A Review of Routing Protocols in Wireless
Body Area Networks
Samaneh Movassaghi, Mehran Abolhasan
School of Communication and Computing, University of Technology, Sydney, Australia
Email: Seyedehsamaneh.Movassaghigilani@student.uts.edu.au and Mehran.Abolhasan@uts.edu.au
Justin Lipman
Intel IT Labs, China
Email: justin.lipman@intel.com
Abstract—Recent technological advancements in wire-
less communication, integrated circuits and Micro-Electro-
Mechanical Systems (MEMs) has enabled miniaturized, low-
power, intelligent, invasive/ non-invasive micro and nano-
technology sensor nodes placed in or on the human body
for use in monitoring body function and its immediate
environment referred to as Body Area Networks (BANs).
BANs face many stringent requirements in terms of delay,
power, temperature and network lifetime which need to be
taken into serious consideration in the design of different
protocols. Since routing protocols play an important role in
the overall system performance in terms of delay, power
consumption, temperature and so on, a thorough study
on existing routing protocols in BANs is necessary. Also,
the specific challenges of BANs necessitates the design of
new routing protocols specifically designed for BANs. This
paper provides a survey of existing routing protocols mainly
proposed for BANs. These protocols are further classified
into five main categories namely, temperature based, cross-
layer, cluster based, cost-effective and QoS-based routing,
where each protocol is described under its specified category.
Also, comparison among routing protocols in each category
is given.
Index Terms—IEEE 802.15.6, Body Area Networks,
BANs, Wireless Sensor Networks, Mobile Ad Hoc Networks
I. INTRODUCTION
Sensors in BANs can either be implanted in the human
tissue (in-body) or strategically placed on the body (on-
body). Either approach requires considering the effect of
radiation emitted by wireless transceivers on the body
tissue for human safety. In the in-body case, relaying
or transmission of data to neighbor nodes may lead to
average temperature rise which may have undesirable
effects on human tissue given prolonged operation of
the sensor nodes [1]. One solution is to disseminate data
transmission in the entire network instead of relaying on
some predefined routes. This avoids a dramatic increase
in the temperature of sensors located in specific areas.
However, such a solution increases overall system over-
head and system complexity that should be minimized
for BAN. Additionally, the severe path loss of radio
signals in the surrounding of a human body necessitates
the need of multihop communication in BANs as their
direct transmission will come at high communication
costs [1, 2].
Routing protocols in WSNs [3] and MANETs [4]
have been excessively studied in the past few years.
However, the stringent requirements of BANs imposes
certain constraints on the design of their routing protocol
which leads to novel challenges in routing which have
not been met through routing protocol in WSNs and
MANETs. WSNs consider minimal routing overhead and
maximal throughput more significant than minimal energy
consumption [5]. On the other hand, energy efficient
routing protocols in MANETs consider finding routes
to minimize energy consumption in cases with small
energy resources. Unfortunately, they do not consider the
required energy to receive and transmit a symbol over a
wireless link and operations required for memory access,
data processing and measurements [5]. WSNs assume
homogeneous nodes comprise the network, whereas BAN
nodes are heterogeneous and have varying capability with
respect to data rate and available energy [6]. Mobility
in WSNs may be on the order of meters to tens of
meters, whereas in BANs movement is on the order of
tens of centimeters [5, 6]. Additionally, BAN routing must
consider variations in body movement, effects of radiation
on tissue heating and limited energy resources to provide
efficient usage of available resources to further reduce the
intervals of battery charging, enhance network lifetime
and develop a user-friendly system. Hence, even though
the general characteristics of BANs are somehow similar
to MANETs and WSNs, the unique differences amongst
them with BANs requires novel solutions in their routing
protocols.
In the past decade, several routing protocols have been
proposed for BANs that can be classified with respect
to their aims. The first category is temperature based
routing protocols which are mainly designed to minimize
the local or overall system temperature rise. In fact, the
idea behind these protocols is to route data from different
routes to avoid a dramatic temperature rise in some
sensors leading to human tissue damage and depletion of
the node. However, these protocols suffer from system
complexity and overhead which dramatically increases
with higher number of nodes. The second class is cluster-
JOURNAL OF NETWORKS, VOL. 8, NO. 3, MARCH 2013
559
© 2013 ACADEMY PUBLISHER
doi:10.4304/jnw.8.3.559-575
based routing protocols which try to divide nodes in
BANs into different clusters and assign a cluster-head for
each cluster and route data from sensor to the sink through
the cluster-heads. These protocols aim to minimize the
number of direct transmissions from sensors to the base
station. However, the large amount of overhead and delay
required for cluster selection are main drawbacks of these
protocols.
