Content uploaded by Muhammad Tahir

Author content

All content in this area was uploaded by Muhammad Tahir on Jun 16, 2015

Content may be subject to copyright.

Noise Filtering, Channel Modeling and Energy

Utilization in Wireless Body Area Networks

B. Manzoor, N. Javaid, A. Bibi, $Z. A. Khan, M. Tahir

Department of Electrical Engineering, COMSATS

Institute of Information Technology, Islamabad, Pakistan.

$Faculty of Engineering, Dalhousie University, Halifax, Canada.

Abstract—Constant monitoring of patients without disturbing

their daily activities can be achieved through mobile networks.

Sensor nodes distributed in a home environment to provide

home assistance gives concept of Wireless Wearable Body Area

Networks. Gathering useful information and its transmission

to the required destination may face several problems. In this

paper we ﬁgure out different issues and discuss their possible

solutions in order to obtain an optimized infrastructure for the

care of elderly people. Different channel models along with their

characteristics, noise ﬁltering in different equalization techniques,

energy consumption and effect of different impairments have

been discussed in our paper. The novelty of this work is that

we highlighted multiple issues along with their possible solutions

that a BAN infrastructure is still facing.

Index Terms—Health, Care, Monitoring, Wireless, Body, Area,

Sensor, Networks

I. INTRODUCTION

Global trends in growth of elderly people attracts re-

searchers to formulate wireless health monitoring systems.

The system use sensors for continuous monitoring of patients

with less interaction to doctors. Wearable Wireless Body

Area Networks (WWBAN) gained a tremendous attention in

providing such facilities to keep in view the elderly age’s,

as well as giving proper weight to the economical issues.

Moreover, they are providing a reasonable platform to the

caregivers in recognition of patterns of diseases through their

databases.

The data acquired by the central device is transmitted

to the medical servers or to the care givers through wire-

less network. This communication have many issues to be

addressed. 1) Channel models, 2) effect of different inter-

ferences/disturbances, 3) noise ﬁltering and 4) energy con-

sumption are some of the issues which have been addressed

separately and there are a number of possible solutions are

proposed for these issues.

Proper estimation and equalization of transmitted signal is

a key concern in wireless communication. For this purpose,

different channel models are used which describes different

issues like; path loss, amplitude distortion, clustering and inter

arrival-time characteristics for proper estimation of the signal.

Signal can be degraded by Additive White Gaussian Noise

(AWGN). Therefore noise ﬁltering is also required in different

equalization schemes for better performance.

Simultaneous communication of sensors yields in different

types of interferences and disturbances (ISI, MUI, and NOISE)

which affect strongly the performance of received signal.

These disturbances emerge to be a great hurdle in signal

reception and can destroy partially or fully the information

being sent to the destination. Mean while, retransmission of

signals for proper data acquisition can utilize more energy

which is not suitable in any case. Energy utilization in wireless

BAN is a critical issue due to its infrastructure.

II. RELATED WORK AND MOTIVATION

Rapid advancements in technology devices many possi-

bilities to address routine issues and hence making life as

simpler and cureable as it could be. WWBAN introduces a

unique solution for the cure of elderly peoples throughout the

world by continuous monitoring of the patients anywhere and

anytime . Although it’s a broad domain. However, there is a

lot, to be discussed and optimized for making this approach

more effective and goal oriented.

In WWBAN different sensors are working simultaneously

to attain information to transmit it to the desired destination.

Energy utilization in wireless transmission are the main issues

which should be given equal importance to extend the life of

sensors. In order to extract useful information in the presence

of different interferences e.g. Inter Symbol Interferences (ISI),

Multi User Interference (MUI), and noise certain algorithms

have been proposed. MUD receivers provide an appropriate

platform to mitigate the effect of MUI caused by wireless

nodes which are communicating asynchronously [1]. Thus,

energy utilization for retransmission is optimized.

Improved performance of the system by allowing nodes to

use Medium Access Control protocol (MAC) among different

modulation schemes gave birth to the concept of novel Link

Adaptation (LA) strategy [2]. By considering the effects like

large SNR and low SIR data rates can be adapted dynamically

by reducing the time required for transmission and conse-

quently the interference caused by it [2].

Full bandwidth utilization of the channel without transmis-

sion of the training signal presents an idea of blind equalization

scheme. Constant Modulus Algorithm (CMA) emerges to be

a suitable blind wireless channel equalizer. The equalizer is

used to remove ISI which is produced by dispersive channels

[3]. Further studies states that use of dither input signal to the

equalizer by a distributed signal before a sign operation being

applied can be more goal oriented.

Fig 1 deﬁnes the hierarchy of the Wireless Body Area Net-

works (WBANs), its related issues and some of the possible

2012 IEEE 14th International Conference on High Performance Computing and Communications

978-0-7695-4749-7/12 $26.00 © 2012 IEEE

DOI 10.1109/HPCC.2012.264

1746

2012 IEEE 14th International Conference on High Performance Computing and Communications

978-0-7695-4749-7/12 $26.00 © 2012 IEEE

DOI 10.1109/HPCC.2012.264

1754

Fig. 1. Issues and Possible Solutions in WWBAN Infrastructure

solutions. The information obtained from the sensor nodes

is transmitted to the personal digital assistant (PDA). This

information is then transmitted to the care givers or medi-

cal server for continuous monitoring of the patients through

internet for further processing. The transmission and reception

of this information through any medium may ﬁnd certain

problems like MUI, ISI and noise which can be resolved

through different techniques like by using MUD receivers,

DES-CMA and LA techniques. In our work we present an

overview of different issues which a BAN infrastructure still

have to deal with. We highlighted those issues along with some

of the possible solutions.

