Opportunistic routing algorithm for relay node selection in wireless sensor networks 1

112 IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. 11, NO. 1, FEBRUARY 2015
Opportunistic Routing Algorithm for Relay Node
Selection in Wireless Sensor Networks
Juan Luo, Member, IEEE, Jinyu Hu, Di Wu, Member, IEEE, and Renfa Li, Senior Member, IEEE
Abstract—Energy savings optimization becomes one of the
major concerns in the wireless sensor network (WSN) routing pro-
tocol design, due to the fact that most sensor nodes are equipped
with the limited nonrechargeable battery power. In this paper, we
focus on minimizing energy consumption and maximizing network
lifetime for data relay in one-dimensional (1-D) queue network.
Following the principle of opportunistic routing theory, multihop
relay decision to optimize the network energy efficiency is made
based on the differences among sensor nodes, in terms of both their
distance to sink and the residual energy of each other. Specifically,
an Energy Saving via Opportunistic Routing (ENS_OR) algorithm
is designed to ensure minimum power cost during data relay and
protect the nodes with relatively low residual energy. Extensive
simulations and real testbed results show that the proposed solu-
tion ENS_OR can significantly improve the network performance
on energy saving and wireless connectivity in comparison with
other existing WSN routing schemes.
Index Terms—Energy efficiency, one-dimensional (1-D) queue
network, opportunistic routing, relay node, wireless sensor
network (WSN).
I. INTRODUCTION
W IRELESS sensor network (WSN) offers a wide range
of applications in areas such as traffic monitoring,
medical care, inhospitable terrain, robotic exploration, and
agriculture surveillance [1]. The advent of efficient wireless
communications and advancement in electronics has enabled
the development of low-power, low-cost, and multifunctional
wireless sensor nodes that are characterized by miniaturization
and integration.
In WSNs, thousands of physically embedded sensor nodes
are distributed in possibly harsh terrain and in most applica-
tions, it is impossible to replenish energy via replacing batteries.
In order to cooperatively monitor physical or environmental
conditions, the main task of sensor nodes is to collect and
transmit data. It is well known that transmitting data consumes
much more energy than collecting data [2]. To improve the
Manuscript received October 09, 2013; revised June 30, 2014, September 18,
2014, and October 28, 2014; accepted November 15, 2014. Date of publication
November 24, 2014; date of current version February 02, 2015. This work is
supported in part by the National Key Technology R&D Program under Grant
2012BAD35B06, in part by the National Natural Science Foundation of China
under Grant 61370094, in part by the Natural Science Foundation of Hunan
under Grant 13JJ1014, and in part by the Program for New Century Excellent
Talents in University under Grant NCET-12-0164. Paper no. TII-14-0361.
J. Luo, J. Hu, and R. Li are with the College of Computer Science and
Electronic Engineering, Hunan University, Changsha 410082, China (e-mail:
juanluo@hnu.edu.cn; jinyuhu@hnu.edu.cn; lirenfa@hnu.edu.cn).
D. Wu is with the Department of Computer Science, University of California,
Irvine, CA 92697-3435 USA (e-mail: dwu3@ics.uci.edu).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TII.2014.2374071
energy efficiency for transmitting data, most of the existing
energy-efficient routing protocols attempt to find the mini-
mum energy path between a source and a sink to achieve
optimal energy consumption [3]–[5]. However, the task of
designing an energy-efficient routing protocol, in case of sen-
sor networks, is multifold, since it involves not only finding
the minimum energy path from a single sensor node to des-
tination, but also balancing the distribution of residual energy
of the whole network [6]. Furthermore, the unreliable wireless
links and network partition may cause packet loss and multiple
retransmissions in a preselected good path [7]. Retransmitting
packet over the preselected good path inevitably induces sig-
nificant energy cost. Therefore, it is necessary to make an
appropriate tradeoff between minimum energy consumption
and maximum network lifetime.
We focus on one-dimensional (1-D) queue network, which
has been designed and developed for a wide variety of industrial
and civilian applications, such as pipeline monitoring, electrical
power line monitoring, and intelligent traffic. Fig. 1 shows an
example, illustrating a pervasive traffic information acquisition
system based on 1-D queue network platform, where the nodes
are linearly deployed along the road. Most of the existing tradi-
tional traffic information acquisition systems are implemented
without power-saving management. With the demands of var-
ious sustainable developments in smart city, an energy saving
optimization solution for smart traffic information acquisition
should be taken into account. In our solution, when a motion
sensor node detects a vehicle in its sensing range, it will acquire
traffic information, such as traffic volume, vehicle velocity, and
traffic density. Sensor nodes will send the collected data to
relay sensor nodes, and then the relay sensor nodes forward
traffic information along the energy-efficient path to the sink
node that is one or more hops away. Finally, comprehensive
traffic information will be established by the sink node and
sent to the traffic management center. Meanwhile, traffic man-
agement center will select appropriate information and offer
it to the clients via the network. This smart traffic informa-
tion acquisition solution can be used to extend the lifetime of
1-D queue network in the need of energy saving in WSN-based
Information Technology (IT) infrastructure.
In this paper, we propose an energy-efficient routing algo-
rithm for above 1-D queue network, namely, Energy Saving
via Opportunistic Routing (ENS_OR). ENS_OR adopts a new
concept called energy equivalent node (EEN), which selecting
relay nodes based on opportunistic routing theory, to virtually
derive the optimal transmission distance for energy saving and
maximizing the lifetime of whole network. Since sensor nodes
are usually static, each sensor’s unique information, such as the
1551-3203 © 2014 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution
requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
www.redpel.com+917620593389