Cross layer routing which is the third category of
BAN routing protocols discussed in this paper, combines
the challenges in routing with medium access issues.
Although these protocols achiever high throughput, low
energy consumption and a relatively fixed end-to-end
delay, they cannot provide high performance in cases
of body motion and high path loss in some scenarios.
Cost-effective routing protocols periodically update a cost
function based on cost-effective information and find their
route amongst routes with minimum cost. These protocols
suffer from large number of transmissions required for
updating cost-effective information. The last category
is QoS-based routing protocol which mainly provides
separate modules for different QoS metrics that operate in
coordination with each other. Hence, they provide higher
reliability, lower end-to-end delay and higher packet
delivery ratio. These protocols mainly suffer from high
complexity due to the design of several modules based
on different QoS metrics.
We provide a detailed review of each protocol in its
specified category, compare protocols within each cate-
gory and describe their main advantages and drawbacks.
As temperature routing protocols try to minimize the
overall or local temperature rise in BANs and do not
consider link quality or other system parameters, they may
not satisfy all the requirements in BAN routing. Cross-
layer and cluster based protocols require a large amount
of overhead to exchange network information between
nodes and do not consider temperature effects of the
protocol on the skin and so do not fulfill all requirements
of routing in BANs. Cost-effective protocols can not
provide high throughput without minimum overhead and
energy consumption. QoS routing protocols require too
much information that leads to high energy consumption
and huge overhead. In fact, each classification of routing
protocols only tries to satisfy a specific requirement in
BANs. This encourages us to find new routing protocols
that meet all requirements of BANs. This paper takes the
first step in this regard by providing a detailed review on
existing routing protocols in BANs which is essential to
gain the overall knowledge of challenges in BAN routing
and possible solutions in each case.
The rest of this paper is organized as follows. Section
II provides background information on BANs. Section III
describes challenges of routing in BANs. BAN specific
temperature routing protocols are described in Section
IV. Section V describes cluster based routing protocols
in BANs. Cross-layer routing protocols are described
in Section VI. Section VII and Section VIII describes
cost-effective and QoS-based routing protocols in BANs,
respectively. In Section IX, we provide a comparison of
routing in BANs with WSN and MANET routing. Section
X concludes the paper.
II. BACKGROUND
BANs have a huge potential to revolutionize the future
of health care monitoring by diagnosing many life threat-
ening diseases and providing real-time patient monitoring
[7]. Demographers have predicted that people age 65
and over in 2025 will double the 357 million population
in 1901 and become 761 million. This implies the fact
that by mid-century, medical care will become a major
issue. By 2009, the health care expenditure in the United
States was about 2.9 trillion and is estimated to become
4 trillion by 2015, almost 20% of the gross domestic
product. Moreover, based on the advances in technology
in microelectronic miniaturization, integration, sensors,
the Internet and wireless networking; the deployment and
service of health care services will be fundamentally
changed and modernized. Via the use of BANs, health
care systems can be augmented to manage illness and
react to crisis rather than just wellness [8, 9].
A node in a body area network is referred to an
independent device with communication capability. Nodes
in BANs can be classified into three different categories
based on their functionality, implementation and role in
the network. In terms of functionality, there are the three
types of nodes: a) Sensors that measure certain parameters
in one’s body internally or externally and gather and re-
spond to data on a physical stimuli, process necessary data
and provide wireless response to information. b) Actuator
which interacts with the user once it receives data from
the sensors [6]. c) Personal Device (PD) which collects
all information received from sensors and actuators and
handles interaction with other users.
In terms of implementation nodes are classified into
three classes of Implant Node, Body Surface Node and
External Node; which are implanted in the human body
and, 2cm away from the body and farther away from
the it, respectively [10, 11]. Nodes in BANs can also
be classified into three types based on their role in the
network: a) Coordinator which is a gateway to the outside
world or another BAN, b) End Nodes which are only
capable of performing their embedded application, c)
Routers are intermediate nodes which have a parent node
and a few child nodes through which they relay messages.
Based on the IEEE 802.15.6 working group nodes in
BANs are considered to operate in either a one-hop or
two-hop star topology with the node in the center of the
star being placed on a location like the waist [12, 13]. As
for communication architecture, BANs can be separated
into three different tiers as follows: Intra-BAN (tier-1),
Inter-BAN (tier-2) and Extra-BAN (tier-3) shown in Fig.