III. NOISE FILTERING

This section describes noise ﬁltering in different equaliza-

tion techniques for WBAN. We discuss and compare noise

ﬁltering in three different selected scenarios; 1) MUD recievers

2) Novel Link Adaptation, 3) Blind Equalization.

A. Noise Filtering In Multiuser Detection Based Recievers

Signal are commonly effected by InterSymbol Interference

(ISI) and Additive White Gaussian Noise (AWGN). ISI is

minimized through equalization, whereas, noise is still a

problem [1].

𝑟=𝑠∗ℎ+𝑛(1)

where, 𝑟=recived signal, 𝑠=transmitted signal, ℎ=channel re-

sponse and 𝑛=noise. Equalizer parameters can be calculated

using Wiener-hopf equation, which is presented in [1] as:

𝑤=Υ

𝑟𝑟Γ−1𝑟𝑟 (2)

There is a tradeoff between ISI and noise. Two types of

receivers are deﬁned in [1];

A) Linear Multi User Detector based MUD receivers

B) Non-linear Multi User Detector based MUD receivers

1) Linear MUD Receivers: In MUD based receivers, there

is only ﬁxed equalizer parameters [1]. Therefore, these are not

suitable for fast varying channels. By implementing Minimum

Mean Square Error (MMSE), ISI can be compensated. How-

ever, effect of noise is still there. Linear MUD tries to mitigate

the effect of MUI and ISI. Whereas, it does not distinguish

between noise and MUI. Therefore, another approach named

as non-linear MUD is introduced.

2) Non-linear MUD Receivers: In such MUD based re-

ceivers there is no ﬁxed equalizer parameter. This is because

DFE is used to adjust equalizer parameter according to de-

tected signal quality. Hence, it is more suitable for fast varying

channels. Noise becomes less problem when someone try to

completely remove ISI (zero forcing) that enhances noise,

effect of noise is catered by implementing Minimum Mean

Square Error (MMSE) criteria through Non Linear MUD

Receivers.

B. Noise Filtering In Link Adaptation For IEEE 802.15.4

As, it is discussed in previous section that noise ﬁltering is

a problem in interference environment. Link Adaptation is a

technique which is used in [2] to minimize the ISI. The basic

idea behind this technique is discussed in [2] which is based

on, the selection of modulation technique according to channel

quality. However, by increasing order of modulation high data

rates can be achieved. In wireless environment, degradation is

caused due to multi-path. Channel quality depends upon Signal

to Noise Ratio (SNR) and Signal to Interference Ratio (SIR).

If SNR and SIR is low then channel is not good. Therefore,

lower order modulation is chosen, however, this selection leads

to increase ISI. If SNR is below some threshold then lower

order modulation is chosen which causes decrease in data rate.

Otherwise, higher order modulation is selected which causes

increase in data rate. If S/N is above some threshold and Packet

17471755

0 5 10 15 20 25 30 35 40

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

generated random sequence

bit index

binary value

(a)

−5 0 5

−5

−4

−3

−2

−1

0

1

2

3

4

5

Quadrature

In−Phase

Received Signal

Received Signal

Signal Constellation

(b)

−5 0 5

−5

−4

−3

−2

−1

0

1

2

3

4

5

Quadrature

In−Phase

Received Signal

Received Signal

Signal Constellation

(c)

Fig. 2. Performance Evaluation of Channel Models in the Presence of AWGN

Delivery (P.F) is below threshold then higher order modulation

is avoided.

We compare performance evaluation of channel models

in the presence of AWGN. For this purpose Quadrature

Amplitude Modulation modulation (16-QAM) scheme along

with AWGN channel for the transmission of data stream is

used. The general idea is to change modulation or coding

schemes according to response of channel. The response can

be addressed through SNR. The channel is considered ’bad’

when SNR is low, due to path loss, shadowing, or fading

problems. Whereas considered ’good’ when SNR is high. The

ratio of bit energy to noise power spectral density is set to

10dB and is then converted to corresponding SNR. A random

bit is generated, as shown in ﬁg 2a and then converted in to

random symbols that have to be transmitted through AWGN

channel.

The system Bit Error Rate (BER) is computed using 16-

QAM modulation scheme, as shown in ﬁg 2b. Moreover, the

effect of changing S/N on the quality of received signal is

showninﬁg2c.

C. Noise Filtering in Blind Equalization

For equalization of biological signals using adaptive Blind

Equalization Algorithm called as Constant Modulus Algorithm

(CMA) [3] is not suitable for noise ﬁltering. Because, training

sequence is not sent with information. Therefore, FEC cannot

be implemented. All the effect of multipath and noise is

on actual information. The detection of signal is based on

stochastic information of signal that is stored at receiver. In

case of blind equalization, if receiver does not know about

channel response then noise ﬁltering becomes difﬁcult. Setting

equalizer parameters periodically according to errors received

in previous is discussed in [3] minimizes noise. If we go

towards zero forcing solution than it leads to enhance noise.

