Wireless Body Area Network (WBAN): A Survey on Reliability,

Fault Tolerance, and Technologies Coexistence

MARWA SALAYMA, AHMED AL-DUBAI, and IMED ROMDHANI, Edinburgh Napier University
YOUSSEF NASSER, American University of Beirut

Wireless Body Area Network (WBAN) has been a key element in e-health to monitor bodies. This technology
enables new applications under the umbrella of different domains, including the medical field, the entertain-
ment and ambient intelligence areas. This survey paper places substantial emphasis on the concept and key 3
features of the WBAN technology. First, the WBAN concept is introduced and a review of key applications fa-
cilitated by this networking technology is provided. The study then explores a wide variety of communication
standards and methods deployed in this technology. Due to the sensitivity and criticality of the data carried
and handled by WBAN, fault tolerance is a critical issue and widely discussed in this paper. Hence, this
survey investigates thoroughly the reliability and fault tolerance paradigms suggested for WBANs. Open
research and challenging issues pertaining to fault tolerance, coexistence and interference management and
power consumption are also discussed along with some suggested trends in these aspects.
CCS Concepts: rComputer systems organization → Sensor networks; Reliability; Fault-tolerant
network topologies; Sensors and actuators; r
Hardware → Wireless devices; Wireless integrated
network sensors; Fault tolerance; Sensors and actuators; Sensor devices and platforms; Analysis and
design of emerging devices and systems
Additional Key Words and Phrases: Wireless body area networks, QoS, medical, channel access, fading,
WBAN standards
ACM Reference Format:
Marwa Salayma, Ahmed Al-Dubai, Imed Romdhani, and Youssef Nasser. 2017. Wireless body area network
(WBAN): A survey on reliability, fault tolerance, and technologies coexistence. ACM Comput. Surv. 50, 1,
Article 3 (March 2017), 38 pages.
DOI: http://dx.doi.org/10.1145/3041956

1. INTRODUCTION
The first decades of the last century heralded a revolution in wired communication: they
brought about an almost magical technological evolution especially sensing data and
communications. It did not take long for people to realise however that this technology
was inefficient, especially when it comes to wiring costs, mobility and independent
connections. Such inefficiencies have been the key driving forces towards the evolution

Authors’ addresses: M. Salayma, School of Computing, Edinburgh Napier University, 10 Colinton Road,
Edinburgh, EH10 5DT, UK; email: [email protected]; A. Al-Dubai, School of Computing, Edin-
burgh Napier University, 10 Colinton Road, Edinburgh, EH10 5DT, UK; email: [email protected];
I. Romdhani, School of Computing, Edinburgh Napier University, 10 Colinton Road, Edinburgh, EH10 5DT,
UK; email: [email protected]; Y. Nasser, ECE Department, American University of Beirut, Beirut,
Lebanon; email: [email protected].
Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted
without fee provided that copies are not made or distributed for profit or commercial advantage and that
copies show this notice on the first page or initial screen of a display along with the full citation. Copyrights for
components of this work owned by others than ACM must be honored. Abstracting with credit is permitted.
To copy otherwise, to republish, to post on servers, to redistribute to lists, or to use any component of this
work in other works requires prior specific permission and/or a fee. Permissions may be requested from
Publications Dept., ACM, Inc., 2 Penn Plaza, Suite 701, New York, NY 10121-0701 USA, fax +1 (212)
869-0481, or [email protected].
c 2017 ACM 0360-0300/2017/03-ART3 $15.00
DOI: http://dx.doi.org/10.1145/3041956

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:2 M. Salayma et al.

of wireless technology [Akyildiz et al. 2002; Boukerche 2005]. Wireless communication

has revolutionised our daily lives as it pervades most technological applications such as
controlling, tracking, monitoring and automation [Selavo et al. 2007; De Oliveira et al.
2009]. It was not long before the wireless sensor networking (WSN) revolution shifted
its direction to a technology that suited human mobility by devising a technology that
can be wearable or even implanted in the human body. This technology is characterized
by low-cost, energy-constrained, tiny, heterogeneous sensor nodes that form a special
type of WSN, namely, the Wireless Body Area Network (WBAN) [Movassaghi et al.
2014]. A WBAN comprises sensors that capture physiological information and send
it to a central base station through wireless communication. WBAN replaces complex
and wired healthcare equipment as it is able to continuously monitor the body’s vital
statistics [Movassaghi et al. 2014]. WBAN sensor devices are supposed to provide
real-time feedback without causing any discomfort, thereby providing a greater deal
of flexibility and mobility to the user [Otto et al. 2006]. More importantly, the data
provided by WBAN gives doctors a better view of a patient’s situation as this data is
gathered during a patient’s normal activities in his/her natural environment. WBAN
devices are characterised by their heterogeneity [Mahapatro and Khilar 2011], i.e.,
they vary in their capabilities, tasks, sizes, sampling rates, required resources and
levels of intelligence. The criticality of WBAN applications, the dynamic environment
within which they operate (limited to the human body) and the heterogeneity of the
deployed sensor devices confirm that WBAN has special characteristics that impose
key challenges in designing an efficient and resilient WBAN. For instance, a WBAN
has to be reliable as any fault could be life threatening for the person dependent
on this technology. One of the requirements of a reliable system is that it has to
be fault tolerant, that is, it needs to have the ability and the capability to self-heal
if a fault occurs regardless of the type or nature of fault [Liu et al. 2009]. Despite
ongoing research studies that tackle technical issues in WBAN, such as energy, mobility,
security, routing and sensor design, however, fault detection and recovery in WBANs is
still in their infancy. These observations motivate us to place a great deal of emphasis
on this concept in this study.
In fact, there have been several surveys pertaining to WBANs. We briefly introduce
these surveys in chronological order and then we demonstrate how our survey differs
from existing literature surveys. For instance, the MAC layer protocols of WBANs are
surveyed in Gopalan et al. [2010]. The survey provided in Latré et al. [2011] covers
the network layers and some existing projects related to the WBANs deployment.
The survey conducted by Barakah and Ammad-uddin [2012] focuses on the role of the
virtual doctor server (VDS) in current WBAN architecture. The healthcare applications
are surveyed in Crosby et al. [2012] and the WBAN layers are covered in Ullah et al.
[2012]. As for the routing protocols, [Movassaghi et al. 2013] provided a review of
protocols deployed in WBANs. The design properties of WBANs are covered in Cavallari
et al. [2014] and the coexistence problem was explored in Hayajneh et al. [2014].
The architecture, routing, channel modelling, PHY layer, MAC layer, security and
applications are covered in Movassaghi et al. [2014]. The security related issues have
been tackled in the survey of Mainanwal et al. [2015]. Moravejosharieh and Lloret
[2015] have focused on IEEE 802.15.4 standard. Although the previous studies shed
some light on various important issues and concepts in WBANs, this technology is still
unexplored in different aspects and dimensions. To the best of our knowledge, there
has not been any survey study reported on fault tolerance, reliability and interferences
management. To fill in this gap, this survey makes these major contributions.

—Unlike existing surveys, this survey emphasizes on two major issues lightly ad-
dressed before, namely, fault-tolerance- and reliability-related issues.

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:3

Fig. 1. Statistics of population aging in developed countries [Rutherford and Socio 2012].

—It provides an overview of some of the key challenges presented in the recent litera-
ture regarding reliablity and provides a summary of related research work. Moreover,
we explore both the hardware and software faults.
—It presents the need for better reliable solutions and integration among different
layers. It presents the interferences management and illustrate how the different
mobility models can affect the network reliability and possible solutions to mitigate
interferences.

The structure of this survey is as follows: Section 2 presents an overview about

WBAN, it discusses the possible motivation for adopting this technology, its potential
applications, the proposed standards and its devices. Section 3 illustrates the chal-
lenges of the WBAN technology. It focuses on fault tolerance, as it is one of the major
WBAN reliability requirements, it classifies faults and failures and illustrates the fault
tolerance life cycle. Section 4 presents the proposed mechanisms and schemes that tar-
geted towards achieving a reliable WBAN. This survey is ended in Section 5 with a
summary of the main points revealed in the survey and a conclusion of the literature
reviewed.

2. WIRELESS BODY AREA NETWORKS (WBANs)

Our world is facing a rapid growth in population which is accompanied with an increase
in the average lifetime expectancy of individuals, especially in the developed countries,
leading to an increasing number of people who are over 65 years old [Rutherford and
Socio 2012]. Given the population chart shown in Figure 1, 23.6% of United Kingdom’s
population is expected to be over 65 by 2050. Figure 1 reveals that not only the United
Kingdom is characterised with population aging, but the majority of the developed
countries lead this phenomenon. The ratio of people of age 16–64 to those who are over
65 will be 2.5:1 in the United Kingdom.
Unfortunately, chronic and fatal diseases such as cancer and cardiovascular and
asthma diseases are often diagnosed too late. Consequently, this increases the death
rate of individuals who are diagnosed with such diseases. Early detection would mit-
igate the impact of such diseases and increase the sufferers’ life expectancy. Further-
more, traditional monitoring systems do not provide a complete picture of a patient’s
status as the bodily functions are monitored too infrequently. The lack of early detection
and effective monitoring of diseases affect increases health care cost and adds a huge

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:4 M. Salayma et al.

Fig. 2. Typical WBAN.

load to health care systems with limited financial resources. This calls for more afford-
able and scalable systems that are able to move current health care in the direction
of early disease diagnosis and proactive wellness management. Wireless technologies,
including WSN and Wireless Personal Area Networks (WPAN), nanotechnologies, and
the Internet provide the means through which economical solutions could be found
for health care systems. Such technologies suggest the idea of using tiny, smart, low-
power, microsensors and actuators to sample physiological data and then forward it to
a remote server through wireless communication. While WPAN devices operate within
very small range area (less than 10m), the communication range of WBAN devices
would be narrower than that of WPAN as they are limited to the human body area
where they could be worn or implanted [Latr’e et al. 2011]. Actually, WBAN is consid-
ered to be the next generation of WPAN. A typical WBAN is shown in Figure 2. Though
WBAN comprises the typical characteristics of both WSN and WPAN, it also has its
own characteristics and thus its own requirements. WBAN saves lives by allowing
early detection of abnormal situations through wearable and implanted monitoring
devices. It allows continuous and real-time monitoring with no human or mechanical
intervention, which improves the quality of the achieved results. Moreover, WBAN of-
fers patients the ability to carry out their normal activities without interruption while
their vital signals are monitored, as they are no longer required to stay in hospital
or stick to a medical service [Movassaghi et al. 2014]. The adoption of WBAN should
reduce the cost of healthcare, by minimizing the need for expensive in-hospital care
monitoring. Such benefits have motivated practitioners in other fields, such as sports,
the military and the entertainment field to adopt it in their systems [Patel and Wang
2010].
A classification of the various applications that can be facilitated by WBAN is pre-
sented next, followed by the four popular standards proposed for WBAN, the main
WBAN devices and finally the most important WBAN challenges featured in the liter-
ature to its special nature.

2.1. Applications of WBAN

WBAN applications span from the health care and entertainment fields to sport and the
military among others. According to (TG6) [2012], WBAN applications are categorised
as either medical or non-medical. This section classifies WBAN applications according
to their target domain of application. Each application is further classified into medical,

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:5

Table I. Fields and Applications of WBAN

WBAN fields Applications types Examples of applications
Healthcare Medical Wearable Electrocardiogram (ECG),
electroencephalogram (EEG),
Electromyography (EMG), Saturation of
Peripheral Oxygen (SPO2), temperature,
blood pressure, drugs delivery
Implant Diabetes control
Non-medical Motion detection
Secure authentication
Military and Medical wearable Asses solider fatigue, detect life threatening situations
defense
Non-medical wearable Fire detection, poisonous gas
Sports Medical wearable heartbeat, temperature, blood pressure, motion sensor
Entertainment Non-medical wearable Real time streaming: Video streaming by camera, audio
streaming by headsets
Consumer electronics: MP3 player, microphone, camera
Gaming purposes, virtual reality, ambient, intelligence
areas, personal item tracking and social networking

non-medical, implanted, and wearable. WBAN fields of applications are summarised in

Table I. They are further classified into medical and non-medical applications. Table I
also gives some examples for each WBAN field.

2.1.1. Healthcare. This is one of the most promising fields for using WBAN. Implanted
and wearable sensors are used to collect biomedical signals remotely and continu-
ally [Malik and Singh 2013; Cavallari et al. 2014]. This continual monitoring allows a
proactive fatal and anomalies detection which is vital for diagnosing heart and brain ac-
tivities. Actuators help in automatic drug delivery. Some applications such as cochlear
implants, hearing aids, and artificial retinas help enhance the life style of human
beings [Barakah and Ammad-Uddin 2012; Malik and Singh 2013]. Additionally, given
that medical accidents can and do happen, WBAN applications help to reduce them and
increase public safety by using profiles of previous medical accidents to alert medical
personnel before similar accidents occur. Consequently, WBAN is expected to improve
the management of illnesses and reaction to crisis which will increase the efficiency
of health care systems. WBAN health care applications can be further classified as
follows.

2.1.1.1. Medical Applications. WBAN medical applications enable the continual mon-
itoring of physiological parameters such as the heartbeat, the body temperature, and
blood pressure [Movassaghi et al. 2014]. The data collected can be sent through a cell
phone, which acts as a gateway, to a remote location such as an emergency centre so
that the relevant action can be taken [Barakah and Ammad-Uddin 2012; Malik and
Singh 2013; Movassaghi et al. 2014]. WBAN is considered key to the early detection and
treatment of patients with serious cases such as diabetes and hypertension. Medical
applications of WBAN can be further divided according to the position of the medical
sensors as follows:

—Wearable Applications. Medical wearable healthcare applications include tempera-

ture monitoring, blood pressure monitoring, glucose level monitoring, ECG, EEG,
EMG, SpO2 , drugs delivery. Other applications can be found in Movassaghi et al.
[2014].

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:6 M. Salayma et al.

