Computers and Electrical Engineering 56 (2016) 441–455

Contents lists available at ScienceDirect

Computers and Electrical Engineering

journal homepage: www.elsevier.com/locate/compeleceng

A genetic algorithm based distance-aware routing protocol for

wireless sensor networksR
Tarunpreet Bhatia∗, Simmi Kansal, Shivani Goel, A.K. Verma
Thapar University, Bhadson Road, Patiala, Punjab 147004

a r t i c l e i n f o a b s t r a c t

Article history: A Wireless Sensor Network (WSN) is an aggregation of sensor nodes which are remotely
Received 30 January 2016 deployed in large numbers, operate autonomously in an unattended environment and have
Revised 14 September 2016
limited energy resource. In most of the hierarchical routing protocols, the cluster head (CH)
Accepted 14 September 2016
selection is on the basis of random probability equation. There is a scope to reduce the
Available online 20 September 2016
energy dissipation by improving CH selection procedure. The proposed scheme, coined as
Keywords: GADA-LEACH, makes use of evolutionary genetic algorithm for improving CH selection in
Wireless sensor networks (WSNs) legacy LEACH routing protocol in sensor networks. The concept of relay node is introduced
Leach which acts as an intermediary between CH and base station (BS) to ease the communica-
Genetic algorithm tion between the CH and BS. The simulation results obtained supports that our proposed
Relay node algorithm is efficient in terms of network lifetime.
GADA-LEACH © 2016 Elsevier Ltd. All rights reserved.
Network lifetime

1. Introduction to wireless sensor networks (WSNs)

Wireless Sensor Networks (WSNs) consist of mobile or stationary sensor nodes which are spatially distributed for sensing
and processesing the data. Sensor nodes sense the event according to the intended application. WSN is used for monitor-
ing physical and environmental conditions such as temperature, sound, pressure etc. Some of its applications include area
monitoring, air pollution monitoring, disaster management, security surveillance, healthcare monitoring, industrial monitor-
ing, forest fire monitoring, landslide monitoring etc [1]. Due to mobility in sensor nodes, the topology of network changes
dynamically [2,3]. The architecture of WSN is shown in Fig 1.
Some of the limitations from which WSNs suffer are limited battery source, lower data rates, slower computing speeds,
smaller memory and limited communication range. Sensor nodes are battery powered and have to operate in an unattended
environment for a longer period of time so it becomes cumbersome to change or recharge their batteries. Due to resource
constrained nature of WSNs, routing has become a challenging task. These limitations should be given due weightage while
designing the routing protocols for WSNs as they directly impact functioning of whole network [4]. The focus of routing
protocols in WSN should be more in finding out the routes that may result in prolonged lifetime of the network apart from
considering other parameters such as shortest distance, minimum delay or maximum bandwidth.
There are various types of routing protocols proposed in literature for WSNs such as flat, hierarchical and location based
protocols. Among all, hierarchical or cluster based routing protocols are one of the most efficient due to high network scala-
bility, energy efficiency, low data retransmission, efficient sharing and optimum use of constrained resources [5]. The major

R
Reviews processed and recommended for publication to the Editor-in-Chief by Guest Editor Dr. M. S. Kumar.
∗
Corresponding author. Fax: 0175 2393005.
E-mail addresses: tarunpreetbhatia@gmail.com (T. Bhatia), simmikansal877@gmail.com (S. Kansal), shivani@thapar.edu (S. Goel), akverma@thapar.edu
(A.K. Verma).

http://dx.doi.org/10.1016/j.compeleceng.2016.09.016
0045-7906/© 2016 Elsevier Ltd. All rights reserved.
442 T. Bhatia et al. / Computers and Electrical Engineering 56 (2016) 441–455

Fig. 1. Wireless sensor network architecture.

design issue for such protocols is formation of clusters and selection of CHs to reduce the energy consumption. LEACH, one
of the popular hierarchical routing protocols [6], though uses cluster based routing to maximize energy consumption dis-
tributes energy consumption non-uniformly as CH selection doesn’t consider residual energy and distance of nodes selected
as CH. The major challenge in the design of routing protocols for WSNs are minimizing energy dissipation and maximiz-
ing network lifetime. In the proposed work, the cluster formation and CHs communication procedure of LEACH protocol
has been modified in order to improve network lifetime. The proposed algorithm makes use of Genetic Algorithm (GA) for
optimized CH selection and relay node for Distance Aware (DA) routing. It is named as GADA-LEACH.
This paper is organized as follows: Section 2 discusses related work. Section 3 gives an overview of Genetic Algorithm.
Section 4 discusses the proposed algorithm GADA-LEACH followed by Section 5 presenting the simulation environment and
experimental results of GADA-LEACH in comparison to various LEACH based protocols such as LEACH, LEACH-E, LEACH-EX,
M2NGA, GP-LEACH and LEACH-H and finally conclusion and future scope is presented in Section 6.

