2012

worldwide, and this number reached more than 4 million by 2015. When cellular networks need

to be extended to remote areas, off-grid BSs need to be deployed because of the unavailability of

electrical grids in those areas. Off-grid BSs cost ten times more to run in comparison to their on-

grid counterparts, since they generally depend on fuel, which is a costly and unreliable power

source [Correia, et al., 2010]. On the other hand, hydrocarbon energy, one of primary

conventional energy resources that provides 85% of primary energy usage in the United States

and releases large amounts of greenhouse gases when combusted, is proved not sustainable and

expected to be exhausted in the near future [Brochures, 2008].

62
2.14 Energy-efficient radio resource management in WSN

Access NetworkRadio resource management involves strategies and algorithms for controlling

parameters such as transmit power, user allocation, data rates, handover criteria etc. ina way such

that the limited radio-frequency spectrum resources and radio network infrastructure can be

utilized as efficiently as possible. Energy-efficient radio resource management is one of the

effective ways to reduce energy consumption of wireless systems. Most current network

dimensioning is peak-load oriented to satisfy the users QoS requirements. In fact the daily traffic

loads at BSs vary widely over time and space. Because of this, a large amount of power wastage

occurs when the traffic load is low. This issue was already recognized by both vendors and

operators and actions were taken. For instance, Alcatel-Lucent announced that a new feature of

their software upgrades dynamic power save (DPS), bring up to 27% energy saving for BSs

deployed by China Mobile (Feng, 2013). OPERA-Net project (Esnault, 2008) proposed energy-

saving solutions through cell-size breathing and sleep modes based on the traffic loads. Optimal

power-saving schemes using cell switch-off under a trapezoidal traffic pattern and a measured

traffic pattern are analysed in (Marsan et al., 2013), proven that a 25 − 30% energysaving is

possible by simply switching off the active cells during the periods when traffic is low. However,

no studied regarding the effect of the switch-off state on coverage. In Gong et al. (2010), a

traffic-aware BS mode (active or sleeping) switching algorithm, based on a blocking probability

requirement, is introduced. To avoid frequent BSmode switching, a minimum mode holding time

is suggested. It is shown that the effect of changing holding time over a specified range have

little performance change on either energy saving or blocking probability (Gong et al., 2010). In

(Oh and B. Krishnamachari, 2010), demonstrates that energy saving will increase with the BS

density and the variance-to-mean ratio of the traffic load. Energy saving should not only exploit

63
the traffic load variations, but alsothe diversity of the QoS requirements. The trade-off between

energy consumption and delay on the Internet has been extensively studied (Feng, 2013). In the

case of cellular network, little research has been done due to the limited service types (mainly

voice communications) were available in the early systems (1G, 2G systems). However,the

evolution of cellular systems and the popularity of smart phones, more and more diverse

applications will appear on cellular networks. The ability to differentiate real time services and

delay tolerant services is beneficial, since it is essential for scaling the energy consumption with

the traffic type. Recently, some researchers have exploited the service latency of applications to

reduce the energy consumption in cellular networks (Feng,2013). Effective capacity, which

provides the maximum arrival rate that a wireless channel can sustain while satisfying statistical

QoS constraints, is considered. Spectral efficiency bit energy trade-off isanalyzed in the low-

power and wideband regimes by employing the effective capacity formulation, rather than the

Shannon capacity. Based on the research, energy requirements under QoS constraints are

identified. In low-power regime, minimum bit energy required under QoS constraints; same as

that attained when there are no such limitations. The minimum bit energy and wideband slope

expressions were obtained. The required bit energy levels are found to be strictly greater than

those achieved when Shannon capacity is considered in this regime. Overall, a characterization

of the energy-bandwidth-delay trade-off is provided. Energy-efficient radio resource

management can provide significant energy savings (Feng, 2013), but several important issues

are still exposed: The collaboration between neighboring cells should be further studied since the

cell mode switching changes the coverage and handoff issues. The effect of these changes on EE

should be evaluated. When the diversity of QoS requirements for different applications is

exploited, a more general and practical QoS requirement model, as well as the fairness issues

64
between users, should be considered. For example, since both the channel condition and the

traffic flow are time-varying in wireless networks, it is possible that a traffic flow has a higher

transmission priority according to its QoS requirement, but the corresponding channel condition

is bad. Thus, we should balance the EE gain based on the diversity of QoS requirements and the

QoS requirements themselves.

2.15 Fundamentals in Markov Decision Process

Each day people make a number of decisions that have both immediate and long term

consequences. Decisions cannot be made in isolation. Today’s decision impactson tomorrow and

tomorrows decisions impact on the following days. By not accounting for the relationship

between present and future decisions, and present and future outcomes, a good overall

performance might not be achieved. For example, in a long race, the decision to sprint at the

beginning might deplete energy reserves quickly and result in a poor finish. MDP, also referred

to as stochastic dynamic programs or stochastic control problems, are models for sequential

decision making when outcomes are uncertain. Probability theory states that a Markov Model is

defined as a stochastic model that is used to model randomly changing systems where the

assumption of future states depend only on the current state and not on the events that occurred

before it is held valid. There are different types of Markov model (MM): Markov Chains and

Hidden Markov Models (HMMs)(Puterman, 2014). HMMs are models where the states are only

partially observable. Observations are related to the state of the system, but they are typically

insufficient to accurately determine the state. This means that the user enters the current state

which is not completely tangible with the expected state. An extension to such HMMs is the

inclusion of actions performedat each state that leads to the next possible state with a reward.

Such cases are known as Markov Decision Processes (MDPs). However, the extension of an

65
HMM leads to a Partially Observable Markov Decision Process (POMDP) due to the states

consisting of partial information. At each time interval, the agent gets to make some observations

that depend on the state. The agent only has access to the history ofobservations and previous

actions when making a decision. It cannot directly observethe current state, hence unable to

acquire complete information regarding the current state. The MDP model consists of decision

epochs, states, actions, rewards, and transition probabilities. Choosing an action in a state

generates a reward and determines the state at the next decision epoch through a transition

probability function. Policies or strategies are prescriptions as to which action to choose under

any eventuality atevery future decision epoch. Decision makers seek policies that are optimal in

some sense. The following example gives us a clear idea on how MDP can be helpfull in taking a

decision. An agent (or robot) exists in a certain environment. The environment consists of states,

and the agent moves between states in this environment. Based on the current state information

the agent decides which state to move to next. Depending on the agents action, the environment

returns new state information and some reward. The goal in this problem is to maximize the

agents reward. The above interaction between the agent and the environment can also be shown

as Fig.2.9. Here, the agent starts at state 𝑆0 and initially executes action 𝑎0. This gets the agent

reward 𝑅0 and takes him to state 𝑆1. At state 𝑆1, he executes action 𝑎1, which gets him reward

𝑅1 and takes him to state 𝑆2 and so on. This environment obeys the Markov property, i.e.,

everything in the past can be summed up in the current state; or, in other words, the future

depends only on the current state. It is also assumed that the environment has a definite number

of states and that the goal of the MDP is to find an optimal way to act in this environment. MDPs

are powerful modelling tools that allow controlling a Markov chain by creating optimal policies

that dictate what action to take as a response to the current state (Puterman, 2014). MDPs have

66
been successfully applied in a diverse range of industries, from revenue management to a variety

of real life systems including inventory management. A wide range of computer,

manufacturing,and communications systems can also be modeled using the MDP. Moreover,

MDP models have been applied to a number of equipment maintenance and replacement

problems (Puterman, 2014)

Figure 2.9 States, Action and Reward of a MDP (Source: Islam, 2017)

2.16 Handover

In the heterogeneous wireless environment, a mobile user is able to enjoy an uninterrupted

service using the mobile device while moving from the coverage area of one BS to another BS.

This process is called a handover, by which a mobile terminal keepsits connection active as it

migrates from the coverage of one network access point to another (Nasser et al, 2006).

Handovers are called to be seamless if the handover is transparent to the user of the available

network (Arumugam, 2015). A seamless handover is defined as a handover scheme that

maintains the connectivity of all applications on the mobile device when the handover occurs

(Chen et al.,2004)

67
Figure 2.10: Horizontal handover. (a) Hard handover; (b) Soft handover (source: Islam,

2017)

2.17 Related Works

Islam et al. (2020) proposed a smart healthcare system in IoT environment that can monitor a

patient’s basic health signs as well as the room condition where the patients are now in real-time.

In this system, five sensors are used to capture the data from hospital environment named heart

beat sensor, body temperature sensor, room temperature sensor, CO sensor, and CO2sensor. The

error percentage of the developed scheme is within a certain limit (< 5%) for each case. The

condition of the patients is conveyed via a portal to medical staff, where they can process and

analyze the current situation of the patients. The developed prototype is well suited for

healthcare monitoring that is proved by the effectiveness of the system. However, the system was

too bulky to handle and cannot proficiently handle cases of diabetes and respiratory problems.

Tamilselvi et al.(2020) developed a health monitoring system that can monitor basic symptoms

of a patient like heartmrate, percentage of oxygen saturation, body temperature, and eye

movement in IoT network. For this purpose, the system used Heartbeat, SpO2, Temperature, and

Eye blink sensors as capturing elements and Arduino-UNO as a processing device. The

developed system was implemented but no specific performance measures are described for any

68
patient. Acharya et al. (2020) introduced a healthcare monitoring kit in IoT environment. The

developed system monitored some basic parameters of human health like Heartbeat, ECG, body

temperature, and Respiration. The major hardware components which are used here are pulse

sensor, temperature sensor, BP sensor, ECG sensor, and raspberry pi. The data were collected

from sensors and sent it to raspberry pi for processing and again transmitted it to IoT network.

The major drawback of the system is that no interfaces for data visualization are developed.

Banerjee et al.(2016) proposed a pulse rate detection system based on a noninvasive technique.

The proposed system used plethysmography process and correspondingly displayed the output

digitally that made it a real-time monitoring device. The method has proved as reliable for the

patient compared to other invasive techniques. Gregoski et al. (2012) introduced a smartphone-

based heart rate monitoring system. The system used a mobile light and camera to track finger

blood flow and calculated blood flow-based cardiac output. The developed system described an

integrated device that wirelessly transmitted a person’s pulse to a computer, empowering people

to test their heart rate by merely looking at their phones instead of using hands each time. This is

an excellent design but it is not feasible if continuous heart monitoring is needed. Oresko et al.

(2010) mentioned a fully functional cardiovascular disease sensing system for smartphones,

identifying a tool that is developed to be the same given sufficient time and monetary resources.

The developed prototype only tracked coronary rhythm in real-time, did not track heart rate over

time, and could not detect any cardiovascular disease. Trivedi et al. (2017) suggested a mobile

device regulated Arduino-based health parameter surveillance framework. The collected sensor

data are analog and sent it to the board of Arduino Uno. By the integrated analog to digital

converter, the recorded analog values are converted into digital data. Bluetooth transmitted the

physical qualities to the developed device. The Bluetooth device used a module not covering a

69
wide area. Kumar et al. (2017) developed an adaptive IoT safety monitoring device. The

configuration of the framework is separated into 3 layers such as the control layer, the device

layer, and the transport layer. A DS18B20 sensor was used for the measurement of body

temperature in the control segment and a pulse sensor is used for the pulse measurement. The

data were loaded from Arduino into the cloud through the Wi-Fi module and Ethernet shield on

the transport layer. The framework layer finally collected the server details. However, Arduino

Uno was used here, and hence, many sensors cannot be treated properly. Desai et al. (2017)

developed a wireless sensor network (WSN) to track smart homes and heartbeat. Here, Spartan3

is used in parallel data processing with FPGA architecture. The MCU results are shown by an

LCD, and all sensors are connected with a microcontroller. Nevertheless, all machine elements

are not integrated in one unit.

2.18 Overview of Massive Mimo Technology

Multiple-antenna (MIMO) technology is fetching matured for wireless communications and has

been merged into wireless broadband standards like LTE and Wi-Fi (Lipfert, Hermann, 2007).

Ultimately, the more antennas the transmitter/receiver is equipped with, the more the possible

signal paths and the better the performance in terms of data rate and link reliability. The price to

pay is increased complexity of the hardware and the complexity of energy consumption of the

signal processing at both ends.

Multiple-input and multiple-output, or MIMO, Means using several antennas at both the

transmitter and receiver to advance communication performance (Ngo et al.,2013). To

accomplish an array gain that incrementally advances the spectral proficiency and/or

accomplishing a diversity gain that advances the link consistency, multiple antennas can be used

70
to accomplish smart antenna functions. Though, currently the term “MIMO” generally states to a

scheme for enlarging the capacity of a radio link by manipulating multipath propagation

(Hoydiset al., 2013).

Fig. 2.11 Understanding of SISO, SIMO, MISO and MIMO:

According to MIMO systems, a transmitter drives numerous streams by numerous transmit

antennas. Those transmitted streams pass over a matrix channel that contains of all N t N r paths

among the N t transmit antennas at the transmitter and N r receive antennas at the receiver. After

that the receiver collects the signal vector and decodes that collected signal to get the original

evidence. A narrowband flat fading MIMO system is demonstrated as

Where y and x represents the receiver and transmitter vectors, and H and n represents the channel

matrix and the noise vector, respectively.

Multi-user MIMO (MU-MIMO) is the advanced form of MIMO, in multiple-input and multiple-

output, technologies’ the available antennas are spread above a multitude of independent access

71
points and independent radio terminals, all having one or multiple antennas. While comparing,

single-user MIMO uses a single multi-antenna transmitter that communicate with a single multi-

antenna receiver. To improve the communication competencies of all terminals, MU-MIMO uses

an comprehensive form of (SDMA) to permit multiple transmitters to send distinct signals and

multiple receivers to collect separate signals instantaneously in the same band Mezghani, et al.,

2010).