LUO et al.: OPPORTUNISTIC ROUTING ALGORITHM FOR RELAY NODE SELECTION 113
Fig. 1. Smart traffic information acquisition system.
distance of the sensor node to the sink and the residual energy
of each node, are crucial to determine the optimal transmission
distance; thus, it is necessary to consider these factors together
for opportunistic routing decision. ENS_OR selects a forwarder
set and prioritizes nodes in it, according to their virtual optimal
transmission distance and residual energy level. Nodes in this
forwarder set that are closer to EENs and have more residual
energy than the sender can be selected as forwarder candidates.
Our scheme is targeted for relatively dense 1-D queue networks,
and can improve the energy efficiency and prolong the lifetime
of the network.
The main contributions of this paper include the following.
1) We calculate the optimal transmission distance under the
ideal scenarios and further modify the value based on the
real conditions.
2) We define the concept of EEN to conduct energy optimal
strategy at the position based on the optimal transmission
distance.
3) We introduce the forwarder list based on the distances to
EEN and the residual energy of each node into EEN for
the selection of relay nodes.
4) We propose ENS_OR algorithm to maximize the energy
efficiency and increase the network lifetime.
The remainder of this paper is organized as follows.
Section II describes the related work. Section III introduces
1-D queue network and an energy models. Section IV pro-
poses the concept of EEN and initiates theoretical analysis of
the optimal transmission distance. To address the problem of
unbalanced distribution of residual energy, a new opportunis-
tic routing mechanism based on optimal energy strategy is
devised in Section V. While Section VI evaluates the integrated
performance of ENS_OR algorithm compared with existing
routing protocols. Finally, the conclusion and future directions
are drawn in Section VII.
II. RELATED WORK
In recent years, there are several studies on routing-related
parameters, like connectivity-related parameters and density of
the distributed nodes, in 1-D queue networks. Previous works
[8] and [9] studied the connectivity probability of two certain
nodes versus the entire network. Other work in [10], [11] inves-
tigated on uniformly and independently distribution under the
assumption that the transmission range is fixed among sensor
nodes.
Some energy-efficient approaches have been explored in
the literature [12]–[14]. As transmitting data consumes much
more energy than other tasks of sensor nodes, energy sav-
ings optimization is realized by finding the minimum energy
path between the source and sink in WSNs. In [12], the the-
oretical analysis about the optimal power control and optimal
forwarding distance of each single hop was discussed. There is
a tradeoff between using high power and long hop lengths and
using low power and shorter hop lengths. With this in mind,
minimum energy consumption can be achieved when each sen-
sor node locates with the optimal transmission distance away
from others in dense multihop wireless network. The most for-
ward within range (MFR) [13] routing approach has also been
considered in 1-D queue networks, which chooses the farthest
away neighboring node as the next forwarder, and eventually
results in less multihop delay, less power consumption. Another
approach proposed in [14] reduces the total consumed energy
based on two optimization objectives, i.e., path selection and
bit allocation. Packets with the optimum size are relayed to the
fusion node from sensor nodes in the best intermediate hops.
Surprisingly, the benefit of optimal bit allocation among the
sensor node has not been investigated in 1-D queue networks.
The unreliable wireless links makes routing in wireless
networks a challenging problem. In order to overcome this
problem, the concept of opportunistic routing was proposed in
[15]. Compared with traditional best path routing, opportunis-
tic routings, such as extremely opportunistic routing (ExOR)
[16], geographic random orwarding (GeRaF) [17], and efficient
QoS-aware geographic opportunistic routing (EQGOR) [18],
take advantage of the broadcast nature of the wireless medium,
and allow multiple neighbors that can overhear the transmission
to participate in forwarding packets. However, these routing
protocols did not address exploiting OR for selecting the appro-
priate forwarding list to minimize the energy consumption, and
optimize the design of an energy-efficient OR protocol for wire-
less networks. However, these routing protocols did not address
exploiting OR for selecting the appropriate forwarding list to
minimize the energy consumption, and optimize the design of
an energy-efficient OR protocol for wireless networks. Mao
et al. [19] introduced an energy-efficient opportunistic routing
strategy called energy-efficient opportunistic routing (EEOR),
which selects a forwarder set and prioritizes them using energy
savings optimization solution of forwarding data to the sink
node in WSNs.
While all of these routing methods to improve the energy
efficiency of individual node or the whole network can min-
imize energy consumption, it is equally important to focus on
other objectives such as network lifetime and residual energy of