1. These communication tiers cover multiple design issues
in facilitating an efficient, component-based system for
BANs [14]. As shown in Fig.1, the devices of BANs
are scattered all over the body in a centralized network
560
JOURNAL OF NETWORKS, VOL. 8, NO. 3, MARCH 2013
© 2013 ACADEMY PUBLISHER
Fig. 1. Communication Tiers in a Body Area Network
architecture where the precise location of a device is
application specific [14].
III. ROUTING CHALLENGES IN BANS
BANs span a wide area of medical and non-medical
applications from sport and entertainment to ubiquitous
health care, military and many more. The main goal of
all BAN applications is to improve one’s quality of life.
However, BANs applications have different architectures,
technological requirements, constraints and goals. This
Section covers a general view of challenges in different
BAN applications.
1) Postural Body Movements: The link quality be-
tween nodes in BANs varies as a function of time due
to postural body movements [15]. Thus, the proposed
routing algorithm should be adaptive to different topol-
ogy changes. In this regard, the authors of [16] have
considered BANs to be in the category of Delay Toler-
ant Networks (DTN) due to disconnection and frequent
partitioning concluded from postural body movements.
Moreover, body segments and clothing have been shown
to negatively intensify RF attenuation to signal blockage.
2) Efficient Transmission Range: Low RF transmission
range leads to disconnection and frequent partitioning
among sensors in BANs which leads to similar perfor-
mance to DTNs [16]. More specifically, if the transmis-
sion range of sensor nodes in a BAN is less than a
threshold value, the choice of the next sensors for routing
is reduced which causes higher number of transmissions
to obtain a route leading to an overall average temperature
rise. Moreover, the lower the number of neighbors the
less the probability for packets to arrive at the destina-
tion within a certain hop count. Hence, packets would
take longer to arrive at the destination and the average
temperature of the network will increase [1].
3) Limitation of Resources: The bandwidth in BANs
is limited and varies with interference, noise and fading.
Hence, the proposed routing protocol needs to be aware
of the limitation on network control, energy and data
gathered as the nodes in BANs may deplete due to
unavailable memory, battery and bandwidth which may
affect Quality of Service (QoS) [5].
4) Interference and Temperature Rise: In terms of
computing power and available energy, the energy level
of nodes needs to be taken into account in the proposed
routing protocol. The transmission power of nodes needs
to be extremely low in order to avoid tissue heating and
minimize interference [5].
5) Limitation of Packet Hop Count: Based on the
IEEE standard draft of IEEE 802.15.6 [17], only one-
hop or two-hop communication is defined for BANs.
Multi-hopping will increase overall system reliability by
providing stronger links. However, the larger number
of hops the higher the energy consumption [2]. Most
proposed BAN routing protocols have not considered the
limitation of number of hops.
6) Local Energy Awareness: The proposed routing
algorithm should not rely on one route and one node in
the network but has to further disperse its communication
data to avoid total power usage of a specific nodes leading
to node failure.
7) Global Network Lifetime: Network lifetime in
BANs is defined as the time interval between which the
network starts working to the time the first node dies
[15]. Network lifetime is of greater importance in BANs
compared to WSNs and Personal Area Networks (PAN) as
devices are expected to operate over a longer period e.g.
charging and battery replacement is not feasible in im-
plantable medical devices [12]. In this regard, simulation
results in various papers have clarified the improvement
of network lifetime through multihop relay networks [18].
8) Heterogenous Environment: Nodes in BANs can be
heterogenous. More specifically the memory and power
consumption of nodes may be different from one another,
which imposes several challenges to QoS in BANs [6].
IV. TEMPERATURE BASED ROUTING
Radio signals generated through wireless communica-
tion generate magnetic and electric fields. The exposure
of electromagnetic fields results in radiation absorption
of the human tissue leading to temperature rise [19]. This
will reduce blood flow and cause thermal damage to more
sensitive organs. Prolonged temperature rise inside the
human body tissue can lead to damage, growth of certain
types of bacteria, effect enzymatic reactions and reduce
blood flow in some organs [20]. The amount of radiation
energy absorbed by human tissue given in (1) is referred
as the Specific Absorption Rate (SAR) [19].
SAR =
σ|E|
2
ρ
(W/kg) (1)
where σ is the electrical conductivity of tissue, E is the
electric field induced by radiation and ρ is the density
of tissue. Experiments have shown exposure to SAR
of 8 W/kg for 15 minutes can cause significant tissue
damage [19]. Hence, BAN routing protocols must actively
decrease temperature and radiation emission. More specif-
ically, even routes with short delay and light traffic might
not be efficient in terms of temperature which makes
routing and forwarding intolerable for the nodes. The
JOURNAL OF NETWORKS, VOL. 8, NO. 3, MARCH 2013
561
© 2013 ACADEMY PUBLISHER
common objective of all temperature routing protocols
reviewed in this section is to maintain low temperature
among sensor nodes by avoiding routing on hot spots.