Adaptive algorithm is deﬁned in [3] as:

𝑓(𝑛+1)=𝑓(𝑛)+𝜇𝑟(𝑛)𝜓𝑑𝑠𝑒−𝑐𝑚𝑎(𝑦(𝑛)) (3)

where, 𝑓(𝑛+1)=next equalizer parameter, 𝑓(𝑛)=current equal-

izer parameter, 𝜇=step size and 𝑟(𝑛)=received signal.

IV. CHANNEL MODELING

In this section, we discuss different channel models and

characteristics of these models for wireless communication

in different standards of WBAN. These models are used

for estimation and equalization of transmitted signals. We

compare and discuss three channel models for WBAN.

A. Channel Model for IEEE 802.15.4a

Channel model for MUD based receivers is recently pro-

posed for IEEE 802.15.4a is given in [1]. One of the rea-

sons for ISI is channel memory. Traditional receiver (DFE,

MMSE, MLSE) knows about channel characteristics. Analysis

shows that electromagnetic propagation at UWB range (3.1-

10.6) GHz is negligible through human body. Transmitted

signals reaches at receiver in two different ways: 1: Around

human body, 2: Through reﬂection from surrounding. There-

fore, WBAN channel posseses signiﬁcantly different path

loss, amplitude distortion, clustering and inter arrival-time

characteristics. We summarize simulation results for WBAN’s

channel model as: (A) For an outdoor channel (Outdoor-

BAN), measurements [1] indicate that there are always two

17481756

clusters of multipath components due to the initial wave

diffracting around the body, and reﬂections from the ground.

Therefore, number of clusters is always two and does not

need to be deﬁned as a stochastic process as deﬁned for

other scenarios. Moreover, depending upon exact position of

transmitter on human body intercluster arrival time distribution

is deterministic. (B) A human wearing WBAN devices in

an indoor environment is deﬁned as an Indoor channel. Two

deterministic clusters due to stochastic nature of multipath

(reﬂections from the indoor environment) should be taken into

account, as discussed in [1]. Channel model corresponding to

the propagation for outdoor environment [1]:

ℎ𝑏𝑜𝑑𝑦(𝑡)=

∞

𝑘=0

𝛽𝑘𝑒𝑥𝑝(𝑗𝜙𝑘)𝛿(𝑡−△𝑘)(4)

where, 𝛽𝑘is the 𝑘𝑡ℎ element of vector Y, 𝜙𝑘is uniformly

distributed random variable and △is size of bin.

Model parameters are taken for ’front’, ’side’ and ’back’

positions of receiver on human body. Reﬂections from ground

introduces a delay 𝜏𝑔𝑟𝑜𝑢𝑛𝑑 in channel model and thus, overall

channel model is given as :

ℎ𝑔𝑟𝑜𝑢𝑛𝑑 (𝑡)=

∞

𝑘=0

𝛽𝑘𝑒𝑥𝑝(𝑗𝜙𝑘)𝛿(𝑡−△𝑘−𝜏𝑔 𝑟𝑜𝑢𝑛𝑑)(5)

By assuming that reﬂections from ground are uncorrelated

with components diffracting around body, then complete chan-

nel response and is represented in [1] as:

ℎ𝑜𝑢𝑡𝑑𝑜𝑜𝑟−𝐵𝐴𝑁(𝑡)=ℎ𝑏𝑜𝑑𝑦 (𝑡)+ℎ𝑔𝑟 𝑜𝑢𝑛𝑑(𝑡)(6)

For an Indoor-BAN channel, reﬂections from surroundings

create multipath. Multipath distribution is stochastic in nature

therefore, clusters are modeled using Poisson or Weibull dis-

tributed random variables. Whereas, intra-cluster distribution

is dense, so multipath arrivals within a cluster are assumed

to be uniform [1]. The channel model due to reﬂections from

surroundings is given in [1] as:

ℎ𝑟𝑒𝑓 (𝑡)= 𝑋

√𝐸

∞

𝑙=0

∞

𝐾=0

𝛽𝑙,𝐾𝑒𝑥𝑝(𝑗𝜙𝑙,𝑘)𝛿(𝑡−△([𝜏𝑙/Δ] + 𝑘)) (7)

20𝑙𝑜𝑔10(𝛽𝑙,𝑘)=Γ

𝜏𝑙+Υ(𝜏𝑙+Δ

𝑘)+𝜎Γ𝑛𝑙+𝜎Υ𝑛𝑘(8)

𝑝(𝜏𝑙/𝜏𝑙−1) = 1

𝛽𝑒−(𝜏𝑙−𝜏𝑙−1/𝛽)(9)

where, 𝛽𝑙,𝐾represents the amplitude in 𝑘𝑡ℎ bin and in 𝑙𝑡ℎ

cluster modeled in [1], 𝜏𝑙is the cluster arrival time modeled

using a Poisson distribution, 𝛽represent the mean arrival rate.

Average energy of the clusters decays exponentially at a rate

of Γ𝑑𝐵/𝑛𝑠 and the terms within the cluster decay at the rate

of 𝛾𝑑𝐵/𝑛𝑠. Lognormal fading in the clusters and within the

clusters is modeled using 𝜎Γand 𝜎𝛾, respectively. Here, 𝑛𝑙

and 𝑛𝑘represent uncorrelated normal random variables with

unit mean and unit variance. 𝜙𝑙,𝑘are uniformly distributed

random variables and Δis bin size [1].