—Implant Applications. These applications comprise nodes implanted either under the
skin or in the stream of the blood such as in diabetes control systems, cardiovascular
diseases and cancer detection.
2.1.1.2. Non-Medical Applications. These applications are considered to fall within
the wearable sensor class of applications and include two applications which are
Movassaghi et al. [2014]:
—Motion Detection. This application is used to detect, capture, recognise and identify
body gestures and motions and send alerts to the owner of the application. For
example, fear increases heartbeat, which leads to sweating and other symptoms.
Thus, emotional status can be measured and monitored [Movassaghi et al. 2014].
—Secure Authentication. This is a very promising WBAN application as it is the core of
both multi-modal biometrics and electroencephalography. This application harnesses
physiological and behavioural human body biometrics such as fingerprints and facial
patterns.
2.1.2. Sports. Sport activities and fitness can be improved by keeping a log of vital
physiological data such as temperature, heart beat and blood pressure. The data can
be used to avoid sport accidents and injuries and to plan for future training [Barakah
and Ammad-Uddin 2012; Malik and Singh 2013; Cavallari et al. 2014]. According to
Movassaghi et al. [2014], WBAN sport applications are considered to be medical wear-
able applications. Such applications enhance professional and amateur sport training
especially for athletes. For example, they provide the necessary information to enable
training schedules of professional athletes to be adapted to make them more effective.
2.1.3. Entertainment. Entertainment is also a very promising field for WBAN. The film
industry, for example, benefits from motion capturing and post production mechanisms
to produce movies in which actors perform the objects roles [Cavallari et al. 2014]. Using
the on-body accelerometers and gyroscopes for capturing motions facilitates the possi-
bility of tracking the different positions of body parts [Cavallari et al. 2014]. According
to Movassaghi et al. [2014], WBAN entertainment applications are considered to be
wearable non-medical applications. WBAN can be used in three types of entertainment
applications, presented below.
—Real-Time Streaming. This includes video streaming, as well as audio streaming
such as voice communication for headsets that are used for listening to explanations,
illustrations, and multicasting (for example, conference calls).
—Consumer Electronics. These applications include appliances/devices such as mi-
crophones, MP3-players, cameras and other advanced interfaces such as neural
interfaces.
—Gaming, virtual reality, ambient intelligence areas, personal item tracking and social
networking [Malik and Singh 2013].
2.1.4. Military and Defense. WBAN provides new capabilities to improve performance
of individual and teams of soldiers in military situations. To avoid threats at the
individual tier, a group of sensors sample important information on the surrounding
emerging actions and environment. At the team level, the taken information enables
the commander to coordinate team tasks efficiently. Inter-WBAN communications
and security play a key role in preventing critical data from being hacked by enemies
[Cavallari et al. 2014]. WBAN applications can be considered as either medical
wearable or non-medical wearable as follows:
2.1.4.1. Medical Military WBAN Applications. These types of applications are
used to assess soldier fatigue and battle readiness and for safeguarding uniformed

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:7

personnel. For example, sensors surrounding soldiers, firefighters or policemen can

foresee a life-threatening situation by monitoring the level of air toxins.
2.1.4.2. Non-Medical Military WBAN Applications. Such applications involve off-
body sensors (on buildings) that are used for emergencies. Such sensors are capable of,
for example, detecting a fire in the home or a poisonous gas and must directly send this
information to on- and in-body devices to notify the wearer of the emergency situation.

2.2. WBAN Communication Standards

There are several reasons behind the motivation of wireless communities to standardise
their technologies. Standardisation allows interoperability which enables wide use of
the products since manufacturers depend on common fixed specifications in developing
their products. Additionally, customers need not to depend on a certain vendor. This
saves the costs for both the vendors and customers [Cavallari et al. 2014]. It is worth
to mention that WBAN often follows a star topology and due to the body nature,
the majority - if not all- the WBAN challenges are related to the reliability of the
channel access mechanisms. In this regard, this section presents the main technologies
that are proposed to serve WBAN and focuses on the MAC techniques adopted by
those technologies to support WBAN. A comparison between the WBAN supported
technologies is presented in Table II.
2.2.1. IEEE 802.15.4 Standard. The IEEE 802.15 Task Group 4 (TG4) proposed a com-
munication standard that is geared toward WPANs [TG4 2003]. This technology, called
IEEE 802.15.4, has become the de facto standard that supports both WSN and WBAN.
Many studies on the design of WSN power-aware algorithms and standards based
on IEEE 802.15.4 have emerged recently [Salayma et al. 2013a, 2013b]. Notice that,
the IEEE 802.15.4 standard considers both physical and MAC layers. Due to WBAN
challenges, the design of these networks necessitates the need for new protocols. In
this regard, IEEE 802.15 community has proposed amendments to the physical and
MAC layers of IEEE 802.15.4 protocol stack to overcome the drawbacks of the legacy
IEEE 802.15.4 in achieving the requirements of WBAN, which are the IEEE 802.15.4a
[TG4a 2007] and IEEE 802.15.4j [TG4j 2012]. In addition, IEEE 802.15 has proposed
a new standard that is geared towards WBAN, specifically, namely the IEEE 802.15.6
this standard among with the two amendments on IEEE 802.15.4 are presented in the
following subsections.
2.2.2. IEEE 802.15.6 Standard. The existing standards (e.g., IEEE 802.15.4) do not
meet the regulations of medical communication as they fail to support the needs
of applications in terms of key issues such as reliability, low power, the variety of
traffic flows and coexistence. The IEEE 802.15 Task Group 6 (TG6) has proposed
a communication standard that is geared toward applications in the vicinity of, or
inside, the body, such as in medical, sports and military applications [TG6 2012]. It
supports low cost, low complexity, very short range, and highly reliable and ultra-low
power wireless communication. It aims to support an array of applications with a
range of requirements such as data rates and channel bandwidths. To support various
applications, the standard offers three bandwidths defined in three different physical
layers: Narrow Band (NB), Human Body Communication (HBC) and UWB [Cavallari
et al. 2014; Movassaghi et al. 2014]. These physical layers share only one MAC layer.
The supported data rate ranges between 75.9Kbps in NB and 15.6Mbps in HBC. As the
range, it is limited to 3m for in-body communication and has to be at least 3m for body-
to-body communication patterns. The standard allows star and 2-hops tree topologies
[Movassaghi et al. 2014]. IEEE TG6 provides two classifications for devices, according
to their position in the body: implanted nodes, body surface nodes and external nodes.

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:8 M. Salayma et al.

Table II. A Comparison between the Aforementioned Technologies Discussed in this Section
Standard/
Criteria IEEE802.15.4 IEEE802.15.4a IEEE802.15.4j IEEE802.15.4.6
Network type WPAN UWB WBAN MBAN WBAN: NB, UWB and
HBC
MAC modes Beacon/Non- Superframe Superframe Superframe with
Beacon mode Beacon mode Beacon mode beacon mode
with/without enabled enabled Superframes with
superframe non-beacon mode
Non-beacon mode
without only
Access Random access Random access Random access Contention-based
mechanisms (with Contention: (Contention- (with contention random access:
CAP) Scheduled based: CAP) CAP) Scheduled EAP1,RAP1,EAP2,
access (without Scheduled access RAP2,CAP)
Contention: CFP) access (Contention free: TypeI/II)
(Contention CFP) Connection oriented
free: CFP) contention-free:
TypeI/II)
Contention Slotted and Slotted or Slotted CSMA/CA/a slotted
access un-slotted unslotted Aloha CSMA/CA Aloha
resource CSMA/CA in (CSMA is (in CAP) (in EAP1,RAP1,EAP2,
allocation beacon mode and optional) RAP2,CAP)
procedure in non-beacon (in CAP)
mode (in CAP)
Contention Schedule access by Schedule access A multi-periodic Two access modes:
free access allocating static by allocating scheduled access 1-Schedules the
resource GTS Allocation static GTS by allocating allocation of slots: one/
allocation persists in the Allocation static GTS multiple upcoming
procedure upcoming persists in the (m-periodic) superframes
superframes. upcoming (in CFP) single-periodic or
(1-periodic) (in superframes. m-periodic)
CFP) (1-periodic) (in allocations.
CFP) 2-In unscheduled
access: unscheduled
polling/posting
Specific No No No Yes exclusive access
Access for (EAP1,EAP2).
prioritised
(emergency)
traffic

More detail about this classification is provided in TG6 [2012]. The IEEE TG6 proposes
only one shared MAC to manage the channel access above the three physical layers. It
combines both contention and contention-less access techniques to support the variety
of data flaws that might occur in WBAN, such as burst, continuous and periodic traffic.
The coordinator splits the time or the channel into a successive number of Superframes
[TG6 2012; Cavallari et al. 2014; Movassaghi et al. 2014]. To access the channel, the
coordinator chooses one of three access modes, these are illustrated in the following:
(1) Beacon Mode with Beacon Superframe Periods. Here, the coordinator sends suc-
cessive beacon frames to specify the beginning and the end of the Superframe,
which is referred to as the beacon period. The Superframe structure of this mode
is presented in Figure 3. Table II provides more details about the MAC eaccss
mechanisms that are offered by this access mode.
(2) Non-Beacon Mode with Superframes. No beacon frame is used in this access mode,
and the superframe may only comprise either Type I phase or Type II phase as
expained in Movassaghi et al. [2014].

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:9

Fig. 3. IEEE802.15.6 beacon mode superframe structure.

Table III. WBAN Devices

WBAN devices Functionality Examples
Sensor node Samples and Wearable: added to Spo2, ECG, EEG
communicates clothes or placed on the
physiological attributes body to collect vital signs.
and provides a response Implantable: injected In Parkinson’s
to the information under the skin or in the disease, sensors
through wireless blood stream. send electrical
communication for impulses to the
anybody, anywhere and brain through
anytime. neural simulators
Actuators Administer medicine to a patient when a sensor detects Control blood
an abnormality according to the doctor’s decision. pressure, the body’s
temperature and to
treat many other
illnesses
Personal Device (PD) Set up communication between a cellular phone Can be a specialised
sensors, actuators wirelessly. dedicated unit, PDA
or a smart phone

(3) Non-Beacon Mode without Superframes. A coordinator follows an unscheduled al-

location using unscheduled Type II poll allocation. Thus, each node specifies its
time schedule in a distributive manner considering either Exclusive Access Pe-
riod1 (EAP1) or Random Access Period1 (RAP1) as access phases, during which it
competes for channel access following CSMA/CA.
From this brief description, it can be seen that these diverse channel access mech-
anisms offer the flexibility to support a variety of WBAN applications. However, the
parametrization of the Superframe and the selection of the optimal solution is not an
easy task and requires further study.

2.3. Types of WBAN Devices

As the name indicates, WBAN comprises tiny devices with communication capabilities.
Based on their functions and roles, these devices are divided into three classes. This
section presents a brief taxonomy of WBAN devices according to their functionality.
This taxonomy is summarised in Table III.
2.3.1. Wireless Sensor Node. It comprises four components: transceiver, battery, micro-
processor, and the sensor component. WBAN sensor nodes provide wireless monitoring
for anybody, anywhere and anytime. These nodes can be physiological sensors, ambient
sensors or bio kinetic sensors [Barakah and Ammad-Uddin 2012].
2.3.1.1. Wearable Sensors. These devices are added to clothes or positioned on the
body to gather vital signs, such as the SpO2 that measures the oxygen saturation

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:10 M. Salayma et al.

level in the human blood, which coincides with the cardiac cycle. The ECG sensor
that investigates the heart function by sampling the heart muscle propagation electric
waveform with respect to time. The EEG sensor that detects brain electrical activity
and the motion detection sensors that combine both accelerometer and a gyroscope to
monitor and analyse a person’s movements [Khan and Yuce 2010; Crosby et al. 2012].
2.3.1.2. Implantable Sensors. These devices are injected under the skin or in the
blood stream [Crosby et al. 2012]. In Parkinson’s disease, for example, these sensors
are used to send electrical impulses to the brain through neural simulators. Other
applications for implantable sensors can be found in Garg et al. [2004] and Maloney
and Santini Jr [2004].
2.3.2. Actuators. Actuators are used to administer medicine to a patient. The required
drug is administered directly in a predefined manner when a sensor detects an abnor-
mality or when it is triggered by an external source, according to the doctor’s decision.
Similar to a sensor node, an actuator consists of a transceiver, battery, memory, and the
actuator hardware that holds and manages the drug. The drugs could be used to control
blood pressure, the body’s temperature and to treat many other illnesses. The actuator
is activated upon receiving data from the sensors [Barakah and Ammad-Uddin 2012;
Movassaghi et al. 2014].
2.3.3. Wireless Personal Device (PD). It is responsible for establishing communication
between sensors, actuators and a cellular phone in a wireless fashion. Its main com-
ponents are: a transceiver, a rich power source, a large processor and a large memory
[Barakah and Ammad-Uddin 2012; Movassaghi et al. 2014].

3. CHALLENGES OF WBAN
As WBAN is a special type of WSN, it inherits many of its challenges. However, a
number of new challenges characterise WBAN and a number of problems require
better solutions. A survey of the differences between WBAN and WSN is given in
Latr’e et al. [2011]. Practical adoption of WBAN could not be achieved without tackling
the various technical, ethical and social challenges this type of networks faces. The
main objective is to achieve a reliable, fault-tolerant network with minimum delay and
maximum throughput while considering power consumption by reducing unnecessary
communication. User requirements such as privacy, safety, ease of use, security and
compatibility are also of great importance. The most challenging issues concerning
WBAN are detailed in this section. This survey considers a wide range of challenges
in WBANs. However, it is worth indicating that this survey will put a great deal of
emphasis on the reliability challenges.

3.1. Power Consumption

Devices in WBAN are generally battery powered. WBAN has fewer and smaller nodes
with smaller batteries compared to the other WSN, which adds more constraints on
power consumption in communication [Wu et al. 2010]. The power required by nodes
in WBAN varies according to the application type. All implanted nodes are required
to operate for multiple years. Pacemakers, for example, need to operate for at least
5 years [Cavallari et al. 2014]. Therefore, it is essential to design ultra-low power radio
transceivers. WBAN protocols have to be able to minimise power consumption without
sacrificing reliability. A common technique is to allow devices to sleep for most of the
time and thus lower the duty cycle. However, balancing between power consumption
and average end-to-end delay should be considered. The first point to consider when
choosing a wireless technology for WBAN is the power usage. WBAN peak power de-
mands in idle mode vary between 0.001mW and 0.1mW and requires up to 30mW in

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:11

active mode [Movassaghi et al. 2014]. Wireless technologies focus on minimising the
average current drawn from the battery by different means and techniques. Further
improvements are necessary to reduce the drawn peak current in sensing technologies,
radio hardware and integrated circuits. The issue of minimising interference and in-
creasing WBAN lifetime by adopting transmit power control requires further attention.
Some studies reviewed in the literature focus on scavenging energy from body heat or
motion. Power consumption challenge is deeply discussed in Cavallari et al. [2014] and
Movassaghi et al. [2014].
3.2. Heterogeneity of Devices
Since sensors in WBAN capture different kinds of data, reliability is a key issue. For
instance, sensors vary in their sensed traffic rate as this depends on the application
type and data to be sent. Bit rate values vary between less than 1kbps to 10Mbps
[Cavallari et al. 2014]. Inherently, some sensors sense more critical data than others.
Moreover, the same sensor might be in different states that vary in their criticality.
Hence, the reliability grade may change dynamically at runtime [Wu et al. 2010].
For example, human temperature might be normal and requires a normal level of
reliability, but when the temperature suddenly goes over or under the natural limit, the
reliability requirement becomes much more rigorous. As a consequence, WBAN needs
to dynamically guarantee reliability for the sensor nodes. Assuring a dynamic level of
reliability for different sensors is a challenge of great importance. These factors, as well
as sensor limitations, breakdowns and interference change the network operational
conditions, which consequently leads to incomplete and erroneous sensor data [Ullah
et al. 2012].
3.3. Reliability
A WBAN demands a high degree of reliability as it directly affects the quality of patient
monitoring. Undetected life-threatening situations can lead to fatality. A main require-
ment is that the health care professionals receive the monitored data correctly. Thus,
reliability is a crucial issue in WBAN. Reliability can be measured by the quality of
the link or by the efficiency of end-to-end communication. In order to meet user expec-
tations and achieve a reliable network, there are three basic characteristics that any
network technology, programmed service and protocol needs to address: fault tolerance,
QoS and security [C.N.A.P Staff 2013]. Designing protocols for unforeseen problems is
an essential element of WBAN design because it is necessary that the WBAN operates
continuously for users who rely on it. This requires that the architecture of a WBAN
and its design techniques should be fault tolerant. A fault-tolerant network is the one
that limits the effect of a failure, so that the fewest number of network components are
affected [Kshirsagar and Jirapure 2011; Mishra et al. 2012; C.N.A.P. Staff 2013]. It is
also built in a way that enables quick recovery when such a failure occurs. However,
WBAN medical applications create higher expectations for the quality of the deliv-
ered services as for such applications any constant breaks, pauses, delays or packet
loss could be fatal especially in emergency situations [Khan and Yuce 2010; C.N.A.P.
Staff 2013]. Hence, QoS becomes an ever-increasing requirement of WBAN. Reliable
QoS could be acquired by proposing well-designed protocols that can prioritise network
traffic. Reliable QoS also means that that all packets arrive on time and in their correct
order. This requires mechanisms that avoid or could manage traffic congestion. Net-
work bandwidth measures the capacity of the medium to carry data that is the amount
of transmitted information through the channel during a specific time.
When there are simultaneous multiple attempts in using the medium, the demand
for the bandwidth outstrips its availability as the medium has to carry more than
what the bandwidth can deliver. This leads to channel congestion. In most cases, when

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:12 M. Salayma et al.

the volume of packets is greater than what can be carried across the medium, data
can be queued or saved in temporary memory until the resources become available.
However, if the memory or queues become full, packets will be dropped causing packet
loss. Packet loss could occur due to collisions in simultaneous packet transmission. For
a WBAN, packet loss might lead to erroneous output and a life-threatening failure. It
is understood that achieving the required fault tolerance and QoS by managing faults,
the delay and the packet loss parameters on a WBAN becomes the secret to providing
a successful solution for end-to-end application quality [Khan and Yuce 2010; C.N.A P.
Staff 2013]. It can be concluded that, for WBAN, fault tolerance and QoS are correlated,
and that in order to provide a successful and reliable solution for WBAN end-to-end
applications, both fault tolerance and QoS should be considered side by side. The third
requirement that should be addressed to achieve a reliable WBAN is security. Security
in WBAN is fully addressed in Barakah and Ammad-uddin [2012], Chin et al. [2012],
and Movassaghi et al. [2014].