2. Related work

Recent advances in WSNs have led to designing many new routing protocols specifically clustering or hierarchical proto-
cols, supporting longer network life. The energy efficiency of routing protocols in WSN still remains an open challenge for
the research community [6]. There are several hierarchical routing protocols proposed in the literature for sensor networks
[7–10], their main objective is to generate stable clusters in mobile environment. Many of such techniques focus on node
reachability and route stability leaving other critical factors such as network longevity and coverage. The energy consump-
tion of the sensor nodes is mainly dominated by data transmission and reception. Therefore, routing protocols for WSNs
should focus on energy efficient routing for extending the lifetime of individual sensors and thus network.
LEACH, one of the first hierarchical routing protocols, was proposed by Heinzelman [11] in the year 20 0 0. The process of
generating clusters of nodes where one node acts as a cluster head is performed in periodical manner, dividing every round
into two phases: a) Cluster Building Phase: One of the node is elected as a cluster head by following cluster head election
process. Then this CH sends message to non-cluster nodes informing about its new role. The non-cluster nodes sends joining
request to the Cluster head on the basis of received signal strength. b) Stable Data Communication Phase: Cluster member’s
sends data to the CH, and then CH aggregates the received data and forwards it to the sink node.
Kumar et al. [12] proposed a protocol EEHC (Energy Efficient heterogeneous clustered scheme for WSN) that aims at
increasing the stability and network lifetime in presence of heterogeneous nodes. Nodes in the network are divided into
three categories: normal nodes, advanced nodes and super nodes which are referred to as heterogeneous nodes. In this, it
is assumed that energy of advanced nodes is more than energy of normal nodes and energy of super nodes is more than
 T. Bhatia et al. / Computers and Electrical Engineering 56 (2016) 441–455 443