Fig. 2. Queuing model of relay with maximal transmission range of R and
minimal transmission range dmin.
relay nodes. Therefore, it is reasonable to take residual energy
of sensor nodes as a primary metric into consideration.
III. NETWORK AND ENERGY MODELS
In this section, the network model and energy model will be
described.
A. Network Model
We consider a multihop WSN in a 1-D queue model as
shown in Fig. 2. We assume that our scheme is targeted for
relatively dense network, i.e., each relay node has plenty of
neighboring nodes. Nodes have some knowledge of the loca-
tion information of their direct neighboring nodes and the
position of the source node and the sink node. Every wire-
less sensor node has fixed maximum transmission range R and
minimal transmission range dmin. The 1-D queue network is
then constructed by a connected graph G = (V, E), where V
is a set of sensor nodes aligned on a single line and E is a
set of directed links between communication nodes. We set
the indices {0, 1, 2, . . . , h, n, . . . , M − 1, M} from left to right,
and two specific nodes with index 0 and index M among them
as the source node and the sink node. Let N (h) represents as
the neighbor set of a node h, i.e., n ∈ N (h). Each directed
link (h, n) has a nonnegative weigh w (h, n), which denotes the
total energy dissipation in transmission and receiving required
by node h to its neighboring node n.
B. Energy Model
In this work, we refer to a simplified power model of radio
communication as it is used in [20] and [21]. The energy
consumption can be expressed as follows:
ET = (Eelec + εampdτ
) B (1)
where Eelec is the basic energy consumption of sensor board
to run the transmitter or receiver circuitry, and εamp is its
energy dissipated in the transmit amplifier. d is the distance
between transmitter and receiver, τ is the channel path-loss
exponent of the antenna, which is affected by the radio fre-
quency (RF) environment and satisfies 2 ≤ τ ≤ 4. ET denotes
the energy consumption to transmit a B-bit message in a
distance d.
On the other hand, the energy consumption of receiver ER
can be calculated as follows:
ER = EelecB. (2)
In our model, since the noise and environmental factor are
constant, only the transmitter can adjust its transmission power
to make ET reach a minimum value.
IV. OPTIMAL TRANSMISSION SCHEMES
In this section, energy consumption analysis is conducted
on the proposed 1-D model, where data are delivered to sink
node through hop-by-hop connected relay nodes. Our objective
is to design an energy-efficient opportunistic routing strategy
for each relay node that ensures minimum power cost and pro-
tects the nodes with relatively low residual energy. Theorem
1 proves the optimal transmission distance dop of sensor node
under large-scale 1-D queue network.
Theorem 1: In a large-scale WSN where nodes are uniformly
and independently distributed in a 1-D queuing model, the posi-
tion of the sensor nodes h is xh (xh M), according to (1)
and (2), the optimal transmission distance dop for node h is
dop = M−xh
nop
= {(2Eelec)/[(τ − 1) εamp]}
1/τ
.
Proof: To illustrate this point, consider node h shown in
Fig. 2, the distance between hth node and the sink node is
d(h, m) = M − xh =
n
i=1
(xi − xi−1), where n represents the
number of hops that hth node relay data to sink. Thus, the total
consumed energy (Ch) of node h can be expressed as follows:
Ch =
n
i=1
ET +
n−1
i=1
ER
=
n
i=1
{[Eelec + εamp(xi − xi−1)
τ
] B} +
n−1
i=1
(EelecB) .
(3)
In order to minimize Ch, we use the average value inequality
to derive inequality
Ch ≥ (2n − 1) EelecB +
εamp
n
i=1
(xi − xi−1)
τ
B
nτ−1
. (4)
According to inequality (4), we have
Cmin
h (n) = (2n − 1) EelecB +
εamp(M − xh)
τ
B
nτ−1
. (5)
One way to optimize the overall energy consumption during
data relay is to take a derivative with respect to hop. We take
the first derivative of Cmin
h with respect to n as
∂Cmin
h /∂n = 2EelecB − (τ − 1)
εamp(M − xh)
τ
B
nτ
= 0.
(6)
This global minimum/maximum can be calculated as
follows:
nop =
[(τ − 1) εamp]
1/τ
(M − xh)
(2Eelec)
1/τ
. (7)
Then, we take the second derivative of Cmin
h with respect to
n as
∂2
Cmin
h
∂n2
n=
[(τ−1)εamp]1/τ
(M−xh)
(2Eelec)1/τ
= τ (τ − 1)
εamp(M − xh)
τ
B
nτ+1
> 0. (8)

