Medical Body Area Networks : [602947]

Medical Body Area Networks :
Mobility and Channel Modeling
Rafik Braham, Ferdaws Douma, Amina Nahali
Prince RG, ISITCom – University of Sousse , Tunisia
[anonimizat] – [anonimizat] – [anonimizat]

Abstract— Body Area Networks (BANs) have become
the focus of both research and industry in the last five
years [1-3]. Furthermore, IEEE has published the
official standard in 2012 [4]. BANs come in several
variants such as Medical (or hospital) BANs ( MBANs)
and wireless BANs (WBANs). They are also located at
the crossing of several domains fo r example wearables,
Internet of Things (IoT), sensors nets, and so on. In this
paper we propose contributions at two BAN levels. We
first present a system level solution for MBAN mobility
and handover (HO). Second, we give a BAN channel
mode l which consis ts of an extension of our work
formerly applied to Wi -Fi and Wimax.
Keywords – IoT, BAN, propagation model, HO, mobility
I. INTRODUCTION
The market of BANs is growing and expected to
continue doing so for coming years. According to recent
data from manufacturers [5-6], revenues are estimated at
$180 billion in 2014 and could reach more than $1,000
billion by 2020 with 6.4 billion devices connected just at
this year’s end (2016) based on predictions from Gartner
Group [6].
In this paper, we provide an extension of our work
on wireless signal propagation modeling to the case of
BAN channels. We also discuss system level issues,
always in the context of BANs and IOT.
II. INTERNET OF THINGS (IOT)
The eight billion connected BAN devices expected by
2020 are just a small fraction of a giant network that will
connect billions of physical digital objects and make them
accessible from the world Internet : it’s the Internet of
Things (IoT) [7]. These objects range from simple
elect ronic tags (RFID) to powerful smart phones, vehicles,
homes, through various sensors providing an interface
with the physical world including Humans. Extending
mobile Internet access to these objects will be made
possible mostly with wireless connections a nd diverse
standards in a heterogeneous networking environment.
But IoT is not just an extension of traditional
Internet. It should be a smart Internet made with
presumably low -cost intelligent interconnections
collecting data and disseminatin g intelligen t services and
applications. At the end of the day, IoT should improve services for our health, education, and more generally our
standard s. IoT can offer a huge number of applications but
only few are currently available to our society [1]. The
domains and the environments in which new applications
can improve human life conditions and make many tasks
easie r include the following :
a. Prediction of natural disasters: The combination of
sensors and their autonomous coordination and simulation
will predict the occurrence of landslides and other natural
disasters so that appropriate measures are taken in
advance.
b. Smart environments and industry applications: for
example IoT may serve in managing a fleet of cars for an
organization. IoT can help track their envir onmental
performance and process data to determine for instance
maintenance needs.
c. Transportation domain: With IoT, vehicles will be
able to communicate with many actors. They will first
communicate with road infrastructures such as tolls,
parking and traf fic flows. The driver will know which
route to take depending on traffic, costs, etc.
d. Healthcare domain: Some works proposed to use
IoT to provide effective and appropriate healthcare to
elderly and disabled. Authors in [2] and [3], for example,
introduce an approach based on IoT in medical
environments to reach a global connectivity with patients,
sensors and everything around. The goal of this approach
is to make patient’s life easier and clinical process more
effective.
III. IOT RESEARCH CHALLENGES
Before th e full promise of IoT will happen,
technical and technological challenges IoT is faced with
must be resolved. Areas concerned include networking
and communications, information gathering, storing and
processing, sensing, security and privacy. Research efforts
have to find satisfactory solutions to major IoT hurtles.
Some of the pressing subjects include the following .
As IoT is expected to generate data traffic with
patterns that are different from those used in the current
Internet , traffic characterization and QoS support are open
IoT issues. Consequently, new QoS requirements and
support schemes will need to be defined.
Data integrity and privacy of information pose a
serious problem because important quantit ies of private
and sens itive information about a person can be collected

