ORE Open Research Exeter
TITLE
Energy-aware Dual-path Geographic Routing to Bypass Routing Holes in Wireless Sensor Networks
AUTHORS
Huang, H; Yin, H; Min, G; et al.
JOURNAL
IEEE Transactions on Mobile Computing
DEPOSITED IN ORE
06 November 2017
This version available at
http://hdl.handle.net/10871/30154
COPYRIGHT AND REUSE
Open Research Exeter makes this work available in accordance with publisher policies.
A NOTE ON VERSIONS
The version presented here may differ from the published version. If citing, you are advised to consult the published version for pagination, volume/issue and date of
publication
IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. XX, NO. XX, XX 20XX 1
Energy-aware Dual-path Geographic Routing to
Bypass Routing Holes in Wireless Sensor
Networks
Haojun Huang, Hao Yin, Geyong Min, Junbao Zhang, Yulei Wu, and Xu Zhang
Abstract—Geographic routing has been considered as an attractive approach for resource-constrained wireless sensor networks
(WSNs) since it exploits local location information instead of global topology information to route data. However, this routing approach
often suffers from the routing hole (i.e., an area free of nodes in the direction closer to destination) in various environments such as
buildings and obstacles during data delivery, resulting in route failure. Currently, existing geographic routing protocols tend to walk
along only one side of the routing holes to recover the route, thus achieving suboptimal network performance such as longer delivery
delay and lower delivery ratio. Furthermore, these protocols cannot guarantee that all packets are delivered in an energy-efficient
manner once encountering routing holes. In this paper, we focus on addressing these issues and propose an energy-aware dual-path
geographic routing (EDGR) protocol for better route recovery from routing holes. EDGR adaptively utilizes the location information,
residual energy, and the characteristics of energy consumption to make routing decisions, and dynamically exploits two node-disjoint
anchor lists, passing through two sides of the routing holes, to shift routing path for load balance. Moreover, we extend EDGR into
three-dimensional (3D) sensor networks to provide energy-aware routing for routing hole detour. Simulation results demonstrate that
EDGR exhibits higher energy efficiency, and has moderate performance improvements on network lifetime, packet delivery ratio, and
delivery delay, compared to other geographic routing protocols in WSNs over a variety of communication scenarios passing through
routing holes. The proposed EDGR is much applicable to resource-constrained WSNs with routing holes.
Index Terms—Wireless sensor networks, geographic routing, energy-aware routing, anchor list, routing hole.
F
1 INTRODUCTION
E
NERGY conservation and load balance are two im-
portant goals in designing routing protocols for
Wireless Sensor Networks (WSNs) due to two challenges
[1], [2]. First, the sensor nodes are usually powered only
by batteries but expected to operate for a long period;
Second, it is infeasible and costly to replace or recharge
batteries once sensor nodes have been deployed. Notice
that the routing holes, referring to an area free of nodes
closer to destination [2], [3], [4], [5], [6], [7], [11], [16],
[38], are hardly avoided in WSNs in various actual geo-
graphical environments such as puddles, obstacles, and
buildings, and this incurs additional energy expenditure
used for data delivery. In this paper, therefore, we focus
on designing energy-aware geographic routing protocols
regarding how to bypass routing holes for resource-
constrained WSNs, which can achieve both energy ef-
ficiency by selecting the energy-optimal forwarders and
• H. Huang is with the School of Electronic Information, Wuhan University,
Wuhan 430072, Hubei, China. E-mail: hhj0704@163.com.
• H. Yin is with Department of Computer Science and Technology, Tsinghua
University, Beijing 100081, China. E-mail: h-yin@mail.tsinghua.edu.cn.
• G. Min and Y. Wu are with the College of Engineering, Mathematics and
Physical Sciences, University of Exeter, Exeter EX4 4QF, U.K. E-mail:
g.min@exeter. ac.uk, and Y.L.Wu@exeter.ac.uk.
• J. Zhang is with the Department of Computer Science and Technology,
Zhongyuan University of Technology, Zhengzhou 450007, Henan, China.