Fig. 3. Real nodes and EEN in 1-D queue model.
From (8), we deduced that (7) is the global minimum with
respect to the energy consumption of node h. Hence, the cor-
responding optimal transmission distance dop for node h is
given by
dop =
M − xh
nop
= {(2Eelec)/[(τ − 1)εamp]}
1/τ
dmin < dop ≤ R.
(9)
Therefore, the proof of Theorem 1 is finished. However,
Theorem 1 is an ideal model for multihop 1-D queue net-
work. However, the distance between optimal next relay node
to source node could not actually equal to dop. Fig. 3 depicts
a realistic environment, where the optimal next relay node of
node h based on Theorem 1 would possibly be set between
two real relay nodes. To solve the problem, we further address
Theorem 1 that uses the idea of EEN to select the optimal next
relay nodes.
Definition 1: EEN is a virtual relay node that the relay func-
tion is realized by several real nodes and its energy consumption
equals to the total amount of energy of these real nodes.
In this paper, we only focus on the behavior of transmitter
for data relay in our model. We replace real nodes with EENs
and then obtain the minimum relay energy consumption of each
node according to Theorem 1. The illustration of this process is
shown in Fig. 3.
V. OPPORTUNISTIC ROUTING ALGORITHM FOR RELAY
NODE SELECTION
In this section, we further analyze the energy consumption of
large-scale network under 1-D model.
A. Problem of Optimal Energy Strategy
In order to acquire the minimum energy consumption during
data transmission in whole network, we introduce the concept
of EEN to conduct energy optimal strategy at the position based
on the optimal transmission distance dop. However, the optimal
energy strategy does not explicitly takes care of the residual
energy of relay nodes in the network. For instance, in the case
of hop-by-hop transmissions toward the sink node, the relay
nodes lying closer to the EENs tend to deplete their energy
faster than the others, since dop is a constant. As a conse-
quence, this uneven energy depletion dramatically reduces the
network lifetime and quickly exhausts the energy of these relay
nodes. Furthermore, such imbalance of energy consumption
eventually results in a network partition, although there may
be still significant amounts of energy left at the nodes farther
away. Therefore, we should readdress the optimal energy strat-
egy for large-scale network from Theorem 1. Inspired from the
opportunity routing approach, EEN is formed by jointly consid-
ering the distribution of real nodes and their relay priority. The
specific algorithm to choose EEN is described in the following
section.
B. Forwarder Set Selection for Optimal Energy Strategy
In the proposed Theorem 1, we conclude that the energy con-
sumption function (5) is convex with respect to the number of
hops n. We can achieve optimal energy strategy by choosing
optimal hops nop to determine optimal transmission distance
dop. In addition, factors such as energy-balanced of a network
and the residual energy of nodes are also considered while
selecting the available next-hop forwarder.
We assume that node h is sending a data packet to sink,
and h + i is one of neighbors of node h. If it is closer to
the estimated result in (9) and has more residual energy, the
neighboring node h + i can be a forwarding candidate, then the
network can obtain better energy usage. Moreover, these eli-
gible candidates rank themselves according to their distances
from the EEN and the residual energy of each node as
P(h + i) =
(dh+i − dh) 1
|dh+i−dop| + (Eh+i − ζ)
(h + i) ∈ F (h) , −R ≤ i ≤ R
(10)
where dh+i − dh is the distance between node h and neighbor
node h + i, Eh+i denotes the residual energy of node h + i, and
ζ denotes the value of energy threshold. F(h) (F(h) ⊆ N(h))
is the selected forwarding candidate set of node h. The larger
the value of P(h + i) is, the higher priority of the node will
be. Only the forwarder candidate with the highest priority is
selected as the next forwarder.
We use above forwarding candidate set to decide corre-
sponding energy saving strategy, which is specifically achieved
through the following opportunistic routing algorithm, called
ENS_OR.
C. ENS_OR Algorithm
In this section, we will describe how to select and priori-
tize the forwarder set using optimal energy strategy on each
node, and how to choose the optimal relay node among poten-
tial forwarders that respond in a priority order. In addition, the
transmitted data can be naturally classified into two categories:
1) the former is the collected data of its own; and 2) the latter
is the relay data from other nodes. Obviously, we should distin-
guish incoming data (the data of second category) by tracing the
ID of sender. Eventually, we introduce ENS_OR algorithm for
energy saving to select the next relay node which has the high-
est priority in forwarder set to forward the incoming ENS_OR
algorithm. Algorithm 1 depicts the pseudocode of ENS_OR
algorithm.