MU-MIMO is the advanced form of MIMO that can be used in different ways as a multiple

access strategy. A prominent alteration is, the performance of MU-MIMO depend on precoding

competency than OFDMA, if the transmitter does not use precoding, the performance benefit of

MU-MIMO is not attainable.

Fig. 2.12 Multiuser MIMO System (Qi, and Aissa, 2012).

Massive MIMO (also acknowledged as Large-Scale Antenna Systems, Very Large MIMO,

Hyper MIMO, Full-Dimension MIMO and ARGOS) creates a big difference with conventional

MIMO systems by using large number of antennas (e.g., hundreds or thousands) which are

functioned entirely coherently and adaptively (Qi, and Aissa, 2012). To focus the signal

transmission and reception energy into smaller regions of space, spare antennas are helpful. That

72
fetches massive enhancements in throughput and energy efficiency, when predominantly joined

with instantaneous scheduling of a large number of user terminals (e.g., tens or hundreds).

Massive MIMO was formerly proposed for (TDD) operation, but now potentially can also be

functional in (FDD) operation.

Some other advantages of massive MIMO are, the broad use of low-cost low-power equipment,

reduced latency, generalization of the (MAC) layer, and robustness to interference and

intentional jamming (Aueret al., 2012). The estimated throughput depends on the propagation

environment providing asymptotically orthogonal channels to the terminals, and

experimentations have not revealed any restrictions till now in this respect. Whereas massive

MIMO condenses many traditional research problems inappropriate, and exposes completely

new complications that immediately need consideration; for example, the challenging task of

making several low-cost low-precision components that work efficiently together, the need for

efficient acquisition scheme for channel state information, resource allocation for newly-joined

terminals, the manipulation of additional degrees of freedom given by spare service antennas,

reducing internal power consumption to attain overall energy efficiency reductions, and finding

new placement situations.

73
 Fig. 2.13 Massive MIMO Technology
Source: International Journal of Engineering Works, (2014)

2.19 MIMO (Multiple Input Multiple Output)

When talk about more than one antenna at both transmitter and receiver, that system is known as

MIMO system. MIMO stances for Multiple-Input, Multiple-Output can be used to provide

improvements in both channel robustness as well as channel throughput (Miao, 2013).

MIMO is an important factor of wireless communication standards such as IEEE 802.11n (Wi-

Fi), IEEE 802.11ac (Wi-Fi), 4G, 3GPP Long Term Evolution, WiMAX and HSPA+.

Fig. 2.14 MIMO - Multiple Input Multiple Output (Miao, 2013).

74
There are different types of MIMO Configurations that we can use, named as, SISO, SIMO,

MISO and MIMO. These formats requires different number of antennas and have different

complexity level.

Hence, there other forms of single / multiple antenna links are :

i. SISO - Single Input Single Output

ii. SIMO - Single Input Multiple output

iii. MISO - Multiple Input Single Output

iv. MIMO - Multiple Input multiple Output

MU-MIMO is also used for a multiple user type of MIMO .

Fig. 2.15: Downlink of a multi-cell massive MIMO system.

Source: International Journal of Engineering Works, (2014)

Massive Multiple-input multiple-output (Massive MIMO) also known as Large scale Antenna

system or Large scale MIMO is a technology where a large number of antennas are utilized to

multiplex messages for different devices on each time-frequency resource, where the radiated

75
energy is focused towards the desired direction and at the same time intra and inter cell

interference is minimized. In a Massive MIMO system, the number of antennas at the base

station is way much larger than the total number of devices per signalling resource (Marzetta,

2010). With regards to energy efficiency, massive MIMO has been shown to decrease the

radiated power in a base station by a factor proportional to the square root of the total number of

antennas deployed and at the same time keeping the information rate unaltered (Ngo et al.,2013).

As the demands for wireless throughput and the number of wireless devices is on the increase,

together with the increasing concern for energy consumption in wireless networks, three

requirements are expected to e satisfied by the 5G technology, these requirements include

possessing less energy consumption, possessing high throughput and serving many users

simultaneously. The massive MIMO technology where the base station is equipped with a

massive number of both collocated and distributed antennas serving many users in the same

time-frequency resource can satisfy these requirements. Spectral efficiency(bit/s/Hz/cell)of

cellular networks can be improved by orders of magnitude without needing more spectrum or

additional base stations. The arrays of antennas are processed coherently so as to improve the

signal quality in the uplink and the downlink. In massive MIMO, the transmitter and receiver are

equipped with a huge number of antenna elements, where the transmit antenna may be

distributed or collocated in different applications, also the large number of receive antennas can

be of one device or distributed to a lot of devices. Research has shown that massive MIMO can

significantly enhance spectral efficiency as well as energy efficiency increasing the capacity 10

times or more and at the same time improving the radiated energy efficiency in the order up to

100 times (Ngo et al., 2013). Looking at what happens in massive MIMO, as the number of

antennas at the base station is large, owing to the law or large numbers, the channels changes to

76
be more favourable. And because of this, linear processing is very close to being optimal. With

simple linear processing, the multiplexing and array gain can be obtained. (Sivasankari and

Sridevi, 2017)

2.20 Benefits

Some initial advantages from the massive MIMO systems are stated as:

i. Inter-user interference is simply diminished by the extraordinary beam forming

determination.

ii. Low-complexity signal processing processes are asymptotically optimal.

iii. Propagation damages are diminished by a large array gain due to coherent beam
forming/combining.

iv. Interference-leakage because of channel estimation blunders is exterminated

asymptotically in the large-dimensional vector space.

Table 2.1: comparing traditional MIMO and Massive MIMO

Parameters MIMO Massive MIMO

Number of Antennas 2*2 – 8*8 Above 8*8

Number of Users Limited Unlimited

Spectral Efficiency Low High

Throughput Minimum Maximum

Delay High Low

Energy Consumption More Reduced

Application WiMax, LTE Future generation

Table 4.1 shows the comparison between traditional MIMO and Massive MIMO, it is seen that

the Massive MIMO system is equipped with more number of antennas when compared to the

77
traditional MIMO which enables it to have an unlimited number of users. It also shows that the

Massive MIMO has better spectral efficiency, throughput, delay and energy consumption.

According to the research by (Sivasankari1 and Sridevi, 2017), a massive MIMO system is

equipped with a lot more antennas, usually tens or hundreds, when compared to the conventional

MIMO systems. Owing to such a massive number of antennas, it has been shown that a massive

MIMO system is able to provide a lot of benefits, like it has greatly increased the capacity,

simplified scheduling design in the frequency domain, and has averaged interference according

to the massive number theorem. Generally, a massive MIMO system can be said to be an

enhanced version of conventional MIMO that utilizes an enormous number of antennas.

Therefore as a result of this, the system performance under massive MIMO, in terms of

efficiency, capacity and reliability, is significantly better than in conventional MIMO systems.

Some of the key technologies of the Massive MIMO system are:

2.21.1 Time Division Duplex

Under time division duplex (TDD) protocol, the overhead needed to estimate the channel is

independent on the number of base station antennas M (Shen et al., 2017). Therefore TDD

protocol is preferred for use in massive MIMO. Exploiting the channel reciprocity can to a large

extent reduce the overhead required for CSI acquisition. The resources required for channel

estimation are not affected by the increase in the number of base station antennas when TDD is

used.

78
2.21.2 Linear Processing

At the terminals in massive MIMO, it is very mandatory the signal processing should be able to

handle dimensional channels. Therefore, one of the merits of massive MIMO is linear decoding

and precoding (Marzetta, 2010). The uplink data transmission can be decoded with simple

matched filter and downlink data transmission is able to be pre-coded with conjugate

beamforming.

2.21.3 Favourable Propagation

The channel between the terminals and the base station can be well conditioned as a result of the

law of large numbers. Hence, massive MIMO uses the assumption that the channel vectors are

almost orthogonal. This phenomenon where only linear processing is required for optimal

performance is known as favourable propagation. Using simple linear detector, the interference

and noise can be ruled out on the uplink while the base station can make use of linear

beamforming techniques to beamform various streams of data to numerous users that do not have

mutual interference.

2.21.4 Array Size

A major feature of massive MIMO is that the antenna array occupies a small space because they

are physically small. This means that for instance if the spacing betweenantennas is about 6 cm

at 2.6 GHz, therefore 128 antennas occupy a cylindrical array will have a dimension of a

28cm×29cm only. Also an array which is designed for carrier frequency 3.7 GHz containing 160

patch antennas which are dual-polarized will have panel size to be is 60*120 cm and the spacing

between the antenna elements is 4 cm which leaves enough space for adding more antenna

elements.

79
2.21.5 Scalability

Massive MIMO is a scalable technology because the base station acquires the channel via UL

pilot when operating in TDD protocol, the time spent on channel estimation is independent on

the number of base station antennas. Therefore, the number of antennas at the base can be

increased without adding more time to the estimation process. Also, because multiplexing and

demultiplexing are required at the user ends, signal processing on each terminal does not depend

on the other user.

2.22 Working Principle of Massive Mimo

(a) Uplink Operation

(b) Downlink Operation

Figure 2.16: Uplink and Downlink Operations of a massive MIMO system

(Ngo et al., 2013).

80
The set-up of the uplink UL and downlink DL operations of a massive MIMO is as shown in

figure 4.2 (a and b). The setup could represent a single cell site, or cell brought out of a network.

A huge number of user equipment UE K inside the cell are usually served via an array of

antennas in the base station. Every terminal possesses a single antenna while the other cells are

served by different base stations which do not cooperate among each other except for pilot

assignment and power control (Albdran et al., 2016). All terminals make use of the full

frequency-time resources simultaneously for UL/DL transmissions. Individual signal sent by the

terminals are recovered at the base station on the UL. The base station on the downlink ensures

that each user equipment UE only receives the signal that is intended for it.

The available knowledge of channel state information CSI makes Multiplexing/demultiplexing

processing at the base station possible.

An arrow beam is created by the base station towards the direction of the terminal under line of

sight (LOS) propagation environment. As the number of antennas is increased, the concentration

of these beams tends to be more accurate (that is they become narrower).

In a scenario where there is an existence of a local scattering, the signal received at any UE is

made up of the superposition of many independent components due to scattering and reflections

which can add up either destructively or constructively. If the transmitted waveforms are

perfectly selected, these components add up constructively exactly at the location of the user. In

order to increase the precision of the power concentration to a certain terminal more antennas

needs to be added to the base station. Hence, having a CSI that is sufficiently accurate to focus

the power at the base station is very important.

81
In massive MIMO, TDD operation as illustrated in figure 2.3 is normally used. The coherence

period is divided into three different operations which include channel estimation (UL/DL

training), Uplink data transmission and Downlink data transmission.

Figure 2.17: TDD protocol of Massive MIMO transmission (Ngo et al, 2013).

2.23 Channel Estimation

A very important function of the base station is to detect the signals transmitted by users on the

Uplink and to precode the downlink signals. Therefore, the base station needs the CSI which can

be gotten by using the UL training. Terminals which are assigned orthogonal pilot signal each

send these pilots to the base station. The pilot sequences transmitted from all terminals are

already known to the base station. Therefore, the base station can estimate the channels by

making use of these pilot signals.

Besides, partial knowledge of CSI might be needed at every terminal for coherent detection of

the transmitted signals from the base station. This partial knowledge can be gotten either by the

use of DL training or via some algorithm which has the ability to blindly estimate the channel.

To detect its intended signal, the terminal only needs the effective gain of the channel because

82
the signals performing are conducted by the use of linear precoding techniques at the base

station.

2.23.1 Uplink Data Transmission

The uplink data transmission occupies part of the coherence interval. In the uplink, the base

station receives the transmitted data from all K terminals in the same frequency-time resource.

The base station detects the signals transmitted from all terminals making use of the channel

estimates and the linear combining techniques.

2.23.2 Downlink Data Transmission

The base station transmits the downlink data to all the terminals on the same frequency/time

resource. Particularly, the base station creates M pre-coded signal and feed them to M antennas.

This can be done through the use of the estimated channel and the symbol intended for the Kth

user.

As a result of new technologies with improved capabilities and features coming up to replace the

older ones, more and more devices and users are being connected which has resulted in an

increase in the percentage of energy consumption of the telecommunication industry, this has

drawn the attention of researchers in looking and devising efficient ways to enhance energy

efficiency. This section reviews some literatures on related topics in this regard:

(Vergados and Sgora, 2010) evaluated several route selection strategies in terms of average

message transmission durations and network connectivity lifetimeexamined the effects of energy

aware routing together with congestion aware routing on the network performance in terms of

network throughput and lifetime. The simulation results from this research showed that the

83
strategy that is accounting for both throughput capacity and energy efficiency can be of great

help in a wireless multi-hop network because it results to an increase in throughput which is as a

result of the more number of available routing paths which in turn results to an increase in the

network lifetime and also causes smaller congestion. The authors however did not take into

consideration the number of antennas in their route selection.

(Koutitas et al., 2010) investigated the concept of green communication in telecommunication

networks. They presented the concept of green telecommunication wireless networks providing

information about the power consumption within wireless and fixed line communication

networks. The significance of energy efficiency in modern and future wireless networks was

outlined. They proposed the use of low power electronics, efficient battery technology, not

‘always on’ devices and recycling at the device level.This research however did not provide the

model for the different techniques proposed.