If we assume that reﬂections from indoor environment and

ground are uncorrelated with components diffracting around

the body, then complete channel response from [1] is:

ℎ𝑖𝑛𝑑𝑜𝑜𝑟−𝐵𝐴𝑁(𝑡)=ℎ𝑏𝑜𝑑𝑦 (𝑡)+ℎ𝑔𝑟 𝑜𝑢𝑛𝑑(𝑡)+ℎ𝑟𝑒𝑓 (𝑡)(10)

B. Channel Model for IEEE 802.15.4

Channel model that is used for a proposed novel Link Adap-

tation (LA) strategy [2], where nodes select their modulation

schemes according to the experienced channel quality and

level of interference. The general idea is to change modulation

and/or coding scheme in such a way that it lowers bit rates

(more robust modulation/coding when the overall bandwidth

is set) in case of ”bad” channel conditions. Whereas, high bit

rates when the channel is ”good”. The bad channel signiﬁes

low SNR, due to path loss, shadowing or fading problems,

or when SIR is low. In both cases, the use of a more robust

modulation/coding is supposed to reduce probability of error.

The channel loss, denoted as 𝐴, is modeled according to 3GPP

proposal for WBANs, as discussed in [2]:

𝐴(𝑑)[𝑑𝐵]=𝐴0+10𝑛log(𝑑/𝑑0)+𝑠(11)

where 𝑑is the distance between transmitter and receiver,

𝐴0deﬁnes path loss in 𝑑𝐵 at a reference distance, 𝑑0,

𝑛speciﬁes path loss exponent and 𝑠represents a random

Gaussian variable having zero mean and standard deviation

𝜎.

Algorithm 1 Novel Link Adaptation

1: 𝑆𝑒𝑡 𝑜𝑓 𝑛𝑜𝑑𝑒𝑠 𝑖𝑛 𝐵𝐴𝑁 ←𝑁

2: 𝑅𝑒𝑐𝑒𝑖𝑣𝑒𝑑 𝑏𝑒𝑎𝑐𝑜𝑛 𝑝𝑜𝑤𝑒𝑟 ←𝐵𝑅𝑃

3: 𝐹 𝑎𝑖𝑙𝑢𝑟𝑒 𝑃𝑟 𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑟 𝑒𝑐𝑒𝑖𝑣𝑒𝑑 𝑝𝑎𝑐𝑘 𝑒𝑡 ←𝑃𝐹

4: 𝑇 ℎ𝑟𝑒𝑠ℎ𝑜𝑙𝑑 𝑓𝑜𝑟 𝑓𝑎𝑖𝑙 𝑢𝑟𝑒 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙 𝑖𝑡𝑦 𝑝𝑐𝑘𝑒𝑡 ←𝑇𝐻

𝑃𝐹

5: 𝐵𝑖𝑡 𝑟𝑎𝑡𝑒 ←𝐵𝑅

6: 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑠𝑒𝑛𝑠𝑜𝑟 𝑛𝑜𝑑𝑒𝑠 𝑖𝑛 𝐵𝐴𝑁 ←𝑛

7: for all 𝑛∈𝑁do

8: 𝐹𝑖𝑛𝑑 𝐵

𝑅𝑃 𝑓𝑜𝑟 𝑒𝑣𝑒𝑟𝑦 𝑛;𝐵𝑅𝑃 (𝑛)

9: 𝑆𝑒𝑡 𝐵𝑅𝑃 (𝑛)←𝑃𝑛

10: if 𝑃𝑛−SNR <𝑇𝐻

𝑆𝑁𝑅 then

11: 𝐵𝑅=𝐵−−

𝑅

12: end if

13: if 𝑃𝐹>𝑇𝐻

𝑃𝐹 then

14: 𝐵𝑅=𝐵++

𝑅

15: else

16: 𝐵𝑅=𝐵𝑅

17: end if

18: end for

C. Geometrical Based Hyperbolically Distributed Scatters

channel model

Geometrical-Based Hyperbolically Distributed Scatterers

(GBHDS) channel model is discussed in [3], for macro cell

environments is considered in this section. A comprehensive

17491757

study of this model proved to be more realistic than other

models, as tested against practical data [3]. GBHDS for

macro cell environment channel model assumes that scatterers

are arranged within a circle of radius R around mobile.

The distances 𝑟between Mobile Station (MS) and scatterers

are distributed according to hyperbolic Probability Density

Function (pdf). The Geometrical Scatterers Density Function

(GSDF) for this model, 𝑓𝑟(𝑟),isgivenin[3]as:

𝑓𝑟(𝑟)={𝑎

tanh(𝑎𝑟)cosh

2(𝑎𝑟)(12)

where, 𝑅is the radius of the circle enclosing the scatterers, and

applicable values of 𝛼lie in interval (0, 1) [3]. GBHDS results

show that there is a good match between model result and

measurement data which is reported for Outdoor environment.

Fig 4 shows a comparison of the results for the Direction

of Arrival (DOA) pdf for the GBHDS channel model, DOA

pdf for the Geometrical Based Single Bounce Macro Cell

(GBSBM) channel model and Gaussian Scatterer’s Density

(GSD) channel model. From ﬁg 4 it is clear that there is a

good match between the GBHDS channel model result and

measurement data. Whereas, GSD model fails in the DOAs

close to Line-of-Sight (LOS).