Algorithm 1. ENS_OR Algorithm
Require: di, dh, dop, Ei, ζ, where i ∈ F (h)
Ensure: the position of next forwarder dn.
Event: Node h has a data packet to send to the sink node.
/ ∗ Steps ∗ /
1: start a retransmission timer
2: select the forwarder set F(h) from neighboring nodes
N (h);
3: for each node i ∈ N (h) do
4: if ((d(i, dop) < d(h, dop)) ∪ (Ei > ζ)) then
5: add i to F(h);
6: end if
7: end for
8: prioritize the forwarder set using Optimal Energy Strategy;
9: for each node i ∈ F (h) do
10: P(i) = (di − dh) 1
|di−dop| + (Ei − ζ)
11: end for
12: broadcast the data packet;
13: for each node i ∈ F (h) do
14: receive the data packet;
15: checks the sender ID and start a timer and time(i) =
α
P (i) ;
16: end for
17: if node n which has the highest-priority receives the data
packet successfully then
18: reply an ACK to notify the sender;
19: for each node i ∈ F (h) except n do
20: discard the data packet and close timer;
21: end for
22: else
23: if the priority timer expire then
24: set n = n , node n has the lower-priority;
25: goto 17;
26: end if
27: end if
28: if no forwarding candidate has successfully received the
packet then
29: if the retransmission timer expire then
30: drop the data packet;
31: else
32: goto 2;
33: end if
34: end if
35: return
VI. PERFORMANCE EVALUATION IN DIFFERENT METRICS
A. Simulation Scenario Experiments
1) Simulation Environment: We conduct the simulation
experiments using MATLAB with 100 nodes uniformly and
independently distributed over a line. Each node has the same
frequency B = 1 Mbit/s, and firmware character Eelec and
εamp in (1) is set as 50 × 10−9
J/bit and 100 × 10−12
J/bit/m2
,
respectively. Path-loss exponent of environment τ is 2. Hence,
TABLE I
SIMULATION PARAMETERS
TABLE II
ABBREVIATIONS AND CORRESPONDING FULL NAMES
the value of optimal transmission distance dop in (9) is approx-
imately equal to 31.6 m. Since Eelec and εamp, τ are fixed, no
matter how the distance between two nearest nodes changes,
dop still will be 31.6 m, without change. The longest transmis-
sion distance of a single hop is 50 m and the initial energy
is 720 mJ. Other simulation parameters are listed in Table I.
In this one-source-one-sink topology, a node can only act
as a relaying node. In this paper, we ignore the interference
among the generated signals of each node. To fully analyze
the performance of ENS_OR, we compared it with the meth-
ods GeRaF and minimum transmission energy (MTE) which
represent the transmission power strategy with minimum trans-
mission power, to satisfy quality of service (QoS) requirement
of reception. The abbreviations and corresponding full names
are listed in Table II.
2) Performance Metrics: We define four main measurable
metrics to evaluate the effectiveness of ENS_OR algorithm for
data forwarding in 1-D queue networks.
1) Average of residual energy (ARE): Relay nodes left with
more average residual energy indicates that all the relay
nodes are alive for longer time, which would help to
prolong network lifetime.
2) Standard deviation of residual energy (SRE): We use SRE
as a metric to quantify the energy balance characteristic of
the routing protocol, we have noticed that high standard
deviation in the estimations of residual energy implies the
unbalanced energy dissipation among sensor nodes, and
lowering SRE is important for the routing protocol.
3) Receiving packets ratio (RPR): RPR is defined as the ratio
of the amount of packets received by the sink to the total
amount of packets sent by the source. In order to effec-
tively avoid the network partition, the sink should receive
most of the packets sent from the source, and eventually
results in a good connectivity of the network.

Fig. 4. Comparison of ARE according to time.
4) First dead node (FDN): We define this metric to eval-
uate the influence of the network connectivity. As the
first energy exhausted node appears, the probability of
network partition increases, and the connectivity of the
network goes bad.
5) Network lifetime (NL): The network lifetime of a 1-D
queue network is defined as the time when the sink is
unable to receive packet sent from the source. The net-
work lifetime is closely related to the energy consumption
and network partition. The higher the network lifetime is,
the more effectively the balance of energy consumption
will be achieved, and the more likely the network partition
is going to happen.
3) Evaluation of Relay Algorithm: Fig. 4 describes the
average residual energy as a function of time, when system
is fully operated. As we can see, in general, the total residual
energy decreases as the simulation time increases. This can be
explained by (1) and (2), where packet size grows incremen-
tally over time can communicate with more energy over a given
distance. ENS_OR can achieve higher average residual energy
compared with GeRaF and MTE, because of its energy optimal
strategy and opportunistic routing scheme. ENS_OR always
keeps the energy consumption at the lowest level. Due to the
lower energy consumption, a longer lifetime can be achieved as
well by ENS_OR method.
From Fig. 5, we notice that ENS_OR has a lower stan-
dard deviation of residual energy compared with GeRaF. MTE
has the lowest value, because MTE always deliver data to
sink node hop-by-hop, which implies that energy dissipation
of MTE strategy is balanced among relay nodes. However the
total energy consumption of delivery is maximal as shown
in Fig. 4. Thus, according to Figs. 4 and 5, we can also
infer that ENS_OR strategy is a better equilibrium energy
strategy.
Fig. 6 reports the RPR under different minimum distance
between two nearest nodes. Here we also observe from Fig. 6
that initially data received at sink node in ENS_OR is greater
than that in GeRaF. However, when the distance between
two nearest nodes exceeds 15 m, the difference between two
Fig. 5. Comparison of SRE according to time.
Fig. 6. Comparison of RPR according to the minimum distance between two
nearest nodes.
methods is rather small. Thus, ENS_OR receive more pack-
ets sent from the source than GeRaF, which can effectively
avoid the network partition and has a good connectivity of the
network. The more the number of data is transmitted means
more energy will be consumed. So there is a direct relationship
between the number of data received and energy consumption.
Therefore, the results are cross-checked by plotting (Fig. 4)
energy consumed in network over time.
There is a very strong correlation between FDN and NL. The
longer the network lifetime is, and the more slowly the first dead
node is going to appear. As shown in Figs. 7 and 8, the result
shows that the time that the first dead node appears in ENS_OR
is much later than that in MTE and GeRaF, and the life time of
ENS_OR is much longer. Since the optimal energy strategy will
especially protect the low energy nodes, ENS_OR performs the
best. Thus, ENS_OR guarantees both the extensive lifetime and
the largest conservation of energy.