3.3.1. Threats: Faults, Error, Failure. A threat is a violation of one or more of the system
requirements. Some threats could be noticed without affecting the functionality of the
system. Others could be noticed but negatively affect the system functionality as they
violate the needs of the system. To understand the concept of fault tolerance, three
types of threats should be differentiated, that is fault, error and failure. Any threat
starts out as fault, it could be a physical hardware defect or a software defect. It could
occur intentionally or accidentally [Mishra et al. 2012; Raghunath and Rengarajan
2013; Alrajei and Fu 2014].
If the fault is noticed, then it is called an active fault. One example of an active fault is
the dead battery of a sensor [Alrajei and Fu 2014]. If the fault cannot easily be noticed,
then it is called a passive fault, such as a bug in the code [Alrajei and Fu 2014]. If a
fault occurs and remains unconsidered, it is possible that it extends and affects other
system components and consequently becomes an error. The error is a noticed threat,
which, once it occurs drives the system into the state of behaving wrongly [Bellalouna
and Ghabri 2013; Raghunath and Rengarajan 2013; Alrajei and Fu 2014]. In other
words, an error is an active fault. If errors propagate, they can cause system failure.
In this case, the system does not achieve the correct service it is supposed to offer. A
service is a number of coherent system’s external states. Suppose that one external
state, at least, of the system deviates from the correct service state, a service failure is
generated and the deviation is the error. For example, when a node depletes its energy
due to a dead battery fault, a loss of connection with other nodes error occurs [Alrajei
and Fu 2014]. The dead node is now not able to operate in the network and other nodes
are not able to receive data from it. This means that an error is propagated within the
network, leading to an abnormality in the network behaviour, which causes network
failure. However, not all errors lead to an overall system failure as they are not able
to affect the external state of the system, leading to other forms of failure that cause
various modes of service failure.
Though the occurrence of faults does not necessary lead to system failure, it is es-
sential to control faults from propagation to avoid system failure especially if system
recovery, maintenance and repair are complex or impossible. When the system fails to
offer a functional requirement, then it fails to offer one or more of its non-functional
requirements [Alrajei and Fu 2014]. Thus, it is essential that the system continue work-
ing according to its non-functional requirement despite the occurrences of functional
faults. This is the definition of fault tolerance.
To get a complete understanding of the concept of fault tolerance in WBAN, this
section focus on the fault tolerance. It discusses and differentiates between different
terms used in this field, namely faults, errors and failures. It presents a taxonomy of

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:13

Fig. 4. Faults at node level in wireless communication.

different types of faults and failures that could occur in wireless communication. The
management of the fault tolerance lifecycle is also illustrated in detail.
3.3.2. Classification of Faults. Like other wireless networks, a WBAN consists of nodes
and communication links. Nodes are built from hardware components and software
components. Limited resources and a harsh environment are just two examples of
factors that can make sensor nodes prone to failure [Wu et al. 2010]. For a better
understanding of the types of faults that might occur in a WBAN, faults are categorised
into two general classes according to where they might occur: the node level and the
channel level.
3.3.2.1. Node-Level Faults. The node is the basic network entity that is comprised of
hardware components such as the transceiver, memory and microcontroller as well as
the software components, such as the actual application programme installed in the
microcontroller, MAC protocols, routing algorithms and so on [Mahapatro and Khilar
2011; Alrajei and Fu 2014]. Faults associated with nodes in wireless networks include
data faults, hardware faults and software faults [Yu et al. 2007; Mishra et al. 2012;
Bellalouna and Ghabri 2013; Raghunath and Rengarajan 2013; Alrajei and Fu 2014].
Each one of those faults is depicted in Figure 4.
(A) Data Faults. This fault is related to an abnormality in sensor readings [Galzarano
et al. 2012; Mishra et al. 2012; Raghunath and Rengarajan 2013]. Various reasons
might cause data faults, for instance:
—Short Fault. This fault has a minor impact on the system and is noticed in the
presence of a spike. It is called short because it is just a short-lived oscillation on
the sensor reading signal [Mishra et al. 2012; Raghunath and Rengarajan 2013;
Alrajei and Fu 2014].
—Constant Fault. This is an invariant repetition of random values, which are
noticed as constant flat signals of sensor readings.
—Noise. This kind of fault makes sensors lose their data input and thus reduces
the signal to noise ratio.
—Drift. When reading signals deviate from the original specification and persist,
the error is considered as a drift error, which distorts the sensor data.
Data faults can be reduced by applying signal processing mechanisms that
provide services to validate sensor readings [Galzarano et al. 2012].
(B) Hardware Faults. Hardware faults can be categorised in various ways, for example
according to their duration. They can be classified into permanent, intermittent,

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:14 M. Salayma et al.

transient and potential faults [Mahapatro and Khilar 2011; Mishra et al. 2012;
Bellalouna and Ghabri 2013; Raghunath and Rengarajan 2013; Alrajei and Fu
2014]. These faults are described below:
—Permanent Faults. Permanent faults are continuously present and their effect
remains stable until the source of the fault is removed, fixed or replaced. Manu-
facturing defects on sensor hardware components is an example of such faults.
—Intermittent Faults: These faults do not persist continuously, they occur and
vanish frequently. Such faults cause a repeated faulty state of the network and,
thus, diagnosing them is not a trivial process. They normally occur due to non-
environmental conditions, for example, when a sensor loses link communication
with others due to an aging sensor component.
—Transient Faults also known as “soft faults”. These faults occur due to environ-
mental reasons such as temperature, humidity, pressure and so on. They occur
less frequently than the intermittent faults as they present themselves one time
and then vanish after which the system behaves normally. For instance, a sent
packet might not be received by the sink due to some environmental reason, but
it is very likely to be received successfully if it is retransmitted.
—Potential Faults. These faults occur when any of the node hardware resources
diminish. They disable the functionality of the system as a whole. The system
will not work again unless the deleted resource is substituted. An example of a
potential fault is when a node dies when its battery is depleted and is considered
useless unless its battery is recharged or replaced [Raghunath and Rengarajan
2013].
(C) Software Faults. The hardware of a sensor device works through software compo-
nents. Analysing software faults is necessary as any bug in software functionality
may disallow the sensor from completing the desired purpose [Raghunath and
Rengarajan 2013]. However, such faults are rarely considered in the literature. In
general, software faults are classified into six types of faults:
—Software Assignment Fault. This happens when the initialisation phase is han-
dled incorrectly.
—Software Build/Package/Merge Fault. One example of this fault is the fault that
occurs in system libraries.
—Software Functional Fault. Such faults happen due to errors in the design process
of the system that allows it to misbehave, failing to offer the required functions.
—Software Interface Fault. This is a communication error between transmitter
and receiver.
—Software Programming Fault. This is the deadlock which occurs in simultaneous
operations.
—Software Checking Fault. This occurs due to wrong data code validation or bugs
in software math.

3.3.2.2. Channel-Level Faults. Different factors can affect the communication chan-
nel between sensor devices. This includes interference, obstacles, weather conditions
and signal strength [Wu et al. 2010]. The human body absorbs RF electromagnetic rays,
this energy absorption leads to channel fading which adversely affects propagation
paths [Ullah et al. 2012]. Transmission paths are also affected by reflection, diffrac-
tion, shadowing that occurs due to rapid body movement, body structure and posture.
Body motion causes frequent changes in the network topology [Ren and Meng 2006;
Ullah et al. 2012]. All of these factors lead to communication link errors which cause
channel impairment. Channel impairment might lead to insufficient use of channel.
This effect increases in an emergency because physiological information is normally
correlated and in an emergency situation a group of sensors might be involved in

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:15

transmission. Even if every component of the WBAN is working properly when the
emergency occurs, simultaneous transmission leads to collision and consequently to
packet loss and maybe to a fatality [Liu et al. 2011; Rezvani and Ghorashi 2013]. As
a conclusion, faults associated with channel in WBAN can be classified according to
their causes into two boarder classes: channel faults due to human body nature and
channel faults due to interference.
Human Body Nature. The human body can absorb RF electromagnetic waves that
heat the surrounding tissues [Ullah et al. 2012; Movassaghi et al. 2014]. The body
usually absorbs RF at a certain rate, namely the Specific Absorption Rate (SAR) that
should be recognized under the umbrella of the regulations of the Federal Communica-
tions Commission (FCC) [Ntouni et al. 2014]. Besides tissue heating, energy absorption
causes fading, which adversely affects propagation paths. Transmission paths are also
affected by reflection, diffraction, shadowing due to rapid body movement, body size
and posture [Ren and Meng 2006; Ullah et al. 2012; Movassaghi et al. 2014]. Body
movement causes frequent change in network topology, which is a very complicated is-
sue, as nodes might move with regard to each other due to the correlation between some
moving parts of body. Fading effects lead to channel impairments. This affects sensor
channel allocation strategies and even the common coding mechanisms mentioned in
the literature to mitigate interference might not be sufficient. Channel impairments
increase Bit Error Rate (BER) and cause unreliable data transmission as critical data
might not be sent as expected and the doctor might mistakenly diagnose the patient
which could be fatal [Movassaghi et al. 2014]. Moreover, the packet loss increases data
retransmission which increases power consumption. In order to avoid this eventuality,
the transmitting power must be as low as possible. Another important challenge is the
antenna design in terms of height, size and material shape.
(A) Interference and Coexistence. Interference is one of the drawbacks of WBAN that
crucially needs to be addressed. The requirement to support per body up to 256
devices in the WBAN, and the coexistence of up to 10 WBANs in 6×6×6 meters
(TG6) [2012] lead to critical interference scenarios in WBANs due to the presence of
diifrent sensors the belong to different bodies within the same range [Alam and Ben
Hamida 2015]. Another interference issue in WBANs rises from the coexistence of
other signals related to, for instance, Wi-Fi, Zigbee, and Bluetooth etc. This can
take place in a small volume and within the same frequency band used by WBANs,
e.g., ISM band [Lo Bello and Toscano 2009]. Therefore, limited number of channels
has to be shared between these different technologies and this will lead to severe
inter-network interference that may affect the neteork reliability and deteriorate
the performance of the network when used in E-Health related applications [Dakun
et al. 2015]. Hence, the life of a patient whose health status is being monitored
may be in danger in emergency scenarios where a triggered alarm should be sent
from the sensors connected to the body of the patient to a monitoring application at
a remote location. This issue motivates the need to fully understand the network
coexistence problem and the impact of Interference on WBANs in different bands
of frequency used by the IEEE 802.15.6 standard in order to provide efficient and
effective interference mitigation and coexistence schemes [Wang and Cai 2011].
Erroneous outputs due to channel faults might result in life-threatening situations.
These conditions adversely affect WBAN reliability and performance.
3.3.3. Classification of Failures. As explained in Section 3.3.1, failures are a normal con-
sequence of faults that might occur at node and channel levels [Mishra et al. 2012;
Bellalouna and Ghabri 2013; Raghunath and Rengarajan 2013; Alrajei and Fu 2014].
The overall system might stop working due to one of the following failures:

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:16 M. Salayma et al.

Fig. 5. Types of failures in wireless communication.

—Crash Failure. When there is no response to a certain request, it is likely that the
network has suffered a crash failure. This might occur due to messages loss or due
to physical damage, which disjoins the sensors from the network. When this type of
failure occurs, there is no further output as the process remains halted.
—Omission Failure. This occurs as a result of transmission faults when the system
fails to respond to incoming queries due to a limited buffer size. Incoming and out-
going messages fail to be processed. Receive and send omissions are called channel
omissions.
—Hardware Failure. This is a consequence of hardware faults that occur in hardware
components mostly due to environmental reasons as mentioned previously. Hardware
failure might lead to node software failure.
—Time Failure. This is a periodic operational behaviour of the network. In this case,
the nodes successfully respond with a message but the message is received too late
or too early according to the specified message time interval. Thus, the synchronous
real-time system responds outside the range of the required time interval.
—Response Failure. This failure occurs when nodes send messages successfully and in
time but the message contains incorrect reading information. This might happen due
to malicious attacks, noise, software malfunction and many other reasons as specified
previously. Consequently, the accumulated results would be inaccurate which will
reduce application reliability. There are two types of response failure. If the system
responds with faulty replies, then this is a value failure. If the system responds with
unintended action at wrong time to handle a collapsed control flow, then this is called
a state transition failure. It generates irrelevant information as a response.
—Byzantine Failure. If the system generates random values at random times, then it is
considered as byzantine failure. Managing such failures is messy, as during process-
ing, this failure neglects the required processing measures and follows unintended
processing. This consequently produces corrupted messages or concurrent channel
usage due to multiple message transmission. Types of failures are summarised in
Figure 5.

3.3.4. Fault-Tolerance Life Cycle. Eliminating the occurrences of threats in the system
could not be achieved without predicting the occurrences of faults that could potentially
happen in the system [Alrajei and Fu 2014]. Eliminating faults will prevent system
failure. Fault tolerance requires fault prediction. First, the possibility of a fault occur-
ring needs to be predicted and an attempt made to prevent it [Yu et al. 2007; Mishra
et al. 2012; Raghunath and Rengarajan 2013; Alrajei and Fu 2014]. If it cannot be pre-
vented, then at least a guarantee should be built into the system to allow it to operate
in the presence of that fault. This is the second level and requires the system to be
aware of how to react in the presence of faults so as to avoid the failure of the overall
system services. This level comprises fault detection, isolation, fault identification and
fault recovery [Yu et al. 2007; Mishra et al. 2012; Raghunath and Rengarajan 2013;
Alrajei and Fu 2014].