(Najmeh and Hoang, 2011) presented a new approach that can be used for minimizing the overall

energy consumption of wireless sensor network applications using the Hierarchy Energy Driven

Architecture (HEDA). Components of each part of the HEDA were identified particularly. A

model for each of the constituents and components in terms of their parameters (or dominant

factors) was extracted. A formulation for the overall energy cost function in terms of their

constituent was proposed. The model could be used for optimizing the total energy consumption,

determining the contribution of each constituents and their relative significance, this was

demonstrated by the simulation results for lifetime and residual energy of a test network with a

different radius, transmission radius and selective and random networks. It was noted that many

important issues were still to be explored but the paper only suggested an outline model that can

be used for each constituent and giving a detailed energy model for each of the constituent

84
HEDA.The authors however left out many important issues in terms of each of the constituent

HEDA.

(Abbas, 2012) proposed and analyzed mathematically solutions that will work by switching off

most components that are energy consuming. In order to discover the network capacity and

optimal placement of the relay node, a continuous space analytical technique for analyzing a

hybrid network that has uniform node distribution was proposed and investigated. For this

analysis, two schemes (one deterministic and one Finite Markov Decision Process (FMDP

based) probabilistic) were presented. Results from the analysis showed that in order to save the

energy of the mobile station battery, that different links can be used for uplink and downlink

traffic. Also proposed were a DTMC-based and an FMDP-based method for the prediction of

energy consumption and optimization of a mobile station. As part of limitation of this study,

joint optimization of data transfer speed and MS energy consumption for heterogeneous traffic

still needs to be looked into. However, this research was done based on LTE standards only.

(Ziaul and Frank, 2012) proposed a novel approach which can switch off a base station under

light traffic conditions so that the power consumption of such network can be conserved. Two

teletraffic-based power saving schemes that can be used to analyze the transmission and power

consumption of the network for a simple network with micro cells were proposed. The first

scheme analyzed the expected sojourn times of separate channel occupancies and then it

switches off the base station deterministically at the instance when the switching thresholds are

met. The second scheme works by switching off the base station probabilistically based on a

policy that was designed using Finite Markov Decision Process (FMDP). The numerical results

obtained demonstrated that in the sojourn time-based scheme, the hysteresis-based technique

saves total network power for an acceptable number of extra transmission power consumption

85
cost. The numerical results of the FMDP-based scheme shows that using this analysis, a lower

limit on the long-term network transmission power can be obtained.The research however, did

not consider some of the few users that may want to transmit data at the low traffic conditions.

2.24 Knowledge Gap

Energy efficiency in wireless communication has become one of the most attractive topics in

current contemporaries.

The review of relevant literatures relating to energy issues has been carried out, it is however

noteworthy that all the scholarly works in this area focused mainly on network performance and

reduction in transmission energy of the network.

Due to increase in the awareness of 5G network, LTE- Advance heterogenous networks, there is

therefore a need to have a tradeoff between energy efficiency and deployment cost for the 5G

network. Also, the area of energy efficiency cell-size design, emerging heterogenous networks

(mix of macro-cell, microcells, pico-cells and femto-cells) needs serious consideration.

This research shall focus on the application of massive multiple-input multiple-output (MIMO)

model to reduced the total energy consumption of the network.

Table 2.2:Summary of Identified Gaps

Author Year Research Technique Limitation

Islam et al 2020 Proposed a smart healthcare Five sensors are used to capture the The system was too
system in IoT environment data from hospital environment bulky to handle and
that can monitor a patient’s named heart beat sensor, body cannot proficiently
basic health signs as well as temperature sensor, room handle cases of
the room condition where the temperature sensor, CO sensor, and diabetes and
patients are now in real-time. CO2 sensor. The error percentage of respiratory problems.
the developed scheme is within a
certain limit (< 5%) for each case.
The condition of the patients is

86
 conveyed via a portal to medical
staff, where they can process and
analyze the current situation of the
patients.
Acharya et 2020 Developed a health For this purpose, the system used The developed system
al. monitoring system that can Heartbeat, SpO2, Temperature, and was implemented but
monitor basic symptoms of a Eye blink sensors as capturing no specific
patient like heartmrate, elements and Arduino-UNO as a performance measures
percentage of oxygen processing device are described for any
saturation, body temperature, patient.
and eye movement in IoT
network.
Gregoski et 2012 Introduced a smartphone- The system used a mobile light and This is an excellent
al. based heart rate monitoring camera to track finger blood flow and design but it is not
system. The developed calculated blood flow-based cardiac feasible if continuous
system described an output. heart monitoring is
integrated device that needed.
wirelessly transmitted a
person’s pulse to a computer,
empowering people to test
their heart rate by merely
looking at their phones
instead of using hands each
time.
Oresko et al. 2010 Developed a fully functional Identifying a tool that is developed to The developed
cardiovascular disease be the same given sufficient time and prototype only tracked
sensing system for monetary resources. coronary rhythm in
smartphones, identifying a real-time, did not track
tool that is developed to be heart rate over time,
the same given sufficient and could not detect
time and monetary resources. any cardiovascular
disease.
Trivedi et al. 2017 Developed a mobile device The collected sensor data are analog The Bluetooth device
regulated Arduino-based and sent it to the board of Arduino used a module not
health parameter surveillance Uno. By the integrated analog to covering a wide area.
framework. digital converter, the recorded analog
values are converted into digital data.
Bluetooth transmitted the physical
qualities to the developed device.
Desai et al. 2017 Developed a wireless sensor Spartan3 is used in parallel data Nevertheless, all
network (WSN) to track processing with FPGA architecture. machine elements are
smart homes and heartbeat. The MCU results are shown by an not integrated in one
LCD, and all sensors are connected unit.
with a microcontroller.

87
 Kumar et al. 2017 Developed an adaptive IoT The configuration of the framework The framework layer
safety monitoring device is separated into 3 layers such as the finally collected the
control layer, the device layer, and server details.
the transport layer. A DS18B20 However, Arduino
sensor was used for the measurement Uno was used here,
of body temperature in the control and hence, many
segment and a pulse sensor is used sensors cannot be
for the pulse measurement. The data treated properly.
were loaded from Arduino into the
cloud through the Wi-Fi module and
Ethernet shield on the transport layer.

3.5 KNOWLEDGE GAP

No doubt, energy efficiency in WSN is a major challenge in next generations cellular network

Technologies. However, some of the researchers concentrate mainly on:

i. Improving energy efficiency at the user’s end

ii. Energy efficiency in term of through put.

iii. Energy efficiency in a renewable.

iv. Energy efficiency source in an homogeneous networks

v. Spectral efficiency and capacity

From the literatures review not much have been done in the area of energy efficiency for mobile

communication system in a real-time and periodic-time mode. The research will work in how to

improve base status energy efficiency in a heterogeneous network considering various nodes and

cooperative communications network.

of the wireless communication network both at User Equipments (UEs) and Base Stations (BS).

It is important to note that the future 5G will have higher traffic load, the increase in power

consumption of the network is therefore a great issue of concern, which will invariably increase

carbon-12 emission of the Atmosphere. Increasingly the research will address the efficiency with
88
which each Joule of energy required to transmit a data packet must grow by a factor of X100 or

more without necessarily contributing to carbon-12 emission at reduced cost, also bearing in

mind a minimal power cases on the hardware infrastructure.

Table 2.3:Research Timelines

Activities Expected duration

 Research proposal January 2023

 Collection of Data March 2023 –July 2023

 Progress Report August 2023

 Statistical and Numerical Analysis of Data September 2023 – December 2023

 Simulation of Results March 2024 – June 2024

 Progress Report August 2024

 Drawing conclusion of the Research December 2024

3.7 EXPECTED CONTTRIBUTION TO THE BODY OF KNOWLEDGE

Massive MIMO Technology if deployed will enhance the energy efficiency of the wireless

sensor network.

The outcome of the study will definitely improve the quality of service of Telemedicine on

cardiovascular cases considering the limited network bandwidth, especially now that there is

sharp increase in incidences of Cardiovascular Diseases (CVD) across the globe.

89
 CHAPTER 3

MATERIALS AND METHODS

Uncertainty and difficulty in interpreting ECG patterns in cardiovascular related diseases (CVD)

are the major causes of unnecessary medical intervention or a failure to intervene when needed.

The analysis and interpretation of changes in the ECG are carried out by visual inspection which

is very subjective.

Improvements in the quality of the instrumentation, better understanding of the patterns,

improved recognition of the position of ECG and the current status of ECG was a problem in the

past. The major weaknesses are formed by its application. Differences in paper speed, lack of

uniform and exact definitions, uncertainties regarding how, when and whom to monitor,

problems with storage of ECG tracings, uncertainty if and when micro blood sampling should be

applied, limited training of people who are directly involved with Cardiovascular monitoring and

the lack of guidelines on how to handle clinical circumstances are the problems which have

weakened the position of ECG as a monitoring technique.

A computerized wireless sensor system for the detection, operation and successful application of

telemedicine to cardiovascular problems, which incorporates digital processing and storage of

all cardiovascular related diseases (CVD) data has provided the following advantages: the

availability of quantified information, easy retrieval of relevant ECG traces for education

purposes and decreased medico-legal vulnerability concerning lost ECG traces.

The current improvements in ECG can lead to the following information: A future understanding

of the physiology, Knowledge of technical pitfalls, Uniform paper speed and vertical scaling,

Knowledge of factors is influencing cardiovascular related diseases (CVD), Knowledge of all

relevant clinical data, Discipline in reading, classification and interpretation of ECG patterns.

90
Recognition that ECG is an indirect parameter of cardiovascular related diseases (CVD)

condition, Recognition that ECG is only a screening technique, Disposition of maximal possible

information (micro blood sampling), Availability of quantified and objective cardiovascular

related diseases (CVD) analyses, Digital storage of heart rate data, and Systematic pre-and

postgraduate training.

To implement the wireless sensor system for the detection, operation and successful application

of telemedicine to cardiovascular problems, it is necessary to obtain data that gives adequate

reports on the cardiovascular problems within the study area through the application of energy

efficient sensor.

3.2 Research Method

The research will adopt a modified waterfall model of project management are as shown

below in fig 3.1;

Fig 3.1: Methodological Framework

91
3.2.2 Design Method

The research data on symptoms of cardiovascular diseases will be collected from data from

medical records and Electrocardiogram (ECG) across the map out area, this information will be

subjected various statistical and numerical analysis to determine the spartial and normal

population distribution density of cardiovascular diseases of the selected area.

The research will make use of the beam-forming technique together with antenna array and the

RF architecture for massive MIMO technology. The modeling and simulation of the Massive

MIMO for LTE(4G), 5G communication network will be executed by MATLAB/Simulink.

Graph will also be generated for performance evaluation.

3.3 Materials Requirement for the project

The materials for this research shall be as outlined.

3.3.1 Base station module

This is a short-range transceiver which connects wireless phone, computer, or other wireless

device to a central hub and allows connection to a network.

3.3.2 Power Amplifiers

An amplifier is an electronic device used to increase the magnitude of voltage/current/power of

an input signal. It takes in a weak electrical signal/waveform and reproduces a similar stronger

waveform at the output by using an external power source.

92
3.3.3 Antenna Interface:

An active antenna device comprises digital processing modules, transmission and reception

radio-frequency modules and an antenna array in sequence. The device further comprises

combiners, a passive antenna feeder phase-shift network and antenna air interfaces.

3.3.4 Power supply unit

This is a power source for powering the entire system

3.3.5 User equipment module (UE)

In the Universal Mobile telecommunications System (UMTS), 3G, 4G, Long Term Evolution

(LTE) and 5G networks, user equipment (UE) is any device used directly by an end-user to

communicate. It can be a hand-held telephone, a laptop computer equipped with a mobile

broadband adapter, or any other device.

3.3.6 Set of sensor nodes

Set of sensor node is provided sense various physical parameters of human body such as

Temperature. Blood pressure, Heart rate, motion etc. The sensed information is then reported to

the collector node/Agregator.

3.3.7 Medical Server (MS) is a remote medical healthcare monitoring equipment where

medical decisions are carried out.

3.3.8 An electrocardiogram (ECG) records the electrical signals from the heart to check

various heart condition. These transmitted signals are shown as waves on a computer monitor,

93
3.3.9 Computation and Software

A laptop computer with the following specification will be cited for the computation (HP,

window 7, intel (R) core (TM) 13 CPU M370 AT 2.4 GH, 6.00GB RAM). The Softwares that

will be used in data analysis are Microsoft Excel 2013, SPSS for Analysis of covariance

(ANOVA), standard variation. MATLAB/Simulink R2007 version for the system simulation.

The block diagram of the proposed WSN is shown in figure 3.2, the WSN model contains the

sensor nodes, the interfacing and the communication unit.

Figure 3.2: WSN for cardiovascular operation

3.4 Real time cardiovascular (CDV) detection model using ECG and QSR algorithm

This section shows how to detect the QRS complex of cardiovascular (CDV) signal in real-time.

Model based design is used to assist in the development, testing and deployment of the

algorithm.

94
The electrocardiogram (ECG) is a recording of body surface potentials generated by the

electrical activity of the heart. Clinicians can evaluate an individual's cardiac condition and

overall health from the ECG recording and perform further diagnosis(Real-Time ECG QRS

Detection - MATLAB & Simulink, 2022).

A normal ECG waveform is illustrated in the following figure 3.3. Because of the physiological

variability of the QRS complex and various types of noise present in the real ECG signal, it is

challenging to accurately detect the QRS complex used for CDV detection.

Figure 3.3: normalized ECG waveform

3.4.1 ECG Signal Source

The ECG signals used in the development and testing of the biomedical signal processing

algorithms are mainly from three sources: 1) Biomedical databases (e.g., MIT-BIH Arrhythmia

95
Database) or other pre-recorded ECG data; 2) ECG simulator; 3) Real-time ECG data

acquisition.

For operation purposes and implementation in the wireless sensor system, the following pre-

recorded and simulated ECG signals are used in the detection of cardiovascular problems. The

signals all have sampling frequencies of 360 Hz.