energy of normal nodes. In this protocol, the lifetime and performance of network is improved by using weighted probability
of election of CH.
Researchers have also explored the use of evolutionary computing technique such as Genetic algorithm and have pro-
posed various solutions. One of the initial GA based solution was given by Bari et al. [13] for improving the network lifetime
in which high power relay nodes are used as cluster heads. The lifetime of relay node determines the lifetime of network.
In the proposed scheme, genetic algorithm is used for scheduling the data gathering of relay node by which lifetime of relay
node increases. The fitness function uses initial energy of all relay nodes which is known beforehand and maximum energy
dissipated by any relay node. Once the sensor nodes sends the data to relay nodes, these relay nodes further transmits the
gathered data to either base station directly or to the other relay node which is more closer to it.
Gautam et al. [14] proposed a protocol DAIC (Distance Aware Intelligent Clustering Protocol for WSN) that involves di-
viding the network into number of tiers with each tier having gateway CH except the tier farthest from base station. This
protocol works in rounds which is further split into phases: Setup phase and Steady State phase. In setup phase there is
network setup at first followed by routing path construction and schedule creation phase. Network setup phase includes
formation of tiers, selection of CH and formation of clusters. Liu et al. [15] proposed a protocol LEACH-GA which consists of
setup phase and steady state phase for each round with an additional phase named preparation phase before the beginning
of the first round. In preparation phase, CH selection procedure is followed by the nodes by sending messages to BS such
as their status of being CH or not, its node id and its distance from BS. After receiving the messages from the nodes, the
BS computes the optimal probability of node by using genetic algorithm to become CH. Then the BS sends the values of
computed optimal probability values to all the nodes so as to form clusters during the setup phase.
Naregal has proposed LEACH-E and LEACH-EX protocol [16] which is an improvement over LEACH. In LEACH-E, the resid-
ual energy of nodes is also considered during the second round of the selection of the cluster head thereby making the
network more energy-efficient than LEACH. In LEACH-EX, formula for calculation of threshold in LEACH-E has been simpli-
fied by taking the ratio of current energy and initial energy, instead of taking the square root of this ratio which results into
reduced computational complexity and increased overall probability of node to be selected as CH. Karimi et al. proposed GP-
LEACH [17] which optimizes CH selection in WSNs using GA by partitioning the network and considering residual energy
and nodes’ position information for optimization. It ensures CH selection from each partition for better network stability.
The simulation results show that GP-LEACH is more efficient than P-LEACH [18].
Peiravi et al. proposed M2NGA [19], a multi-objective two-nested genetic algorithm based clustering algorithm to opti-
mize network lifetime and delay. The top level GA considers energy consumption for sending one bit to CH and delay in
terms of hop count as fitness function to optimize network lifetime. The lower level GA is used within the cluster to op-
timize communication from sensor nodes to CH. Their approach works only for the homogeneous and static WSNs. It also
requires BS to have complete knowledge of geographic location of all other nodes in the network which increases commu-
nication overhead.
Salim et al. proposed an Intra-balanced routing protocol (IBLEACH) [20] for balancing the energy consumption in WSNs
which is an improvement to LEACH in terms of the minimization of energy consumption and the lifetime of the network.
In this protocol, a new phase i.e. pre-steady phase is introduced between the setup phase and steady phase of round. There
is random election of CH and dynamic formation of the clusters similar to the LEACH protocol in setup phase. In the pre-
steady phase, the cluster’s workload is calculated in one frame, and the cluster member (CM) is selected from the cluster
nodes for handling the process of data aggregation in the round through all the frames. If any such type of node doesn’t
exist, then CM is elected for handling the aggregation process for each frame in the round and the frames that do not have
aggregators, their aggregation process is handled by the CH itself. The sensor nodes by the end of this phase are divided
into two categories i.e. CH and CM in each frame. The third phase i.e. the steady state phase is divided into frames [21]. All
the nodes transfer their data in each frame to the aggregator in their time slots. Once the aggregator receives all the data
from the nodes, it gathers the whole data and forwards it to base station.
Balamurugan et al. [22] proposed a Fitness Based Routing Protocol for the optimization of energy efficient data transmis-
sion so as to increase the network lifetime. In this protocol, reduction in consumption of energy by the nodes is done by
selecting the nodes with minimum hop count and distance. Fitness value of the node is computed on the basis of distance
between node and the base station. In this scheme, there is considerable increase in the throughput, reduction in end-to-
end delay and the packet loss. Lee et al. [23] proposed a protocol to balance the clusters by considering the importance of
selection of cluster head in the network. In this scheme, the sensor nodes are partitioned into two layers and CH selection
is done based on the distance between nodes and base station. Each layer is sequenced according to the distance between
nodes and BS. The second layer is the sub layer of the first layer. Without communicating with other nodes, the selection of
CH is made in the same sequence of second layer.
Miao et al. proposed LEACH-H [24] as an improvement over LEACH based on genetic algorithm for WSN. It used residual
energy of nodes, number of neighboring nodes and the distance between nodes and base station for calculating fitness.
The authors compared LEACH-H with LEACH only and gave results that lifetime increased by 35%. The major limitation of
this work was not considering distance between CHs and BS for optimization. In order to design energy efficient hierarchal
routing protocols, the cluster formation phase needs consideration/attention. Many parameters such as network topology,
data forwarding techniques, distance between the nodes and base station etc. have to be focused upon. The previous studies
conducted consider some of the parameters for increasing the efficiency of the network [15,22,23] but some of them works
for homogeneous networks only [19] or did not consider distance between CH and BS for optimal routing [24].
444 T. Bhatia et al. / Computers and Electrical Engineering 56 (2016) 441–455

Fig. 2. Flowchart of genetic algorithm.

3. Overview of genetic algorithm

Genetic Algorithm is an evolutionary algorithm used to generate solutions to optimization and search problems [25]. It
uses techniques inspired by natural selection to obtain the optimized results. Such techniques are selection, crossover and
mutation. The general flowchart of genetic algorithm is given in Fig 2. In the first step, initial population is generated ran-
domly and the size of population is dependent on the nature of the problem. The second step is selection in which the
population from the initialized population is selected based on the defined fitness function and survival of the fittest. The
individuals selected are parents and are used in next generation to contribute in the generation of the population. Fitness
function is defined on the basis of particular problem and varies from problem to problem. It is used to select individ-
ual solutions and to measure the quality of the solution. There are two genetic algorithm operators: crossover operator
and mutation operator. Crossover operator combines two parents to generate new children which takes the characteristics
from both parents and are considered to be better than the both parents. Mutation operator forms new children by ap-
plying random changes to the individual parents. The generation of new population stops when a termination criterion is
reached.