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:17

Fig. 6. Fault tolerance life cycle.

Fault tolerance aims to eliminate threats of the system by finding ways to allow the
system to survive and to continue operating in the presence of errors according to its
design goals without deteriorating the performance of the services it is supposed to
offer. Fault tolerance could be achieved through two methods. The first method is fault
masking, that is to hide faults and prevent them from extending to errors [Alrajei and
Fu 2014]. The second method is fault isolation that is to remove the fault or the source
of fault from the system either manually or automatically. Each step in the lifecycle of
fault tolerance as well as in choosing the method is considered to be a research topic
on its own and depends on the application and the criticality of faults. This subsection
illustrates each step in the fault tolerance life cycle. The steps of fault tolerance life
cycle in wireless communication are depicted in Figure 6.
3.3.4.1. Fault Prevention. This avoids the occurrences of faults and takes place in
three stages of the WBAN application, as follows [Mishra et al. 2012; Raghunath and
Rengarajan 2013]:
—Specification Phase. This phase must ensure that incomplete and equivocal specifi-
cations are avoided.
—Design and Deployment Phase. The design of the network topology, connectivity
and link coverage must be insured during this phase. This phase also includes the
necessity of choosing RF and hardware components with a suitable wireless standard
in terms of quality, reliability and flow.
—Monitoring Phase. To constantly watch out for the status of the network for functional
degradation and incorrect usage of resources by monitoring node status, link quality
and the level of congestion and take reactive action when necessary.
3.3.4.2. Fault Detection. Due to the dynamic nature of wireless communication, fault
prevention does not guarantee 100% prevention of faults [Yu et al. 2007; Mishra et al.
2012; Raghunath and Rengarajan 2013; Alrajei and Fu 2014]. Therefore, a detection
phase is required to identify the unexpected fault in the network. Basically, fault
detection can be categorized as online and offline detection.
—Offline Detection. This is often applied in wired networks using special programmes
during the idle time of the network.
—Online Detection. Online detection is real-time detection and is adopted in wireless
networks. It is classified into explicit and implicit detection.
(A) Explicit Detection. This is an application-specific fault-detection technique. It is
performed by sending an alarm when the event/action is detected. For exam-
ple, in WSN, there is a predefined threshold used to detect the misbehaviour

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:18 M. Salayma et al.

of sensing applications [Yu et al. 2007]. These conventional threshold tests are
unsuitable for WBAN fault detection, since WBAN is characterised by the com-
plexity of a heterogeneous sensing environment [Mahapatro and Khilar 2011].
The hybrid resource capabilities of nodes in terms of their resource capabili-
ties, installed software, and data rates limits the adoption of a fixed threshold
to check their status in WBAN. For example, ECG and SpO2 sensors sample
different readings and detection of the faulty nodes by comparing energy levels
might deliver inaccurate results. Moreover, medical sensors might be introduced
under various situations, such as normal situations and emergency situations.
For example, temperature readings might be higher or lower than the aver-
age normal temperature, which could be life, threatening. Consequently, WBAN
has two levels of variation in the reliability requirements of its sensors: amongst
heterogeneous sensors and within the same sensor, as it might run in different
situations. In addition, high threshold values lead to several missed detections,
while values that are too low result in false positives. Thus, adopting a constant
or fixed threshold value for overall WBAN components may not function well if
the properties of the sensors vary, or if the scene and the environment change,
because it generates faulty observations, which produce unsatisfactory results.
For WBAN, explicit detection can be classified in to two types which are:
—Anomaly Detection of Abnormality due to Malfunction of Sensor Node. Detec-
tion can be done by measuring the average receiving rate of the previous data
and comparing the result to a predefined dynamic threshold. It is necessary
to classify the level of critically of the malfunction sensor because action is
taken according to its level [Jeong et al. 2014].
—Anomaly Detection due to Human Body Abnormality. This kind of detection
reflects that there is a problem in the human body that could be life threaten-
ing. This requires a detailed set up since each medical situation has its own
parameters and requires a different response. This kind of detection can be
further classified into: detection of an accidental change in the body medical
data and the detection that medical data reached a threshold or a reference
value [Jeong et al. 2014].
(B) Implicit Detection. This type of detection requires overall network management
because it targets faults that happen due to abnormal phenomena such as
vulnerable environment conditions [Yu et al. 2007]. There are two ways to adopt
implicit detection: active and passive detection which are discussed below:
—Active Model (Proactive). In this model, nodes send continuous messages to a
central controller to indicate that they are still alive. If a certain specified time
period has elapsed without receiving a message from a node, the controller
can tell that the node is dead [Yu et al. 2007; Mishra et al. 2012; Alrajei and
Fu 2014].
—Passive Model (Reactive). In this model, a node sends the alarm to the con-
troller only when a fault or network corruption is detected. In this case, nodes
save energy as they are requested to send keep alive messages constantly to
the controller, thus they stay inactive most of the time.

Figure 7 summarises different types of online fault detection in WBAN in particular.

3.3.4.3. Fault Isolation. After the fault is detected, it is necessary to analyse the
nature of that fault in terms of its type, behaviour, and characteristics. A procedure
should be followed to isolate that fault [Yu et al. 2007; Mishra et al. 2012; Raghunath
and Rengarajan 2013].

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:19

Fig. 7. Fault detection in WBAN.

3.3.4.4. Fault Diagnosis. Fault diagnosis is essential to properly identify the isolated
faults in terms of their characteristics and behavioural nature. Faulty nodes should
also be distinguished from suspicious nodes and from other irrelevant triggers. This
could be achieved by applying historic data and sharing knowledge. The state-of-art
fault solution models either manage faults at the component level of the node or at link
communication level between nodes. However, although there are several accurate
fault detection techniques, there is no comprehensive model to comprise accurate fault
diagnosis and fault recovery approaches [Yu et al. 2007; Abedi et al. 2011; Mishra et al.
2012; Raghunath and Rengarajan 2013].
3.3.4.5. Fault Recovery. In this phase, the network is restructured and reconfigured
to minimise or stop the detected faults from generating further impact on the network.
Fault recovery techniques are classified into two broad areas according to the type of
fault: the recovery of faults that might occurred at sensor node level and the recovery of
faults that might occur at channel level [Yu et al. 2007; Mishra et al. 2012, Raghunath
and Rengarajan 2013; Alrajei and Fu 2014].
(A) Node-level recovery: In this case, the first approach is to directly isolate the failed
node from the network, such as isolating it from routing communication [Yu et al.
2007; Raghunath and Rengarajan 2013; Alrajei and Fu 2014]. In clustered net-
works, if the failure occurs in the cluster head (CH), then its nodes members join
the neighbour clusters. Conventional WSN fault tolerance techniques follow sensor
replication by deploying redundant sensors through following appropriate redun-
dancy mechanisms. For example, WSN was commonly used to sense a region and
emit the sensed data to the sink. If a node fails to provide the correct data, the
sink receives the correct data from the redundant deployed nodes [Liu et al. 2009].
However, unlike WSN, WBAN is considered a sparse communication network be-
cause of the short area communication space and the limitations on the number
of nodes, thus, there is no way to establish packet redundancy. Moreover, adding
extra sensors increases the interference among sensors and might be stressful
[Mahapatro and Khilar 2011]. Nevertheless, it is worth indicating that our types
of redundancies that are commonly followed in WSN, are as follows:
—Hardware Redundancy. This is achieved by adding extra physical components,
such as additional sensors, additional batteries and so on [Raghunath and
Rengarajan 2013; Alrajei and Fu 2014].

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:20 M. Salayma et al.

—Software Frequency. This is done by replicating software code. For example, soft-
ware components can be allowed to reconfigure the management of the software.
Another technique is using mobile code technology.
—Time Redundancy. This is done by repeating the required processes multiple times.
—Information Redundancy. Two ways to achieve information redundancy are:
—Active Replication. This technique allows all replicas to process all requests. A
failed node is recovered by duplicating its information by distributing redundant
sensed data to other nodes. Thus, each node requires extra memory space to buffer
its own data and even the data and the replicas for other nodes. Redundant copies
are used to recover the failed sensor data. Beside the extra memory hard require-
ment, this approach is prone to high message communication overhead in sending
the update messages [Mishra et al. 2012; Raghunath and Rengarajan 2013].
—Passive Replication: This allows only a single replica to process a single request, and
other replicas are allowed to process the request only when the current replica fails to.
(B) Channel-Level Recovery. A fault-tolerant system should be able to tolerate channel
impairments. Finding ways to coordinate channel usage among nodes according to
their criticality is one solution. Distinguishing nodes’ criticality means assigning
them different priority levels [Wu et al. 2010]. Sensors that are used in WBAN are
usually classified into two parts: medical sensors that are used to sense physio-
logical information and non-medical sensors that are used to sense environmental
information [Liu et al. 2011; Rezvani and Ghorashi 2013]. It is always the case
that, while some medical sensors work in their normal situations, other medical
sensors might suddenly require high QoS and exclusive resources when emergent
situations occur. In this case, it does not make sense to provide all sensors in WBAN
identical conditions and resources because less important data might be achieved
before the critical one resulting in unreliable and an untrustworthy, erroneous
WBAN. To avoid this, nodes in emergency situations must transmit their data as
fast as possible and should be assigned enough bandwidth to allow the complete
information achieved in a real time, even if this would be at the expense of other
less important information sensed by those nodes which work in normal situations
or which sense environmental information. What makes the problem worse is that,
despite the heterogeneity of WBAN sensor nodes, their sensed data is correlated.
For example, both the ECG and hemodynamic signals, such as blood pressure,
have information mutually correlated due to the physiological inter-relation of
the mechanical and electrical functions of the heart [Mahapatro and Khilar 2011].
Moreover, physiological information also has a direct correlation with environmen-
tal information. It is mentioned early in this section that due to this correlation
nodes might produce erroneous results in emergency situations as nodes will try
to transmit simultaneously. This all-at-once data transmission leads to collisions,
which consequently increases packet loss and produces erroneous results. This
could lead to fatal situations. This emphasises the necessity of prioritising nodes
according to their situation, and according to the type and the criticality of the
data they sense [Wu et al. 2010]. Various approaches can be adopted to assign
priorities for heterogeneous nodes in different situations. The commonly used one
is to check the nodes’ situation according to a specified threshold [Mahapatro and
Khilar 2011]. To overcome the channel interference problem, other adaptive chan-
nel allocation approaches are required to enable successful coexistence between
the various communication technologies that might coexist in the area [Zhou et al.
2011; Wu et al. 2010]. Dynamic channel switching can be followed according to
the standards adopted. Adaptive channel allocation to solve coexistence problem
is beyond the remit of this research.

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:21

Fig. 8. Fault tolerance solutions in WBAN.

4. SOLUTIONS FOR FAULT TOLERANCE

WBAN should provide measurable, predictable and, when necessary, guaranteed ser-
vices. To ensure this and improve the reliability, a tremendous amount of work has
been proposed in the literature. Studies to date typically seek to improve QoS and find
ways to achieve a secure WBAN, however, few works addressed the fault tolerance
issue. Each proposed approach aims to tackle fault tolerance from its own perspective.
Nevertheless, there is no work standing firmly enough to address all types of faults or
the entire life cycle of fault tolerance in WBAN. The taxonomy of WBAN fault-tolerance
solutions which are going to be illustrated in this section are depicted in Figure 8. This
section examines recent literature pertaining to achieving a fault tolerant and high QoS
WBAN. Interference is classified earlier as one of the sources that causes channel faults
and which has a huge impact on WBAN reliability. Interference however, is a wide topic
and has huge research directions by its own. Therefore, interference management so-
lutions are added as separated subsection in this section. As detailed in Section 3, fault
prevention is considered in reference to network specification and topology deployment
during the early design stage when developing a network [Raghunath and Rengarajan
2013; Alrajei and Fu 2014]. The proposed protocols aim to prevent faults by monitoring
node status, link quality and level of channel congestion [Alrajei and Fu 2014]. The
majority of work focuses on proposing fault detection techniques, omitting the necessity
of fault recovery [Zhou et al. 2011; Otal et al. 2009; Ali et al. 2010; Wu et al. 2010; Liu
et al. 2011; Rezvani and Ghorashi 2013]. Literature varies between proposing generic
algorithms [Zhou et al. 2011; Wu et al. 2010; Mahapatro and Khilar 2011; Galzarano
et al. 2012; Abarna and Venkatachalapathy 2014], MAC protocols and enhancements
to existing wireless standards such as IEEE 802.15.4 [Otal et al. 2009; Rezvani and
Ghorashi 2013], IEEE 802.15.4a [Liu et al. 2011], and IEEE 802.15.6 [Liu et al. 2011].
A number of the proposed fault tolerance approaches are targeted towards node faults
in WBAN, others are geared towards coping with channel vulnerability regardless of
nodes abnormality, but very little consider both node and channel levels impairments.
According to the literature reviewed, this survey classifies fault tolerance approaches in
WBAN into three broadbands according to the type of fault level they consider, namely
fault tolerance approaches at node-level, fault tolerance approaches at channel level,
and fault tolerance approaches at both node and channel levels. Each one of these
three classes is further classified into two major categories: Centralised approaches
and Distributed approaches. Several techniques that follow the centralised manner
have been suggested, such as performing data fusion [Mahapatro and Khilar 2011],

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:22 M. Salayma et al.

adopting a specialised layer [Zhou et al. 2011; Mahapatro and Khilar 2011, Galzarano
et al. 2012], following clustering techniques [Abarna and Venkatachalapathy 2014]
and developing context awareness [Liu et al. 2011; Rezvani and Ghorashi 2013]. Yet,
there are also a distributed middleware and context aware techniques that are clas-
sified under the category of the distributed approaches. A new trend that is neither
centralised nor distributed, which is based on the cloud is also considered in this survey
as one of the fault tolerance techniques proposed to cover node-level faults.
Due to the limited communication space of WBAN, it is inappropriate to adopt sensor
replication as a recovery technique, thus, the majority of those techniques consider
isolating and replacing faulty nodes as a recovery method [Mahapatro and Khilar 2011;
Galzarano et al. 2012; Abarna and Venkatachalapathy 2014; Jeong et al. 2014]. This
review emphasises that achieving both fault tolerant and high QoS is a key concern
in WBAN. It also suggests some trends to adopt in order to achieve more accurate,
feasible, real time and efficient fault tolerance framework for WBAN.