 One set of recorded real ECG data sampled from a healthy volunteer with a mean heart

rate of 82 beats per minute (bpm). This ECG data was pre-filtered and amplified by the

analog front end before feeding it to the 12 bit ADC.

 Four sets of synthesized ECG signals with different mean heart rates ranging from 45

bpm to 220 bpm. ECGSYN is used to generate synthetic ECG signals in MATLAB.

3.4.2 Modeling the ECG signal processing and filtering

A real-time QRS detection algorithm can be developed in matlab with the assumption that the

sampling frequency of the input ECG signal is always 200 Hz (or 200 samples/s). However, the

recorded real ECG data may have different sampling frequencies ranging from 200 Hz to 1000

Hz, e.g., 360 Hz in this example. To bridge the different sampling frequencies, a sample rate

converter block is used to convert the sample rate to 200 Hz. A buffer block is inserted to ensure

the length of the input ECG signal is a multiple of the calculated decimation factor of the

sample-rate converter block.

The ECG signal is filtered to generate a windowed estimate of the energy in the QRS frequency

band. The filtering operation has these steps(Shwetha & Sathisha, 2015):

96
1. FIR Band pass filter with a pass band from 5 to 26 Hz

2. Taking the derivative of the band pass filtered signal

3. Taking the absolute value of the signal

4. Averaging the absolute value over an 80 ms window

Step 1: Modeling the band pass filter

Consider a complex z-plane with sampled with band pass from 5 to 26Hz as described in

equation

z=[ 1 , 0.60+0.66 i ] (3.1)

Along with power spectral density of the filter.

Step 2: carrying out the derivative using a derivative function

The derivative block contains the sum and product function as shown in equation 3.2

−1 −3 −4
2+1 z + z 2 z (3.2)

By substituting the value of the complex signal z in the sum and product function equation, we

have

−1 −3 −4
2+1 [ 1 , 0.60+0.66 i ] + [ 1 ,0.60+ 0.66 i ] 2 [ 1, 0.60+ 0.66 i ] (3.3)

Finding the derivative of the above equation will result in

−2 −8
−[ 1 ,0.60+ 0.66 i ] −14 [ 1 ,0.60+ 0.66 i ] (3.4)

97
Step 3: find the absolute value of the signal

Next, we fine the absolute value of the derivative of the signal,

|− [ 1 , 0.60+0.66 i ]−2−14 [ 1 , 0.60+0.66 i ]−8|(3.5)

[ 1 ,0.60+ 0.66 i ]−2−14 [ 1 ,0.60+ 0.66 i ]−8 (3.6)

3.4.3 Real-Time QRS Detection of CDV Signal

The QRS detection block detects peaks of the filtered CDV signal in real-time. The detection

threshold is automatically adjusted based on the mean estimate of the average QRS peak and the

average noise peak. The detected peak is classified as a QRS complex or as noise, depending on

whether it is above the threshold.

The following QRS detection rules reference the PIC-based QRS detector implemented in (Pan

and Tompkins, 1985).

Rule 1. Ignore all peaks that precede or follow larger peaks by less than 196 ms (306bpm).

Rule 2. If a peak occurs, check to see whether the raw signal contains both positive and negative

slopes. If true, report a peak being found. Otherwise, the peak represents a baseline shift.

Rule 3. If the peak is larger than the detection threshold, classify it as a QRS complex. Otherwise

classify it as noise.

Rule 4. If no QRS has been detected within 1.5 R-to-R intervals, but there is a peak that was

larger than half the detection threshold, and that peak followed the preceding detection by at least

360ms, classify that peak as a QRS complex.

98
The QRS detection algorithm can be enclosed using a math function that detects cardiovascular

problems by detection of heart beat rates and heart beat peak values

3.4.4 Pre-Processing: The Signal Enhancement Scheme for QRS detection

The signal processing constraints for QRS detection are more relaxed than ST waveform analysis

thereby enabling a more aggressive signal enhancement scheme to be used. The QRS complex

can still be detected and the heart-rate can still be measured even if the quality of the data is very

poor and many of the ECG features cannot be used. The only constraint is that the general shape

of the QRS complex should be preserved to the extent that it differs from the background noise.

The primary purpose of pre-processing is to remove all the power-line noise and the low

frequency from the signal so that the QRS portion of the ECG waveform can be located

accurately and reliably. The filter must attenuate the signal energy in the following bands:

0Hz – 5Hz Mostly DC, baseline shifts PR and ST waveform.

49Hz – 51Hz Mains power-line noise energy.

99Hz – 101Hz Harmonics of the power-line noise produced by neighboring electrical

equipment.

This technique works very well in the absence of high frequency noise and transient shifts.

Differentiation worsens the effects of higher frequency noise because the derivative of a signal is

proportional to the frequency.

Consider a sinusoidal noise source n(t).

n(t) = A Sin(ωt + ϴ) --------------------------------------------------- (3.7)

Where A is the amplitude, ω is the angular frequency and ϴ is an arbitrary phase shift.

Differentiating eqn. (3.7) gives

dn(t) = Aω cos(ωt + ϴ) -------------------------------------------------- (3.8)

99
Equation (3.8) shows that the amplitude of the derivative data is proportional to the frequency.

Therefore the effects of high frequency noise are increased and this can affect the important

features.

3.4.5 Non-Linear Prediction Filter

This technique uses a multi-layered perception (MLP), with sigmoid activation functions in the

hidden layer and a linear output as a non-linear prediction filter. The MLP parameters are

adapted on-line to predict the unwanted lower-frequency components of the signal, such as the

baseline shift, PR and ST segments of the ECG waveform. The error between the predicted

signal and the actual signal is computed, revealing sharp features such as the R peak and some of

the noise. The output of this is then passed to a pattern recognition algorithm.

The output of the network approximates to the low-frequency components of the FECG signal,

including the P wave and T wave. The output error of the filter is the remaining higher-frequency

components which include the sharp QRS complex and high frequency noise. The output error is

then squared and smoothed using a moving average to locate the candidate QRS complex.

This technique does not suffer from the effects of high frequency noise as much as the

differentiation technique, but is difficult to obtain consistent results because the training

algorithm for the MLP can become stuck in local minima and can produce unpredictable results.

The non-linear prediction filter inspired another technique that achieves the same objectives with

consistent results.

100
3.4.6 Linear Filtering design by least squares approximation

This technique uses least-squares curve fitting to approximate the slow moving components of

the FECG signal such as the baseline shift, ST and PR segments. The difference is computed

which retains mostly the sharp QRS complex and high frequency noise.

A low order curve of the form

f(t) = C0 + C1(t) + C2(t)2 + . . . Cntm - - - - - - - - - (3.9)

Where t = {-L, -L+1 . . . L}, is fitted to a window of 2L+1 raw fetal ECG data points x (t)

The degree of approximation, m, and the window size, 2L + 1 are chosen such that only the slow

moving parts of the data (baseline shift, PR and ST segments are accurately approximated but the

sharp QRS portions of the data are very poorly approximated. The Coefficients c is easily

computed directly with no need for training algorithms. Equation (3.9) can be written for all t in

matrix notation as

H. C = y . . . . . . . . . . . . . (3.10)

Where Ht,k = tk is a (2L + 1) by m rectangular matrix and y1 = f(t), k = {0 . . . . m},

t = {0, 1 . . . 2L}, multiplying both sides of equation (3.10) by H+ gives

C = H+ y . . . . . . . . . . . . . . . (3.11)

Where H+ is the pseudo-inverse is by using the singular value decomposition algorithm, where H

is factored as H = UΣVT, where U and V are orthogonal matrices and Σ is a diagonal. This can be

computed with mathematical computer software such as MATLAB. The pseudo-inverse is

computed from this as

H+ = V Σ-1 UT . . . . . . . . . . . (3.12)

In equation (3.10) H is a constant matrix, therefore H+ is also a constant matrix.

101
From equation (3.9), the mid-point of the curve f(0) is taken as the low-frequency

approximation, the window is then advanced by on sample, and the process is repeated. Form

equation (3.9), f(0) = C0, which is simply computed by

C0 = h0+y . . . . . . . . . . . . . . (3.13)

Where h0+ is the first row of H+ and is constant row vector. Given that y is shifted by one sample,

then equation (3.13) is simply a convolution operation and can be written as a fixed non-causal

fixed FIR filters.

2L

f(0) = Σht0 (n) yn . . . . . . . . . . . . (3.14)

n=0

where h0+ are the filter coefficients. This is known as a Savitzky-Golay smoothing filter.

3.4.7 COMB Filtering

Most of the fetal ECG lies approximately between 2.5Hz and 65Hz. The QRS portion of the

ECG waveform lies mostly between 10Hz and 40Hz. The magnitude response of a filter

designed to maximize the QRS signal to noise power ratio would be a band-pass filter.

To implement this directly with an optimal FIR filter requires at least 141 coefficients, so it is

initially designed at a lower sampling frequency. The design of this filter starts with the desired

filter specification design with an effective sampling frequency of 50Hz.

The Remez exchange algorithm is used to design this filter with L = 24 coefficients hi, where i =

0 . . . . . . . . . L – 1. (3.15)

102
These coefficients are interpolated with zeros to restore the sampling frequency to 50Hz. This

produces a filter with a repeating frequency . The filter can be implemented very efficiency with

the following difference equation.

y1(n) = Σ hi.x(n-10i) . . . . . . . . . . . . (3.16)

X(n) are the raw data samples, hi are the filter coefficients, and y 1(t) are the filtered output

samples.

The delay through the filter is 120 samples. To remove the unwanted frequency bands a second

band selection filter is designed at a sampling frequency of 250Hz with coefficients hi, where j =

0 ……. M and M = 23.

The coefficients are again interpolated with zeros to restore the sampling rate to 500Hz which

produces a filter with a repeating frequency. The delay through this filter is 24 samples and is

implemented using the following difference equation.

y2 (n) = Σ hi.y1 (n – 2i) . . . . . . . . . . . (3.17)

Where y1 is the output of the first filter and y2 is the cascaded filter output. These filters are

cascaded to produce a linear band-pass filter of the desired specification. The overall delay

through the filter is 144 samples which equates to 288ms.

3.4.8 Comparison of Pre-processing Techniques

The pre-processing of the fetal ECG data requires the removal of baseline shifts and power-line

noise. The QRS is the only feature that needs to be preserved so that the constraints on the signal

enhancement algorithms are more relaxed than the signal enhancement for ST analysis. ST

waveform analysis and CTG are both dependent on the accurate location of the R peak. Beat-by-

103
beat heart rate analysis is routinely used and so there is a great advantage to measure heart-rate

even if the ST waveform cannot be analyzed.

In this thesis, a separate enhancement schemes is used for QRS detection and heart rate analysis.

To remove baseline shifts, differentiation method is unsuitable because of the unwanted effect of

amplifying higher frequency noise. The non-linear prediction filter is an improvement, but

computationally expensive and not always reliable.

The Savitzky-Golay filter and the multi-band filter have similar merits as the prediction filter.

They are both stable and have predictable performance. The multi-band filter does have the

added advantage of inherently removing the power-line noise which makes it the preferable

choice.

3.5 QRS Detection in ECG

The process in cardiovascular related diseases (CVD) analysis is to determine the location of

each ECG waveform from the raw and noisy ECG data. There are many problems in detecting

the ECG pattern due to the effects of noise and the presence of cardiovascular ECG waveforms.

The QRS complex is the only component of the ECG waveform that does not change during

cardiovascular related diseases (CVD) analysis unlike the ST and PR waveforms which can

change slowly throughout the process. For this reason, only the QRS portion of the ECG

waveform is used for detection purposes.

The first task is to minimize the effects of noise in the ECG signal, which is known as pre-

processing. The second task is to employ a pattern recognition algorithm to locate each QRS

complex.

104
A. Data acquisition from subjects

The ECG data is collected from Federal Medical Center Umuahia Abia State, sampled at 360Hz

for performance evaluation of the proposed cardiac disorder detection platform.

Four subjects with age of 24-35 years are taken to record signals for 10 minutes each. According

to the report one subject with normal sinus rhythm (heart rate of 70 beats/ minute), one having a

trial fibrillation, and other two with arrhythmia are considered.

The bandpass filter (2-250Hz) is used to reduce the influence of noise such as power line

interference, baseline wander and motion artifacts which are generally embedded with acquired

signal. Then, the QRS complexes of ECG signals are detected.

The block diagram of the proposed cardiac disorder detection technique is shown in Fig. 3.4.

Fig. 3.4. Block diagram of proposed cardiac disorder detection technique.

B. Signal processing in MATLAB GUI to determine PSD

Lead II is considered for the analysis. The obtained signal has noise due to high frequency,

motion artifact, or power line interference. The significant QRS peaks of the signal are extracted

105
through a band pass filter at 1-300 Hz and sampled at 200 Hz to obtain the signal x(t). The

signal, x(t) is segmented into three part x i (t) according to the three frequency spectrum, i = 1, 2,

3. After the x i (t) is obtained, WT is used to compute its wavelet power coefficients, WT (t , a) as

follows:

WT ( b , a )=
1
√a
∫ xi ( t ) φ
¿ t−b
a
dt( ) (3.18)

WT ( t , a ) =√ a∫ e
jwt ¿
x i ( w ) dt φ ( aw ) dw (3.19)

where w , x (t), φ (t)represents angular frequency, the signal in the ith state, and Morlet wavelet

respectively. The scaling factor, a and shifting factor or space, b are real and a > 0. Actually, the

WT is not a function of time and frequency but time b and the scaling factor a. As scale is related

to frequency, PWT ( t ,a ) defines spectral density of time and scale.