without the person being aware. Furthermore, control on
the diffusion and broadcast of all such information is
impossible with current solutions.
Although these and other concerns cover a large
scale of sc enarios, we focus in this paper on the
consequences for medical and health areas within which
our work lies primarily. Indeed to be m ore specific, our
investigations are targeting BANs and some of the open
issues they face. Of these issues, we are particul arly
interested in two linked subjects: mobility and channel
modeling as we explain in subsequent sections.
IV. MOBILITY FOR BODY AREA NETWORKS
Body Area Network (BAN s), compo sed of a number o f
wireless devices in, on or near t he human body, have been
in an intense expansion for the last decade. A good survey
of this subject and related research issues may be found in
[8, 22]. Nevertheless several technological challenges are
ahead for this technology to be developed and deployed
like mobile phones for example . Market prospects are
numerous and concern di verse sectors such as health,
sport, leisure, security personnel, etc.
Body Area Networks (BANs) refer to a
subcategory of sensor networks mainly used for measuring
or monitoring vital body parameters. In this type of
networks, a plurality of sensors and electronic equipments
that may be implanted within the body, carried on the body
or placed in the vicinity of the body, collecting
information. They are also likely to exchange data, store
and transmit them rem otely. Individuals, equipped with
sensors and actuators, can indeed interact more effectively
with their immediate environment, while ensuring
connectivity via the global mobile Internet.
BANs are useful in the Healthcare . For example
patients can be moni tored while they are at home. High –
risk p ersons (pregnant women, elderly, persons with
diabetes, blood pressure, etc.) may receive periodic
medical attention without actual hospitalization and thus
reduc ing costs and painful journeys .
One of the challeng es encountered with BAN is
patient mobility management. In mobility scenarios, if a
node loses contact with its corresponding AP, bad
consequences might result. This behavior occurs because
the system is not aware for a valid period of time that a
node can be out of communication range of its registered
AP.
Many works have been done on IP mobility
management systems to ensure the handover o f MNs.
Existing methods are based on the host or on the network.
Network mobility protocols proved to be very useful in the
6LoWPAN mobility. These networks consist of devices
with limited memory and energy resources and low
computing power.
For example, in a Hospital Wireless Sensor Networks
(HWSN) or MBAN , monitoring patient s’ health continuously is necessary. Patients are presumed
autonomous and mobile.
In HWSN, the sensors on patients’ bodies generate
critical data about health parameters such as pulse rate,
ECG, sugar level, body temperature, blood pressure, etc.
These data are time critical and should be sent to hospital
monitoring station s (and personnel) without packet loss or
delay . A health specialist at the monitoring station must
give immediate advice by interpreting received data.
So in an HWSN /MBAN setting, the handover (HO)
operation must be done efficiently and reliably. It is well
know that mobile communication systems (cellular
telephone networks) assure users to move freely (global
mobility) by the provision of mechanisms including the
concepts of Home L ocation Registers (HLR) and Visitor
Location Registers (VLR). In a nutshell, these are data
structures to keep track of customers and their respective
services. HOs come in various types such as hard HO
(with call interruption) and soft HO which are transp arent
to the user.
The execution of an HO is necessary to provide
mobility and continuous services between adjacent cells
(or covering areas). Unfortunately, an HO is usually
accompanied by delays and even call disruption and
packet loss. The need for op timized and efficient HO is
clearly noticed in recent standards f or example the 2012
WiMAX release [15].
For this reason , we propose a new handover
management algorithm. This algorithm is based on a
multitude of handover decision making criteria such as
speed of mobile node, bandwidth, delay and Received
Signal Strength Indicator (RSSI ). This is another r eason
for having a good model for electromagnetic propagation
in BAN channel s whether for simulation or actual
experimentation. Fig. 1 below sho ws a summary of system
components and HO delay.

Fig. 1. Handover mechanism with reduced delay . MN NAG PAG HHS
Attachment
trigger
Tunneling Location
update
request
Location
update
response
Location
update Handover
Delay