4.1. Fault Tolerance Approaches at Node Level

Approaches that fall within this band are typically limited to those which occur in
WBAN nodes. Those approaches do not consider channel vulnerability due to body
or surrounding environment, they however, detect and deal with abnormalities that
occur at the node level due to data faults, hardware faults as well as the abnormality
in body physiological information. Approaches followed in literature in this realm are
categorised into centralised, distributed, and cloud-based approaches. WBAN studies
that have adopted such techniques are presented in the following subsections.
4.1.1. Centralised Node Fault Detection using Fusion Centre. In the centralised approaches,
network faults are managed by one node, which is the central controller, sink or base
station. In this way, complex computations and processes are shifted from resource
constrained nodes and performed by a resource rich node, in order to increase power
efficiency [Yu et al. 2007; Kshirsagar and Jirapure 2011]. Active detection is normally
considered a centralized approach as the central node receives periodic and constant
messages that indicate the situation of the nodes and network by sending queries to
make inferences about the network and the individual nodes’ statuses used by the
controller to identify both failed and suspicious nodes.
One centralised node fault detection is based on data fusion techniques. In this
approach, a specific node aggregates the decisions from the remaining network nodes
[Yu et al. 2007]. This so-called data fusion node can determine the abnormal status
of other nodes by comparing data from a set of nodes. This technique is very efficient,
especially for data centric applications that require detection precision, to manage the
number of recovered sensors and the communication costs. However, there must be a
balance between the accuracy of the detection and the network’s energy consumption.
The work in Mahapatro and Khilar [2011] is the only work found in literature
that is based on a centralised data fusion fault detection in WBAN. A PDA provides
continuous data transmission and performs real-time analysis of physiological sensor
data. PDA derives relevant physiological information from sensor data, according to
which it diagnoses faulty nodes. Time redundancy can be used to address the detection
of intermittent and permanent errors. Mahapatro and Khilar [2011] observed a
tradeoff between choice of the time detection parameter T, and the latency and
accuracy of the error detection. However, it is important to note that there must be
some level of context awareness before the data fusion centre can generate results.
For example, an increase in heart rate does not necessarily mean a patient is having
a cardiac episode, the patient might be engaged in some form of sport (for example,
running) and that causes a change in their physical activity level. Such critical issues

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:23

were neglected in this work. Changes in the internal and external states of the body
can be analysed according to triggering events [Korel and Koo 2007]. Moreover, as a
centralised technique, this technique is prone to a single point of failure, as the error
detection is performed by a centralised PDA, and no discussion was made of what
would happen if the PDA crashes or produces faulty results. There can be several
fusion sensors, the number of which depends on the requirements of the area, in
terms of the communication range between nodes and the region of the application
of interest. Although the physiological information extracted is used to decide the
occurrences of faulty nodes, it could be argued that the information extracted could
be used for temporary recovery, such as by substituting one piece of information for
another, allowing a temporary flow of information. This temporary recovery continues
until permanent action can be taken such as node isolation and replacement.
Although centralisation is considered an accurate and efficient approach, its major
drawback is the failure of a single central node to manage everything. This is because
the central node is prone to data traffic concentrations, which leads to rapid energy
depletion in the controller [Yu et al. 2007; Kshirsagar and Jirapure 2011]. Moreover, in
multi-hop communication, nodes close to the central node will forward more commu-
nication messages to the central node, which will also cause early energy depletion in
those nodes leading to what is called the hot spot problem. The single point of failure
problem requires adoption of a distributed fault detection model.
4.1.2. Distributed Node Fault Detection. To relax the pressure of duties, the central con-
troller is responsible for avoiding the problem of single point of failure, level of fault
detection is shifted to local nodes by allowing them to report the awareness of their sta-
tus, thereby enabling them to take decisions [Yu et al. 2007; Kshirsagar and Jirapure
2011]. Obviously, this minimizes the number of messages the nodes are required to
transmit to the central node, which reduces the energy consumption and network con-
gestion. Generally, distributed techniques can be fully adopted selflessly at each node,
by allowing the node to diagnose malfunctions in its physical hardware components.
Distributed fault detection can be adopted by following clustering techniques. A com-
mon distributed detection approach in WSN, is the coordination between neighbours.
However, such a technique has yet to be considered in WBAN. The three distributed
fault tolerance techniques in the literature are given in the following discussions.
4.1.2.1. Node Self-Detection. Nodes are allowed to self-detect faults, taking decisions
about their own status without the need to consult the coordinator [Yu et al. 2007;
Kshirsagar and Jirapure 2011]. One approach presented in literature for distributed
self-detection in WBAN is by adopting a middleware layer in each node to be respon-
sible for detecting and possibly recovering faulty status, without the need to rely on a
certain node to perform complex computations. A middleware is a layer that acts as a
bridge between the conventional network stack layers, to perform some functionality
and provide some services that are not provided by the current stack layers, without
violating the traditional layers’ tasks [Wikipedia The Free Encyclopedia 2015]. In this
technique, a set of software modules are used in each sensor node to perform tasks
beyond those the legacy stack layers perform.
One work that uses a separated layer for node fault detections is presented in
Galzarano et al. [2012]. Galzarano et al. [2012] concentrated on data faults in WBAN,
by analysing the level of decreased accuracy after applying four types of data faults to
readings acquired by sensors in a human activity recognition application. Galzarano
et al. [2012] studied the impact of faults on the quality of that application. Those
faults are: short faults, noise, constant faults, and accumulative faults. Galzarano
et al. suggest that the reliability of WBAN can be enhanced by adding autonomic
elements that perform self-healing operations, and thus enhance data fault tolerance.

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:24 M. Salayma et al.

The idea is realised by adopting a self-healing layer with the aim of detecting and
recovering from such situations. The layer is based on the Signal Processing In-node
Environment (SPINE) framework [Bellifemine et al. 2011], which is built on top of
the existing SPINE2 framework [Galzarano et al. 2012]. Experimental results confirm
that the defined autonomic elements can mitigate faults affecting the sensor data,
and improving the system behavior correctness [Galzarano et al. 2012]. However, this
technique is based on image-processing algorithms that require complex processing
capabilities, and resource rich devices which are not suitable for energy constrained
sensor nodes. Moreover, Galzarano et al. [2012] considered only one type of data fault,
short faults. The accuracy and performance of the self-healing layer should be proven
by testing it in reference to the occurrence of other types of data faults.

4.1.2.2. Clustering Detection. This technique distributes fault management, by divid-

ing the network into groups of clusters where distributed fault management is adopted
for each group. Thus, a virtual Skeleton communication is created [Yu et al. 2007]. Cen-
tralisation is adopted in each cluster head (CH), which is then becomes responsible for
detecting faults by exchanging messages with the members of its group. Identification
of faulty nodes can be achieved by applying a pre-defined failure detection rule to each
CH. The information detected can be propagated to all other clusters through a gate-
way, which is considered a neighbour to two CHs from different clusters. Hence, the
overall network is managed through self-aware clusters [Yu et al. 2007]. A study fol-
lowing this approach in WBAN is presented in Abarna and Venkatachalapathy [2014].
Abarna and Venkatachalapathy [2014] tried to alleviate a drawback of single of point
of failure, as presented in Zhou et al. [2011] and Wu et al. [2010]. To achieve this
aim, Abarna and Venkatachalapathy [2014] followed a cluster-based approach to fault
tolerance in WBAN. Abarna and Venkatachalapathy [2014] applied a hierarchal archi-
tecture, where each group of physiological nodes are connected to one CH. The CHs are
interconnected and connected with a Wireless Local Gateway (WLG), which is linked
with a Hospital Gateway (HG). Then, each CH assigns priority values to its nodes that
reflect their fault tolerance level according to three pieces of aggregated information:
physiological collected information, environmental information, and the node’s physi-
cal status. The three values collected are then compared with three threshold values
and node priorities are assigned accordingly. Nodes with a high priority are processed
first. A scheme for CH fault detection and recovery is also proposed. However, although
this work considers three threshold values, it has not been shown why, how and what
the chosen values are. Once again, the threshold test is not suitable for fault detection
in WBAN, as it is characterised by the complexity of heterogeneous sensors that might
follow different critical cases requiring different dynamic thresholds. Moreover, it con-
siders WBAN as a homogenous network comprising homogenous sensors in terms of
data rate, packet size, capabilities and physical status. For this reason, Abarna and
Venkatachalapathy [2014] suggested that, in case of a CH failure, its members can
associate with any other CH, as all CHs achieve identical processing and capabilities
and are all provided with the same threshold values. However, this assumption is irra-
tional for WBAN, as it has a characteristically heterogeneous nature. On homogenous
networks, when a fault occurs, all the CHs might discover the faulty node, and thus
will send failure information simultaneously, leading to collision and packet loss. The
increased use of control packets between CHs adds an extra overhead, which might
cause congestion on the channel.
While Abarna and Venkatachalapathy [2014] claimed that their technique reduced
control overheads and energy consumption rates significantly, they have not considered
the power consumption in their performance evaluation. Clearly, this work adopts a
level of redundancy to recover the controller failure by adopting several CHs. Adopting

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:25

such an approach in WBAN, which is considered a very short scale network, is not
trivial. Indeed, an intensive study and in depth analysis is required concerning the
optimal number of CHs, their position on the body, as well as the number of maximum
sensors in each cluster. In addition, this work does not specify the action to be taken
after assigning priorities to nodes, if the only action was to wait for faulty node isolation,
there would be a huge delay before the faulty node could be replaced which could be
fatal in an emergency situation. Thus, there must be a real-time recovery mechanism in
place, even if this is considered as temporary solution, such as the previously presented
data fusion technique.
4.1.3. Cloud-Based Node Fault Detection in WBAN. Cloud-based fault detection in WBAN
is neither a centralised nor a distributed approach, as both the node and the controller
are not responsible for tracking the node’s status or performing a reactive response.
A third party, outside the WBAN, achieves fault tolerance, i.e., the cloud-computing
Infrastructure as Service (IaaS)-based application. Jeong et al. [2014] proposed a visual
monitoring system (VMSFT) to monitor patient’s conditions and to tolerate sensor
failures immediately. The work presented in Jeong et al. [2014] differs from other works,
as it considers a large number of WBAN adopted by multiple patients simultaneously,
therefore, it is considered a local monitoring system for a group of people within target
area. VMSFT obtains patient’s data through the IaaS application and according to
this data, it can infer sensor failure. This work performs two types of explicit anomaly
detection, anomaly because of sensor failure, and anomaly because of a problem in the
human body. VMSFT can provide three responses according to the criticality of the
detected sensor anomaly. The level of criticality is assigned according to the position
of the sensor in the body. A high level is assigned for sensors close to the heart, and
a low level is assigned to hands and leg sensors. All other sensors are assigned to
the middle level. If VMSFT is connected to an emergency room, it is possible to care
for patients proactively. High-level cases are sent to emergency room as well as to
managers, middle-level incidents are sent to managers, but if a delayed response is
expected according to a specified time, calls are sent to the emergency room, whereas
a low-level response incidents are monitored periodically by the manager. However,
for abnormalities in the human body, if data changes abruptly, status is sent as an
emergency directly, but if the body data approaches a specified reference threshold,
the GP is informed to prompt a suitable proactive response. A hierarchical scheme is
suggested for management within a cloud service infrastructure. However, Jeong et al.
[2014] considered the position of the sensor as a criterion to decide the criticality of
the data readings. Sensors positioned on the body extremities have a lower priority
in triggering an emergency response, as their readings are of lower criticality than
other sensors, for example, those placed close to the heart. It is very popular, however,
for sensors being used to detect heart rate to be placed on the hand. Thus, other
criteria must be applied to decide the level of criticality. Moreover, cloud computing–
based applications for WBAN cannot be adopted, unless there is a full guarantee of
trustworthiness, which is still a major drawback of cloud computing, as level of privacy
and security are not wholly determined.

4.2. Fault Tolerance Approaches at Channel Level

The approaches that fall within this band are typically limited to channel-level im-
pairments that occur in response to the body and environmental changes, and does not
consider node-level faults. In other words, they consider nodes as fault free. Approaches
found in literature for detecting and dealing channel vulnerability are generally
centralised. Some approaches enable the controller to perform some level of context
awareness to classify traffic according to the importance of the sampled data, to

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:26 M. Salayma et al.

improve network QoS and to tolerate channel impairments [Ali et al. 2010; Liu et al.
2011; Rezvani and Ghorashi 2013]. In other approaches a controller might adopt an
extra layer or middleware that prioritises traffic, tolerate and manage the channel’s
vulnerability [Zhou et al. 2011; Wu et al. 2010]. Such techniques are presented in the
following subsections.

4.2.1. Centralised Context Aware Fault Detection. According to Korel and Koo [2007], con-
text refers to any information that characterises an entity’s situation. Context aware-
ness is the ability to diagnose entities’ internal or external states. To achieve a fault
tolerant WBAN, the entity for which the context needs to be analysed is the sensor node.
More generally, context-aware WBAN is a network in which wearable and implanted
sensors’ situations can be described in accordance with the surrounding environment,
according to which, their behaviour can be modified. As stated previously, WBAN has
several challenging characteristics, one of which is its vulnerability to context and en-
vironmental changes [Wu et al. 2010]. Contextual changes include the body’s motion
and activity, and the temperature of the surrounding environment among others. Such
factors affect the reliability of the network. Context awareness can assist in the in-
terpretation of any physical and physiological data being monitored, and accounts for
the current situation as it affects the body and the surrounding environment. Thus, it
can be used as a method to determine tolerance to impairments that might occur in
the network channel. Therefore, context awareness plays a key role in WBAN, by en-
hancing reliability and helping to facilitate long-term system monitoring. Supervision
of the context awareness process in terms of data gathering, classification of context,
transitions and events, and extraction of context recognition, can either be centralised
at a single resource rich sensor node (usually the controller), or implemented in a dis-
tributed manner to a certain level at the resource constrains nodes. This subsection
presents literature that has adopted centralised context awareness for fault tolerance
in WBAN at the channel level.
Rezvani and Ghorashi [2013] studied the channel impairments that might occur in
WBAN, because of fading and body postures which cause packet loss in a star topology.
Rezvani and Ghorashi [2013] considered both medical and other types of applica-
tions, proposing an adaptive channel allocation superframe-oriented MAC protocol for
WBAN. The proposed protocol supports the heterogeneous characteristics of different
applications by separating medical and non-medical applications. Moreover, Rezvani
and Ghorashi [2013] considered correlations that may occur in medical traffic in emer-
gencies and splits that traffic into normal and emergency medical traffic. The separa-
tion between non-medical, normal medical, and emergency medical traffic is adaptive,
according to channel condition and the user’s medical situation. As medical data is usu-
ally periodic, and in order to improve the efficiency, medical nodes can follow TDMA
synchronisation. However, because non-medical data is flexible, it follows that there is
a contention-based access mechanism for traffic transmission. Moreover, according to
the condition of channel deep-fading, the interval between consecutive transmissions
changes adaptively. To summarise, the proposed protocol tolerates channel faults by
adapting the channel allocation to the classified sampled data that consider fading
effects. The proposed algorithm outperforms both IEEE 802.15.4 and IEEE 802.15.6
MAC in terms of reliability, channel utilisation, and power consumption; this is because
of the successful data packets transmission due to the adaptive resource allocation.
However, although Rezvani and Ghorashi [2013] considered emergency traffic, it
does not explain the classification process applied for different types of traffic (for
example, if they followed a threshold comparison mechanism). The proposed protocol
still follows the CSMA mechanism in emergency situations, and while it is considered
a smart solution for WSN, it does not guarantee collision avoidance in WBAN, which