1 2
PWT ( t ,a )= 2
|WT (t ,a)| (3.20)
2 ΠC a

Where the constant C is chosen to obtain the energy by using WT,

2
|φ̇(w)|
C=∫ dw (3.21)
|w|
r
w
A reference frequency, w r is chosen from w= to obtain a time-frequency density. Finally,
a

wavelet time frequency analysis can be expressed by eq. (3.22).

PWT ( t , w )=
wr
w
2 ( ww )
PW T t ,
r
(3.22)

| ( )|
2
1 wr
PWT ( t ,a )= WT t , (3.23)
2 ΠC w r w

106
Then, the instantaneous power, P (t, a) WT is obtained from Eq. (3.24), which contains the

messages of time and frequency. For assessing the instantaneous power of independent

frequency components of signal, the power spectrum at different frequency is divided into three

parts:

(i) The power spectrum for very low frequency (VLF, 0.003–0.6 Hz) is:
0 .6
PVLF ( t , a )= ∫ PWT ( t , f ) df (3.24)
0 . 003

w
where f = frequency interval is related to the scaling factor a.
2Π

(ii) The power spectrum for low frequency (LF, 0.6–1.7Hz) is:
1 .7
P LF ( t ,a )=∫ P WT ( t , f ) df (3.25)
0 .6

(iii) And for high frequency (HF, 1.7–250 Hz) the power spectrum is:
250
P HF ( t , a ) =∫ PWT ( t , f ) df (3.26)
1 .7

C. Statistical features extraction for disorder detection

Different statistical features (heart rate (HR), form factor (FF), kurtosis, skewness, and LF/HF

ratio) are extracted to differentiate among infected and non-infected heart beats.

1) Average heart rate: Instantaneous heart rate values are obtained by the inverse of the RR

interval of each beat.

IHR=
[ 60 60 60
, ,
tR R tR R tR R
1
,……………
2
60
tR R 3 n
] (3.27)

w h ere t R R , t R R , t R R , … … … … … … … .t R R are time instants at which the QRS complexes occur

1 2 3 2

in the ECG signal.

107
 60
AHR= (3.28)
t RR Average

2) Form Factor: It is the ratio of the mobility of the first derivative of the signal to the mobility

of the signal itself. It discriminates between normal and infected heart beats.

3) Kurtosis: Kurtosis is a measure for the degree of flatness or peakedness in the variable

distribution.

E [ ( X −μ )3 ]
Kurt [ X ]= (3.29)
2 2
E [ (X−μ) ]
High kurtosis tends to have a distinct peak near the mean, decline rather rapidly, and have heavy

tails. While low kurtosis tends to have a flat top near the mean rather than a sharp peak.

4) Skewness: Skewness can reflect the degree of asymmetry in the histogram of instantaneous

power:

[( ) ] E [ X 3 ] −3 μ σ 2−μ 3
3
X−μ
Skew [ X ] =E = 3 (3.30)
σ σ

Where, E is the expectation operator. If the histogram is symmetrical, the skewness is zero. If the

left hand tail is longer, the skewness will be negative. If the right hand tail is longer, the

skewness will be positive.

5) LF/HF ratio: Instantaneous power of independent frequency components PVLF (t), P LF (t), and

P HF (t) , are calculated from eq. ((3.24))–(3.26). The ratio of LF and HF is

LF P LF (t)
ratio= (3.31)
HF P HF (t)

108
3.6 The power model

An empirical formula for propagation path loss model is derived from Okumura’s report in order

to put his propagation prediction method to computational use. Okumura’s report is very

practical because it carefully arranges field strength and service area. Not only is the report used

as comparison data with authors’ reports, but also the propagation prediction methods in the

report have become standard for planning in today’s land mobile system in Japan.

The Okumura-Hata model was proposed to be used in the frequency range of 500 MHz to 2

GHz, although most relevant research studies use this path loss model up to 3 GHz. The model is

typically used to estimate channel path loss because of its simplicity and availability of

correction factors. It is commonly restricted to macro cells. The model is created from a number

of representative path loss models for the small and medium, urban, suburban and rural areas

environments, expressed as:

P L= A+ B∗log ( d ) +C 3.32

where A , B , and C are factors that depend on frequency and antenna height:

A=69.55+26.16 log ( f c )−13.82 log ( h bs) −a hm 3.33

B=44.9−6.55 log ⁡(hbs ) 3.34

with:

d is the distance between the user and base station (km),

h bsis the height of the base station (m), while the function a h m and factor C depend on the

environment.

Small and medium size cities:

For a built up small city or a medium town with small buildings and houses, a h m and C are

defined as:

109
a h m=( 1.1 log ( f c )−0.7 ) hue −(1.56 log ( f c )−0.8) 3.35

C=0 3.36

Where:

f c is carrier frequency (MHz)

hue is height of UE.

Urban areas:
This category is for a built up city or a large town with large buildings and houses with two or

more stories, or a large village with close, tall, thickly grown trees. In this case, a h m and C are

defined as:
2
a h m=3.2∗(lo g 10 ( 11.75∗h ue ) ) −4.97 3.37

C=0 3.38

Suburban areas:

This category is for a village or highway scattered with tree and houses with a few obstacles near

the UE. In this case,a h mis defined as equation (3.37) and C is defined as:

( ( ))
2
fc
C=−2 log −5.4 3.39
28

Rural areas:

This category is for rural areas with no tall trees or buildings in the propagation path, or a plot of

land cleared for 200-400m, e.g. farmland, rice fields and open fields. In this case, a h mis defined

as equation (3.37) and C is defined as:

2
C=−4.78 ( log ( f c ) ) +18.33 log ( f c ) −40.98 3.40

3.7 UE Model Design

110
The UE’s power consuming physical layer components are examined one by one. The purpose is

to determine how the components affect the total power consumption. Figure 3.1 illustrates the

LTE physical layer components and the UE model parameters. The envisioned UE model shall

depend on received (Rx) and transmit (Tx) power levels, uplink (UL) and downlink (DL) data

rate and RRC mode. In the following sections the parts in figure 3.5 are showed to determine if

and how they depend on the aforementioned parameters.

Figure 3.5 LTE UE physical layer

Transmit Baseband: In the LTE Tx baseband (BB) the main task is to turbo encode user data

with Forward Error Correction codes (FEC). Turbo coding relies on conventional encoding and

generates a bit stream with code rate 1/3. The turbo coding complexity scales linearly with the

amount of data to encode which is set by the Transport Block Size (TBS) i.e. the UL data rate,

but is independent of the UL Tx power.

Transmission Radio Frequency: In general the Radio Frequency (RF) will not depend on the UL

data rate, but when the modulation format is changed the Peak to Average Power Ratio (PAPR)

is affected. This entails the Power Amplifier (PA) will adjust its performance to comply with the

111
Tx emission requirements such as the Adjacent Channel Leakage Ratio (ACLR) and this may

affect the power consumption. The Tx RF will obviously on the UL Tx power. A single PA only

has one output power level where it achieves its maximum energy efficiency and therefore

researchers develop methods to increase the efficiency at other output levels. These include the

use of multiple PAs voltage supply and bias switching and the envelope tacking concept. The

Power Added Efficiency (PEF) is expected to be stepwise increasing with output power as each

of the methods are utilized.

Receive Radio Frequency: The RX RF power consumption is expected to be independent of DL

data rate, but it will depend on the DL Rx power level. The reason is that the RF contains Gain

Controls (GA) and Low Noise Amplifiers (LOA), which are used to obtain a certain signal level

at the Analog to Digital Converter (ADC). If the DL Rx power level is high the gain in the

aforementioned circuits can be reduced and they may be powered off to reduce the power

consumption.

Receive Baseband: The majority of the BB processing tasks are complexing, e.g. channel

estimation and equalization are independent of the DL data rate. To decode the received user

data the UE applies turbo coding, which is an iterative algorithm and the most computational

complex task in the digital BB. To support the high data rate of LTE a highly parallelized turbo

decoder architecture is required. The complexity and thus the power consumption scale linearly

with DL data rate.

UE Power Consumption Model: Base on the review of the four physical layer parts, the model is

defined as follows:

Ptot =midle . P+midle .{P con+ mTx . mRx . p Rx+Tx +¿

m Rx . [ P Rx + P RxRF ( S Rx ) + P RxBB ( R Rx ) +m 2 CW . P2 cw ]

112
 mTx . [ PTx + PTxRF ( ST x ) + PTxBB ( R Tx ) ] }[W ] 3.41

where P is the power consumption the scrip tot defines the total consumption, idle and con the

consumption in RRC idle and connected mode, RxRF and TxRF is the consumption of the RF

part in the Rx and Tx chains respectively, RxBB and TxBB is the consumption of BB parts and

2 cw is related to increased consumption when using two code words (CW) in DL. The

parameters P Rx, PTx and P Rx+Tx+ ¿are included to model the base power the Rx and Tx chains

consume when active. The logical variable m is the mode, which can be the RRC idle,

transmitting, receiving and indicate the use of 2 CW . The Rx and Tx power levels are designed

by S and R is the Rx and Tx data rate.

3.8 Power versus Energy

A battery is usually stated to have a certain capacity in ampere-hours at a specific voltage supply.

Therefore, it contains energy

U . I . t=P . t=E [J ] 3.42

Where:

U is the battery voltage supply

I is the current drawn from the battery

I is the time of use

P is the instantaneous power drawn from the battery

E is the energy drawn from the battery

Given the battery’s specifications, the first thing the designer will need to ensure is that the

amount of current drawn by the UE does not exceed a certain limit defined by the battery’s

internal resistance. If the current is to high it may lead to battery voltage drop, which may cause

the Power Management Unit (PMU) to power down the UE to protect it from unexpected

113
behavior. The current draw is typically at its peak when many components are powered “ON” at

once, and therefore the PMU can apply a power ON sequence. Provided that this fundamental

requirement is fulfilled designer can start optimizing the components to achieve longer battery

life.

One important metric, which relates the immediately power consumption to data transfer is the

Energy Efficiency (EE), which is defined as:

R b
EE= = 3.43
P P .t

Where:

EE is the energy efficiency [J/bit]

R is the data rate [bit/s]

P is the power consumption [W]

t is the time to transfer [s]

t is the number of bits transferred

Open loop parameters:

The setting of the UE transmit power, PTX , for the uplink transmission in a sub-frame of the

Physical Uplink Shared Channel is defined as:

PT =min ⁡¿
X
3.44

where:

Pmax , is the maximum allowed UE transmit power (class 3) of 23 dBm.

Po is a cell/UE-specific parameter signalled by Radio Resource Control (RRC). However, we

assume that Po is cell specific. Po is the eNodeB sensitivity.

M is the bandwidth of the UE uplink resource assignment, expressed in number of PRB for each

TTI.

114
α is the path loss compensation factor. It is a three-bit cell specific parameter in the LTE

standard, α belongs to the set {0, 0.4:0.1:1} signalled by the RRC.

P L is the downlink path loss estimated and calculated in the UE which is used to compensate the

received power at eNodeB in uplink power control.

Closed loop

δ m is a cell/UE-specific Modulation and Coding Scheme (MCS) parameter defined in 3GPP

specification for LTE and has been set to 0 in this work.

f (∆¿ ¿ i)¿ is a correction value provided by the Transmit Power Control (TPC) command. The

TPC commands are sent from an eNodeB after the OLPC has set the initial transmit power using

the desired 𝛼𝛼 and 𝑃𝑃𝑜𝑜 values to adjust the target SINR for different UEs. In CLPC,

f (∆¿ ¿ i)¿ can be used to improve system performance controlled from eNodeB.

In this work, focus only on open loop power control, hence the expression of UE transmit power

allocated by given user to a PRB simplifies to:

PT =min ⁡( Pmax , P o +10 lo g10 ( M ) + α∗P L )

X
3.45

When α = 0 (i.e. no compensation of path loss), no power control is present and all users will use

the maximum allowed transmit power. When α = 1 (full compensation of path loss), the UE

power is set to PT =Pmax . Withα = (0<α <1) (functional compensation of path loss) full
X

functional power control is thus obtained. Therefore the control of α is of primary importance in

this work.

Figure 3.6 highlights the flow diagram of the system.

115
 Start

Sort the clinic based on the number of patient. Each clinic serves in ascending order

Select the clinic with the smallest number of clinics served

Sort the used B.S that have connection to the selected clinic based on the total number
of clinics they can serve in ascending order followed by unused B.S in descending order

Select the first B.S in list to assign patient

Assign the patient of the selected clinic to the selected B.S and update the available
resources of the B.S

No All patients in the selected B.S are served?

Yes

All the clinics are served? No

Yes
Determine the minimum number of candidate nodes to place the processing server

Calculate the energy consumption resulting from placing the servers in the minimum number
of required node(s) considering the transmission of raw health data and feedback traffic

Increase the number of node required to host the server and calculate the energy consumption
resulting from this placement

Yes N node results in lower energy consumption? 116

No
Figure 3.6 highlights the flow diagram of the system.

3.9. PARAMETER SELECTIONS

We consider health monitoring application that differs in data rate. A low data rate ECG

monitoring application. It has been reported that cardiovascular disease (CVD) has emerged as

the top cause for mortality worldwide and is expected to reach 23.3 million by 2030. Therefore,

patients with postoperative atrial fibrillation (AF), a common cardiac case following cardiac

surgery are considered in the ECG monitoring application. Each patient will send a 30-second

ECG signal as recommended in which requires high processing capabilities for processing and

analysis

A. NETWORK LAYOUT

In this study, 35 clinics wards located at Federal medical centre Umuahia, according to

2022/2023 data are selected to monitor patients of the applications. The patients of a clinic are

considered to be located at the clinic due to the uncertainty in their precise locations. Potential

BSs to serve patients are selected by looking into the distance between the clinics and the BSs.

