An Improved Ant-based Opportunistic Routing (AOR) for Non-Stagnant Packet

Transmission in Wireless Sensor Network

Abstract

Wireless Sensor Network (WSN) is a collection of wireless nodes that facilitates environment
monitoring through vigorous observation and sensing. The nodes transmit sensed information to
a sink node that serves as a collection hub of all the transmitted information. Due to the region of
deployment and wireless communication medium, the nodes suffer from non-reliable
transmissions and communication failure. Optimization techniques are proposed to improve the
performance of the nodes despite the prevailing drawbacks. We address the problem of
stagnation in optimization algorithms through Ant based Opportunistic Routing (AOR). This
AOR incorporates the advantages of clustering and Ant Colony Optimization (ACO). The
clustering process requires a light-weight header selection phase for conserving energy in a
traditional manner. The external ant based routing relies on Link Estimation Time (LET) and
Residual Energy of the nodes for identifying optimal routing paths. The process of ants are
altered based on pheromone concentration implicitly and LET explicitly to retain seamless
transmission, alleviating paused communications in the network. The performance of the
proposed AOR is assessed using simulation for the metrics: Throughput, Packet Delivery Ratio,
Delay, Energy Consumption and Alive Nodes.
Keywords: ~ Ant Colony Optimization, Clustering, Energy Efficient Routing, Link Quality
Estimation, Opportunistic Node Selection,

Introduction

Wireless Sensor Network (WSN) is an assortment of wide-range of miniature electro-

mechanical devices with exceptional operational capabilities. These tiny devices called nodes are
self-powered with the built-in battery for their sensing, computation and communication
processes. Application areas of sensor network include habitat monitoring, disaster management
and information systems, weather forecasting, residential ambience appliances, etc. The nodes
are distributed in an indiscriminate manner across a surveillance region. The primary operations
of the nodes include sensing external phenomenon and reporting it through wireless
transmissions [1, 4]. With the advancement in sensor technology, the nodes are equipped with
diminutive processing units to operate on the sensed information. These types of devices
communicate with their direct neighbors to share information through a common shared wireless
medium or link. The information accumulated by the sensor nodes are disbursed for end-user
application support through a sink node. The sink node acts a collection hub, receiving sensor
information at periodic intervals. The sink node will be stationed at a common point so as to
receive multiple node information; a typical sensor node relies on its neighbors for transmitting
the collected information to the common sink [2, 3].
Sensor device operations are supported by the power unit equipped with it. Sensor nodes
are not often deployed in human-friendly regions; rather, it is dispersed in hostile environment
like forests, wild-life monitoring where human access is inflexible. Sensor nodes drain energy
frequently over time, for their power unit has to be replaced or revived; failing which leads to
node malfunction and communication interruption. End-user application demands cannot be met
with such deprived sensor communication based on energy; for which energy depletion of the
node must be optimized. Information relaying in sensor network occurs in a multi-hop fashion to
communicate with the sink node located away from the range of the communicating node.
Therefore, the sensor node must adopt suitable routing protocol to ensure if the information is
communicated in a reliable manner. A routing protocol must take on the responsibilities of route
discovery, energy conservation and data fidelity. The routing protocol must cooperate with the
dual behavior of the node acting as both a router and a host. Researches over the decades
concentrate over improving energy efficiency of the nodes to prolong network’s active span.
Optimization in WSN focuses over the routing and energy conservation designs and application
challenges over the recent years [3-5].
From the evolution of optimization in WSNs, clustering and data gathering are two
prominent solutions put forward by the researchers. In a distributed network, a single sensor
node like gateway, gathers information from the nodes at its range, binds the collected
information and transmits the same to the sink node. The energy expenditure of the gathering
node is obviously high as it is influenced by frequent operations. Aggregation minimizes the
communication cost and computation complexity by creating a dependency on the aggregators.
The goal of using aggregation in sensor network is to improve network lifetime by curtailing
unnecessary energy expenses. In general, an ideal aggregation technique balances energy
utilization of the nodes at the time of data gathering and transmission [2, 3]. Clustering is another
traditional energy conserving method in WSN, an advancement of data gathering non distributed
sensor nodes. The nodes are united by a cluster head (CH) that harmonizes the communications
of its one-hop neighbors. The CH forms alliance with its one-hop neighbors called cluster
members (CM) to accumulate, aggregate and transmit information to the sink. A cluster head acts
as a central controller that prevents the periodic operations of its members so as to save their
redundant energy drain. Preserving the energy of the nodes helps to prolong their active period.
In a clustering process, the operations of a cluster are high, for which the CH depletes energy
faster than its members. Therefore, CH must be reinstating at periodic intervals based on energy,
link stability, reliability, etc. A cluster head solely takes responsibilities of inter and intra cluster
communication, routing, data gathering, transmission and energy management. Hence CH
selection is a crucial task for ensuring flawless WSN communication to serve end-user
application demands [6, 7].