In this figure, we used the following conventions which
are close to those in [19]. In addition to Mobile Node
(MN) which represents in fact the BAN, NAG and PAG
are respectively the new and previous access gateway (an
access point with more resources). The H HS is the
Hospital Home Server which represents the core network
or the central server , normally located in the hospital or in
its service provider facilities for example the clouds .
In classical mechanisms, the HO is delayed until
final approval by H HS. In our system, communication
between MN and its new serving access gateway may
proceed normally as soon as possible and other approval
may be done in “the background.” This may be called
“emergency HO.”
V. BAN CHANNEL MODELS
We have presented previously a generic model for wireless
signal propagation developed by our research group [10].
Since then, this model has attracted a significant attention
from researchers at large as it received 14 independent
citations (Source: G oogle , My Citation s, 2015).
The model as introduced earlier was applied to Wi –
Fi and Wi -MAX wireless technologies. Here we extend it
to BAN channels. Modeling these channels is an important
subject for two main reasons:
(1): BANs are expected to expand w ith IoT especially
those of medical type,
(2): energy levels are very small compared to other
networks such as Wi -Fi and Bluetooth. Although
802.15.6 -2012 [11] standard requires power limits to be
compliant with local regulations, in its “ low power low
duty cycle (LP/LDC) mode” a transmitter operating in
Channel 6 shall be capable of transmitting at most –40
dBm effective isotropic radiated power (EIRP). This is
below 1 µw.
Power limitation is a function of the rate at which
electromagnetic energy is ab sorbed by body tissue, the so –
called Specific Absorption Rate (SAR) [12].
Manufacturing and using BAN devices (wearable
electronics) may cause health hazards so that utmost
precaution is required. Accurate channel modeling is one
such tool that can help.
Although IEEE standard 802.15.6 has already
addressed the subject, more research is needed. Research
findings are required to guide systems designers and
equipment manufacturers for example antennas. We hope
this work will contribute in this direction. But first and
before we present the actual model, some background on
related works is necessary.
For wireless signal propagation within BANs,
several scenarios ought to be considered according to
propagation milieu [9, 20]:
– within the body
– on body su rface (the skin)
– around the body. In this work we focus on wave propagation around
the body and more precisely "creeping waves" as next
section explain s. From a communications point of view,
this propagation scenario is useful in two cases:
1. To link two devices (two wireless sensors) on
different parts or sides of the body for example around
the waist or around the head. In the IEEE jargon [9],
this case corresponds to “ Body Surface to Body
Surface ,”
2. Or to link a sensor and the BAN coordinator
typical ly the network access point AP. Again in the
IEEE terminology, this case may correspond to two
scenarios: either “ Body Surface to Body Surface ” or
“Body Surface to External ” [9].
With respect to propagation scenarios of interest, IEEE
P802.15 standard [9] defines several Channel
Communication (CM) models of which CM3 at 2.4 GHz
is most relevant. It is defined by the equation giving the
path loss (PL) as :
PL (d) [dB] = n * log 10 (d) + b + N (1)
where n is the path loss exponent (see below), b a
coeffi cient of linear fitting, d : Tx -Rx distance, and N :
normally distributed random variable with a certain
standard deviation σN to account for short -term fading
fluctuation (known a s shadowing ). Equation (1)
corresponds to the well -known path loss model, which is
actually one of the simplest expres sions among the
several channel models proposed in the literature. A quick
review of these models is giv en in [10] and i nterested
readers may find det ailed discussions in the authoritative
book by Rappaport [21].
VI. PROPOSED CHANNEL MODEL
The model as introduced earlier was validated for Wi –
Fi and Wi -MAX technologies [10]. Here we extend it to
BAN channels. The original model equation giving the
inverse of the path loss G (in dB) has the following generic
format:
G (dB)=PrPt =A
B∗dα+C∗(log 10d)β+D (2)
In (2), A, B, C, D, α and β are free coefficients optimized
for target scenarios , Pr and P t are received and transmitted
powers respectively .
The proposed model was tested with creeping waves.
This type of waves is extensively used in modeling
wireless propagation around human body. They h ave been
the subject of extensive research investigatio ns in this
context for quite some time [8, 12-14]. These waves occur
around the t orso (trunk) as well as the head . Actual
measurements and simulations have carried out on
humans and phantoms.
Channel characterization is this context is difficult
and has l ed to numerous BAN models because it depends