4. Proposed algorithm

In legacy LEACH protocol, the cluster head selection is random without considering other parameters such as residual
energy or location of the node [6]. Every node produces random number between 0 and 1 and the node announces itself
as CH if this number is less than some threshold value, say T(n). T(n) is set to 0 for the node that has been already been
selected as CH. There is a problem in CH selection procedure as nodes with lower energy may be selected as CH which
causes node’s energy to be exhausted quickly and reduction in overall network lifetime. The nodes that have not been
selected as CH have T(n) possibility of being selected as CH as follows:
 T. Bhatia et al. / Computers and Electrical Engineering 56 (2016) 441–455 445

 p
  ∀n ∈ G
T (n ) = 1 − p r mod 1
p
0 if n ∈
/G

where P is desired percentage of CH that is the probability of CH node election, r is the current round, and G is the set of
nodes that haven’t been selected as CH in last 1/P rounds.
The overall performance of the network needs to be improved by incorporating the network parameters such as the
energy, distance between nodes and cluster heads and the distance between cluster heads for CH selection. The objectives
of our proposed algorithm are two-fold. The first objective is modification in the criteria of selection of CH to improve
network lifetime. Earlier, cluster heads were selected randomly using probability equation, but in this proposed scheme the
selection of cluster head is based on the Genetic Algorithm (GA). The overall performance of the network depends on the
energy of the node, the distance between the node and the cluster head and the distance between the cluster head and the
sink. These factors together contribute for the better efficiency of the network. In recent research trail, the sink was located
out of the network, which increased the distance between the sink and the cluster head, which resulted in early degradation
of the energy of the nodes, and in turn decreased network lifetime.
Most of the GA based clustering methods available in the literature assume that cluster nodes send their data to the
CH and CH to BS. In the proposed scheme, a relay node is introduced between the sink and the cluster head to achieve
second objective of distance aware routing. The cluster head first transfers the information to the relay node if the distance
between cluster head and sink is more and then the relay node further transfers it to the sink. The relay node helps to ease
the communication between the cluster head and sink and improves the energy efficiency. This introduced relay node is
rechargeable.
The proposed energy efficient protocol GADA-LEACH is based on GA and distance aware routing. In this, fitness function
is computed by using parameters such as energy of all nodes, energy of cluster heads, distance of CH with its associated
nodes, number of nodes in cluster, distance of base station from all CH’s and number of CH’s formed. The entire flow of
GADA-LEACH is shown in Fig 3. The steps of the proposed algorithm are described below:

1. The parameters on which the network performance depends are to be initialized in the first step.
2. Then the energy parameters which include initial energy of sensor nodes, energy to run transmitter and receiver, data
aggregation and amplification energy are initialized in the second step.
3. The initial population for the cluster head selection is generated. The cluster heads are selected from some number of
nodes present in the network.
4. In the next step, after the generation of initial population, the fitness of each node is evaluated for the better results so
that the best nodes can be selected as cluster head. Our fitness function includes energy parameters, distance of CH with
associated nodes and distance of BS from CH’s. Fitness function used is as follows:

F it ness f unct ion = [(0.3 ∗ F 1) + (0.35 ∗ F 2) + (0.35 ∗ F 3)]

Energy o f all nodes

F1 =
Energy o f Cluster heads

Eucledian Distance o f CH with its associated nodes

F2 =
Number o f nodes in cluster

Eucledian Distance o f BS f rom all CH  s

F3 =
Number o f CH  s f ormed

5. From the fitness evaluated, the best individuals from the population are chosen using Roulette Wheel selection.
6. The crossover and mutation operations are applied for selecting efficient CHs.
7. After applying GA operations, the fitness of the each individual is evaluated again and is compared with the initial one.
8. Since GA minimizes fitness function, if the evaluated fitness is less than the initial population, the current population is
updated by the corresponding new generation and next iteration takes place. The population is selected first, then it is
updated using genetic algorithm, then again the fitness of the new updated population is evaluated and compared with
the initial population’s fitness.
9. If stopping criteria meets, then the best nodes are chosen as the cluster heads from that population for that round.
10. After selecting the cluster heads, now communication is initiated between cluster head and nodes. Final communication
between cluster head and base station or sink is accomplished by incorporating relay node. This relay node is introduced
between sink and the cluster heads. All the cluster heads selected after GA calculate the Euclidean distance from the
Base Station (BS) and from the relay node. The aggregated data is sent to the either which has the shortest distance
from it.
11. After this the final calculation of parameters is done.
446 T. Bhatia et al. / Computers and Electrical Engineering 56 (2016) 441–455

Fig. 3. Flowchart of GADA-LEACH.

 T. Bhatia et al. / Computers and Electrical Engineering 56 (2016) 441–455 447

Fig. 4. Network structure of GADA-LEACH, comparison of network lifetime with varying initial energy (constant packet size 20 0 0), when initial energy
= 0.3 J/node.