A. Thermal-aware routing algorithm (TARA)
The TARA [19] protocol has been considered for in-
body sensor networks and considers sensor locations
and cluster leadership history to minimize the hazardous
effects of temperature rise on the human tissue. It mea-
sures temperature changes of its neighboring sensor nodes
through monitoring neighbors packet count, calculation of
communication radiation and power consumption. TARA
aims to reduce the possibility of overheating and han-
dles packet transmission in temperature rise by defining
hotspots as areas that exceed a certain temperature due
to data communication. Accordingly, it aims to specify
paths to detour around the hotpots. As can be seen in
Fig.2, in cases where packets arrived at nodes surrounded
by hot spots, they are sent back to the sender and an
alternate path is specified to detour the routes. After
the hot spots have been cooled down to a certain limit,
they can be considered in later routing. TARA uses the
Finite-Difference Time-Domain (FDTD) [21, 22] method
to measure the Specific Absorption Rate (SAR) and
temperature rise of each node. This protocol measures
temperature rise by using the FDTD and Pennes bioheat
Equation shown in (2) [23], by which it discretizes the
problem space into small grids with a pair of coordinates
(i, j).
In (2), σ is the discretized space step (size of grid),
σ
t
is the discretized time step, b is the blood perfusion
constant, ρ is the mass density, C
p
is the specific tissue
heat, K is the thermal conductivity of the tissue, T
b
is
the temperature of the tissue and the blood; and P
c
is the
heat generated from power dissipation of circuitry. Based
on (2), the temperature of grid point (i, j) at time m + 1
is a function of the temperature of its surrounding grid
points (i + 1, j), (i, j + 1), (i −1, j) and (i, j −1) at time
m. TARA has shown to have low maximum temperature
rise and small average temperature rise which makes it
a safe routing protocol for use in in-body BANs. Also,
the thermal-aware capability of TARA leads to better load
balancing and less traffic congestion [19].
However, since TARA withholds packets from hot spot
regions and finds routes through alternate paths, there is
an average increase in the number of transmissions and
overall network temperature. Additionally, TARA only
considers temperature as a metric, has low network life-
time, high end-to-end delay, low reliability, high packet
loss ratio and does not consider power efficiency and link
probability.
B. Least Temperature Routing (LTR)
Bag et. al [24], have proposed the LTR protocol which
is a thermal aware routing protocol for BANs. LTR defines
hot spots as areas which have high temperature due to data
communication focus. Each node in LTR is assumed to
Fig. 2. TARA
Fig. 3. LTR
have knowledge of the temperature of its neighbor nodes,
similar to TARA. As shown in Fig. 3, unlike TARA, LTR
chooses its routes from neighbor nodes with the lowest
temperature. Hence, it sets its path to the coolest neighbor
without involving routing loops. In fact, a hop-count is
specified for each packet and is incremented by the value
of one each time a node forwards a packet. In order to
maintain the network bandwidth constraint, the packet
is discarded if it has exceeded the threshold value of
MAX HOP S, which is relative to the diameter of the
network. LTR also provides its packets with tables that
keep track of the sensor nodes through which the packets
have passed and avoids getting into infinite loops.
However, as nodes in LTR forward packets to nodes
with lowest temperature until the destination is reached,
there is potential for significant power consumption, over-
all temperature rise and waste of bandwidth throughout
the network as most nodes will be involved in routing.
Also, LTR does not ensure that packets are forwarded in
the direction of the destination, consequently the route
towards the destination is less optimal. Additionally, the
temperature of sensor nodes is variable over time which
will increase the end to end delay. LTR is considered a
greedy approach to routing that is not globally optimal,
but may be locally optimal [1].
C. Adaptive least temperature routing (ALTR)
Another temperature based routing scheme was re-
cently proposed in [24], namely ALTR. It is similar
to LTR in specifying MAX HOPs COUNT for packets
being routed to not exceed the MAX HOPs ADAPTIVE.