Note that the locations of clinics and BSs (i.e. latitude and longitude) refer to the actual locations

found in Umuahia, which had been obtained from Google Maps

In this work, LTE-M is considered to serve the health application with a coverage radius of less

than 11km.

117
Hence, patients could be served by a BS within 11km from their registered clinics. As for this

work, 315 BSs are located less than 11km from any clinic. The 26 nearest BSs to the clinics were

selected to serve patients to reduce the model complexity.

Table 3.1 presents the deduced total number of patients registered at each ward who have been

expected to experience postoperative AF.

WARD NO OF PATIENTS

MONITORED

USING ECG

1 20

2 18

3 13

4 23

5 6

6 29

7 13

8 13

9 8

10 10

11 14

12 11

13 27

14 25

15 21

118
16 12

17 44

18 28

19 6

20 68

21 20

22 15

23 16

24 30

25 32

26 8

27 24

28 4

29 16

30 16

31 10

32 9

33 6

34 19

35 14

B. TIME FOR PROCESSING AND ANALYSIS

1) ECG MONITORING APPLICATION

119
For the ECG monitoring application, a 30-second ECG signal is required to be sent to monitor

postoperative AF of cardiac surgical patients. This signal is retrieved from the Arrhythmia

database. Note that, the 30-second ECG signal offers accurate results for the analysis, as

recommended. Such 30-second of un-processed ECG signals have a volume of 252.8 kbits.

The ECG signals are processed using Pan Tompkins algorithm to extract heart rate and QRS

duration for further analysis. The calculation of the heart rate from the 30-second ECG signal is

based on the number of R waves within the 30 seconds and this number is multiplied by 2 to

obtain the heart rate in beats per minute. The QRS duration is obtained based on the time

between Q and S waves found in the ECG signal.

The PS selected in central cloud to process ECG signal is Intel Core i5-4460 with 3.2 GHz CPU

and 500 GByte hard drive. An experiment was conducted using MATLAB with a parallel

processing function to determine the correlation between time and number of patients for

processing and analysis of raw ECG data. This was carried out by performing the processing task

on the 30-second ECG signals generated by 10k to 50k patients in 10k steps. At each 10k step,

the processing operation was repeated 5 times to calculate the average time for the processing

duration. Note that, the 30-second ECG signals are made up of 1 ECG record repeated for all

patients. Also, note that the time to perform the processing using MATLAB consists of both the

time to submit the data for parallel processing and the time to run the algorithm.

C. EQUIPMENT POWER CONSUMPTION

The power consumption of all networking equipment and PS consist of an idle part and a linear

proportional part. The idle power of BS, PS, and content server are obtained from datasheets and

references while the idle power for the other networking devices was considered to be 90% of

120
the power consumption at maximum utilization. The maximum power consumption of the

networking equipment and the PS and their maximum capacity is given by the manufacturers. As

for FMC, the maximum capacity is considered as the summation of the maximum uplink

capacity, i.e. 1.25 Gbps and maximum downlink capacity, i.e. 2.5 Gbps, to obtain Eb. Note that,

the networking devices are shared by multiple applications while the considered PSs and

Ethernet switch are dedicated for the healthcare application. As discussed for the link capacity, in

this work we consider 0.3% of the idle power of the shared devices is contributed by our

healthcare applications while 0.42% for LTE-M BS. Note that, the LTE-M shares capacity,

antenna, radio, and hardware with the legacy LTE networks (20MHz). Due to this, the calculated

idle power of the BS (0.42%) contributed for healthcare applications is based on 7% allocation of

LTE-M network from the legacy LTE network (i.e. 1.4MHz/20MHz) and 6% [70] allocation of

healthcare application from the total M2M application supported by LTE-M network. Note that,

the 6% allocation refers to the estimated total number of RBs that is dedicated for healthcare

applications which gives 360 PRBs per second as there are numerous types of M2M applications

served by LTE-M. However, the maximum idle power is considered for the unshared devices.

Due to cooling, lighting and other overheads in the network, the total power consumed in a site is

higher than the power consumed by the communications and computing equipment. The ratio of

the total power consumed to the power consumed by the communications and computing

equipment is defined as the power usage effectiveness (PUE). PUE is used to describe the energy

efficiency of each site (core node site or building, cloud site or building or fog site).

A PUE of 1.5 is considered for IP over WDM, metro, and access networks. A PUE of 2.5 is

considered for small distributed clouds in this work. In addition, a PUE of 2.5 is set for fog.

121
This section present simulation UE uplink transmits power under open loop power control. The

simulations were performed using MATLAB/SIMULINK. To analyze the UE energy

consumption’s dependency on the path loss compensation factor ( α ) and eNodeB sensitivity ( Po )

are applied.

Figure 3.7 highlights the overall CDV detection system with the ECG signal selector acting as

input to the system.

filteredSignal
1
FilteredSignal
heartRate

num(z) num(z) 2
1 Differentiator u peak
1 1 HeartRate (bpm)

Bandpass Filter moving Avg window threshold 3

Peak
QRS Detection
4
Threshold

Figure 3.7: ECG signal processing unit

122
 Raw and filtered ECG signal
FilteredSignal
HeartRate (bpm)
ECG Source 1 1 1
raw Input Peak
HeartRate
Threshold
ECG Signal Selector Convert Sample Rate to 200Hz ECGSignalProcessing1

HeartRate, Peak and Threshold

Figure 3.8: Overall CDV detection system using ECG signal processing

The ECG signal source is distributed into different mean heart rate for the detection of

cardiovascular problems at various heart beat rate in bits per minutes, figure 3.9 shows the

implementation of the ECG signal source on simulink environment.

1
ecgdata360Hz_hrmean45

Mean Heart Rate = 45 bpm

2
ecgdata360Hz_hrmean60

Mean Heart Rate = 60 bpm

1
ECG Source
3
ecgdata360Hz_hrmean82

Mean Heart Rate = 82 bpm

4
ecgdata360Hz_hrmean160

Mean Heart Rate = 160 bpm

ecgdata360Hz_hrmean220 *, 5
Mean Heart Rate = 220 bpm

Figure 3.9: Simulink implementation of ECG signal source

123
Figure 3.10: Design Resultant Simulink Block

124
 REFERENCES

Abbas, M. Negar, Z. and Meysam, M. (2018). Energy Efficient Power Allocation for Device-to-
Device Communications Underlaid Cellular Networks Using Stochastic Geometry. pp 1 –
20.

Abrol, A. (2018). Energy Efficient Techniques in 5G Networks. International Journal of

Advanced Research in Computer and Communication Engineering, IJARCCE, Vol. 7,
Issue 1, 100 – 104.

Ahmed, A., Ikhlef, R,. Schober, K., and Mallik ,R. (2014). Power allocation for conventional and
buffer-aided link adaptive relaying systems with energy harvesting nodes. IEEE Trans.
Wireless Commun., 13(3), 1182-1195.

Alamouti, S. M. (1998). A simple transmit diversity technique for wireless communications.

EEE Journal on Selected Areas in Communications, 16(8), 1451-1458.
Andrae, A. S. and T. Edler, (2015) “On global electricity usage of communicationtechnology:
trends to 2030,” Challenges, 6(1), 117–157.

Andrews,J. G. Claussen, H. Dohler, M. Rangan, S. and Reed, M. C. (2012). “Femtocells: Past,

present, and future,” IEEE Journal on Selected Areas in Communications, 30(3), 497–508.

Araniti, G. Cosmas, J. Iera,A. Molinaro, A. Orsino,A. and Scopelliti, P. (2014).“Energy

efficient handover algorithm for green radio networks,” in Broadband Multimedia Systems
and Broadcasting (BMSB), 2014 IEEE International Symposium on. IEEE, 1–6.

Arnold, O. Richter, F. Fettweis, G. and Blume, O. (2010). “Power consumption modeling of

different base station types in heterogeneous cellular networks,” in FutureNetwork and
Mobile Summit, 2010. IEEE,. 1–8.

Auer, G. Blume, O. Giannini, V. Godor, I. Imran, M. Jading, Y. Katranaras, E. Olsson, M.

Sabella, D. Skillermark, P. and Wajda, W. (2012). D2.3: Energy efficiency analysis of the
reference systems, areas of improvements and target breakdown. INFSO-ICT-247733
EARTH, ver. 2.0, 2012. [Online]. Available: http://www.ict-earth.eu/

Auer,G. Giannini, V. Desset, C. Godor, I. Skillermark,P. Olsson,M..Imran, M. A Sabella, D.

Gonzalez, M. J. Blume,O. et al., (2011) “How much energy isneeded to run a wireless
network?” IEEE Wireless Communications, 18(5).

Baek, S. Son, J. J. and Choi, B. D. (2011). “Performance analysis of push-to-talk overieee 802.16
e with sleep mode and idle mode,” Telecommunication Systems, 47(3-4), 291–302.

125
Chen,J.-J. and Hu, C.-H. (2015). “Energy efficient dynamic network configuration intwo-tier
lte/lte-a cellular networks,” in Heterogeneous Networking for Quality,Reliability, Security
and Robustness (QSHINE), 2015 11th International Conference on. IEEE, 31–37.

Chen, L.-J. Sun, T. Chen, B. Rajendran,V. and Gerla, M. (2004). A smart decision model for
vertical hand-off," in The 4th ANWIRE International Workshop on Wireless Internet and
Reconfigurability, 2004.

Chen, T. Kim, H. and Yang,Y. (2010). “Energy efficiency metrics for green
wirelesscommunications,” in Wireless Communications and Signal Processing
(WCSP),2010 International Conference on. IEEE, 1–6.

Chen, X. Wang, H. Xiang, X. and Gao, C. (2014). “Joint handover decision and channel
allocation for lte-a femtocell networks,” in Game Theory for Networks(GAMENETS),
2014 5th International Conference on. IEEE, 1–5.

Chu X. and Sethu,H. (2014). “An energy balanced dynamic topology control algorithmfor
improved network lifetime,” in Wireless and Mobile Computing, Networkingand
Communications (WiMob), 2014 IEEE 10th International Conference on.

Cisco, C. V. N. I. (2014). “Global mobile data tra ffic forecast update, 2013–2018,” whitepaper,
2014.

Combes, R. Elayoubi, S. E. Ali, A. Saker,L. and Chahed, T. (2015) “Optimal onlinecontrol for
sleep mode in green base stations,” Computer Networks, 78, 140–151.

Cui,S. Goldsmith, A. J. and Bahai, A. (2005).“Energy-constrained modulation optimization,”

IEEE transactions on wireless communications, 4(5). 2349–2360.

Cui, S. Goldsmith, A. J. and Bahai,A. (2003). “Modulation optimization under

energyconstraints,” in Communications, 2003. ICC’03. IEEE International Conference on,
4. IEEE, 2805–2811.

Dampage, U. and Wavegedara,C. B. (2013). “A low-latency and energy e fficient

forwardhandover scheme for lte-femtocell networks,” in Industrial and Information
Systems (ICIIS), 2013 8th IEEE International Conference on. IEEE, 53–58.

Dang, X. Wang, J.-Y. and Cao, Z. (2012). “Mdp-based handover policy in wireless
relaysystems,” EURASIP Journal on Wireless Communications and Networking,vol. 2012,
no. 1, p. 358, 2012.

De Sanctis,M. Cianca, E. and Joshi,V. (2011). “Energy efficient wireless networkstowards

green communications,” Wireless Personal Communications, vol. 59,no. 3, pp. 537–552,

126
De Vuyst, S. De Turck, K. Fiems, D. Wittevrongel, S. and Bruneel, H. (2009). “Delayversus
energy consumption of the ieee 802.16 e sleep-mode mechanism,” IEEETransactions on
Wireless Communications, 8, 11.

Dou, Z. Zhao, Z. Jin, Q. Zhang, L. Shu, Y. and Yang, O. (2011).“Energy-e fficient rateadaptation
for outdoor long distance wifi links,” in Computer Communications Workshops
(INFOCOM WKSHPS), 2011 IEEE Conference on. IEEE, 271–276.

El Gamal, A. Nair, C. Prabhakar, B. Uysal-Biyikoglu, E. and Zahedi, S. (2002). Energy-e_cient

scheduling of packet transmissions over wireless networks," in INFOCOM 2002. Twenty-
First Annual Joint Conference of the IEEE Computerand Communications Societies.
Proceedings. IEEE, vol. 3. IEEE, 2002, pp. 1773{1782.

El Gamal, A. Nair, C. Prabhakar, B.. Uysal-Biyikoglu, E and S. Zahedi, (2002).“Energy-efficient

scheduling of packet transmissions over wireless networks,”in INFOCOM 2002. Twenty-
First Annual Joint Conference of the IEEE Computer and Communications Societies.
Proceedings. IEEE, vol. 3. IEEE, 1773–1782.

Elmirghani, J. M. Klein,T. Hinton, K. El-Gorashi, T. Lawey, A. Q andDong,X. (2014).

“Greentouch greenmeter core network power consumption modelsand results,” in Green
Communications (OnlineGreencomm), 2014 IEEE On line Conference on. IEEE, 1–8.

Fedrizzi, R. Gomez,K. Kandeepan, S. Rasheed, T. and Saradhi,C. V. (2014). Energy aware

routing in heterogeneous multi-hop public safety wireless networks," inCommunications
Workshops (ICC), 2014 IEEE International Conference on. IEEE, 2014, 218{224.

Fedrizzi, R. Gomez, K. Kandeepan,S. Rasheed,T. and C. V. Saradhi, (2014) “Energyaware

routing in heterogeneous multi-hop public safety wireless networks,” in Communications
Workshops (ICC), 2014 IEEE International Conference on.IEEE, , pp. 218–224.