5. Simulation environment and results

5.1. Simulation environment

The proposed algorithm has been designed in MATLAB. The 100 nodes are randomly distributed in an area of 100 × 100
m2 . Implementation consists of selection of cluster heads on the basis of GA optimization technique and then formation of
clusters is done. Cluster heads (x1, y1) calculate the Euclidean distance from the Base Station (BS) and distance from the
relay node and the aggregated data is sent to the either which has the shortest distance from it. A relay node is deployed
at the centre of the network (50, 50). Consider a CH located at coordinates (x1, y1) and a node (BS or relay node) deployed
at coordinates (x2, y2), then the Euclidean distance between CH and node is calculated as follows:
 
Distance between BS and Nodes = (x2 − x1 )2 + (y2 − y1 )2
So with the use of relay nodes, the energy consumption of the cluster heads is decreased and hence the network lifetime
increases. The proposed algorithm is efficient in both cluster head selection and the routing of data from cluster head to
the base station. GADA-LEACH is evaluated in terms of network lifetime. In the simulation, we have assumed that energy
dissipation for communication is based on the first order radio model [8] which comprises of energy to run transmitter
and receiver (Eelec ), data aggregation energy (Eda ) and amplification energy for shorter and longer distance (Efs and Emp )
respectively, and their constant values along with other simulation parameters are given in Table 1. This energy model
considers both channel models of the free space with d2 power loss and multipath fading with d4 power loss. If distance is
less than threshold distance do , free space model is used otherwise multipath model is used. The transmission energy and
receiver energy consumed in transmitting a n-bit message at a distance d using this radio model is as follows:
 
Eelec ∗ n + E f s ∗ n ∗ d2 , i f (d < do )
ET X (n, d ) =
Eelec ∗ n + Emp ∗ n ∗ d4 , i f (d ≥ do )

where, do = E f s /Emp

ERX (n, d ) = Eelec ∗ n

5.2. Simulation results

Since the results have been taken in simulated environment, it is assumed that position coordinates of the sensor nodes
are fixed and pre-programmed and the external environment is not affecting the radio signals adversely. The sensor nodes
remain stationery for the lifetime. In GADA-LEACH, the BS is situated outside the network and relay node is introduced
448 T. Bhatia et al. / Computers and Electrical Engineering 56 (2016) 441–455

Table 1
Simulation parameters.

Network parameters Values

Network size 100 × 100 m2

Number of nodes 100
Packet size 20 0 0, 30 0 0, 40 0 0 bits
Routing protocol LEACH, LEACH-E, LEACH-EX, M2NGA, GP-LEACH and LEACH-H
Initial energy of node (Eo ) 0.3, 0.4, 0.5 J/node
Energy to run transmitter and receiver (Eelec ) 50 nJ/bit
Data aggregation energy (Eda ) 50 nJ/bit
Amplification Energy to transmit at shorter distance (Efs ) 10 pJ/bit/m2
Amplification Energy for longer distance to transmit at longer distance (Emp ) 0.0013 pJ/bit/m2
Threshold distance (do ) 70 m
Primary population 100
Crossover rate 0.4
Mutation rate 0.006
Selection method Roulette wheel selection
Maximum number of iterations 30

Fig. 5. Plot of comparison of network lifetime of LEACH, LEACH-E, LEACH-EX and GADA-LEACH at Eo = 0.3 J/node.

Fig. 6. Plot of comparison of network lifetime of M2NGA, GP-LEACH and LEACH-H and GADA-LEACH at Eo = 0.3 J/node.
 T. Bhatia et al. / Computers and Electrical Engineering 56 (2016) 441–455 449

Fig. 7. Plot of comparison of network lifetime of LEACH, LEACH-E, LEACH-EX and GADA-LEACH at Eo = 0.4 J/node.

Fig. 8. Plot of comparison of network lifetime M2NGA, GP-LEACH and LEACH-H and GADA-LEACH at Eo = 0.4 J/node, when initial energy= 0.5 J/node.

in the centre of the network. The purpose of using relay node is to decrease the energy degradation of CH while sending
the aggregated data to BS. CH sends data to BS through relay node when its distance from relay node is lesser than its
distance from the BS. The energy of relay node is higher than cluster heads and lesser than BS and they are assumed to
be rechargeable. The less energy of CHs is consumed by the use of relay node so the results obtained are better. Network
structure of GADA-LEACH is shown below in Fig 4. The blue node at centre is relay node and the node represented with
label ‘x’ is the BS. Nodes numbered from 1 to 100 are the sensor nodes.
In Fig. 5 in case of LEACH, LEACH-E, LEACH-EX and GADA-LEACH the first node dies at different round. The first node
died at round 400, 437, 449 and 1170 in LEACH, LEACH-E, LEACH-EX and GADA-LEACH respectively. Lifetime refers to the
time when first node dies. So lifetime of network is highest in case of GADA-LEACH followed by LEACH-EX, LEACH-E and
LEACH.
450 T. Bhatia et al. / Computers and Electrical Engineering 56 (2016) 441–455

Fig. 9. Plot of comparison of network lifetime of LEACH, LEACH-E, LEACH-EX and GADA-LEACH at Eo = 0.5 J/node.