If the number of hops is less than or equal to
MAX HOPs ADAPTIVE, the same rules as the LTR
562
JOURNAL OF NETWORKS, VOL. 8, NO. 3, MARCH 2013
© 2013 ACADEMY PUBLISHER
T
m+1
(i, j) = [1 −
σ
t
b
ρC
p
−
4σ
t
K
ρC
p
σ
2
]T
m
(i, j) +
σ
t
C
p
SAR +
σ
t
b
ρC
p
T
b
+
σ
ρC
p
P
c
+
σ
t
K
ρC
p
σ
2
[T
m
(i + 1, j) + T
m
(i, j + 1) + T
m
(i − 1, j) + T
m
(i, j − 1)] (2)
Fig. 4. ALTR
algorithm apply. Whereas, in cases where the hop count is
higher than MAX HOPs ADAPTIVE, the Shortest Hop
algorithm (SHR) is used [24]. An example of routing
in ALTR is shown in Fig. 4. ALTR also differs from
LTR in being adaptive to different topologies, as it uses a
proactive delay strategy to cool down the temperature of
nodes in a ring topology which tends to increase rapidly
by passing the same path repeatedly. In cases where a
node receives a packet when even its coolest neighbor
has a high temperature, the node delays the packet by
one unit of time before sending it to its coolest neighbor.
Thus, a minor increase in packet delivery delay is traded
off for the average temperature of the network. Even with
a hop count specification in ALTR, network bandwidth is
wasted when routes calculated from SHR go through hot
spots. Also, as ALTR sends packets to neighbors with
minimum temperature, the overall network temperature
and number of hops will eventually increase. In fact,
this algorithm does not guarantee that packets are routed
towards the destination which leads to increase in sensor
temperature and hop count.
ALTR, LTR and TARA do not optimize routing in
terms of reliability, delay or efficiency. More specifically,
the excessive hop count leads to more than 50% packet
loss ratio which results in average network temperature
rise, energy wastage and low packet delivery ratio.
LTR and ALTR have shown to have lower temperature
rise at all packet arrival rates compared to TARA and
SHR. SHR has higher temperature rise as it ignores tem-
perature rise and aims to find the shortest route whereas
LTR and ALTR have better performance even at high
packet arrival rates as they route packets through cooler
nodes from the start [24].
LTR and ALTR have better end-to-end delay than
TARA at higher packet arrival rate. However, ALTR has
considerably lower delay than LTR due to its adaptive
nature. Also, TARA has the highest power consumption
compared to LTR and ALTR as it withdraws packets from
heated regions and detours them which leads to higher
power consumption [24]. Additionally, TARA experiences
larger number of hops and higher packet loss compared
to LTR and ALTR as it reroutes data from heated regions.
D. Least Total Route Temperature (LTRT)
LTRT is a temperature aware routing protocol proposed
in [1] which basically is a smart hybrid of LTR and
SHR. LTRT aims to optimize issues related total tem-
perature rise and redundant hops. Hence, it is designed
to reduce hop count to maintain network bandwidth and
select routes with minimum temperature from sender to
destination. LTRT uses the single source shortest path
(SSSP) algorithms of graph theory, Dijkstra’s algorithm,
to calculate its routes and uses the routes for further
transmission. Basically, LTRT translates the temperature
of sensors into graph weights which eventually lead to
minimum temperature routes. The temperature of each
sensor node is assigned as the weight of that sensor
node. It then transfers the weight of its sensor through
predefined outgoing edges that connect the nodes (Fig. 5).
The step by step procedure of route allocation in LTRT
is as follows:
a. Observe communication activity of neighbor sensor
nodes to assign the temperature of sensor nodes as
the weight of each sensor node.
b. Transfer weight of the sensor nodes to the weight of
outgoing edges connected to the node.
c. Find least temperature routes from sender to destina-
tion nodes by applying single shortest path algorithm
to the configured graph.
d. Update routes periodically to avoid excessive tem-
perature rise of sensor nodes and maintain topology
changes related to node mobility.
Simulation results in [1] have shown LTRT to have
lower average temperature rise, hop count per packet
compared to ALTR and LTR. This is because of spec-
ifying a route to the destination in LTRT before packet
transmission which affects the maximum number of hops
required to reach the destination node and the average
temperature rise in the network. Since LTRT and ALTR
are designed to not drop any packets in the routing
procedure, their packet loss ratio is nearly zero. Whereas,
LTR has a higher packet loss as it discards some packets
and the packets take more time to reach the destination
node which inevitably exceeds the maximum hop count
threshold. Even with increasing the number of nodes,
LTRT has lower average temperature rise compared to
ALTR and LTR.
JOURNAL OF NETWORKS, VOL. 8, NO. 3, MARCH 2013
563
© 2013 ACADEMY PUBLISHER
![Fig. 5. LTRT [1]](/figures/fig-5-ltrt-1-m9e421nt.webp)
![Fig. 8. Link quality matrix Mi of node ni [35]](/figures/fig-8-link-quality-matrix-mi-of-node-ni-35-2y5hh5fi.webp)
![Fig. 6. ANYBODY [28]](/figures/fig-6-anybody-28-fnj9spac.webp)