(a)

Fig. 3. D.O.A Comparison for GBHDS, GBSBM AND GSD Channel Models

V. DATA BROADCAST IN IEEE 802.15.4 AND ZIGBEE

Energy consumption is an important issue in wireless sensor

network and same is the case associated with WBAN. Multiple

sensor nodes are working simultaneously for acquisition of

useful information from the patients in WBAN. The key idea

in this regard is to use IEEE 802.15.4 standard along with

ZigBee network which is considered to be a low data rate and

low cast infrastructure [4]. IEEE 802.15.4 deﬁnes two types

of devices: Full Function Device (FFD) which can serve as a

coordinator or a regular device and a Reduced Function Device

(RFD) which is a simple device that associate and commu-

nicate only with an FFD. Based on IEEE802.15.4, ZigBee

speciﬁes the standards for network and application. ZigBee

network layer is responsible for assigning addresses and builds

a hierarchical tree topology [4]. Network parameters such as

NOISE FILTERINGCHANNEL MODELING

INDOOR

BODY+

GROUND

LINEAR

MUD

MINIMUM NOISE

FILTERED

LINE OF SIGHT MODEL LOW SNR/

L.o.m

REFLECTION

INSIDE CIRCLE

ABSENCE OF

TRAINED SIGNAL

MUD RECIECVERS LINK ADAPTATIONBLIND EQUALIZATION

OUTDOOR

BODY+

GROUND+

REFLECTION

NONLINEAR

MUD

MAXIMUM NOISE

FILTERED

NOISE FILTERINGCHANNEL MODELING NOISE FILTERINGCHANNEL MODELING

High SNR/

H.o .m

CHANNEL MODELING/NOISE

FILTERING

Fig. 4. Hierarchy for Channel Modeling and Noise Filtering

maximum allowable number of children’s, 𝑛𝑐ℎ𝑙 and maximum

level of logical tree, 𝑑𝑙, are managed by a coordinator. A

coordinator acts as a root of the tree with address zero.

On-tree self pruning broadcast algorithm keeps the energy

efﬁcient broadcast in account for ZigBee network and decides

whether to rebroadcast or not after receiving a packet. When a

source node broadcasts a packet, then all its 1-hope neighbors

receives this packet. This may cause collision when multiple

nodes are communicating and accessing the channel at the

same time. To avoid such collisions, each node waits for

random period of time before rebroadcasting. During this

waiting state, forward node, ′𝑥′only needs to cover 1-hope

distance i.e., 𝑁(𝑥)−𝑁(𝑌). If node 𝑥learns that all its 1-

hope neighbors are already covered before time out then it

does not need to rebroadcast [4]. This self pruning algorithm

can perform poorly when applied to zigBee networks because

of unavailibilty of 2-hope neighbor’s information.

An optimal On-tree forward node Selection algorithm

(OOS) resolves the problem and is assumed to guarantee

convergence of the whole network [4]. This algorithm tries to

reduce the size of the candidate forward node set ′𝑆′and to

be covered set ′𝐶′. If node 𝑥is the source or forward node, it

will rebroadcast and cover all its 1-hope neighbors, therefore,

it does not need to be selected as a forward node again, as

well as, all its neighbors need not to be covered again.

Fig. 6𝑎shows ZigBee tree topology with node 𝑣at the

centre and four tree neighbors v1, v2, v3 and v4. u1....u8 are

other eight 1- hope neighbors which can be located anywhere

on the logical ZigBee tree. Given the network addresses of v1

to v4 and u1 to u8. Node V can further identify their parent

and children.

Fig. 6𝑏shows principle OOS which describes set size of the

candidate forward node S and C. OOS algorithm goes through

level by level in a top-to-bottom direction and left to right at

each level.

17501758

v1

v2

v

v4

u3

u8u5 u6 u7

u4

v3

(a)

`

v1

u1

v3

u5 u6 u7

u4

v4

v2

Construction of trees involving S and V

(b)

Fig. 5. OOS Algorithm in Local ZigBee using Tree Topology

VI. ENERGY UTILIZATION

In this section, we discuss and compare energy utilization

in selected three techniques; 1) MUD receivers, 2) Novel Link

Adaptation, and 3) Blind Equalization for modeling channels.

WWBANs use a central device (i.e., hand held computer) and

multiple sensors, simultaneously for communication. Multi

user interference is a phenomenon which leads in the degra-

dation of the received signal and certain receivers based on

RAKE do not cancel this effect [1]. In BAN environment Multi

user based receiver is considered to be an appropriate solution

for the BAN environment where several devices communicate

simultaneously [1].