Fig. 10. Plot of comparison of network lifetime of M2NGA, GP-LEACH and LEACH-H and GADA-LEACH at Eo = 0.5 J/node.

In Fig. 6, we have compared GADA-LEACH with M2NGA, GP-LEACH and LEACH-H. The first node died at round 1087,
1095, 1121 and 1208 in M2NGA, LEACH-H, GP-LEACH and GADA-LEACH respectively. This shows that our GADA-LEACH is
more efficient in terms of network lifetime than M2NGA, GP-LEACH and LEACH-H. Without the concept of relay node in
GADA-LEACH, first node dies at round 1080. So, with the concept of relay node, network lifetime is increased by 11.85%.
When Initial Energy= 0.4 J/node
In Fig. 7, in case of LEACH, LEACH-E, LEACH-EX and GADA-LEACH the first node dies at different round. The first node
died at round 564, 589, 598 and 1441 in LEACH, LEACH-E, LEACH-EX and GADA-LEACH respectively. So lifetime of network
is highest in case of GADA-LEACH followed by LEACH-EX, LEACH-E and LEACH.
 T. Bhatia et al. / Computers and Electrical Engineering 56 (2016) 441–455 451

2000

1500
Number of rounds
LEACH
1000
LEACH-E
LEACH-EX
500
GADA LEACH

0
0.3 J 0.4 J 0.5 J
Initial Energy per node

Fig. 11. Network Lifetime of LEACH, LEACH-E, LEACH-EX and GADA-LEACH when varying Eo, Comparison of network lifetime with varying packet size
(constant initial energy 0.5 J/node), when packet size= 20 0 0.

Fig. 12. Plot of comparison of network lifetime of LEACH, LEACH-E, LEACH-EX and GADA-LEACH when packet size= 20 0 0 bits.

In Fig. 8, we have compared GADA-LEACH with M2NGA, GP-LEACH and LEACH-H. The first node died at round 1381,
1406, 1451 and 1582 in M2NGA, LEACH-H, GP-LEACH and GADA-LEACH respectively. This shows that our GADA-LEACH is
more efficient in terms of network lifetime than M2NGA, GP-LEACH and LEACH-H.
In Fig. 9, in case of LEACH, LEACH-E, LEACH-EX and GADA-LEACH the first node dies at different rounds. The first node
died at round 710, 727, 765 and 1778 in LEACH, LEACH-E, LEACH-EX and GADA-LEACH respectively. So lifetime of network
is highest in case of GADA-LEACH followed by LEACH-EX, LEACH-E and LEACH.
In Fig. 10, we have compared GADA-LEACH with M2NGA, GP-LEACH and LEACH-H. The first node died at round 1751,
1789, 1801 and 1974 in M2NGA, LEACH-H, GP-LEACH and GADA-LEACH respectively. This shows that our GADA-LEACH is
more efficient in terms of network lifetime than M2NGA, GP-LEACH and LEACH-H.
So from above observations it can be concluded in Fig. 11 that higher the value of initial energy is, higher will be the
network lifetime.
452 T. Bhatia et al. / Computers and Electrical Engineering 56 (2016) 441–455

Fig. 13. Plot of comparison of network lifetime LEACH, LEACH-E, LEACH-EX and GADA-LEACH when packet size= 30 0 0 bits.

Fig. 14. Plot of comparison of network lifetime of LEACH, LEACH-E, LEACH-EX and GADA-LEACH when packet size= 40 0 0 bits.