Fig. 7. Comparison of FDN according to the minimum distance between two
nearest nodes.
Fig. 8. Comparison of NL according to the minimum distance between two
nearest nodes.
B. Realistic Scenario Experiments
Characterized by a true system-on-a-chip (SOC) solution tai-
lored for IEEE 802.15.4/Zigbee applications, CC2530 is used to
enable ZigBee nodes to be built to implement our opportunis-
tic routing algorithm in realistic testing environment. In order
to acquire accurate measurements of targets, the wireless sen-
sor nodes are tested outdoor and indoor separately. As shown in
Figs. 9 and 10, we deploy ZigBee nodes along the road in the
1-D queue network. ZigBee nodes in our network are divided
into three types, including coordinator, router and end device.
There is only one end device (source node) residing at the head
of 1-D queue network, and only one coordinator (sink node)
is set at the tail of 1-D queue network, others are all router
nodes (relay nodes). Through the multihop routing algorithm,
the end device transmits data packet every 1 s to the coordina-
tor. The detailed parameters and corresponding values used in
real testbed experiments are summarized in Table III.
Fig. 9. Network topology in the outdoor test.
Fig. 10. Network topology in the indoor test.
To validate and evaluate the performance of our ENS_OR
algorithm in real testbed experiments, we compared it with the
GeRaF and MTE routing algorithm by using the open-circuit
voltage method to measure the residual capacity on battery. In
this paper, we mainly focus on the energy consumption of router
nodes (relay nodes) for data forwarding in our network model,
and give no considerations to the temperature and aging of bat-
tery. Since our energy model only involves packets transmitting
and receiving, we turn off the sleep mode on ZigBee nodes.
As shown in Fig. 11, the effect of time on the discharge pro-
cesses in battery has been investigated by the measured voltage
method over a 5-h period starting from 13:20. The parameters
of the test circuit are given in Table IV. According to Fig. 11,
the discharge curve decreases rapidly after 0.8 V. This means
the low cutoff voltage of our test battery is 0.8 V. Therefore, we
assume that when the battery voltage of node reaches 0.8 V, this

TABLE III
PARAMETERS AND CORRESPONDING VALUES USED IN REAL TESTBED
EXPERIMENTS
Fig. 11. Effect of time on the discharge processes.
TABLE IV
PARAMETERS AND CORRESPONDING VALUES USED IN DISCHARGE
EXPERIMENT
node may run out of energy, and define 0.8 V as the voltage of
first dead node.
Fig. 12 shows the average voltage measured on router nodes
with varying time. The initial average voltage is 3 V. At the
end of the experiment, the initial average voltage of ENS_OR
is 2.847 V, which is the highest value compared with others.
Fig. 12. Average voltage of the batteries according to time in the outdoor test.
Fig. 13. Average voltage of the batteries according to time in the indoor test.
GeRaF is following behind ENS_OR as 2.774 V, and MTE has
the lowest value 2.758 V. As we can see, the energy efficiency
of ENS_OR is better than GeRaF and MTE, which implies the
more average residual energy is left. So far, we have evidences
to conclude that ENS_OR can improve the energy efficiency of
individual node or the whole network, and prolong the lifetime
of whole network even in the realistic scenario.
In Fig. 13, the measured voltage of ENS_OR algorithm is
shown and compared to the measured voltage of other two
algorithms. From this figure, it is clear that the proposed algo-
rithm demonstrates desirable performance in prolonging the
network lifetime. Since the optimal energy strategy will espe-
cially protect the low energy nodes and balance the energy
consumption, ENS_OR performs well. Both GeRaF and MTE
give no consideration to the residual energy of relay nodes, and
their performance is much worse than that of ENS_OR because
low residual energy of relay nodes would run out more quickly
for transmitting large amount of data.
Furthermore, Figs. 12 and 13 display the average voltage
together with the associated confidence interval. As shown in