Related Works
LEACH (Low Energy Adaptive Clustering Hierarchy) [8] protocol designed for WSN improves
lifetime of the networks by integrating the benefits of clustering and data gathering. LEACH
works in two phases: set-up and steady-state. In a set-up phase, the CH the elected arbitrarily
with the assurance of each cluster member will serve as CH in the fore coming rounds. Data
gathering from the cluster members and transmission to the sink node is commenced in the
steady-state phase. A prominent setback of LEACH is that it does not support large-scale
networks.
The authors in [9] proposed a neural network based clustering approach for mitigating the issues
of the centralized approach. With the implication of neural networks, clustering process self-
organizes to levitate the quality of service of the networks. The limitation with this self-
organizing approach is that it is favorable for cluster with smaller communication range.
Shen and Wang [10] proposed a neural network based technique for alleviating issues that arise
due to network dynamics and topology alterations. Fuzzy Hopfield Neural Network (FHNN)
technique proposed distributes the communication time slots among the nodes to reduce
collisions. The TDMA based transmission cycle length is varied through acknowledgements
received post distribution.
General Self-Organized Tree-based Energy-balance (GSTEB) [11] designed for WSN minimizes
energy consumption of the nodes and prolongs network lifetime. GSTEB performs routing by
selecting root nodes with the recommendation from BS to select energy potent paths to the
destination. The BS shares the distance and energy of the nodes to identify shortest and energy
effective paths.
Xia and Jia [12] proposed a hybrid clustering technique to improve the energy efficiency of the
network through organized communications. Initially, the network is partitioned into cluster
wherein the CH is responsible for compressing sensed information to the sink node. The CH
routes to the sink node with the awareness of their location through broadcast.
Energy Efficient Routing Protocol (EERP) proposed by Ghaffari [13] is built using A-Star
algorithm for selecting protracted communication paths in the network. The algorithm works in a
best-first refining search method to construct energy-sufficient routing paths. The refining
process depends on the remaining energy of the nodes; this accomplishes the task of conserving
energy of majority of the nodes.
Amiri et al. [14] proposed Fuzzy Ant Colony Optimization Routing (FACOR) for improving
network performance in terms of energy conservation and delay. In FACOR, ant agents are used
to discover optimal path; the pheromone concentration of the nodes are updated at periodic
intervals. The fuzzy part works on the pheromone concentration of the nodes to identify a better
path to achieve its goal.
A reactive routing protocol developed with the attributes of conventional AOMDV protocol,
EEEMRP (Enhanced Energy Efficient Multipath Routing Protocol) is discussed in [15].
EEEMRP is of two different working phases: multipath discovery and precise path selection.
AOMDV is used to identify multiple paths to the destination and cuckoo search filters the routes
based on available energy. It recommends paths with higher energy for packet transmission. This
protocol improves network throughput requiring lesser energy of the nodes.
Leu et al. [16] proposed Regional Energy Aware Clustering with Isolated Nodes (REAC-IN) for
mitigating the isolated node routing problem in WSN. This clustering method discovers routes
using neighbor CH and isolated nodes in the network. The isolated nodes are driven into the
communication process based on distance factor. The CH is selected by evaluating the remaining
energy of the nodes and energy of the nodes within a specific region. The extensive simulations
of REAC-IN show that it improves path stability and network life span.
Energy Efficient Reliable Routing [17] is designed to enhance the time-to-live period of the
sensor network alongside communication reliability. This is a hierarchical clustering protocol
that resolves earlier link failures to stabilize network communications. In this clustering process,
two deputy clusters are elected to pursue multi-hop relaying at the time of CH replacement. Due
to the distributed decisions, the BS directs the CHs for routing and node replacements.
Broadcasting Combined with negative acknowledgement proposed Mianxiong Dong et al. [18] is
intended to improve the performance of data gathering process in WSN. This cluster based
broadcasting protocol is designed to optimize energy utilization and delay during data gathering
process. With excess acknowledgement scheme, the node minimizes the number of
retransmissions. With a negative acknowledgement, the CH identifies the non-active
transmission nodes after a TDMA slot.
The authors in [19] put forward an analytical framework for determining the movement pattern
of the mobile collectors to improve data gathering rates in WSN. The analytical model is
constructed using Markovian random walk mobility for deciding the movement pattern of the
sink collectors. This model is used to assess network loss.
To address the Geometric Disc Covering problem, the authors in [20] proposed a Local Search
Approximation Algorithm (LSAA). LSAA segregates the nodes into groups initially and then
identifies local neighbor for transmission. The local neighbor is identified in a greedy manner to
minimize time complexity.
For improving data collection reliability, the authors in [21] proposed ReDAST that facilitates
dual mode of communications in a node. This scheme addresses the issues of node isolation and
transmission redundancy. The nodes operate in RF and acoustic modes of communication
depending on their region of coverage and presence of transfaulty nodes. ReDAST is focused to
conserve energy and minimize delay in WSN.
Yuan Zhou et al. [22] proposed a particle swarm optimization (PSO) based routing protocol for
improving network lifetime. This routing is built on clustering process, wherein the cluster
interacts with relay nodes for cutting down energy expenses in a distributed manner. The
protocol accounts the communication distance and energy potency of the nodes for achieving an
optimized solution.
To alleviate the CH overloading at the time of cluster formation, the concept of fuzzy logic is
applied to WSN to facilitate decision making [23]. Fuzzy decisions concentrate in distributing
load over the network in an adoptive and flexible manner. The proposed type 2 fuzzy logic
(T2FL) decision making improves network lifetime, throughput and prolongs the dead time of
the nodes. The fuzzy decision making process is imposed over the energy of the nodes for
making them participate in routing and transmission.
The authors in [24] propose a contact-aware expected transmission (CA-ETX) routing for
improving throughput in WSN with mobile sinks. The proposed routing metric mitigates the
problems of network dynamicity and unpredictable node mobility. The authors defend that the
existing ETX models with a slight alteration can be adopted for mobile senor networks for
achieving delay tolerant network performance. Opportunistic Backpressure Collection is
employed for supporting data collection from fast moving mobile sinks through multipath
scheduling.
From the survey, we concern about the optimization problem in WSN for which we rely on bio-
inspired optimization based on ant behavior. We consider routing optimization, CH selection and
energy conservation issues that prevails in the existing approaches with a modification to the ant
behavior.
Problem Definition
Performance enhancement in WSN requires multi-metric evaluation and stagnant
solutions using a conservative bio-inspired optimization technique. In an ant colony optimization
(ACO), the paths are selected in a recursive manner to relay packets to the sink node. This causes
earlier energy drain of a node due to which ACO results in mediate solutions. The local
pheromone update process though serves as a solution for the above results in frequent node
replacements as the neighbors are identified based on non-uniform evaluation criterion [25, 26].
Considering these factors as a drawback in conventional ACO, we propose an Ant based
Opportunistic Routing (AOR) algorithm with the advantages of ACO. In this routing algorithm,
the process of routing is carried out using opportunistic ants that behave in a self-replenish
manner to assess nodes with similar criterion preventing local search options.
Network Model
A cluster based network as illustrated in Figure 1 is considered for our proposed algorithm.
 Figure 1 Network Model
Our network consists of set of nodes connected through a set of