In Fig. 12, in case of LEACH, LEACH-E, LEACH-EX and GADA-LEACH the first node dies at different round. The first node
died at round 1270, 1341, 1409 and 2325 in LEACH, LEACH-E, LEACH-EX and GADA-LEACH respectively. So lifetime of net-
work is highest in case of GADA-LEACH followed by LEACH-EX, LEACH-E and LEACH. Without the concept of relay node in
GADA-LEACH, first node dies at round 1800. So, with the concept of relay node, network lifetime is improved by 29.16%.
When packet size= 30 0 0
In Fig. 13, in case of LEACH, LEACH-E, LEACH-EX and GADA-LEACH the first node dies at different round. The first node
died at round 915, 966, 992 and 1839 in LEACH, LEACH-E, LEACH-EX and GADA-LEACH respectively. So lifetime of network
is highest in case of GADA-LEACH followed by of LEACH-EX, LEACH-E and LEACH.
When packet size= 40 0 0
In Fig. 14, in case of LEACH, LEACH-E, LEACH-EX and GADA-LEACH the first node dies at different round. The first node
died at round 710, 727, 765 and 1720 in LEACH, LEACH-E, LEACH-EX and GADA-LEACH respectively. So lifetime of network
is highest in case of GADA-LEACH followed by LEACH-EX, LEACH-E and LEACH.
So from above observations it can be concluded in Fig. 15 that size of packet to be transmitted has effect on the network
lifetime in WSN which concludes that higher the size of the packet, higher is the consumption of the energy.
 T. Bhatia et al. / Computers and Electrical Engineering 56 (2016) 441–455 453

2500

2000
Number of rounds

1500
LEACH
LEACH-E
1000
LEACH-EX

500 GADA-LEACH

0
2000 3000 4000

Packet Size (bits)

Fig. 15. Network lifetime of LEACH, LEACH-E, LEACH-EX and GADA-LEACH when varying packet size.

6. Conclusion & future scope

In this paper, GADA-LEACH, Genetic Algorithm based Distance-Aware routing protocol is proposed in which cluster head
selection is optimized by using genetic algorithm. The proposed approach is efficient as it include more number of parame-
ters in fitness function for selecting better cluster heads and also introduction of intermediate node i.e. relay node reduces
the distance between the cluster head and sink and eases the further communication between them. The simulation results
show that GADA-LEACH performs better than other conventional protocols such as LEACH, LEACH-E, LEACH-EX, M2NGA,
LEACH-H and GP-LEACH and can prolong the lifetime of the network. Due to improved CH selection and introduction of
relay node in the network, network lifetime increases so more date is sent to BS.
In future, we would like to validate our results on real world test-beds where actual distance between the nodes is
measured through radio signals taking into account external parameters such as temperature, humidity, etc affecting radio
distance measurement. Further, the improvements can be made in the scheme by integrating sleep and awake concept in
which all the nodes do not remain active throughout. The proposed algorithm uses relay node for routing after CH selection
but in future, we would like to study the improvement in network lifetime by assuring direct communication of sensor
nodes with relay node bypassing communication with CH if the distance between sensor nodes and CH is more than the
distance between sensor nodes and relay nodes.

References

[1] Khedr AM, Osamy W, Agrawal DP. Perimeter discovery in wireless sensor networks. J Parallel Distrib Comput 2009;69:922–9.
[2] Akyildiz IF, Su W, Sankarasubramaniam Y, Cayirci E. A survey on sensor networks. IEEE Commun Mag 2002;40(8):102–14.
[3] Al-Karaki JN, Kamal AE. Routing techniques in wireless sensor networks: a survey. IEEE Wireless Commun 2004;11(6):6–28.
[4] Pottle GJ, Kaiser WJ. Embedding the internet: Wireless integrated networks sensors. Commun ACM 20 0 0;43(5):51–8.
[5] Haneef M, Deng Z. Design challenges and comparative analysis of cluster based routing protocols used in wireless sensor networks for improving
network life time. Adv Inf Sci Serv Sci 2012;4:450–9.
[6] Anastasi G, Conti M, Francesco M, Passarella A. Energy conservation in wireless sensor networks: a survey. Ad Hoc Netw 2003;7(3):537–68.
[7] Nikolaos AP, Stefanos AN, Dimitrios DV. Energy-efficient routing protocols in wireless sensor networks: a survey. IEEE Commun Surv Tut
2013;15(2):551–91.
[8] Guo S, Yang O. Energy-aware multicasting in wireless ad hoc networks: a survey and discussion. In: Computer communications, 30(9). Elsevier; 2007.
p. 2129–48.
[9] Ehsan S, Hamdaoui B. A survey on energy-efficient routing techniques with QoS assurances for wireless multimedia sensor networks. IEEE Commun
Surv Tut 2011;14(2):265–78.
[10] Ihsan A, Saghar K, Fatima. Analysis of LEACH protocol(s) using formal verification. In: Applied Sciences and Technology: IBCAST, 12th International
Bhurban Conference on. IEEE; 2015. p. 254–62.
[11] Heinzelman WR, Chandrakasan A, Balakrishnan H. LEACH: energy efficient communication protocol for wireless microsensor networks. In: proceedings
of Hawai International Conference on System sciences. Maui, Hawaii; 20 0 0. p. 3005–14. 4-7 Jan.
[12] Kumar D, Trilok CA, Patel RB. EEHC: Energy efficient heterogeneous clustered scheme for wireless sensor networks. Comput Commun 2009:662–7.
[12] Bari A, Wazed S, Jaekel A, Bandyopadhyay S. A genetic algorithm based approach for energy efficient routing in two-tiered sensor networks. Ad Hoc
Netw 2009;7(4):665–76.
[14] Gautam N, Jae-Young P. Distance aware intelligent clustering protocol for wireless sensor networks. J Commun Netw 2010;12(2):122–9.
454 T. Bhatia et al. / Computers and Electrical Engineering 56 (2016) 441–455