E-mail: junbao.zhang.uestc@gmail.com.
• X. Zhang is with the School of Electronic Science and Engineering,
Nanjing University, Jiangsu, China. E-mail: xzhang17.cs@gmail.com.
load balance by employing two node-disjoint anchor lists
passing through two sides of the routing holes to shift
routing path.
Geographic routing, also referred to as position-based
[2], [30] or localized routing [16], [19], has been regard-
ed as an attractive approach for resource-constrained
WSNs, since it exploits local location information instead
of global topology information for data delivery. It is
based on the prerequisite that the nodes know their
actual or virtual locations, which can be made available
either through a Global Position System (GPS) receiver
or through some other ways [2], [37], and exchange such
information with neighbors periodically or actively. Be-
ing almost stateless and distributed, geographic routing
does not require dissemination of route establishment
information and maintenance of routing tables at each
node, thus making it efficient, scalable and promising
for WSNs. A recent detailed performance evaluation and
comparison on geographic routing is given in [38].
Generally, geographic routing utilizes greedy mode to
route data packets when it can find a neighbor closer to
the destination than the current forwarder, and switches
to bypass mode once the data packets encounter a
routing hole, where there is no such a node closer to
destination than the current forwarder (e.g., such an
issue exists for node u towards destination v in Fig.
1). To achieve our design goal, there are at least three
issues to be addressed for the current geographic routing
[2], [3], [5], [7], [11], [13], [16], [28]. First, how to detect
IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. XX, NO. XX, XX 20XX 2
v
1
v
2
v
3
w
2
w
3
w
1
v
u
s
(a)
v
u
s
(b)
burstpacketdeliveryinbypassmode
burstpacketdeliveryingreedymode
Fig. 1: Face routing bypassing the routing holes indicated
by red arrows: (a) employing the right-hand rule, and (b)
employing the left-hand rule.
routing holes before data delivery, since the routing holes
occurring in routing will generate long path and hence
consume additional resource. Second, how to bypass
routing holes for load balance. Third, how to select the
energy-aware forwarders and therefore guarantee that
all packets are delivered in an energy-efficient manner
for resource-constrained WSNs.
Currently, most geographic routing protocols, such as
Greedy Perimeter Stateless Routing (GPSR) [4], Greedy-
Face-Greedy (GFG) [6], and Greedy Other Adaptive Face
Routing (GOAFR++) [9], tend to exploit face routing
scheme to bypass routing holes, while doing not detect
them before data delivery. The basic idea is to planarize
the network graph using an algorithm like Relative
Neighborhood Graph (RNG), or Gabriel Graph (GG)
[4] and then forward messages to the destinations by
employing the right-hand rule or left-hand rule [4], [6]
along one or possibly a sequence of adjacent faces which
all locate in the one side of the line from the source
node to destination
1
. Fig. 1 elaborates an example of
the face routing by employing such two rules, which
detours a routing hole by forwarding the data to the
node that is first traversed by the arriving edge of the
packet counterclockwise or clockwise, shown in Fig. 1(a)
and Fig. 1(b), respectively. There are at least two paths
along the two sides of the routing hole, i.e., u-w
1
-w
2
-
w
3
· · · v in Fig. 1(a) and u-v
1
-v
2
-v
3
· · · v in Fig. 1(b), pro-
vided for route recovery for data delivery. However, the
right-hand rule only allows for counter-clockwise bypass
traversal and the left-hand rule only allows for clockwise
bypass traversal, meaning that both of them only walk
along one side of the routing holes for route recovery.
Beyond face routing over planar networks, there are also
other bypass approaches to recover route from routing
holes [7], [11], [16]. A comprehensive survey on various
bypass approaches is given in [5].