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:27

would adversely affect reliability and therefore be fatal in emergency situations,

and thus other channel access mechanisms must be proposed for contention-based
channel access in WBAN. Moreover, context awareness computation is centralised and
performed by a slot sent by the hub itself, this leads to the common problem, namely
a single point of failure. Notice that, the slot in the hub examines the channel to check
if it is busy or not. If it is busy, then it decides there is an emergency situation. This
mechanism, however, must be sufficiently robust to consider all types of nodes in the
network. For example, context awareness can be performed in a distributed manner
throughout all WBAN nodes; this cannot be achieved unless nodes are allowed to
adopt cross layering approaches to a certain level, at all stack layer levels. The core
problem discussed is the occurrence of collisions in emergencies, due to the correlation
in medical data. However, data correlation could also be considered as a feature to
harness to improve WBAN fault tolerance. For example, the data fusion technique
discussed in the previous subsection can be adopted to provide a temporary solution
to recover information from some faulty nodes.
Liu et al. [2011] proposed a hybrid superframe-based MAC protocol (CA-MAC) that
allocates transmission bandwidth according to WBAN context. CA-MAC adjusts the
transmission priority and access strategy according to the traffic and channel status,
respectively. The CA-MAC protocol changes the duty cycle according, adaptively, to
human activity, or environmental conditions, while also considering power consumption
to prolong the network’s lifetime. CA-MAC comprises three parts: the beacon frame,
the contention-based period that adopts the slotted CSMA/CA and which changes
dynamically according to the level of fading in the channel, and the third part is the
contention free period which follows the TDMA mechanism. The TDMA part comprises
two slots types. This first one is schedule-based slots and the second is polling-based
slots. Schedule-based slots are adaptively allocated for nodes based on traffic intensity.
However, polling-based slots are assigned by the nodes’ coordinator, which requests
them through poll messages, a situation which might occur in emergency situation;
thus, such slots are normally inactive. Under normal situations, each node sends data
to the coordinator, and to its allocated slot, after which it enters a sleep state. All the
nodes should wake up at the beginning of a new superframe to check for a new beacon
frame. The coordinator analyses the data, and if it detects anomalies, it sends a new
beacon frame in the next superframe to inform the emergency context. Nodes that
work in a specific context will have an increased sampling rate and will assign more
transmission slots to accommodate emergency requirements. Other sensors, might
acquire fewer slots and a lower sampling rate to reduce energy consumption. The CA-
MAC performance outperforms both 802.15.4 MAC and H-MAC in terms of delay and
power consumption, especially in emergencies.
Apparently, the proposed protocol inherits the basics of the IEEE 802.15.4 beacon en-
abled superframe structure. In addition, although the proposed algorithm can tolerate
an emergency delay, it still does not guarantee reliable data transmission in real time
entirely. This is because nodes are not informed about emergency situations directly,
as they have to wait for subsequent superframe structures, according to which they
can then change their remission slots and sampling rates. Thus, there would be some
delay handling such situations, which might lead to critical packet loss. Moreover, a
centralised coordinator performs the entire context awareness procedure, thus, it is
prone to a single point of failure.

4.2.2. Centralised fault Detection Using Middleware. Zhou et al. [2011] are the first to pro-
pose a middleware layer, called BodyQoS that tolerates channel faults and impairments
by performing an adaptive channel allocation strategy through which the QoS of
WBAN is improved. BodyQoS is mainly a collection of software modules, which reside

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:28 M. Salayma et al.

between MAC and transport layers Zhou et al. [2011]. BodyQoS comprises a virtual
MAC (VMAC) that makes the system radio-agnostic when following an abstraction
technique [Zhou et al. 2011]. It schedules bandwidth adaptively, without knowledge
of the implementation details of the underlying MAC protocols. Thus, the VMAC can
be easily ported from one radio platform to another, allowing it to support an array of
different MACs, including Time Division Multiple Access (TDMA), CSMA, and hybrid
approaches. VMAC re-allocates resources adaptively, to meet QoS requirements.
BodyQoS guarantees a statistical bandwidth for reliable data transmission when
channel impairments occur because of radio interference or body fading effects. If the
wireless resource available is not able to satisfy all QoS reservations, packets with
the higher priority will be processed before the lower priority ones. A resource rich
aggregator is responsible for the majority of channel scheduling computations, and
can be used to minimise the load on resource constrained sensor nodes. BodyQoS per-
formance outperforms conventional solutions, as it tolerates channel impairment with
a minimum overhead. However, although the proposed technique should guarantee
reliable data communication under channel impairments, it does not consider the
dynamic change in the reliability requirements of sensors, as it assigns a specific fixed
priority to sensor nodes based on the level of criticality and data type. In addition,
BodyQoS is prone to single points of failure, as the entire channel’s fault detection,
traffic prioritisation and classification is managed by one central node.

4.3. Fault Tolerance Approaches at both Node and Channel Level

The approaches that fall within this band comprehend abnormalities that might occur
at both nodes and channel levels in WBAN. Two Approaches have been found, one
follows a centralised middleware while the other adopts a context aware technique
implemented in a distributed manner. The WBAN studies that have adopted such
techniques are presented in the following subsections.
4.3.1. Centralised Fault Detection Using Middleware. The work in Wu et al. [2010] aims to
improve BodyQoS that is presented in Zhou et al. [2011]. Wu et al. [2010] suggested
an adaptive and flexible fault-tolerant communication middleware (AFTCS). AFTCS is
made to tolerate channel impairments by adjusting the priority queue of each sensor
node adaptively, according to three pieces of sensor information: physiological, envi-
ronmental and physical status of the sensor. According to the information received,
a central node allocates the channel resources adaptively after considering the pri-
ority level of the sensor nodes and their current data criticality level. The resource
computation and effective bandwidth calculation approach was inspired by the work
in Zhou et al. [2011]. Simulation results revealed that AFTCS can tolerate channel
impairments by reducing the rate of packet loss and the delay in critical data trans-
mission. However, the centralised controller checks the three sources of information
for each node according to a single specified activation threshold, which has the value
0.4, and according to which the priority for each sensor will be adjusted. It was not
demonstrated why this threshold value was chosen. Additionally, a fixed threshold test
is not suitable for fault detection in WBAN, as it is characterised by the complexity of
heterogeneous sensors. Furthermore, each sensor might encounter different situations
that change its criticality. For instance, two settings are considered critical for a body
temperature sensor, e.g., to determine if the temperature is below a specific value and if
it is higher than another certain value. Therefore, choosing as single threshold is not a
realistic option for all cases of sensors. As channel computations and adaptive resource
allocations are performed by one central node, this technique is prone to a single point
of failure. It can be argued that, although the controller receives information about the
node’s physical status, it does not consider the effect of node-level faults.

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:29

4.3.2. Distributed Context Aware Fault Detection. Context awareness can be applied in each
resource constrained sensor nodes, without the need to rely on the controller. Several
approaches can be adopted to achieve this aim. Otal et al. [2009] explored the QoS and
reliability in WBAN and presented a novel cross-layer based on a fuzzy logic technique
to schedule transmissions in WBAN. The Distributed Queuing Body Area Network
(DQBAN) protocol was proposed by Otal et al. to modify the beacon enabled IEEE
802.15.4 MAC protocol, and improve the efficiency in terms of delay, reliability and
power consumption. DQBAN is adaptable to the heterogeneity of traffic load, number
of sensor nodes and interference. A fuzzy-rule–based system is implemented in each
sensor node to handle input variables that relate to node status. DQBAN violates the
first-come-first-served packet transmission policy because it allows the node, according
to its status, to occupy the next frame collision-free “data slot”, even if that node has
already achieved the first location in the queue. A node can also chose to temporar-
ily stop transmission if it detects channel impairments. DQBAN protocol performance
was evaluated according to two different realistic hospital scenarios following a star
topology. This revealed that DQBAN achieves higher reliabilities than other possible
MAC protocols, when considering battery limitations and latency demands. Thus, Otal
et al. ensure that all packets are treated within specific delay parameters, and with
a particular BER, while conserving power. Although the mechanism provides some
solutions for fault-tolerance in WBAN and avoids single point of failure, it does not
consider differences in the reliability and requirements of the heterogeneous sensor
nodes. Moreover, Otal et al. [2009] achieved self-monitoring, by adopting two queues to
detect the physical status of the node. This means deploying specialised hardware and
software to resource constrained nodes, which adds extra resources and increases the
computation complexity of the network. Thus, when using this technique, the trade-off
between the level of required reliability and processing complexity should be investi-
gated. As shown, some primary efforts have been undertaken to try to address fault
tolerance challenges in WBAN. These efforts have varied between designing new pro-
tocols to target network stack layers such as Medium Access Control (MAC) layer [Otal
et al. 2009; Ali et al. 2010; Liu et al. 2011; Rezvani and Ghorashi 2013], and proposing
the injection of a middleware or new layer responsible for fault diagnosis [Zhou et al.
2011; Wu et al. 2010; Galzarano et al. 2012], or new generic techniques that do not
follow the stack-layering concept [Zhou et al. 2011; Wu et al. 2010; Mahapatro and
Khilar 2011; Galzarano et al. 2012; Abarna and Venkatachalapathy 2014]. Others
make enhancements to the proposed standards to improve their fault tolerance capa-
bilities [Otal et al. 2009; Ali et al. 2010; Liu et al. 2011; Rezvani and Ghorashi 2013].
However, the majority of work follows centralised detection techniques [Zhou et al.
2011; Ali et al. 2010; Wu et al. 2010; Liu et al. 2011; Mahapatro and Khilar 2011;
Rezvani and Ghorashi 2013], neglecting the issue of a single point of failure, and while
most consider techniques for an overall fault tolerance approach, fault recovery, which
is a crucial second component of fault tolerance, is not considered, thus, these efforts
cannot be considered complete solutions to fault tolerance in WBAN. Table IV provides
a summary of fault tolerance techniques in WBAN. It classifies the proposed solutions
according to the categorization presented in Figure 8.
Table V categorizes fault tolerance solutions further. For those solutions that are
geared toward node-level faults, it shows whether they consider hardware faults, data
faults or both. It shows which work considers channel-level abnormality, and whether
the proposed solution considers abnormality in emergency situations due to the corre-
lation in medical data.
Table VI classifies the proposed solutions for fault tolerance in WBAN according
to other four criteria. It shows whether a prioritising approach is considered in each
proposed solution discussed in Section 4 along with the priotrisation technique adopted,

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:30 M. Salayma et al.

Table IV. A Summary of Fault Tolerance Techniques in WBAN

Solutions for node-level faults
Technique Solution Summary
[Mahapatro Fusion centre (PDA) derives relevant physiological
and Khilar information from sensor data, according to which it
Centralised (Fusion centre) 2011] (Generic) diagnoses faulty nodes. This technique is efficient for data
centric applications that require detection precision.
[Galzarano Adopts a middleware layer in each node to detect data faults
et al. 2012] and possibly recover faulty status. Suggests adding
Middleware (Generic based autonomic elements that perform self-healing operations.
on SPIN∗ ) Avoids single point of failure and reduces the messages
between nodes and the central node, so reduces energy
Distributed consumption and network congestion.
[Abarna and CH assigns priority values to its nodes that reflect their fault
Venkatachala- tolerance level. Collected values from sensors are compared
pathy 2014] with three threshold values and node priorities are assigned
Clustering (Generic based accordingly. A scheme for CH fault detection and recovery is
on AFTCS) also proposed. Tries to alleviate a drawback of single of point
of failure.
VMSFT [Jeong A visual monitoring system that monitors patient’s
et al. 2014] conditions and tolerates to sensor failures immediately.
Cloud based (IaaS cloud Considers a large number of WBAN adopted by multiple
based patients simultaneously. Collects patient’s data through the
application) IaaS application according to which it can infer sensor
failure. Performs two types of explicit anomaly detection and
provides three responses according to the position of the
sensor in the body.
Solutions for channel-level faults
Technique Solution Summary
[Rezvani and Studies the channel impairments in WBAN due to fading
Ghorashi and body postures in a star topology. Considers correlations
2013] (Based that occur in medical traffic in emergencies and splits that
on traffic into normal and emergency medical traffic adaptively,
IEEE802.15.4) according to channel condition and the user’s medical
situation. Tolerates to channel faults by adapting the
channel allocation to the classified sampled data that
Context aware consider fading effects.
Centralised
CA-MAC [Liu Hybrid superframe-based protocol that adjusts the
et al. 2011] transmission priority and access strategy according to the
(Based on traffic and channel status. It changes the duty cycle of the
IEEE 802.15.6) nodes according to human activity, or environmental
conditions, while considering power consumption.
BodyQoS A middleware layer, called BodyQoS tolerates to channel
[Zhou et al. faults and impairments by performing an adaptive channel
Middleware 2011] (Generic) allocation strategy through which the QoS of WBAN is
improved.
Solutions for hybrid (node and channel) faults
Technique Solution Summary
AFTCS [Wu Each node is assigned a queue with a set of multiple
et al. 2010] priorities. Tolerates to channel impairments by adjusting the
Centralised (Middleware) (Generic based priority queue of each node adaptively, according to three
on BodyQoS) pieces of sensor information. Accordingly, a central node
allocates the channel bandwidth adaptively after considering
the priority level of the nodes and their data criticality level.
DQBAN [Otal A cross-layer and fuzzy logic–based mechanism that
et al. 2009] schedules transmissions in WBAN. Adaptable to the
(Based on heterogeneity of traffic load, number of sensors and
IEEE 802.15.4) interference. A fuzzy-rule–based system is implemented in
each node to handle input variables that relate to node’s
Distributed (Context awareness) status. Adopts two queues that violates the
first-come-first-served packet transmission policy. A node
chooses to temporarily stop transmission if it detects channel
impairments. Avoids single point of failure.

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:31

Table V. A Classification for the Centralized and Distributed Approaches Fault Tolerance Approaches in WBAN
Centralized Distributed
Fault Context Node-self Context Cloud
Type/Solution Data fusion Middleware aware detection Clustering aware based
Node level Hardware AFTCS [Wu [Abarna and [Otal et al. VMSFT
et al. 2010] Venkatacha- 2009] [Jeong et al.
lapathy 2014]
2014]
Data faults [Mahapatro AFTCS [Wu [Galzarano [Abarna and VMSFT
(including and Khilar et al. 2010] et al. 2012] Venkatacha- [Jeong et al.
physiological 2011] lapathy 2014]
information) 2014]
Channel Level Body/ BodyQoS CA-MAC [Otal et al.
interference [Zhou et al. [Rezvani 2009]
2011] and
AFTCS [Wu Ghorashi
et al. 2010] 2013]
Medical correlation CA-MAC [Abarna and [Otal et al. VMSFT
(emergency) [Rezvani Venkatacha- 2009] [Jeong et al.
and lapathy 2014]
Ghorashi 2014]
2013]

it also shows whether the proposed solution follows a recovery approach and whether
the proposed approach is a generic based or based on a standardized protocol.
The following section summarises some interference solutions presented in this sec-
tion. It highlights the strengths and the drawbacks of the presented approaches and
suggests trends that can be adopted in the future to achieve a more reliable WBAN by
overcoming the drawbacks of the literature presented.