[15] Liu JL, Ravishankar CV. LEACH-GA: Genetic algorithm-based energy-efficient adaptive clustering protocol for wireless sensor networks. Int J Mach Learn
Comput 2011;1(1):79–85.
[16] Naregal K. Improved cluster routing protocol for wireless sensor network through simplification. In: 18th annual international conference advanced
computing and communications (ADCOM’12); 2012. p. 1–3.
[17] Karimi M, Naji HR, Golestani S. Optimizing cluster-head selection in wireless sensor networks using genetic algorithm and harmony search algorithm.
In: electrical engineering (ICEE), 2012 20th Iranian conference on. IEEE; 2012. p. 706–10.
[18] Gou H, Yoo Y, Zeng H. A partition-based LEACH algorithm for wireless sensor networks. In: computer and information technology, 2009. CIT’09. Ninth
IEEE international conference on, 2. IEEE; 2012. p. 40–5.
[19] Peiravi A, Mashhadi HR, Hamed Javadi S. An optimal energy-efficient clustering method in wireless sensor networks using multi-objective genetic
algorithm. Int J Commun Syst 2013;26(1):114–26.
[20] Salim A, Osamy W, Khedr AM. IBLEACH: intra-balanced LEACH protocol for wireless sensor networks. Wireless Netw 2014;20(6):1515–25.
[21] AbuBakr B, Lilien L. Extending wireless sensor network lifetime in the LEACH-SM protocol by spare selection. In: proceedings of the fifth interna-
tional conference on innovative mobile and internet services in ubiquitous computing (IMIS ‘11). Washington, DC, USA: IEEE Computer Society; 2011.
p. 277–82.
[22] Balamurugan A. An energy efficient fitness based routing protocol in wireless sensor networks. ICTACT J Commun Technol 2014;5(1).
[23] Lee SL, Park J, Shon JG. A two-layer cluster head selection based on distance in wireless sensor networks. In: Computer science and its applications.
Springer Berlin Heidelberg; 2015. p. 1003–7.
[24] Miao H, Xiao X, Qi B, Wang K. Improvement and application of LEACH protocol based on genetic algorithm for WSN. In: computer aided modelling
and design of communication links and networks (CAMAD), 2015 IEEE 20th international workshop on; 2015. p. 242–5.
[25] Whitley D. A genetic algorithm tutorial. Stat Comput 1994;4(2):65–85.
 T. Bhatia et al. / Computers and Electrical Engineering 56 (2016) 441–455 455

Tarunpreet Bhatia is pursuing PhD in the department of Computer Science and Engineering at Thapar University, Patiala. She received her M.E. degree
from Thapar University in 2013 and B. Tech from Kurukshetra University in 2011, majoring in Computer Science and Engineering. She has published over
15 papers in journals and conferences. Her research interests include wireless networks and network security.

Simmi Kansal received the B.E. degree in Computer Science Engineering from Regional Institute of Management Technology, Mandi Gobindgarh, Punjab
in 2013 and the M.E. degree in Information Security from Thapar University, Patiala. Her research interests include wireless sensor networks and network
security.

Dr Shivani Goel has done her graduation in Computer Science and Engineering (B.E., 1998), post graduation in Software Engineering (M.E., 2006) and PhD
(2012) from Thapar University, Patiala. She has more than 60 publications in conferences and journals. She is a member of IEEE and ACM. Her research
areas are machine learning, software reuse and algorithms.

Dr. A. K. Verma is currently an Associate Professor in the department of Computer Science and Engineering at Thapar University, Patiala. He received his
B.S., M.S. and Doctorate in 1991, 2001 and 2008, respectively, majoring in Computer science and engineering. He has published over 120 papers in referred
journals and conferences. His research interests include wireless networks and cloud computing.