vertices V in a region of . The nodes are grouped as clusters (C) and are controlled by a

cluster head (CH). The nodes that obey their CH are its cluster members (CM). The cluster
members are connected in single-hop to its CH. The sensor nodes relay gathered information to
their CH that in-turn transmits it to the sink node located multi-hops away. The nodes that do not
belong to any clusters are isolated (I) but are not detached from the network. Each of the sensor
node is separated by a distance d with a range of R units.
Energy Model
 The sensor nodes are given an initial energy to perform their communication.

Eventually, the nodes spend energy for transmission and reception . Therefore, the

energy consumed by a node is estimated as in equation (1)

(1)
With time, the node drains its energy for which the node’s residual energy is given by equation
(2)
(2)
Although, the CH is a sensor node, the amount of energy spend by a CH is higher compared to a
normal sensor node. Therefore, the energy model of a CH is computed as in equation (3)

(3)

Where, k is the data collection level, and is the data collection function and rate of data

transmission.

Figure 2 AOR Process Illustration

Methodology
The proposed Ant based Opportunistic Routing is a two-fold process: Light-weight
Clustering and Opportunistic ant routing. The number of iterations in opportunistic ant routing is
high for which maximum energy is required. To prevent excess energy drain in routing, we
administer clustering concept to restore unnecessary energy expenses. In a more conventional
manner, the clustering in WSN restricts unnecessary energy utilization of the nodes. Considering
this advantage of the clustering, we prefer a light-weight clustering approach where a CH is

elected based on two explicit factors: Residual energy and delivery factor . Figure 2 portrays

the process of AOR.