4.4. Interference Mitigation Techniques

For interference analysis and modelling, when building reliable interference models,
a challenging task could be in the assessment of future e-Health applications while
considering the increasing heterogeneity of networks, devices, services, user require-
ments and conditions. Thus, classical deterministic approaches that are characterized
by limited scalability and high complexity can barely characterise such highly dense
networks. As an alternative, emerging stochastic geometry techniques has recently
been used to obtain tractable evaluations and suitable expressions to capture and op-
timize the total effect interference in co-existing scenarios. A very recent application of
stochastic geometry technique in modelling intra and inter WBANs interference was
proposed in Sun et al. [2015]. However, it was limited to unrealistic assumptions where
each node in the network was assumed to follow the same MAC protocol at a given
period of time. This is despite all WBANs use hybrid MAC structure as given in the
IEEE 802.15.6 standard.
There are two major schemes for interference mitigation namely, collaborative and
non-collaborative schemes [Le and Moh 2015]. In the former one, multiple WBANs
interact with each other to manage the co-existence problem while in the latter
WBANs can manage the coexistence issue without any interaction. The work in this
area is very limited and the majority of the proposed schemes depend on unrealistic
assumptions. In addition, they do not consider performance metrics that are essential
to measure and monitor reliability in such networks. Hence, there is a need for new
efficient versions of coexistence schemes where simplicity, reliability are key elements.
For example, in de Francisco et al. [2009], Martelli and Verdone [2012], Dong and
Smith [2013], and Hayajneh et al. [2014], the authors proved that there is a dominant
interferer from other networks in WBANs such as, for instance, IEEE 802.11, IEEE
802.15.1, IEEE 802.15.4, etc. However, their interference analysis was limited to intra-
BAN communication where nodes transmit with the same power to the coordinator.

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:32 M. Salayma et al.

Table VI. Priority and Recovery Mechanisms Proposed by Fault Tolerance Approaches in WBAN
Protocol/ Fault recovery
Criteria Prioritization mechanism Solution Algorithm
BodyQoS [Zhou Specific and fixed priority N/A Dynamic Generic
et al. 2011] based on level of criticality Chanel (VMAC)
of sensor and data type. Allocation
AFTCS [Wu et al. Default priority statically N/A Dynamic Generic Based
2010] configured by the clinician. Chanel on BodyQoS
Based on the perceived Allocation
fault-related information it
adjusts the priority.
[Mahapatro and N/A Isolation Feature-level Generic
Khilar 2011] Fusion
[Rezvani and It adjusts the priority N/A Dynamic MAC based on
Ghorashi 2013] based on the Physician’s Chanel IEEE 80.15.4
recognition and health Allocation and IEEE
condition of the user and 802.15.6
CA-MAC [Rezvani Adaptive priority according N/A Dynamic MAC Based
and Ghorashi to traffic and channel Scheduling on IEEE
2013] status. 802.15.6
[Otal et al. 2009] Adaptive priority according N/A Distributed MAC Based
to node physical status. Queues on IEEE
802.15.4
[Galzarano et al. N/A Isolation Autonomic Generic Based
2012] Elements on SPIN∗
[Abarna and Adaptive priority according Isolation of Dynamic Generic based
Venkatachalapathy to a specified fixed faulty node. Priority on AFTCS
2014] threshold. Nodes under
faulty CH
re-cluster
themselves
and elect new
cluster head.
VMSFT [Jeong Priority according node Isolation by Three levels IaaS
et al. 2014] position on body. emergency of response cloud-based
response based on application
abnormality
criticality

On the other hand, few studies have focused on inter-BAN communication where, in
Davenport et al. [2009], the authors conduct a study on the measurement of coupling
between 10 nodes within a WBAN. This was done in a single room using the 2400–
2500MHz frequency band and within a hospital environment. Note that, although the
measurements were interesting, yet it did not consider the mobility of the nodes which
is an important characteristic of WBANs. In addition, the authors didn’t show the
effect of the measured interference on the degradation of the application performance
and the coexistence strategies proposed were limited to the packet delivery ratio
performance only. Furthermore, the effect of interference in Dotlic [2011] was studied
when using both of the chirp and the sampling receivers. Finally, in Alasti et al. [2014],
the authors addressed co-channel interference between co-located multiple WBANs
where couple of uncoordinated schemes were presented. In both approaches, the co-
ordinator node re-allocates slots in the TDMA scheme. However, such approaches are
limited since they use several unrealistic assumptions including lack of mobility and
absence of computation or estimation of actual interference. Nonetheless, there is still
a need for IEEE 802.15.6-compliant radio transceivers that are available commercially.
In addition, communicating WBANs form a very complex network that cannot be

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:33

Fig. 9. Generalized coordinated coexisting algorithm.

analyzed analytically due to huge number of possible combinations thus simulation-

based approaches are considered.
The performance evaluation of any proposed co-existence mechanism should be done
under realistic environment to simulate real deployment and operating assumptions
[Ben Hamida et al. 2009] where there is a need for accurate intra- and inter-WBAN
mobility and physical layer modeling (e.g., radio link modelling, realistic channel and
path loss modelling) to ensure reliability. A challenging task that affects the inter-
ference in WBANs is the modelling of both intra and inter-WBANs mobility so as to
capture the postural behaviors of real-time motion of human bodies. Perhaps, either
sensor nodes within a WBAN move individually or they may move in a group resulting
in extra interference effect. In addition, several WBANs sharing the same activity may
move in a group into another location where they may enter other WBANs communi-
cation range and interfere with their transmissions or be subject to interference from
dominant interfering signals coming from other communication technologies.
As detailed earlier, WBANs are expected to coexist with other technologies such as
WiFi, Bluetooth, and ZigBee which leads to performance degrading challenges due to
the uncoordinated operation of these multi-Radio Access Technologies (multi-RATs)
and their uncontrolled interference. In general, a solution to the coexistence problem
will follow either coordinated or uncoordinated approaches. In Al-Khansa and Artail
[2015], different coexistence techniques in the context of LAA-WiFi coexistence has
been presented, where LAA stands for “Licensed-Assisted Access to the unlicensed
spectrum” for the Long Term Evolution (LTE) systems [Rupasinghe and Güvenç 2015].
However, this work is not directly applicable to WBAN. Hence, we can summarize the
two approaches as follows.

4.4.1. The Multi-RAT Coordination Approach. Through coordination protocols, the different
technologies exchange traffic information and negotiate system parameters for optimal
system performance. In a WBAN-ZigBee coexistence scenario, for example, the WBAN
coordinator node (a.k.a the hub) will coordinate with the ZigBee coordinator node; co-
ordination occurs through a predefined coordination protocol and dedicated interfaces.
In general, a coordinated system follows the algorithm shown in Figure 9, where the
coordinating technologies switch between regular modes and coexistence modes. This
approach has proven to be useful in several scenarios, however, it is not suitable for

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:34 M. Salayma et al.

WBANs because of the coordination overhead which cannot be tolerated by WBAN

nodes that are of limited range and power capabilities.

4.4.2. The Uncoordinated Approach. The coexisting technologies implement techniques

to avoid interference with each other while operating independently (i.e., without co-
ordinating with each other). For example, cognitive radio, least-congested channel
selection, and power control are uncoordinated techniques. The CSMA/CA protocol is
another example for uncoordinated protocols. A drawback of carrier sense mechanisms
is that they often suffer from energy-detection inaccuracies due to the uncertainties of
noise variance. Moreover, these uncoordinated techniques are time-inefficient, where
most of the modes time is spent in listen-modes sensing the medium and often backing-
off for exponentially growing durations when collisions occur, and the higher the level
of interference, the worse the performance gets. As an illustration of this problem,
Nokia has conducted simulations on LTE-WiFi coexistence [Almeida et al. 2013], and
the results have shown that when LTE interferes with WiFi without implementing any
coexistence techniques, the WiFi nodes consumed “over 99% of the time listening to the
wireless medium and backing off ” [Almeida et al. 2013]. Thus, significant system time
and energy resources are wasted in interfering networks with uncoordinated protocols,
which again is not suitable for WBANs.

5. CONCLUSIONS AND RESEARCH DIRECTIONS

Improved efficiency and reliability are vital to guarantee the success of the next gen-
eration of WBANs and their ability to host a rich portfolio of applications. Although
different surveys have been found in the literature covering different concepts and
applications, resilience and reliability and their related issues have been addressed
lightly. Thus, unlike existing surveys, this article surveys outlines and discusses the
state-of-the-art fault tolerance models and reliability related issues and challenges
such as coexistence and interferences. In addition, this survey emphasized the impor-
tance of the various technology aspects such power efficiency and heterogeneity. Indeed,
the increasing heterogeneity of networks, devices, services, and coexistence of different
WBANs and other communication technologies within the same range impose potential
challenges in reliability of WBANs. The majority of work proposed in the literature for
fault tolerance in both WSN and WBAN focuses only on fault detection techniques, sup-
posing that the only recovery approach is node isolation and replacement [Mahapatro
and Khilar 2011; Galzarano et al. 2012; Abarna and Venkatachalapathy 2014; Jeong
et al. 2014]. Such a solution could be sufficient for WSN where there is node redundancy,
due to the huge number of homogenous nodes. However, in WBAN, waiting for node
isolation and replacement results in a huge delay in response, while critical data needs
to be transmitted immediately. The framework for fault tolerance in WBAN cannot be
considered complete if it only allows fault detection, ignoring real-time fault recovery.
Thus, proposing real time recovery techniques for WBAN is necessary, such that the
critical data of erroneous nodes could be achieved even though the node is faulty.
One technique followed in this work is to exploit the data correlation in sensor
readings. Despite the problems mentioned as possibly occurring due to data correla-
tion in WBAN, existing interconnections can be harnessed by performing data fusion.
Mahapatro and Khilar [2011] was found to be the first and only study to mention
this technique but for fault detection. Data fusion can be considered an initial step
towards fault recovery. In other words, a central node can achieve data of some faulty
sensors by performing some derivations from other available data that has been sent
by other sensors. This could be considered a temporary solution, until node isolation
and replacement takes place.

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:35

Moreover, new technologies might encounter opportunities in this arena, such as

cloud computing and sensor clouds. Through these technologies, fault detection and re-
covery can be achieved in real time simply by exploiting the services such technologies
offer. Virtual sensors in the sensor cloud, for example, could be considered a valu-
able alternative to sensor redundancy [Alamri et al. 2013]. Certainly, adopting such
technologies requires the implementation and installation of fault tolerance applica-
tions in the cloud. Such applications can be used by patients and medical staff, each
with different responsibilities and capabilities. However, privacy and security is still a
primary concern potentially limiting the level of adoption of such technologies at the
current time. Yet it is still beneficial to consider them. It is also found that, to satisfy
the emerging applications, there is a need for new cross-layer interference mitigation
schemes, considering different mobility and coexistence paradigms.
Finally, regarding WSN, we used to claim: the lifetime of a sensor determines the
lifetime of WSN, but in relation to WBAN we claim instead: the behaviour of a sensor
determines lifetime of the human. Hence, if we cannot achieve the highest degree of
reliability with 0% error as an acceptable QoS in WBAN, then it is better to shift to
wired e-health networks despite their limitations on individual’s movements. We must
always remember that the main objective of WBAN is to improve and extend human
life, it is not intended to offer luxury.

REFERENCES
K. T. Meena Abarna and K. Venkatachalapathy. 2014. Cluster based failure detection and recovery technique
for wireless body area networks. Research Journal of Applied Sciences, Engineering and Technology 7,
17 (2014), 3458–3465. DOI:10.19026/rjaset.7.697
Raza H. Abedi, Nauman Aslam, and Sayeed Ghani. 2011. Fault tolerance analysis of heterogeneous wireless
sensor network. In Proceedings of the 24th Canadian Conference on Electrical and Computer Engineering
(CCECE). IEEE, 000175-000179. DOI:10.1109/CCECE.2011.6030433
Ian F. Akyildiz, Weilian Su, Yogesh Sankarasubramaniam, and Erdal Cayirci. 2002. A survey on sensor
networks. IEEE Communications Magazine 40, 8 (2002), 102–114. DOI:10.1109/MCOM.2002.1024422
Rasha Al-Khansa and Hassan Artail. 2015. A semi-distributed LTE-WiFi system design for future LTE-
unlicensed: Deployments in small-cell environments. In Proceedings of the 11th IEEE International
Conference on Wireless and Mobile Computing, Networking and Communications (WiMob). IEEE, 43–
50. DOI:10.1109/WiMOB.2015.7347939
Muhammad Mahtab Alam and Elyes Ben Hamida. 2015. Interference mitigation and coexistence strate-
gies in IEEE 802.15.6 based wearable body-to-body networks. In Proceedings of the International Con-
ference on Cognitive Radio Oriented Wireless Networks. Springer International Publishing, 665–677.
DOI:10.1007/978-3-319-24540-9_55
Atif Alamri, Wasai Shadab Ansari, Mohammad Mehedi Hassan, M. Shamim Hossain, Abdulhameed Alelaiwi,
and M. Anwar Hossain. 2013. A survey on sensor-cloud: Architecture, applications, and approaches.
International Journal of Distributed Sensor Networks 2 (2013), 18. 917923. DOI:10.1155/2013/917923
Mehdi Alasti, Martina Barbi, and Kamran Sayrafian. 2014. Uncoordinated strategies for inter-BAN interfer-
ence mitigation. In Proceedings of the 25th IEEE Annual International Symposium on Personal, Indoor,
and Mobile Radio Communication (PIMRC). IEEE, 2150–2154. DOI:10.1109/PIMRC.2014.7136528
Khaled A. Ali, Jahangir H. Sarker, and Hussein T. Mouftah. 2010. Urgency-based MAC protocol for wireless
sensor body area networks. In Proceedings of the IEEE International Conference on Communications
Workshops. IEEE, 1–6. DOI:10.1109/ICCW.2010.5503911
Erika Almeida, André M. Cavalcante, Rafael C. D. Paiva, Fabiano S. Chaves, Fuad M. Abinader, Robson D.
Vieira, Sayantan Choudhury, Esa Tuomaala, and Klaus Doppler. 2013. Enabling LTE/WiFi coexistence
by LTE blank subframe allocation. In Proceedings of the IEEE International Conference on Communi-
cations (ICC). IEEE, 5083–5088. DOI:10.1109/ICC.2013.6655388
Nancy Alrajei and Huirong Fu. 2014. A survey on fault tolerance in wireless sensor networks. In Proceedings
of the 3rd International Conference on Sensor Technologies and Applications, 2009 (SENSORCOMM’09).
366–371.
Deena M. Barakah and Muhammad Ammad-uddin. 2012. A survey of challenges and applications of wireless
body area network (WBAN) and role of a virtual doctor server in existing architecture. In Proceedings

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:36 M. Salayma et al.