These proposed protocols are simple to be implement-
ed, whereas have a common shortcoming that they walk
along only one side of the routing holes to recover the
1. Note that the right-hand rule and the left-hand rule are equivalent
to traversing the face with the crossing edges removed. It is essential
to planarize the network for face routing. If such a crossing edge exists,
there must cause it failure due to routing loop.
route [2], [3], [4], [5], [6], [7], [9], [10], [11], [16], [20],
[28], [38]. Such approaches will make the traffic load
converged on the boundary of the routing holes, and
consequently achieve suboptimal network performance
such as longer delivery delay and lower delivery ratio
during routing packets. Furthermore, all of them cannot
guarantee that all packets are delivered in an energy-
efficient manner [18], [20], [24], [25], since these protocols
are more inclined to route data along the boundary of
the routing holes or tend to generate long path once
encountering the routing holes during routing data,
thereby consuming additional energy.
In this paper, we propose an energy-aware dual-path
geographic routing protocol called EDGR for better route
recovery from routing holes. The above-mentioned three
issues are taken into account in our routing design,
thus both energy conservation and load balance can
be achieved. The main contributions of this paper are
summarized as follows:
• EDGR establishes dual-path routing following two
node-disjoint anchor lists which pass through two
sides of the routing holes to route data, preventing
data from being forwarded along the boundary of
the routing holes. In this way, each data packet
is routed to destination along two different paths
in greedy mode only instead of bypass mode if
possible, thereby shortening the routing length and
balancing load.
• EDGR proposes a novel alternative approach to find
efficient forwarder in the presence of node failure
in the relay area, by introducing a random shift
to the location of subdestination. Such an approach
is feasible, reasonable, and energy-efficient without
additional communication overhead.
• We prove that EDGR is anchor list node-disjoint and
routing loop-free, and draw out its essential charac-
teristics in terms of time complexity for anchor list
building and successful routing probability.
• We extend EDGR into three-dimensional (3D) sen-
sor networks to provide energy-aware routing for
routing hole detour.
• We evaluate the performance of EDGR and its ex-
tension in a variety of communication scenarios,
including varied communication sessions, network
densities, and routing hole sizes. The results show
that EDGR outperforms the existing energy-aware
geographic routing protocols.
The remainder of this paper is organized as follows:
Section 2 describes the preliminary knowledge that
can benefit the understanding of the proposed EDGR
scheme. The detailed EDGR is given in Section 3. Section
4 presents the theoretical analysis of EDGR. Section 5
then describes the extension of EDGR in 3D sensor
networks for providing energy-aware routing. The sim-
ulation experiments and results are shown in Section
6. Section 7 provides an overview of the related work.
Finally, Section 8 draws the conclusions.
IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. XX, NO. XX, XX 20XX 3
2 PRELIMINARY KNOWLEDGE
In order to present the proposed scheme clearly, this sec-
tion will introduce the preliminary knowledge including
network model, energy model, lemmas and definitions.
2.1 Network Model
It is considered that no two nodes are located at the
same location, as in [24], [26], [27] and so on. All sensor
nodes are distributed in the network following Poisson
distribution. Each sensor node knows its own location
through an internal GPS device or a separate calibration
process, and knows the location of neighbors and their
residual energy within its maximum transmission power
by exchanging beacon messages. The source node can
obtain the location information of packet destinations by
some destination location services [31]. The location of a
node acts as its ID and its network address. Therefore,
there is no need for a separate ID establishment protocol.
We only consider bidirectional links, and assume that
each sensor node can adjust its transmission power from
0 to its maximum transmission power. We consider the
sensor nodes deployed in 2-dimensional (2D) WSNs
in the first stage of our scheme design and analysis,
and then extend this scheme into 3D WSNs to provide
energy-efficient localized routing for routing hole detour.
2.2 Energy Model
Similar to [26] and [28], the energy consumption c(u, v)
used for node u sending a bit data to a neighbor v
consists of four parts and can be characterized as follows:
c(u, v) = c
1
· d(u, v)
α
+ c
2
+ c
3
· d(u, v)
2
, (1)
where, d(u, v) denotes the Euclidian distance between
nodes u and v, α is a path loss constant between 2 and 5
depending on the transmission environment, and c
1
, c
2
and c
3
are some constants that are dependent on the
electronic characteristics and the characteristics of the
wireless devices. The parameters c
1
· d(u, v)
α
represents
the path loss between nodes u and v, c
2
denotes the
energy consumed by nodes u and v to process the signal,
and c
3
· d(u, v)
2
represents the energy used by nodes for
reception in the transmission range of sender. We assume
that each sensor node has the same c
1
, c
2
and c
3
.