Feng,D. Jiang,C. Lim, G. Cimini, L. J. Feng, G. and Li, G. Y. (2013). A survey of energy-
e_cient wireless communications," IEEE Communications Surveys &Tutorials, vol. 15, no.
1, pp. 167{178, 2013.

Feng, D. Jiang, C. Lim,G. Cimini, L. J.. Feng, G and G. Y. Li, (2013). “A survey ofenergy-
efficient wireless communications,” IEEE Communications Surveys & Tutorials, 15(1),
167–178,

Frenger, P. Moberg, P. Malmodin, J. Jading, Y. and G´odor, I. (2011). “Reducing

energyconsumption in lte with cell dtx,” in Vehicular Technology Conference
(VTCSpring), 2011 IEEE 73rd. IEEE, 1–5.

Gomez, K. Kandeepan, S. Vidal, M. M. Boussemart, V. Ramos, R. Hermenier, R.. Rasheed, T

Goratti, L. Reynaud, L. and Grace D. (2016). “Aerial base stationswith opportunistic links

127
 for next generation emergency communications,” IEEECommunications Magazine, vol.
54, no. 4, pp. 31–39.

Gomez, K. Sengul, C. Bayer,N. Riggio, R. Rasheed, T. and D. Miorandi, (2013).“Morfeo:

Saving energy in wireless access infrastructures,” in World of Wireless,Mobile and
Multimedia Networks (WoWMoM), 2013 IEEE 14th InternationalSymposium and
Workshops on a. IEEE, 1–6.

Guo, W. Devine, C. and Wang,S. (2014). “Performance analysis of micro unmannedairborne

communication relays for cellular networks,” in Communication Systems, Networks &
Digital Signal Processing (CSNDSP), 2014 9th International Symposium on. IEEE, 658–
663.

Hasan, Z. Boostanimehr, H. and Bhargava, V. K. (2011).“Green cellular networks: Asurvey,

some research issues and challenges,” IEEE Communications surveys &tutorials, vol. 13,
no. 4, pp. 524–540.

Holtkamp,H. Auer, G. Bazzi, S. and Haas,H. (2014) “Minimizing base station

powerconsumption,” IEEE Journal on Selected Areas in Communications, 32(2), 297–306.

Hoydis, J. ten Brink, S. and Debbah, M. (2013). “Massive MIMO in the UL/DL of cellular
networks: How many antennas do we need?” IEEE J. Sel. Areas Commun., vol. 31, no. 2,
pp. 160–171.

Hussein, Y. S. Ali, B. M. Rasid,M. F. A. Sali, A. and Mansoor, A. M. (2016). “A novelcell-

selection optimization handover for long-term evolution (lte) macrocellusingfuzzy topsis,”
Computer Communications, 73, 22–33.

Islam, N. (2017). Energy Efficiency in Wireless Communication. Ph.D thesis, School of

Engineering College of Science, Engineering and Health. RMIT University, May 2017.

Jallouli, K., Mazouzi, M., Diguet, J., Monemi, A., Jallouli, K., Mazouzi, M., Diguet, J., Monemi,
A., & Mimo-, S. H. (2022). MIMO-OFDM LTE System based on a parallel IFFT / FFT
on NoC-based FPGA To cite this version : HAL Id : hal-03546531 MIMO-OFDM LTE
System based on a parallel IFFT / FFT on NoC-based FPGA.
Jumira, O. and Zeadally, S. (2016).“Energy efficiency in cellular networks,” EnergyEfficiency
in Wireless Networks. 1–16.

Katikala, S. (2014).“Google project loon,” InSight: Rivier Acad. J,10(2).

Komali, R. S. MacKenzie, A. B. and Gilles, R. P. (2008). “E ffect of selfish node behavioron

efficient topology design,” IEEE Transactions on mobile computing, 7(9), 1057–1070,

128
Komali, R. S. Thomas, R. W. DaSilva, L. A. and MacKenzie,A. B. (2010).“Theprice of
ignorance: distributed topology control in cognitive networks,” IEEETransactions on
wireless communications, 9(4).

Kong, L. Wong, G. K. and Tsang, D. H. (2009). “Performance study and system optimization on
sleep mode operation in ieee 802.16 e,” IEEE transactions onwireless communications, 8,
9.

Langedem, T. (2012). “Reducing the carbon footprint of ict devices, platforms andnetworks,”
GreenTouch, Amsterdam, The Netherlands, Nov, 2012.

Li, C.-Y. Peng, C. Lu,S. and Wang,X. (2012). “Energy-based rate adaptation for802.11 n,” in
Proceedings of the 18th annual international conference on Mobilecomputing and
networking. ACM, 2012, pp. 341–352.

Li, C.-Y. Peng, C. Lu, S. and Wang,X. (2012). Energy-based rate adaptation for802.11 n," in
Proceedings of the 18th annual international conference on Mobilecomputing and
networking. ACM, 2012, pp. 341{352.

Li, F. Chen, S. Huang, M. Yin, Z. Zhang, C. and Wang, Y. (2015). “Reliable topologydesign in
time-evolving delay-tolerant networks with unreliable links,” IEEETransactions on Mobile
Computing, 14(6), 1301–1314,

Li, N. Hou, J. C. and Sha,L. (2005). “Design and analysis of an mst-based topologycontrol
algorithm,” IEEE Transactions on Wireless Communications, 4(3), 1195–1206,

Lin, D., Wang, Q., Min, W., Xu, J., and Zhang, J. (2020) A Survey on Energy-
EfficientStrategies in Static Wireless Sensor Networks. ACM Transactions on Sensor
Networks, 17 (1). 3. PP

Lipfert, Hermann (2007). MIMO OFDM Space Time Coding – Spatial Multiplexing, Increasing
Performance and Spectral Efficiency in Wireless Systems, Part I Technical Basis
(Technical report). Institut für Rundfunktechnik

M.-m. Chen, Y. Yang, and Z.-d. Zhong, “Location-based handover decisionalgorithm in lte
networks under high-speed mobility scenario,” in VehicularTechnology Conference (VTC
Spring), 2014 IEEE 79th. IEEE, 2014, pp. 1–5.

Marsan, M. A. Chiaraviglio, L. Ciullo, D. and M. Meo, (2013). “On the e ffectiveness ofsingle
and multiple base station sleep modes in cellular networks,” Computer Networks, 57(17),
3276–3290.

Marsan, M. A. Chiaraviglio, L. Ciullo, D. and Meo, M. (2009).“Optimal energy savings in

cellular access networks,” in Communications Workshops, 2009. ICCWorkshops 2009.
IEEE International Conference on. IEEE, 1–5.

129
Massive MIMO Hybrid Beamforming - MATLAB & Simulink. (n.d.).
Mavrakis, D. (2013).“Small cell market status,” Informa Telecoms&Media,

Mezghani, A Damak, N. and Nossek, J. A. (2010). “Circuit aware design of power-efficient

short range communication systems,” in Proc. Int. Symp. Wireless Commun. Systems
(ISWCS), 869–873.

Mezzavilla, M. Goyal, S. Panwar, S. Rangan, S. and Zorzi, M. (2016).“An mdp modelfor

optimal handover decisions in mmwave cellular networks,” in Networks
andCommunications (EuCNC), 2016 European Conference on. IEEE, 100–105.

Miao, G. (2013). “Energy-efficient uplink multi-user MIMO,” IEEE Trans. Wireless Commun.,
vol. 12, no. 5, pp. 2302–2313.

Micallef, G. Mogensen, P. and H.-O. Scheck, (2010) “Cell size breathing and possibilities to
introduce cell sleep mode,” in Wireless Conference (EW), 2010 European. IEEE, 111–115.

Micallef, G. Saker, L. Elayoubi, S. E. and H.-O. Sceck, (2012). “Realistic energy savingpotential
of sleep mode for existing and future mobile networks,” Journal ofCommunications, 7(10),
740–748.

Murthy,C. R. and Kavitha, C. (2012). “A survey of green base stations in cellularnetworks,”

International Journal of Computer Networks and Wireless Communications (IJCNWC),
2(2), 232–236.

Nahir, A. Orda,A. and Freund, A. (2014). “Topology design of communication networks: A

game-theoretic perspective,” IEEE/ACM Transactions on Networking, 22(2),. 405–414,

Nasser, N. Hasswa, A. and Hassanein,H. (2006). Hando_s in fourth generation

heterogeneousnetworks," IEEE Communications Magazine, 44(10), 96{103, 2006.

Nedevschi,S. Popa, L. Iannaccone, G. Ratnasamy,S. and Wetherall, D. (2008).

Reducingnetwork energy consumption via sleeping and rate-adaptation." in NsDI, 8, 323-
336.

Ngo, H. Larsson, E. and Marzetta, T. (2013). “Energy and spectral efficiency of very large
multiuser MIMO systems,” IEEE Trans. Commun., vol. 61, no. 4, pp. 1436–1449.

Nuggehalli, P. Srinivasan,V. and Rao, R. R. (2002).“Delay constrained energy

efficienttransmission strategies for wireless devices,” in INFOCOM 2002. Twenty-
FirstAnnual Joint Conference of the IEEE Computer and Communications
Societies.Proceedings. IEEE, vol. 3. IEEE,. 1765–1772.

Pan, J and Tompkins, W. (1985) A Real-Time QRS Detection Algorithm, IEEE Transactions on
Biomedical Engineering, 32(3): 230-236.

130
Pan, J. and Zhang, W. (2012). “An mdp-based handover decision algorithm in hierarchical lte
networks,” in Vehicular Technology Conference (VTC Fall), 2012IEEE. IEEE, 2012, pp.
1–5.

Peng, J. Hong, P.and K. Xue,(2014). “Stochastic analysis of optimal base stationenergy saving in
cellular networks with sleep mode,” IEEE Communications Letters, 18(4), 612–615.

Poloha, P., Wieser, V., Haring, J., & Majer, N. (n.d.). MATHEMATICAL MODELS OF MIMO
CHANNEL τ. 66–70.
Pons, X. Mesodiakaki, A. Gruet, C. Naviner, L. Adelantado, F. Alonso, L. andVerikoukis, C.
(2014). “An energy efficient vertical handover decision algorithm,” inGlobecom
Workshops (GC Wkshps), 2014. IEEE, 1145–1150.

Proakis,J. G. (1995). “Digital communications,” McGraw-Hill, New York,

Qi, J. and Aissa, S. (2012). “On the power amplifier nonlinearity in MIMO transmit
beamforming systems,” IEEE Trans. Commun., vol. 60, no. 3, pp. 876–887.

Rajabizadeh, M. and Abouei, J. (2015).“An efficient femtocell-to-femtocell handoverdecision

algorithm in lte femtocell networks,” in Electrical Engineering (ICEE),2015 23rd Iranian
Conference on. IEEE, 213–218.

Real-Time ECG QRS Detection - MATLAB & Simulink. (n.d.).

Reddy, H. Zhou, D. Hu, and P. Agrawal, (2013). “Hando ff management with adelay requirement
in femtocell networks,” in System Theory (SSST), 2013 45 thSoutheastern Symposium on.
IEEE, 119–123.

Rodoplu, V. and Meng,T. H. (2007). “Bits-per-joule capacity of energy-limited

wirelessnetworks,” IEEE Transactions on Wireless Communications, vol. 6, no. 3,

Rohde, S. and Wietfeld, C. (2012).“Interference aware positioning of aerial relays forcell

overload and outage compensation,” in Vehicular Technology Conference(VTC Fall), 2012
IEEE. IEEE, 1–5.

Sadig, M. H. M., & Mustafa, A. B. A. (2018). A Study on The Capacity of Massive MIMO
System with The Aid of MATLAB Simulink. 2(4), 22–25.
Saker, L. and Elayoubi, S. E. (2010).“Sleep mode implementation issues in green basestations,”
in Personal Indoor and Mobile Radio Communications (PIMRC),2010 IEEE 21st
International Symposium on. IEEE, 1683–1688.

131
Saker, L. Elayoubi, S. and H. Scheck, (2009). “System selection and sleep mode for energy
saving in cooperative 2g/3g networks,” in Vehicular Technology ConferenceFall (VTC
2009-Fall), 2009 IEEE 70th. IEEE, 1–5.

Saker, L. Elayoubi, S.-E. and Chahed,T. (2010). “Minimizing energy consumption viasleep
mode in green base station,” in Wireless Communications and NetworkingConference
(WCNC), 2010 IEEE. IEEE, 1–6.

Saker, L. S.-E. Elayoubi, Combes, R. and T. Chahed,(2012). “Optimal control of wakeup

mechanisms of femtocells in heterogeneous networks,” IEEE Journal onSelected Areas in
Communications, 30(3), 664–672.

Schurgers,CAberthorne, . O. and Srivastava,M. (2001). Modulation scaling for energyaware

communication systems," in Proceedings of the 2001 internationalsymposium on Low
power electronics and design. ACM, 2001, pp. 96{99.

Schurgers,C. and. Srivastava,M. B (2002). “Energy efficient wireless scheduling: Adaptive

loading in time,” in Wireless Communications and Networking Conference,2002.
WCNC2002. 2002 IEEE, vol. 2. IEEE, 706–711.

Schurgers, C. Aberthorne, O. and Srivastava, M. (2001). “Modulation scaling for energy aware
communication systems,” in Proceedings of the 2001 internationalsymposium on Low
power electronics and design. ACM, 2001, pp. 96–99.

Schurgers, C. and Srivastava,M. B. (2002) Energy e_cient wireless scheduling: Adaptiveloading

in time," in Wireless Communications and Networking Conference, 2002. WCNC2002.
2002 IEEE, vol. 2. IEEE, 2002, pp. 706{711.