on gender, age, activity (standing, sitting, moving , …)
and on the environment . In this regard, m odels may be
divided into two categories:
– exact and deterministic: derived from physical
equations such as Maxwe ll’s equations and
depend on obstacles; a good example is the
free space model ,
– empirical: for example experimental or
statistical .
In these various models, t he equations may be
simple and valid for a few precise situation s. In other
cases , they are complex and thus unfortunately not so
good f rom a design perspective .
A good illustration of this tradeoff is the case
where w e start from free space model, then allow path
loss exponent n of Equation (1) to be variable, usually
from 2 , corresponding to theoretical free space, to 6 :
indoors with numerous obstacles like walls. For BANs,
usually n runs from 2 to 4.5 .
Model parameters were determined with the trust
region method [17-18]. This is a least -mean square (LMS)
technique more robust than the Levenberg -Marquardt
optimization algorithm that we have employed in [10].
Fig. 2 below gives the algorithm in a pseudo -code format.

Trust Region Algorithm

Begin
Repeat
Define a model 𝑚𝑘 in 𝐵𝑘
Compute 𝑆𝑘 such that 𝑥𝑘+𝑆𝑘 minimise s 𝑚𝑘 :
𝑥𝑘+𝑆𝑘 ∈ 𝐵𝑘
Compute the ratio 𝜌𝑘 such that
𝜌𝑘=𝑓 𝑥𝑘 −𝑓(𝑥𝑘+𝑆𝑘)
𝑚 𝑥𝑘 −𝑚(𝑥𝑘+𝑆𝑘)
If 𝜌𝑘≥𝑛 , 0<𝑛<12
Then {point accept ed → increase radius}
𝑥𝑘+1:=𝑥𝑘+𝑆𝑘
Increase trust region 𝐵𝑘
Else If 𝜌𝑘≥𝑛′ Then
{point accepted but unsatisfactory →
decrease slightly radius}
𝑥𝑘+1:=𝑥𝑘+𝑆𝑘
Decrease trust region 𝐵𝑘

Else
{The point is reje cted. The radius is reduced
sharply }
𝑥𝑘+1:=𝑥𝑘
𝐵𝑘 is decreased
End_If
Until convergence
End

Fig. 2: Pseudo -code for Trust Region Algorithm The experimental set -up to determine parameter values in
the case we show here correspond to experimental data
from [8]. The numerical results are as follows:
A = -213.2 α = -4.575
B = 11 β = -6.947
C = 7.246
D = 3.11.

Fig. 3 shows main results obtained by MathLab code.
Measurements, theoretical model and EBFB -BAN
model are presented for a perimeter value of 70 cm
(waist ). Path gain expression (2) expressed in dB (green
curve). Fig. 3 shows that the new EBFB -BAN model
(green curve) shows a good characterization of the
creeping wave propagation channel around the waist (p =
70cm). We have shown clearly that the EBFB -BAN
model is able to follow the direct and indirect paths
creeping waves. EBFB -BAN model is able to predict the
path gain between two antennas placed around the human
body.
Theoretical model and EBFB -BAN model are
simulated for the same frequency (2.4 GHz) and t he same
antenna gain ( -9.5 dBi) . A comparison between the two
models provide s a validation for the proposed EBFB –
BAN equation .

Fig. 3 : Theoretical and experimental model s vs EBFB -BAN model .
Waist perimeter p=70 cm .

The model proves to be capable of characterizing the
wave propagation around a cylindrical surface. It gives
good results in comparison with the '' Alves" model [8].