2.3 Lemmas and Definitions
Let ξ[c(u, v)] and N represent the energy consumption
and the routing hops, respectively, for delivering one
bit data from current node u to destination v. Let d
o
satisfy 2c
1
(1 − 2
1−α
) − 2c
2
+ c
3
d
2
o
= 0 and d
opt
satisfy
c
1
(α − 1)d
α
opt
− c
2
+ c
3
d
2
opt
= 0. The characteristics of
energy consumption were investigated in [19], [26], [23],
[24], [28] based on the above energy mode or its original
prototype, and two lemmas were given as follows.
Lemma 1. If d(u, v) ≤d
o
, direct transmission is the most
energy-optimal way for data delivery from node u to node v.
56
v
u
opt
d
u
f
u
r
w
Fig. 2: Illustration of the ideal relay location and relay
region for the current forwarder u.
Lemma 2. If d(u, v) >d
o
, ξ[c(u, v)] is minimized when
all hop distances are equal to d(u, v)/d
opt
, and the optimal
routing hops N
opt
is bd(u, v)/d
opt
c or dd(u, v)/d
opt
e.
Lemmas 1 and 2 show that d
opt
is the energy-optimal
forwarding distance for minimizing ξ[c(u, v)]. This obser-
vation motivates us to introduce the concept of energy-
optimal relay region (see Definition 2) for energy con-
servation.
Definition 1 (Ideal Relay Location). The ideal relay lo-
cation f
u
for node u is defined as the location on the straight
line from node u to the anchor node or destination v, where
d(u, f
u
) = d
opt
.
Definition 2 (Relay Region). The relay region r(u, v)
for node u is defined as the circle area centered at f
u
with
radius r(u), where r(u) ≤d(u, f
u
) = d
opt
.
In order to make Definitions 1 and 2 clear, as shown in
Fig. 2, we give an illustration of the ideal relay location,
and the relay region r(u, v) of node u.
3 EDGR: ENERGY-AWARE DUAL-PATH GEO-
GRAPHIC ROUTING
This section will present in detail the proposed EDGR
protocol for bypassing routing holes in WSNs. First, the
EDGR architecture is presented, and then how to obtain
anchor list is introduced. Finally, we formulate how to
deliver messages in an energy-efficient manner.
3.1 EDGR Architecture
The main mechanism of EDGR is to employ two node-
disjoint anchor lists to guide packet delivery and select
the nodes with more residual energy from energy-aware
relay region as forwarders for energy conservation. Thus,
the data packets are likely routed to the anchor nodes
and their destinations along two paths at the energy-
efficient cost.
Fig. 3 illustrates the network architecture of EDGR.
The operation of EDGR is mainly divided into two
phases: anchor list obtaining and data dissemination.
In the first phase, the proposed EDGR scheme uses an
adaptive approach to obtain two anchor lists based on
the projected distance of nodes being involved in bypass
mode. In the latter phase, the proposed EDGR utilizes
geographic information, the residual energy and the
IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. XX, NO. XX, XX 20XX 4
18
EDGR routing
ideal relay location
sensor node
anchor node
the routing of the burst packet
u
v
relay region
Fig. 3: The network architecture of EDGR.
characteristics of energy consumption to make routing
decisions, and then unicasts the packet to the established
next-hop forwarder. If there is no node in the relay
region, the current forwarder then introduces a random
shift to the location of the subdestination to continue
data delivery.