Algorithm 2 Algorithm for MUD receivers

1: 𝐷𝑖𝑠𝑐𝑟𝑒𝑡𝑒 𝑒𝑞𝑢𝑣𝑎𝑙𝑒𝑛𝑡 𝑜𝑓 𝑟𝑒𝑐𝑒𝑖𝑣𝑒𝑑 𝑤𝑎𝑣𝑒𝑓 𝑜𝑟𝑚 ←𝑟1

2: 𝐹𝑟𝑎𝑚𝑒𝑠 𝑠𝑎𝑚𝑝𝑙𝑒𝑠←𝑁𝑠

3: 𝑀𝑒𝑎𝑛 𝑠𝑞𝑢𝑎𝑟𝑒 𝑒𝑟𝑟𝑜𝑟 ←𝑀𝑆𝐸

4: 𝐹 𝑖𝑙𝑡𝑒𝑟 𝑡𝑎𝑝𝑒𝑠 ←𝑊

5: 𝑅𝑒𝑐𝑒𝑖𝑣𝑒𝑑 𝑜𝑢𝑡𝑝𝑢𝑡 ←𝑅𝑜𝑢𝑡

6: if Choice = 1 then

7: 𝑈𝑠𝑒 𝑙𝑖𝑛𝑒𝑎𝑟 𝑀 𝑈 𝐷

8: 𝑟1=𝑁𝜇

𝜇=1 ∞

𝑘=−∞ 𝛼𝜇

𝑘𝜌𝜇(𝑙−𝑘𝑁𝑠)+𝑛1

9: 𝐶𝑎𝑙𝑐𝑢𝑙𝑎𝑡𝑒 ′𝑁′

𝑠𝑓𝑟𝑜𝑚 ′𝑟′

1

10: 𝑤=[𝑤1....𝑤𝑛]

11: 𝐶ℎ𝑜𝑜𝑠𝑒 ′𝑊′𝑠𝑢𝑐ℎ, 𝑡ℎ𝑎𝑡 𝑚𝑖𝑛𝑖𝑚𝑖𝑧𝑒𝑠 ′𝑀𝑆𝐸′

12: 𝑅𝑜𝑢𝑡 =𝐼𝑆𝐼 +𝑀𝑈𝐼 +𝑁𝑜𝑖𝑠𝑒

13: else

14: 𝑈𝑠𝑒 𝑛𝑜𝑛 −𝑙𝑖𝑛𝑒𝑎𝑟 𝑀𝑈𝐷

15: 𝐹 𝑜𝑟 𝐹 𝑒𝑒𝑑 𝐹 𝑜𝑟𝑤𝑎𝑟𝑑 𝐹 𝑖𝑙𝑡𝑒𝑟 (𝐹𝐹𝐹)

16: 𝑅𝑝𝑒𝑎𝑡 𝑠𝑡𝑒𝑝𝑠 8−10

17: 𝐹 𝑜𝑟 𝐹 𝑒𝑒𝑑 𝐵𝑎𝑐𝑘 𝐹 𝑖𝑙𝑡𝑒𝑟 (𝐹𝐵𝐹)

18: (𝑁𝑠−(𝑁𝑠−1)) ←𝑆𝑡𝑎𝑡𝑢𝑠

19: 𝑅𝑜𝑢𝑡 = FFF + FBF

20: 𝑅𝑜𝑢𝑡=ISI +MUI

21: end if

A. Multiuser Detection Based Recievers

A signal consists of following components; a) desired signal

energy spread over several multiple symbols, b) ISI due to

channel memory, c) MUI due to other users signals, and d)

additive white Gaussian noise is received [1]. MUD based

receiver is used to extract effective energy in the presence

of all the components. To effectively extract energy from

different multipaths in the presence of ISI, MUI and AWGN.

MUD receiver is implemented in two phases; 1) Linear MUD

receivers and 2) non-Linear MUD receivers [1].

1) Linear MUD Recievers: A linear MUD receiver uses

Weiner-Hopf equation to minimize mean square error. The

effects of ISI, MUI and noise are mitigated jointly, however,

receiver cannot distinguish between MUI, ISI and noise.

To differentiate between noise and ISI, another approach is

proposed which is known as Non-linear MUD receivers.

2) Non-Linear MUD Recievers: Estimation of previous

symbol could be utilized while estimating the current symbol.

This could be achieved by exploiting decision feedback prin-

ciple. A Non-Linear MUD receiver consists of two ﬁlters: one

is Feed Forward Filter (FFF) and Feed Back Filter (FBF). FFF

resembles Weiner combiner, and the FBF has at its inputs the

sequence of decisions on previously detected symbols. FBF

is intended to remove the part of the ISI from the current

symbol which is caused by previously detected symbol. By

minimizing mean square error through Linear and Non-Linear

receivers, retransmission can be mitigated. Thus, minimizes

energy consumption.

B. Novel Link Adaptation

Novel Link Adaption (LA) mechanism improves perfor-

mance and achieves signiﬁcant result while reporting measured

data to a coordinator with different modulation schemes [2].

17511759

A variable data rate scheme is proposed to reduce the aver-

age power consumption [2]. This scheme considers channel

conditions which would work in a better way. The usage of

lower bit rates (low SNR/SIR) and higher bit rates to reduce

probability of error is well solved in [2]. The use of dynamic

modulation/coding schemes (O-PSK, L-DSPK) by keeping

channel response in view can result beneﬁcially in reduction

of interferences. Thus, minimizes energy consumption through

avoiding re-transmissions.

C. Blind Equalization Algorithm

Conventional equalization techniques rely on the transmis-

sion of training signals. This relay leads to a reduction of

channel bandwidth and sources allocation. Constant Modulus

Algorithm (CMA) is suitable for blind wireless channel equal-

izer, because of its robustness over the violation of perfect

blind equalization condition gives full bandwidth utilization

[3]. Transmission time is always an important factor in terms

of cost and transmission power being utilized. Large data

rate can be achieved through lossless compression algorithm.

In wireless communication systems, most of the energy is

consumed for transmission. Multiple interferences lead to the

effect of retransmission during signal transmissions. In this

study different techniques are discussed which addresses the

minimization of different interferences/disturbances.