VII. DISCUSSION AND FUTURE WORK
In this paper we presented BAN challenges such as
mobili ty supp ort and channel Modeling . The opportunity
for more proposals and related mechanisms for mobility
management in IoT and BANs are active research
directions in spite of the existence of several proposed
schemes .
Our own interest in BANs is two -fold. First we
propose d an efficient handover management algorithm.
This algorithm is based on a number of handover decision
making criteria such as RSS (Received Signal Strength),
speed of mobile node (MN), bandwidth and delay. We
also present ed a new channel model (EBFB -BAN ) for
signal propagation in BAN applications. It is a p ath loss
model for BAN based on the creeping wave theory.
EBFB -BAN model was validated for ON -body
communication around 2.4 GHz frequencies .
More experimental results are w arranted to study
wave propagation around other parts of the body (in
addition to the waist region). An interesting future work
for the entire BAN research community may consist of
utilizing this channel model to study by simulation system
level aspects of handover and mobility mechanisms and
scenarios associated with the first part of this paper.
ACKNOWLEDGMENTS
…
REFERENCES
[1] M. Chen et al. , “Body area networks: a survey ,” Mobile Ne twork
Applications, 16 (), pp. 171 -193, 2011.
[2] Movassaghi et al. , “Wireless body area networks: a survey ,” IEEE
Communications Surveys and Tutorials , 2012
[3] J. Ahmad and F. Zafar , “Review of body area network technology
& wireless medical monitoring ,” International J. of Information
and Communication Technology Research, Vol. 2, No. 2, 2102,
pp. 186 -188
[4] IEEE. Std 802.15.6 -2012.
[5] N. Biet, “Internet of Things: Overview of the market, The Faktory,
Belgium ,” available at http://www.thefaktory.com/ wp-
content/uploads/2015/01/IoT -market -overview -Final.pdf , 2015. [6] R. Towey , “PI Predictions 2016: The Power Of The Internet Of
Things Will Be Revealed ,” PerformanceIN, online Magazine, Jan.
13, 2016.
[7] O. Vermesan and P. Friess, “Internet of Things: Converging
Technologies for Smart Environments and Integrated Ecosystems ,”
Aalborg River Publishers , Denmark, 2013.
[8] S. Movassaghi et al. , “Wireless Body Area Networks: A Survey ,”
IEEE Communications Surveys & Tutorials , Vol. 16, No. 3, pp.
1658 -1686 , 2014 .
[9] T. Alves, B. Poussot, J -M Laheurte, “Analytical Propagation
Modelling of BAN Channels Based on the Creeping Wave
Theory”, IEEE Transactions on Antennas and Propagation, 59(4),
pp 1269 -1274, 2011.
[10] IEEE P802.15 , “Channel Model for Body Area Network ,” 27
April, 2009.
[11] R. Ezzine, R. Braham, A. Al -Fuquaha and A. Belghith, “A New
generic model of signal propagation in Wi -Fi and WiMax
environments,” IFIP Wireless Days Conference, Dubai, 2008.
[12] IEEE standard 802, “Local and metropolitan area networks – Part
15.6: Wireless Body Area Networks ,” Feb. 2012.
[13] Rohit Chandra , Antennas, Wave Propagation, and Localization in
Wireless Body Area Networks. Doctoral Dissertation, Lund
University, 2014.
[14] G. A. Conway , W. G. Scanlon , S. L. Cotton, M. J. Bentum, “An
Analytical Path -Loss Model for On -Body Radio Propagation ,”
URSI Intern ational Symposium on Electromagnetic Theory, pp.
353-356, 2010 .
[15] R. D’errico , R. Rosini , C. Delaveaud , A. Vasylchenko , “Body
Area Networks ,” Project WiserBAN , Report, European
Commission 7th Research Framework Programme, 2009 .
[16] IEEE , “Std 802.16 -2012: IEEE Standard for Air Interface for
Broadband Wireless Access Systems ,” 17 August 2012.
[17] MJD. Powell and Y. Yuan, “A Trust region algorithm for equality
constrained optimization ,” Mathematical Programming 49, pp.
189-211, 1991.
[18] K. Dongmin and D. Suvrit Sra Inde rjit, “A scalable trust -region
algorithm with application to mixed -norm regression ,”
Proceedings of the 27th International Conference on Machine
Learning, Haifa, Israel, 2010.
[19] J. Kim et al., “An ID/Locator Separation -Based Mobility
Management Architecture for WSNs,” IEEE Transactions On
Mobile Computing, Vol. 13, No. 10, pp. 2240 -2214, 2014.
[20] D.B. Smith et al., “Propagation models for Body -Area Networks:
A Survey and New Outlook ,” IEEE Antennas & Propagation, Vol.
55, No. 5, 2013, pp. 97-117.
[21] T. S. Rappaport , Wireless Communications: Principles And
Practice, 2nd Edition. NY: Prentice Hall, 2002 .
[22] R. Cavallari et al. , “A Survey on Wireless Body Area Networks:
Technologies and Design Challenges ,” IEEE Communications
Surveys & Tutorials , Vol. 16, No. 3, pp. 1635 -1657 , 2014 .