Shang, D. Zhang, B. Yao, Z. and. Li, C (2014)“An energy e fficient localized topologycontrol
algorithm 371–377,

Shwetha, H. S., & Sathisha, R. N. (2015). The Performance Analysis of MIMO OFDM System
with Different M-QAM Modulation and Convolution Channel Coding. 4(04), 526–531.
Song, Y. Kong, P.-Y. and Han,Y. (2014). “Power-optimized vertical handover schemefor
heterogeneous wireless networks,” IEEE Communications Letters, vol. 18(2), 277–280.

Suleiman, K. E. Taha, A.-E. M. and H. S. Hassanein, (2016).“Handover-related selfoptimization

in femtocells: A survey and an interaction study,” Computer Communications, vol. 73, pp.
82–98, 2016.

Tolba, F. D. Tolba, C. and Lorenz, P. (2016). “Topology control by controlling mobilityfor

coverage in wireless sensor networks,” in Communications (ICC), IEEEInternational
Conference on. IEEE, 1–6.

132
Verd,_S., (2002). Spectral e_ciency in the wideband regime," IEEE Transactions on Information
Theory, vol. 48, no. 6, pp. 1319{1343.

Verdu´,S. (2002). “Spectral efficiency in the wideband regime,” IEEE Transactions

onInformation Theory, vol. 48, no. 6, pp. 1319–1343.

Wang, Y. Li, W. Yu, P. and Qiu, X. (2012). “Automated handover optimization mechanism for
lte femtocells,” in Communications and Networking in China (CHINACOM), 2012 7th
International ICST Conference on. IEEE, 601–605.

Wu, J. Bao, Y. Miao, G. and Z. Niu, (2014). “Base station sleeping and power controlfor bursty
traffic in cellular networks,” in Communications Workshops (ICC),2014 IEEE
International Conference on. IEEE, 837–841.

Wu, J. Zhang, Y. Zukerman, M. and E. K.-N. Yung, (2015).“Energy-e fficient basestations sleep-
mode techniques in green cellular networks: A survey,” IEEEcommunications surveys &
tutorials, vol. 17, no. 2, pp. 803–826, 2015.

Xenakis,D. Passas,N. Merakos,L. and Verikoukis,C. (2013). “Energy-e fficient

andinterference-aware handover decision for the lte-advanced femtocell network,” in
Communications (ICC), 2013 IEEE International Conference on. IEEE,2464–2468.

Xenakis, D. Passas, N. and Verikoukis, C. (2012). “A novel handover decision policy

forreducing power transmissions in the two-tier lte network,” in Communications(ICC),
2012 IEEE International Conference on. IEEE, 1352–1356.

Xenakis, D. Passas, N. and Verikoukis, C. (2013). “An energy-centric handover decision

algorithm for the integrated lte macrocell–femtocell network,” ComputerCommunications,
vol. 35, no. 14, pp. 1684–1694, 2012.

Xenakis, D. Passas, N. Merakos, L. and C. Verikoukis, (2015). “Advanced mobilitymanagement

for reduced interference and energy consumption in the two-tierlte-advanced network,”
Computer Networks, vol. 76, pp. 90–111, 2015.

Xu, M. Yang, Q. and. Kwak,K. S (2016). “Distributed topology control with lifetimeextension
based on non-cooperative game for wireless sensor networks,” IEEE Sensors Journal,
16(9), 3332–3342,

Xu, P. Fang, X. He, R. and Z. Xiang, (2013).“An e fficient hando ff algorithm basedon received
signal strength and wireless transmission loss in hierarchical cellnetworks,”
Telecommunication Systems, pp. 1–9, 2013.

Xu, Z. Li, G. Y. Yang, C. Zhang, S. Chen,Y. and Xu, S. (2012). “Energy-e fficientpower
allocation for pilots in training-based downlink ofdma systems,” IEEETransactions on
Communications, vol. 60, no. 10, pp. 3047–3058,
133
Yang, H. and. Marzetta, T. L (2013).“Total energy efficiency of cellular large scaleantenna
system multiple access mobile networks,” in Online Conference onGreen Communications
(GreenCom), 2013 IEEE. IEEE, 27–32.
Zarifzadeh, S. Yazdani, N. and Nayyeri, A. (2012). “Energy-e fficient topology controlin
wireless ad hoc networks with selfish nodes,” Computer Networks, 56(2), 902–914.
Zhang, H. Ma, W. Li, W. Zheng, W. Wen,X. and Jiang, C. (2011). “Signalling costevaluation of
handover management schemes in lte-advanced femtocell,” in Vehicular Technology
Conference (VTC Spring), 2011 IEEE 73rd. IEEE, 1–5.
Zhang, X. M. Zhang, Y. Yan, F. and Vasilakos, A. V. (2015). “Interference-based topology
control algorithm for delay-constrained mobile ad hoc networks,” IEEETransactions on
Mobile Computing, 14(4), 742–754,
Zhang, Y. Budzisz, L. Meo, M. Conte, A. Haratcherev, I. Koutitas, G. Tassiulas, L. Marsan, M.
A. and S. Lambert, (2013).“An overview of energy e fficient base station management
techniques,” in Digital Communications-Green ICT (TIWDC), 2013 24th Tyrrhenian
International Workshop on. IEEE, 1–6.
Zuckerberg,M. (2001). “Connecting the world from the sky,” 2014.[46] B. El-Jabu and R. Steele,
“Cellular communications using aerial platforms,”IEEE Transactions on Vehicular
Technology, 50(3), 686–700.

134
 Appendix 1: QRS Dectector based on the band-pass filtered ECG signal

function [heartRate,peak,threshold] = QRSDetection(u)

% Description:
% QRS Dectector based on the band-pass filtered ECG signal.
% The QRS detection rules reference the PIC based QRS detector
% implemented by Patrick S. Hamilton in picqrs.c
% from Open Source ECG Analysis (OSEA)
% http://www.eplimited.com/confirmation.htm
%
% INPUT:
% u : bandpass filtered and moving averaged ECG signal
%
% OUTPUTS:
% heartRate: detected heart rate of cardiovascular ailment(beat per minute)
% peak : detected local peaks (mainly R peaks)
% threshold: the detection threshold is calculated based on
% the mean estiamtes of the average QRS peak and
% the average noise peak.

persistent FS QRSDelay
if isempty(FS)
FS = 200; % Hz, samples per second
QRSDelay = zeros(size(u), 'like', u);
end

heartRate = uint16(zeros(size(u)));
peak = zeros(size(u), 'like', u);
threshold = zeros(size(u), 'like', u);
% Prevents detections of peaks smaller than 150 uV.
MIN_PEAK_AMP = cast(7/2^12, 'like', u(1));

for idx = 1:length(u)

% For a frame of input signal, process one sample at a time
peak(idx) = findpeak(u(idx));

if(peak(idx) < MIN_PEAK_AMP)

peak(idx) = 0 ;
end

% returns adjusted peak

[QRSDelay(idx), threshold(idx), peak(idx)] = qrs_det(peak(idx));
if QRSDelay(idx) ~= 0
% beat per min = samplesPerSec * 60sec/min / samples per beat_oneQRSDalay
heartRate(idx) = int16(60 * FS / QRSDelay(idx));
end

end
end% function QRSDetection

function pk = findpeak(datum)

135
 persistent maxPeak lastDatum timeSinceMax
pk = cast(0,'like',datum);
if isempty(maxPeak)
maxPeak = cast(0,'like',datum);
lastDatum = cast(0,'like',datum);
timeSinceMax = uint16(0);
return
end

if timeSinceMax > 0
timeSinceMax(:) = timeSinceMax + 1;
end

if((datum > lastDatum) && (datum > maxPeak)) % rising slope

maxPeak(:) = datum ;
if(maxPeak > 2) % reset timeSinceMax
timeSinceMax(:) = 1 ;
end
elseif(datum < bitsra(maxPeak,1)) % middle of falling slope
% Less than half the peak height
pk(:) = maxPeak ;
maxPeak(:) = 0 ;
timeSinceMax(:) = 0 ;
elseif(timeSinceMax > MS95)
pk(:) = maxPeak ;
maxPeak(:) = 0 ;
timeSinceMax(:) = 0 ;
end
lastDatum = datum ;
end

function [QRSDelay,thisThreshold, pk] = qrs_det(pk)

persistent count
persistent qpkcnt
persistent lastQRSDelay
persistent threshold
persistent preBlankCnt
persistent tempPeak
persistent initMax
persistent sbpeak
persistent sbcount

if isempty(lastQRSDelay)
count = int16(0);
qpkcnt = int16(0);
lastQRSDelay = int16(0);

threshold = cast(8/2^12, 'like', pk);

tempPeak = cast(0, 'like', pk);
initMax = cast(0, 'like', pk);
preBlankCnt = int16(0);
sbpeak = cast(0, 'like', pk);
sbcount = int16(MS1650);

136
 end

% there can only be one QRS complex in any 196ms (306bpm) window
if (pk==0) && (preBlankCnt==0)
pk(:) = 0;
elseif (pk==0) && (preBlankCnt~=0)
% if we have held onto a peak for 196ms, pass it on for eval
preBlankCnt = preBlankCnt - 1;
if (preBlankCnt == 0)
pk(:) = tempPeak;
else
pk(:) = 0;
end
elseif (pk~=0) && (preBlankCnt==0)
% if there has been no peak for 196ms, save this one and start
% counting
tempPeak = pk;
preBlankCnt = MS196;
pk(:) = 0;
else% (pk~=0) && (preBlankCnt~=0)
% if we were holding a peak, but this one is bigger,
% save it and start counting to 196ms again
if (pk > tempPeak)
tempPeak = pk;
preBlankCnt = MS196;
pk(:) = 0;
else
preBlankCnt = preBlankCnt - 1;
if (preBlankCnt == 0)
pk(:) = tempPeak;
else
pk(:) = 0;
end
end
end

count(:) = count + 1;

% Initialize the QRS peak buffer with the first eight

% local maximum peaks detected
if qpkcnt < 8
if pk > 0
UPDATEQ = true;
threshold = updateQN(pk, UPDATEQ);
qpkcnt = qpkcnt + 1;
if pk > initMax
initMax = pk;
end
if qpkcnt == 8
count(:) = 0;
end

end

else% qpkcnt >= 8

137
 if pk > threshold
UPDATEQ = true;
threshold = updateQN(pk, UPDATEQ);
lastQRSDelay(:) = count;
count(:) = 0;
sbpeak(:) = 0;
elseif pk ~= 0
% if peak is below threshold
UPDATEQ = false;
threshold = updateQN(pk, UPDATEQ);

% If no QRS has been detected within 1.5 R-to-R intervals,

% there was a peak that was larger than half the
% detection threshold, and the peak followed the preceding
% detection by at least 360ms, classify that peak as a QRS
% complex

% persistent sbpeak = 0; sbcount = MS1650

if (count >= MS360) && (pk > sbpeak)
sbpeak = pk;
sbloc = count;
threshold(:) = bitsra(threshold,1);

if (count > sbcount) && (sbpeak > threshold)

sbcount = updateRR(sbloc);
lastQRSDelay(:) = count;
% threshold = threshold >> 1;
count(:) = 0;
sbpeak(:) = 0;
end
end

end
end
QRSDelay = lastQRSDelay;
thisThreshold = threshold;
end

function det_thresh = updateQN(x, Qflag)

persistent QSum
persistent NSum
persistent Qz Nz nQ nN

if isempty(NSum)
nQ = uint8(0);
nN = uint8(0);
Qz = zeros(8,1,'like',x);
Nz = zeros(8,1,'like',x);
QSum = cast(0, 'like', x);
NSum = cast(0, 'like', x);
end

det_thresh = cast(0, 'like', x);

if Qflag %&& x ~= cast(0,'like',x)

138
 Qz(nQ+1) = x;
nQ = bitand(nQ+1,uint8(7));
QSum(:) = sum(Qz);
else
Nz(nN+1) = x;
nN = bitand(nN+1,uint8(7));
NSum(:) = sum(Nz);
end
det_thresh(:) = QSum-NSum ;
det_thresh(:) = NSum + bitsra(det_thresh,1) - bitsra(det_thresh,3);
det_thresh(:) = bitsra(det_thresh, 3);
end

function sbcount = updateRR(x)

persistent RRSum
persistent zR
persistent nR

if isempty(RRSum)
nR = uint8(0);
zR = cast(MS1000*ones(8,1, 'int16'),'like',x);
RRSum = cast(bitsll(MS1000,3), 'like', x);
end

sbcount = cast(0, 'like', x);

zR(nR+1) = x;
nR = bitand(nR+1,uint8(7));
RRSum(:) = sum(bitsra(zR,3));

sbcount(:) = RRSum + bitsra(RRSum, 1); % = 1.5RRSum

sbcount(:) = bitsra(sbcount, 3);
sbcount(:) = sbcount + MS80; % WINDOW_WIDTH
end

function n = MS80
% The number of samples in 95 milli-seconds, where FS is the sample rate
FS = 200;
n = int16(80 * FS / 1000);
end

function n = MS95
% The number of samples in 95 milli-seconds, where FS is the sample rate
FS = 200;
n = int16(95 * FS / 1000);
end

function n = MS196
% The number of samples in 196 milli-seconds, where FS is the sample rate
FS = 200;
n = int16(196 * FS / 1000);
end

function n = MS360
% The number of samples in 360 milli-seconds, where FS is the sample rate

139
 FS = 200;
n = int16(360 * FS / 1000);
end

function n = MS1000
% The number of samples in 1000 milli-seconds, where FS is the sample rate
FS = 200;
n = int16(1000 * FS / 1000);
end

function n = MS1650
% The number of samples in 1650 milli-seconds, where FS is the sample rate
FS = 200;
n = int16(1650 * FS / 1000);
end

140