There are three kinds of packets: beacon packet, burst
packet, and data packet in our scheme. The beacon
packet is used to exchange location information and
residual energy among neighbors, while burst packet
is used for finding the anchor lists. Fig. 4 presents the
format of the burst packet. Specifically, it includes anchor
lists, the locations of source node and destination. The
anchor list contains a series of anchor nodes, and a flag
field which indicates whether this packet bypasses the
routing holes by employing the right-hand rule or left-
hand rule. In addition, it includes a temporary void node
in each bypass mode but is deleted finally if it violates
the determined rules of anchor nodes. Here, the void
node is defined as the node that switches to bypass
mode from greedy mode, i.e., the node (e.g., node u in
Fig. 1) which cannot find a neighbor being closer to the
destination than itself, even though there is a path from
the source node to destination in the network.
3.2 Anchor List Obtaining
Given source node u, it starts preparing for data dissem-
ination by building two anchor lists.
First, it adaptively broadcasts a beacon packet to its
neighbors at the maximum transmission power for an-
nouncing its location information and residual energy.
Once a neighbor receives this packet, then stores such
information and broadcasts a new beacon packet to its
neighbors at the maximum transmission power. This
process will repeat periodically among all nodes in the
network, such that each node can obtain the location
information and residual energy of its neighbors within
their maximum transmission range.
Source node Anchor list Destination
Flag Void node Anchornode1
…
…….
.
Anchornoden
Fig. 4: Illustration of the burst packet format.
Once receiving location information and residual en-
ergy of all neighbors, source node u initiates and sends
a burst packet to destination v. For any forwarder w, it
uses greedy mode to forward this burst packet whenever
possible and switches to bypass mode when encoun-
tering a routing hole. The bypass mode begins from
the void node. For any void node w in the jth bypass
mode, it first adds itself into the burst packet header
(i.e., anchor list) for its downstream forwarders to decide
whether to return to greedy mode. Then, it takes into
consideration the following two cases to make routing
decisions on how to bypass this routing hole.
Case I. If the flag is void, it copies this burst packet
and then simultaneously uses right-hand rule and left-
hand rule to bypass this hole, and sets the flag of two
burst packets r and l, respectively, for their continued
delivery relayed by the subsequent nodes in accordance
with it.
Case II. If the flag is not void, this node and its
subsequent relay nodes use the right-hand rule indicated
by r or the left-hand rule indicated by l to detour this
routing hole.
For right-hand rule or left-hand rule, let
−−−−−−−→
w
f
j
· · · w
l
j
denote the routing of the jth bypass mode from node w
f
j
to node w
l
j
. Given a node w on
−−−−−−−→
w
f
j
· · · w
l
j
, we denote
its projected node as w
0
on the line from source node
u to destination v, and denote d(w, w
0
) as its projected
distance, illustrated in Fig. 5. In the jth bypass mode,
each node calculates the jth anchor node a
j
from its
upstream and downstream forwarders such that
a
j
≡{w
i
|max[d(w
i−1
, w
0
i−1
), d(w
i+1
, w
0
i+1
)]
≤ d(w
i
, w
0
i
), s.t. w
f
j
≤ i ≤ w
l
j
} ∪ {w
f
j
, w
l
j
}
(2)
where node w
f
j
−1
and node w
l
j
−1
both work in greedy
mode, and are w
f
j
’s upstream forwarder and w
l
j
’s
downstream forwarder, respectively. For candidate node
w
i
, its location will be installed into the anchor list and
sent to its downstream forwarder if it satisfies Eq.(2).
There is no additional communication overhead for it to
calculate d(w
i−1
, w
0
i−1
) and d(w
i+1
, w
0
i+1
) since the loca-
tion information of neighbors w
i−1
and w
i+1
is known
to it by beacon exchange.
Once the burst packet arrives at a node k such that
d(k, v)<d(w, v), then switches into greedy mode to con-
tinue delivery and deletes the temporary void node w
(by node k) if node w violates Eq.(2). The anchor list
List(u, v) is obtained as the flag and the union of all a
j
in all bypass modes, where all nodes are sorted in the
increasing order according to their subscripts. It can be