VII. IMPAIRMENTS

Simultaneous communication among sensors and central

devices (hand held computers) yields multiple interferences

like ISI, MUI, and noise. RAKE based receivers do not cancel

these effects, therefore, MUD receivers are used. Whereas,

determination of the coefﬁcients of the combiner is carried

through training sequence method.

A. Linear and Non-Linear Equalizers

By using MMSE criterion to determination of tap coefﬁ-

cients is given in [1] as:

𝑇𝑟

𝑇𝑛 =𝑁𝑠 > 1(13)

Moreover, considering the above condition for sampling rate

in [1] as:

𝑟1=

𝑁𝜇

𝜇=1

∞

𝑘=−∞

𝛼𝜇

𝑘𝜌𝜇(𝑙−𝑘𝑁𝑠)+𝑛1(14)

For u =1

𝑟1=

∞

𝑘=−∞

𝛼1

𝑘𝜌(𝑙−𝑘𝑁𝑠)+

𝜇∕=1

∞

𝑘=∞

𝛼𝜇

𝑘𝜌𝜇(𝑙−𝑘𝑁𝑠)+𝑛1(15)

Representing the MUI as 𝑚1given in [1] as:

𝑟−1=

∞

𝑘=−∞

𝛼1

𝑘𝜌(𝑙−𝑘𝑁𝑠)+𝑚1+𝑛1(16)

Filter tapes 𝑤=[𝑤1....𝑤𝑛]minimize MMSE 𝐸(∥𝑎1

𝑘−

𝑊𝑟

𝑘∥2)) are desired to be chosen, and are obtained through

Weiner-Hopf equation [1]. 𝑤=𝛾𝑎𝑟Γ−

𝑟𝑟1where, Γ𝑟𝑟 =

𝐸[𝑟𝑘𝑟𝑇

𝑘]denotes received vector auto-correlation matrix and

Γ𝑟𝑟 =𝐸[𝑎1

𝑘𝑟𝑇

𝑘]represents cross correlation between known

transmitted symbol and the corresponding received samples.

The receiver tries to mitigate the effect of MUI, ISI and noise

jointly. However, it does not distinguishes MUI and ISI from

noise, thus, the output still contains some amount noise. In

order to mitigate the effect of ISI, a Non-Linear ﬁltering

approach based on FFF and FBF is used. FBF has at inputs

the sequence of decisions on previously detected symbols. The

main aim of this decision is to remove ISI from the current

symbol caused by previous symbols.

MMSE criterion is applied to optimize the coefﬁcients of

the ﬁlters. It is essential to note that input samples to FFF are

spaced 𝑇𝑟 seconds apart while input samples of the FBF are

spaced 𝑇𝑠 seconds apart. The equalizer output is represented

in[1]isgivenas𝑥𝑘=𝑤𝑓𝑓𝑟𝑘+𝑤𝑓𝑏 where, the row vector 𝑤𝑓𝑓

denotes 𝑁𝑠 length of FFF and 𝑤𝑓𝑏 is the 𝑁𝑏 length vector

representing FBF. The set of past decisions is represented by

𝑎1

𝑘=[𝑎1

𝑘, ....𝑎1

𝑘−𝑁𝑏]𝑇as given in [1].

B. Novel Link Adaptation

Another optimized proposed solution for the removal of ISI

is use of dynamic modulation/coding schemes according to

channel response as given in [2]. Considering 3GPP proposal

for WBANs, the channel loss can be calculated from [2] as:

𝐴(𝑑)[𝑑𝐵]=𝐴𝑜+10𝑛𝑙𝑜𝑔(𝑑/𝑑𝑜)+𝑠(17)

where 𝑑is the distance between transmitter and receiver,

𝐴0is the path loss in dB at a reference distance; d0 (A0

= 35.2 dB for d0 = 0.1 m), 𝑛is path loss exponent (n

= 3.11), and 𝑠is a random Gaussian variable, having zero

mean and standard deviation 𝜎=6.1𝑑𝐵. for proper reception

of packet, we assume that following two conditions should

be achieved: a) 𝑃𝑅 > 𝑃𝑅

𝑚𝑖𝑛, where PR is the received

power given by: 𝑃𝑅[𝑑𝐵𝑚]=𝑃𝑡𝑥[𝑑𝐵𝑚]−𝐴[𝑑𝐵],𝑃𝑡𝑥 and

𝑃𝑅𝑚𝑖𝑛 are transmit power and receiver sensitivity, respectively

[2]. Total power received from interfacing nodes is denoted

by 𝐼, whereas, 𝐶represents the power received from useful

transmitter. 𝐶/𝐼 ∣𝑚𝑖𝑛 is the minimum SIR ensuring the correct

reception of a packet. These values are function of modulation

schemes being used.

C. Blind Equalization Algorithm

A bandwidth limited channel with high data rates is al-

ways effected by ISI. This is due to adjacent symbols on

the output of the channel which overlap each other and

causing degradation of error performance. An equalization

ﬁlter attempts to extract the transmitted symbol sequence

by de-convolving them with the inverse version of channel.

Meanwhile, multiplexing techniques emerge to be a powerful

source for minimizing system bit error rate.