Fig. 14. Times of FND and NL.
Fig. 12, confidence intervals for ENS_OR, GeRaF, and MTE
increase over time, as well as the magnitude of the sample sizes.
However, confidence interval for MTE in Fig. 13 decreases
after 88 h. This means that the difference between the voltage
variation of nodes becomes smaller.
Fig. 14 depicts the times of first dead node and network life-
time in the whole network. It is noticed that the time when
the dead node first appeared in ENS_OR is later than that in
MTE and GeRaF as well as the network lifetime. Because of
the increasing number of the energy exhausted nodes, the end
device (sink) is unable to receive packet sent from the coordi-
nator (source), i.e., 1-D queue network is not working, packet
data cannot be forwarded through other routers (relay nodes).
ENS_OR has the energy optimal strategy and opportunistic
routing scheme to reduce the number of energy exhausted
nodes, and can prolong the lifetime of the network.
VII. CONCLUSION
WSN has been widely used for monitoring and control appli-
cations in our daily life due to its promising features, such as
low cost, low power, easy implementation, and easy mainte-
nance. However, most of sensor nodes are equipped with the
limited nonrechargeable battery power. Energy savings opti-
mization, therefore, becomes one of major concerns in the
WSN routing protocol design.
In this paper, we focus on minimizing energy consump-
tion and maximizing network lifetime of 1-D queue network
where sensors’ locations are predetermined and unchangeable.
For this matter, we borrow the knowledge from opportunistic
routing theory to optimize the network energy efficiency by
considering the differences among sensor nodes in terms of
both their distance to sink and residual energy of each other.
We implement opportunistic routing theory to virtually real-
ize the relay node when actual relay nodes are predetermined
which cannot be moved to the place according to the opti-
mal transmission distance. This will prolong the lifetime of the
network. Hence, our objective is to design an energy-efficient
opportunistic routing strategy that ensures minimum power is
cost and protects the nodes with relatively low residual energy.
Numerous simulation results and real testbed results show that
the proposed solution ENS_OR makes significant improve-
ments in energy saving and network partition as compared with
other existing routing algorithms.
In the future, the proposed routing algorithm will be extended
to sleep mode and therefore a longer network lifetime can be
achieved. Apart from that, an analytical investigation of the new
energy model include sleep mode will be performed.
REFERENCES
[1] D. Bruckner, C. Picus, R. Velik, W. Herzner, and G. Zucker, “Hierarchical
semantic processing architecture for smart sensors in surveillance net-
works,” IEEE Trans. Ind. Informat., vol. 8, no. 2, pp. 291–301, May
2012.
[2] G. J. Pottie and W. J. Kaiser, “Wireless integrated network sensors,”
Commun. Assoc. Comput. Mach., vol. 43, no. 5, pp. 51–58, 2000.
[3] L. LoBello and E. Toscano, “An adaptive approach to topology manage-
ment in large and dense real-time wireless sensor networks,” IEEE Trans.
Ind. Informat., vol. 5, no. 3, pp. 314–324, Aug. 2009.
[4] D. Hoang, P. Yadav, R. Kumar, and S. Panda, “Real-time implementa-
tion of a harmony search algorithm-based clustering protocol for energy
efficient wireless sensor networks,” IEEE Trans. Ind. Informat., vol. 10,
no. 1, pp. 774–783, Feb. 2014.
[5] D. Zhang, G. Li, K. Zheng, X. Ming, and Z.-H. Pan, “An energy-balanced
routing method based on forward-aware factor for wireless sensor net-
works,” IEEE Trans. Ind. Informat., vol. 10, no. 1, pp. 766–773, Feb.
2014.
[6] F. Ren, J. Zhang, T. He, C. Lin, and S. K. Ren, “EBRP: Energy-
balanced routing protocol for data gathering in wireless sensor networks,”
IEEE Trans. Parallel Distrib. Syst., vol. 22, no. 12, pp. 2108–2125,
Dec. 2011.
[7] A. Behnad and S. Nader-Esfahani, “On the statistics of MFR routing in
one-dimensional ad hoc networks,” IEEE Trans. Veh. Technol., vol. 60,
no. 7, pp. 3276–3289, Sep. 2011.
[8] A. Ghasemi and S. Nader-Esfahani, “Exact probability of connectiv-
ity one-dimensional ad hoc wireless networks,” IEEE Commun. Lett.,
vol. 10, no. 4, pp. 251–253, Apr. 2006.
[9] A. Behnad and S. Nader-Esfahani, “Probability of node to base station
connectivity in one-dimensional ad hoc networks,” IEEE Commun. Lett.,
vol. 14, no. 7, pp. 650–652, Jul. 2010.
[10] P. Piret, “On the connectivity of radio networks,” IEEE Trans. Inf. Theory,
vol. 37, no. 5, pp. 1490–1492, Sep. 1991.
[11] P. Santi and D. M. Blough, “The critical transmitting range for connec-
tivity in sparse wireless ad hoc networks,” IEEE Trans. Mobile Comput.,
vol. 2, no. 1, pp. 25–39, Jan./Mar. 2003.
[12] V. Ramaiyan, A. Kumar, and E. Altman, “Optimal hop distance and
power control for a single cell, dense, ad hoc wireless network,” IEEE
Trans. Mobile Comput., vol. 11, no. 11, pp. 1601–1612, Nov. 2012.
[13] S. Dulman, M. Rossi, P. Havinga, and M. Zorzi, “On the hop count
statistics for randomly deployed wireless sensor networks,” Int. J. Sensor
Netw., vol. 1, no. 1, pp. 89–102, 2006.
[14] Y. Keshtkarjahromi, R. Ansari, and A. Khokhar, “Energy efficient decen-
tralized detection based on bit-optimal multi-hop transmission in one-
dimensional wireless sensor networks,” in Proc. Int. Fed. Inf. Process.
Wireless Days (WD), 2013, pp. 1–8.
[15] H. Liu, B. Zhang, H. T. Mouftah, X. Shen, and J. Ma, “Opportunistic
routing for wireless ad hoc and sensor networks: Present and future
directions,” IEEE Commun. Mag., vol. 47, no. 12, pp. 103–109, Dec.
2009.
[16] S. Biswas and R. Morris, “Exor: Opportunistic multi-hop routing for
wireless networks,” in Assoc. Comput. Mach. SIGCOMM Comput.
Commun. Rev., 2005, vol. 35, no. 4, pp. 133–144.
[17] M. Zorzi and R. R. Rao, “Geographic random forwarding (geraf) for ad
hoc and sensor networks: Multihop performance,” IEEE Trans. Mobile
Comput., vol. 2, no. 4, pp. 337–348, Oct./Dec. 2003.
[18] L. Cheng, J. Niu, J. Cao, S. Das, and Y. Gu, “Qos aware geographic
opportunistic routing in wireless sensor networks,” IEEE Trans. Parallel
Distrib. Syst., vol. 25, no. 7, pp. 1864–1875, Jul. 2014.
[19] X. Mao, S. Tang, X. Xu, X. Li, and H. Ma, “Energy efficient opportunistic
routing in wireless sensor networks,” IEEE Trans. Parallel Distrib. Syst.,
vol. 22, no. 11, pp. 1934–1942, Nov. 2011.