of the 3rd International Conference on Intelligent Systems Modelling and Simulation. IEEE, 214–219.
DOI:10.1109/ISMS.2012.108
M. Bellalouna, and A. Ghabri. 2013. A priori methods for fault tolerance in wireless sensor networks, In
Proceedings of the 2013 World Congress on Computer and Information Technology (WCCIT), IEEE, 1–6.
Monia Bellalouna and Afef Ghabri. 2013. A priori methods for fault tolerance in wireless sensor networks.
In Proceedings of the 2013 World Congress on Computer and Information Technology (WCCIT). IEEE,
1–6. DOI:10.1109/WCCIT.2013.6618654
Fabio Bellifemine, Giancarlo Fortino, Roberta Giannantonio, Raffaele Gravina, Antonio Guerrieri, and Marco
Sgroi. 2011. SPINE: A domain-specific framework for rapid prototyping of WBSN applications. Software:
Practice and Experience 41, 3 (2011), 237–265. DOI:10.1002/spe.998
Lucia Lo Bello and Emanuele Toscano. 2009. Coexistence issues of multiple co-located IEEE 802.15.4/Zig-
Bee networks running on adjacent radio channels in industrial environments. IEEE Transactions on
Industrial Informatics 5, 2 (2009), 157–167. DOI:10.1109/TII.2009.2018541
Elyes Ben Hamida, Guillaume Chelius, and Jean-Marie Gorce. 2009. Impact of the physical layer mod-
eling on the accuracy and scalability of wireless network simulation. Simulation 85 (2009), 574–588.
DOI:10.1177/0037549709106633
Azzedine Boukerche. 2005. Handbook of Algorithms for Wireless Networking and Mobile Computing. CRC
Press.
Riccardo Cavallari, Flavia Martelli, Ramona Rosini, Chiara Buratti, and Roberto Verdone. 2014. A survey on
wireless body area networks: technologies and design challenges. IEEE Communications Surveys and
Tutorials 16, 3 (2014), 1635–1657. DOI:10.1109/SURV.2014.012214.00007
Craig A. Chin, Garth V. Crosby, Tirthankar Ghosh, and Renita Murimi. 2012. Advances and chal-
lenges of wireless body area networks for healthcare applications. In Proceedings of the 2012
International Conference on Computing, Networking and Communications (ICNC). IEEE, 99–103.
DOI:10.1109/ICCNC.2012.6167576
C. N. A. P. Staff. 2013. Network Basics Companion Guide. Cisco Press.
G. V. Crosby, T. Ghosh, R. Murimi, and C. A. Chin. 2012. Wireless body area networks for healthcare: A
survey. International Journal of Ad hoc, Sensor & Ubiquitous Computing (IJASUC) 3, 3, 1–19.
Garth V. Crosby, Tirthankar Ghosh, Renita Murimi, and Craig A. Chin. 2012. Wireless body area networks
for healthcare: A survey. International Journal of Ad Hoc, Sensor and Ubiquitous Computing (IJASUC)
3, 3 (2012), 1–19. DOI:10.5121/ijasuc.2012.3301
David M. Davenport, Budhaditya Deb, and Fergus J. Ross. 2009. Wireless propagation and coexistence of
medical body sensor networks for ambulatory patient monitoring. In Proceedings of the 6th International
Workshop on Wearable and Implantable Body Sensor Networks. IEEE, 41–45. DOI:10.1109/BSN.2009.8
Ruben de Francisco, Li Huang, and Guido Dolmans. 2009. Coexistence of WBAN and WLAN in medical
environments. In Proceedings of the 70th IEEE Vehicular Technology Conference Fall (VTC 2009-Fall).
IEEE, 1–5. DOI:10.1109/VETECF.2009.5378807
Jie Dong and David Smith. 2013. Coexistence and interference mitigation for wireless body area networks:
Improvements using on-body opportunistic relaying. arXiv preprint arXiv, 1305.6992.
Igor Dotlic. 2011. Interference performance of IEEE 802.15.6 impulse-radio ultra-wideband physical layer.
In Proceedings of the 22nd IEEE International Symposium on Personal Indoor and Mobile Radio Com-
munications (PIMRC). IEEE. 2148–2152. DOI:10.1109/PIMRC.2011.6139895.
Du Dakun, Hu Fengye, Wang Feng, Wang Zhijun, Du Yu, and Wang Lu. 2015. A game theoretic approach
for inter-network interference mitigation in wireless body area networks. China Communications 12, 9
(2015), 150–161.
Stefano Galzarano, Giancarlo Fortino, and Antonio Liotta. 2012. Embedded self-healing layer for detecting
and recovering sensor faults in body sensor networks. In Proceedings of the IEEE International Confer-
ence on Systems, Man, and Cybernetics (SMC). IEEE, 2377–2382. DOI:10.1109/ICSMC.2012.6378098
Satish K. Garg, Sherwyn Schwartz, and Steven V. Edelman. 2004. Improved glucose excursions using an
implantable real-time continuous glucose sensor in adults with type 1 diabetes. Diabetes Care 27, 3
(2004), 734–738.
Sai Anand Gopalan and Jong-Tae Park. 2010. Energy-efficient MAC protocols for wireless body area net-
works: Survey. In Proceedings of the International Congress on Ultra Modern Telecommunications and
Control Systems and Workshops (ICUMT). IEEE, 739–744. DOI:10.1109/ICUMT.2010.5676554
Thaier Hayajneh, Ghada Almashaqbeh, Sana Ullah, and Athanasios V. Vasilakos. 2014. A survey of wireless
technologies coexistence in WBAN: Analysis and open research issues. Wireless Networks 20, 8 (2014),
2165–2199.

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
Wireless Body Area Network (WBAN) 3:37

Hongliang Ren and Max Q.-H. Meng. 2006. Understanding the mobility model of wireless body sensor
networks. In Proceedings of the 2006 IEEE International Conference on Information Acquisition. IEEE,
306–310. DOI:10.1109/ICIA.2006.306015
Young-Sik Jeong, Hyun-Woo Kim, and Jong Hyuk Park. 2014. Visual scheme monitoring of sensors for fault
tolerance on wireless body area networks with cloud service infrastructure. International Journal of
Distributed Sensor Networks 2014. DOI:10.1155/2014/154180
Jamil Y. Khan and Mehmet R. Yuce. 2010. Wireless body area network (WBAN) for medical applications.
New Developments in Biomedical Engineering. INTECH (2010), ISBN 978-953-7619-57-2, 592–627.
DOI:10.5772/7598
Barbara T. Korel and Simon G. M. Koo. 2007. Addressing context awareness techniques in body sensor
networks. In Proceedings of the 21st International Conference on Advanced Information Networking and
Applications Workshops (AINAW’07) 2 (2007), IEEE, 798–803. DOI:10.1109/AINAW.2007.69
Ravindra V. Kshirsagar and Ashish B. Jirapure. 2011. A survey on fault detection and fault tolerance in
wireless sensor networks. In Proceedings of the International Conference on Benchmarks in Engineering
Science and Technology (ICBEST (1)). 3 (2011), International Journal of Computer Science. 130–138.
Benoı̂t Latré, Bart Braem, Ingrid Moerman, Chris Blondia, and Piet Demeester. 2011. A survey on wireless
body area networks. Wireless Networks 17, 1 (2011), 1–18. DOI:10.1007/s11276-010-0252-4
Thien T. T. Le and Sangman Moh. 2015. Interference mitigation schemes for wireless body area sensor
networks: A comparative survey. Sensors 15, 6 (2015), 13805–13838. DOI:10.3390/s150613805
Bin Liu, Zhisheng Yan, and Chang Wen Chen. 2011. CA-MAC: A hybrid context-aware MAC protocol for
wireless body area networks. In Proceedings of the 13th IEEE International Conference on e-Health Net-
working Applications and Services (Healthcom). IEEE, 213–216. DOI:10.1109/HEALTH.2011.6026748
Hai Liu, Amiya Nayak, and Ivan Stojmenović. 2009. Fault-tolerant algorithms/protocols in wireless sensor
networks. In Guide to Wireless Sensor Networks, Springer, London, 261–291. DOI:10.1007/978-1-84882-
218-4_10
Arunanshu Mahapatro and Pabitra Mohan khilar. 2011. Online fault detection and recovery in body sen-
sor networks. In Proceedings of the World Congress on Information and Communication Technologies
(WICT). IEEE, 407–412. DOI:10.1109/WICT.2011.6141280
Vikash Mainanwal, Mansi Gupta, and Shravan Kumar Upadhayay. 2015. A survey on wireless body area
network: Security technology and its design methodology issue. In Proceedings of the International
Conference on Innovations in Information, Embedded and Communication Systems (ICIIECS). IEEE,
1–5. DOI:10.1109/ICIIECS.2015.7193088
Bhavneesh Malik and V. R. Singh. 2013. A survey of research in WBAN for biomedical and scientific appli-
cations. Health and Technology 3, 3 (2013), 227–235. DOI:10.1007/s12553-013-0056-5
Jhon. M. Maloney and John T. Santini, Jr. 2004. Implantable microchips for controlled drug delivery. In
Proceedings of the 26th IEEE Annual International Conference Engineering in Medicine and Biology
Society (IEMBS’04). 1 (2004), 2668–2669. DOI:10.3109/10717544.2014.903579
Flavia Martelli and Roberto Verdone. 2012. Coexistence issues for wireless body area networks at 2.45 gHz.
In Proceedings of the 18th European Wireless Conference 2012, EW, VDE. 1–6.
Sushruta Mishra, Lambodar Jena, and Aarti Pradhan. 2012. Fault tolerance in wireless sensor networks. In-
ternational Journal of Advanced Research in Computer Science and Software Engineering (IJARCSSE),
2, 10, 146–153.
Amirhossein Moravejosharieh and Jaime Lloret. 2015. A survey of IEEE 802.15.4 effective system parame-
ters for wireless body sensor networks. International Journal of Communication Systems (2015). 1–24.
DOI:10.1002/dac.3098
Samaneh Movassaghi, Mehran Abolhasan, and Justin Lipman. 2013. A review of routing protocols in wireless
body area networks’. Journal of Network and Computer Applications 8, 3, (March 2013), 559–575.
10.4304/jnw.8.3
Samaneh Movassaghi, Mehran Abolhasan, Justin Lipman, David Smith, and Abbas Jamalipour. 2014. Wire-
less body area networks: A survey. IEEE Communications Surveys and Tutorials 16, 3, (2014), 1658–
1686. DOI:10.1109/SURV.2013.121313.00064
Georgia D. Ntouni, Athanasios S. Lioumpas, and Konstantina S. Nikita. 2014. Reliable and energy-efficient
communications for wireless biomedical implant systems. IEEE Journal of Biomedical and Health
Informatics 18, 6, (2014), 1848–1856.
Horacio Antonio Braga, Fernandes De Oliveira, Azzedine Boukerche, Eduardo Freire Nakamura, and An-
tonio Alfredo Ferreira Loureiro. 2009. An efficient directed localization recursion protocol for wireless
sensor networks. IEEE Transactions on Computers 58, 5, (2009), 677–691. DOI:10.1109/TC.2008.221

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.
3:38 M. Salayma et al.

Begonya Otal, Christos Verikoukis, and L. Alonso. 2009. Fuzzy-logic scheduling for highly reliable and
energy-efficient medical body sensor networks. In Proceedings of the 2009 IEEE International Conference
on Communications Workshops. IEEE, 1–5. DOI:10.1109/ICCW.2009.5208088
Chris Otto, Aleksandar Milenkovic, Corey Sanders, and Emil Jovanov. 2006. System architecture of a wireless
body area sensor network for ubiquitous health monitoring. Journal of Mobile Multimedia 1, 4, (2006),
307–326.
Maulin Patel and Jianfeng Wang. 2010. Applications, challenges, and prospective in emerging body
area networking technologies. IEEE Wireless Communications Magazine 17, 1, (2010), 80–88.
DOI:10.1109/MWC.2010.5416354
K. M. Karthick Raghunath and N. Rengarajan. 2013. Investigation of faults, errors and failures in wireless
sensor network: A systematical survey. International Journal of Advanced Computer Research 3, 3,
(2013), 151.
Sanaz Rezvani and Seyed Ali Ghorashi. 2013. Context aware and channel-based resource allocation for wire-
less body area networks. IET Wireless Sensor Systems 3, 1 (2013), 16–25. DOI:10.1049/iet-wss.2012.0100
Nadisanka Rupasinghe and İsmail Güvenç. 2015. Reinforcement learning for licensed-assisted access of LTE
in the unlicensed spectrum. In Proceedings of the 2015 IEEE Wireless Communications and Networking
Conference (WCNC). IEEE, 1279–1284. DOI:10.1109/WCNC.2015.7127653
Tom Rutherford and A. Socio. 2012. Population ageing: Statistics. House of Commons library (Standard not.
Retrieved Jan 2, 2013, From: www. Parliament. uk/Topics/PopulationArchive (2012)
Marwa Salayma, Wail Mardini, Yaser Khamayseh, and Muneer Bani Yasin. 2013a. Optimal beacon and
superframe orders in WSNs. In Proceedings of the 5th International Conference on Future Computational
Technologies and Applications, Topology 6 (2013), 49–55.
Marwa Salayma, Wail Mardini, Yaser Khamayseh, and Muneer Bani Yasin. 2013b. IEEE 802. 15.4 Perfor-
mance in Various WSNs Applications. 2013. In Proceedings of the 7th International Conference on Sensor
Technologies and Applications. 139–144.
Leo Selavo, Anthony Wood, Qing Cao, Tamim Sookoor, Hengchang Liu, Aravind Srinivasan, and
Yafeng Wu. 2007. LUSTER: Wireless sensor network for environmental research. In Proceed-
ings of the 5th International Conference on Embedded Networked Sensor Systems. ACM, 103–116.
DOI:10.1145/1322263.1322274
TG4. 2003. IEEE 802.15 WPANTM Task Group 4 (TG4). Retrieved Juanuary 31, 2016 from http://www.
ieee802.org/15/pub/TG4.html.
TG4a. 2007. IEEE 802.15 WPAN Low Rate Alternative PHY Task Group 4a (TG4a). Retrieved Juanuary 31,
2016 from http://www.ieee802.org/15/pub/TG4a.html.
TG4j. 2012. TG 4j Amendment Draft. Internal Report, unpublished.
TG6. 2012. IEEE standard for local and metropolitan area networks: Part 15.6: Wireless body area networks,
802.15.6-2012. Retrieved Juanuary 31, 2016 from https://standards.ieee.org/findstds/standard/802.15.6-
2012.html.
Wen Sun, Yu Ge, and Wai-Choong Wong. 2015. A stochastic geometry analysis of inter-user interference in
IEEE 802.15. 6 body sensor networks. In Proceedings of the 2015 IEEE Wireless Communications and
Networking Conference (WCNC). IEEE, 1912–1917. DOI:10.1109/WCNC.2015.7127760
Sana Ullah, Henry Higgins, Bart Braem, Benoit Latre, Chris Blondia, Ingrid Moerman, Shahnaz Saleem,
Ziaur Rahman, and Kyung Sup Kwak. 2012. A comprehensive survey of wireless body area networks.
Journal of Medical Systems 36, 3, (2012), 1065–1094. DOI:10.1007/s10916-010-9571-3
Xuan Wang and Lin Cai. 2011. Interference analysis of co-existing wireless body area networks.
In Proceedings of the Global Telecommunications Conference (GLOBECOM 2011). IEEE, 1–5.
DOI:10.1109/GLOCOM.2011.6133624
Wikipedia The Free Encyclopedia. 2015. Middleware. Retrieved April 14, 2015 from http://www.ieee802.
org/15/pub/TG4.html.
Guowei Wu, Jiankang Ren, Feng Xia, and Zichuan Xu. 2010. An adaptive fault-tolerant communication
scheme for body sensor networks. Sensors 10, 11 (2010), 9590–9608. DOI:10.3390/s101109590
Mengjie Yu, Hala Mokhtar, and Madjid Merabti. 2007. Fault management in wireless sensor networks. IEEE
Wireless Communications 14, 6 (2007), 13–19. DOI:10.1109/MWC.2007.4407222
Gang Zhou, Qiang Li, Jingyuan Li, Yafeng Wu, Shan Lin, Jian Lu, Chieh-Yih Wan, Mark D. Yarvis, and
John A. Stankovic. 2011. Adaptive and radio-agnostic qos for body sensor networks. ACM Transactions
on Embedded Computing Systems (TECS) 10, 4 (2011), 48. DOI:10.1145/2043662.2043672

Received March 2016; revised October 2016; accepted December 2016

ACM Computing Surveys, Vol. 50, No. 1, Article 3, Publication date: March 2017.