Cyclic preﬁx (longer then channel length) within OFDM

systems by converting linear convolution into cyclic one

17521760

0 500 1000 1500 2000 2500 3000 3500 4000

−30

−25

−20

−15

−10

−5

8−QAM: SNR = 30dB

CMA: μ = 0.0008

Proposed: μ = 0.0006

−2 0 2

−3

−2

−1

0

1

2

3

CMA

−2 0 2

−3

−2

−1

0

1

2

3

β.CMA

0 500 1000 1500 2000 2500 3000 3500 4000

−30

−25

−20

−15

−10

−5

16−QAM: SNR = 80dB

CMA: μ = 0.00015

Proposed: μ = 0.0003

−5 0 5

−6

−4

−2

0

2

4

6

CMA

−5 0 5

−6

−4

−2

0

2

4

6

β.CMA

Fig. 6. Blind Equalization Using CMA, 8-QAM and 16-QAM

mitigates the effect of ISI signiﬁcantly [3]. To utilize full

bandwidth of the channel adaptive blind equalization systems

are introduced in order to minimize the effect of ISI [3].

Transmission of training signals in conventional equalization

techniques leads to reduction in bandwidth. Whereas blind

equalization do not rely on the training sequence, hence,

results in a better consumption of allocated resources.

Because of robustness over violation of Perfect Blind Equal-

ization (PBE) conditions, CMA is an attractive choice for the

researchers. Dithered Signed-Error Constant Modulus Algo-

rithm (DSE-CMA ) changes the input signal to the equalizer

by a non-subtractive sinusoidally-distributed signal before a

sign operation is applied [3]. The output of the equalizer using

multi-rate model from [3] is given as:

𝑦(𝑛)=𝑟𝐻(𝑛)𝑓(𝑛)+𝑊𝐻(𝑛)𝑓(𝑛)

=𝑋𝐻(𝑛)𝐻𝑓(𝑛)+𝑊𝐻(𝑛)𝑓(𝑛)

where, 𝑋=[𝑥(𝑛),𝑥(𝑛−1), ...., 𝑥(𝑛−𝑁𝑥+1)]

𝑡denotes a

ﬁnite-length symbol vector. The length of this symbol vector

is given by 𝑁𝑥=(𝑁ℎ+𝑁𝑓−1)/2,𝑓represents a column

vector of fractionally-spaced equalizer coefﬁcients with a

length 𝑁𝑓. The column vectors 𝑟(𝑛)and 𝑤(𝑛)represent

the time decimated 𝑁𝑓received samples and white Gaussian

noise, respectively. A Hermitian operator is denoted by (.)H,

and matrix transposition is symbolized by (.)T. Matrix H

symbolizes a 𝑁𝑥∗𝑁𝑓time-decimated channel convolution

of the GBHDS channel model.

Signals when pass through a channel undergo various forms

of distortions. Most common is ISI. Most of the times, in prac-

tical systems, channel characteristics are not known as prior.

Therefore, adaptive equalizers are used. Blind equalization as

an adaptive equalizer can compensate amplitude and delay

distortion of a communication channel only by using channel

output samples and knowledge of basic statistical properties

of the data symbol. We compared simulated results for blind

equalization using CMA. We used picchiprati channel model

for our experiments. The transmitted signals are modulated

with 8-QAM and 16-QAM, respectively. From the ﬁg. 8 given

it is clear that step size 𝜇=0.0006 and 𝜇=0.0003 for 8-

QAM and 16-QAM, respectively are sufﬁciently enough to

minimize MSE at steady-state.Thus, convergence is achieved.

VIII. CONCLUSION

In this paper, we addressed some of the problems that

a body area network still has to deal with for efﬁcient

performance in order to provide an optimized solution for the

cure of elderly people. We investigated some of the issues and

tried to sort out their possible solutions. Channel modeling,

effect of interferences/disturbances, energy consumption, and

noise ﬁltering are some of the issues which should be given

importance on the equal basis to attain better performance.

We also discussed some simulated results describing the use

of different modulation/coding schemes. Decision is taken on

the basis of channel characteristics.

REFERENCES

[1] Puduru.Viswanadha. Reddy, et al., ”performance of Multi User Detection

based receivers for UWB Body Area Network” in 978-1-4244-2281-

04/08/2008.

[2] Flavia.Martelli, et al., ”Link Adaptation in IEEE 802.15.4-Based Wireless

Body” in IEEE S1st international symposium on personal, indoor and

mobile radio communications workshop 2010.

[3] Seedahmed.Mahmoud.S, et al., ”A Blind equalization algorithm for

biological signal transmission” in digital signal processing (2012) 114-

123.

[4] Jang.Ding et al,. ”Tree-Based Data Broadcast in IEEE 802.15.4 and

ZigBee Networks”.

[5] Hairong.Yang et al,. ”Experimental e-Health Applicaton in Wireless

Sensor Networks’ in International conference communication and Mobile

computing.

[6] Emil.Jovanov et al,. ”Body Area Networks for Ubiquitous Healthcare

Application: Opportunities and challanges” Febuary 17, 2011.

[7] k.Vac et al,. ”Mobile patient Monitoring: the Mobihealth System” in

31st Annual international conference of the IEEE EMBS Minneapo-

lis,Minnesota,USA,September 2-6, 2009.

[8] Hairong.Yan et al,. ”Wireless Network Based E-Health System Imple-

mentation and Exerimental Results” in IEEE Transction on Consumer

Electronics, Vol.56, No.4, November 2010.

17531761