[20] M. Bhardwaj, T. Garnett, and A. P. Chandrakasan, “Upper bounds on the
lifetime of sensor networks,” in Proc. IEEE Int. Conf. Commun. (ICC’01),
2001, vol. 3, pp. 785–790.
[21] R. Min, M. Bhardwaj, N. Ickes, A. Wang, and A. Chandrakasan, “The
hardware and the network: Total-system strategies for power aware wire-
less microsensors,” in Proc. IEEE CAS Workshop Wireless Commun.
Netw., Pasadena, CA, USA, 2002, pp. 36–12.
Juan Luo (M’10) received the Bachelor’s degree in
electronic engineering from the National University
of Defense Technology, Changsha, China, in 1997,
and the Master’s and Ph.D. degrees in communica-
tion and information systems from Wuhan University,
Wuhan, China, in 2000 and 2005, respectively.
Currently, she is an Associate Professor and
Doctoral Supervisor with the College of Computer
Science and Electronic Engineering, Hunan
University, Changsha. She is also the Director
with the Department of Network and Information
Security, Hunan University. From 2008 to 2009, she was a Visiting Scholar
with the University of California at Irvine, Irvine, CA, USA. She has
authored/coauthored more than 40 papers. Her research interests include
wireless networks, cloud computing, and wireless sensor networks.
Dr. Luo is a Member of the Association for Computing Machinery (ACM)
and a Senior Member of China Computer Federation (CCF).
Jinyu Hu received the B.E. degree in communica-
tion engineering from Hunan University, Changsha,
China, in 2009, and the M.E. degree in computer
applications technology from Chongqing Jiaotong
University, Chongqing, China, in 2013. Currently,
she is pursuing the Ph.D. degree at the College
of Computer Science and Electronic Engineering,
Hunan University.
Her research interests include wireless networks.
Di Wu (M’14) received the Bachelor’s degree in com-
munication engineering and the Master’s degree in
communication and information systems from Hunan
University, Changsha, China, in 2004 and 2007,
respectively, and the Ph.D. degree in computer sci-
ence from the University of California at Irvine,
Irvine, CA, USA, in 2012.
His research interests include wireless networks
and mobile computing, cyber-physical transportation
systems, and mobile social networks.
Renfa Li (M’05–SM’10) received the Bachelor’s
and Master’s degrees in electrical engineering and
automation from Tianjin University, Tianjin, China,
in 1982 and 1987, respectively, and the Ph.D. degree
in computer science from Huazhong University of
Science and Technology, Wuhan, China, in 2003.
He is a Distinguished Professor and Doctoral
Supervisor in Embedding Computing and Wireless
Network with Hunan University, Changsha, China.
He is the editor of eight books. His research inter-
ests include wireless network, embedding system,
system-on-a-chip (SOC) design, network security, and confrontation. Currently,
he is the Chairman of Enbedded and Network Laboratory, Hunan, China.
Dr. Li is a Senior Member of the Association for Computing Machinery
(ACM) and China Computer Federation (CCF).

Opportunistic routing algorithm for relay node selection in wireless sensor networks 1

Recommended

More Related Content

What's hot (20)

Viewers also liked (6)

Similar to Opportunistic routing algorithm for relay node selection in wireless sensor networks 1 (20)

More from redpel dot com (20)

Recently uploaded (20)

Opportunistic routing algorithm for relay node selection in wireless sensor networks 1