Light-Weight Clustering
In a light weight clustering, the CH is selected by evaluating an explicit communication

function . The explicit function is defined to select a CH that is profitable inside and

outside the cluster. For a seamless operation within the cluster, the head node must possess
higher energy and for an optimal routing, the path outside the cluster must be consistent.
Therefore, we evaluate a cluster head candidate node for its remaining energy internally and
delivery factor externally. The delivery factor is computed for an isolated node present outside
the cluster. The CH identifies the capacity of the neighbor in delivering maximum packets. The
delivery factor is a concave metric that is applicable for the non-cluster members. Delivery factor
of a non-cluster member is computed using equation (4)
(4)

where, h is the hop count and is the packet loss ratio of the isolated intermediate node.
As the isolated nodes are located at one-hop distance from the CH but outside the cluster, h=1
and therefore, (4) is given as in equation (5)
(5)
Cluster head replacement takes place when a CH fails to satisfy energy equality function post

each transmission. We define an energy equality function for changing the CH with

respect to the residual energy of the active CH. The average energy of all the nodes for k

transmissions is computed as in equation (6)

(6)

The equality function is valid until the remaining energy of the active CH is equal to the average
energy of the nodes. The constraint for a node to continue as CH is represented in (7)
(7)
If equation (7) fails, CH election takes place by identifying the next optimal CH. The CH is

elected using for which the delivery factor and residual energy of the nodes are required.

The communication function is an union between energy and packet delivery where in both the
constraints have to be satisfied by the candidate nodes to become the next CH. The function for
selecting a new cluster head is given by equation (8)

(8)

The candidate node exhibiting maximum is preferred as the next CH. The Algorithm1

describes CH selection with equality constraint is given as follows.

Algorithm 1. CH Selection Process

Input: n, X*Y
 Output: new CH

Opportunistic Ant Routing

The CH initiates a set of ants to traverse through the nodes to reach the sink. The nodes
visit each of the intermediate nodes to reach the sink. The ants that trail to the sink are called
forward ants. The number of ants depends on the number of intermediates available for the
cluster. The ants on reaching the sink, retraces its path to the CH through the previously visited
nodes. The ants update the pheromone of the nodes so as to inform the CH. We define the Link
expiration Time (LET) and residual energy of a node as the ant pheromone that is updated by the
nodes at the end of each traversal. The process of opportunistic routing based on ants is shown in
Figure 4.

Figure 4 Opportunistic Ant Routing Process

The probability that an ant m, visits an intermediate i from the CH is computed using (9)

(9)
 is the amount of pheromone concentrated over the link connect CH and intermediate node I,

is the energy LET heuristic.

The concentrated pheromone disintegrated with time, for which the concentration at a time t is
estimated using equation (10)
(10)
Where,
(1-ω) is a pheromone disintegrating constant, is the amount of pheromone observed by the

nth ant over the link between CH and i.

Critical Analysis 1: Pheromone concentration need not be updated .
The CH identifies the path outside the cluster for which the pheromone concentration
over the link that connects the intermediate node needs to be computed. Contrarily, the nodes
that are in range of CH include both cluster members and isolated nodes. The vertices between
CH and CM are always connected and the pheromone of these vertices need not be updated.
Therefore, equation (10) is modified with the range constraint as in equation (11)
(11)
The number of pheromone concentration update is more equivalent to the isolated nodes count.

If equation (11) has to be satisfied, then .

The pheromone concentration is estimated using LET and residual energy of the visiting nodes.
The routing protocol discovers shortest path to the sink node that expires with the ceasing node
energy requiring further route discovery process. To minimize the impact of stagnation that is
caused due to communication pause post path failures, LET of a link needs to be known
forehand. The ants update the LET of the visiting nodes to inform the CH that it decides upon
further transmissions or reinforcing a new path. Therefore, the link expiration time is estimated
as in equation (12)

(12)

Where,
 , , and .

and are the position coordinates of the CH and the isolated node i. and

are the velocities of the CH and the isolated nodes respectively. and are the angle of

deviation corresponding to the CH and the isolated node. For updating the pheromone over the
link, the ants require a maximization function, the maximization function is estimated for LET

and . The maximization function is logically estimated as in equation (13)

(13)
There is no need for adjusting the pheromone concentration with weights as both the metrics
needs to be satisfied by the isolated nodes for establishing a reliable route to the sink node. This
pheromone concentration decreases with time with respect to the residual energy of the nodes.
LET is proportional to the node energy therefore; priority is given for residual energy of the
nodes. The ants update the residual energy initially followed by the LET update.
The CH selects the best path for relaying information to the sink node; in other words, the node
with higher pheromone (residual energy and LET) is selected by the CH. The other ants
traversing through the isolated nodes are drawn towards the CH selected ant to pursue
transmission. The pheromone of the nodes is updated over periodic intervals; the interval is the
communication time.
Critical Analysis 2: Change in routing path by the CH is dependent on link delay . Round

Trip Time is the time taken for a packet to reach the sink by traversing through the path nodes
and returning back to the CH. The return packet is always an acknowledgement the CH
communicating the successful delivery of the packet. The relationship between link delay and

round trip time is given as in equation (14)

(14)

The LET updated by the ant must be less than the for transmitting the packet successfully. In

this context, if , then the CH ahs to re-instigate path discovery through the next set of
ants. Figure 3 illustrates the normalized pheromone update and varying pheromone update with
respect to time.

Figure 4 Varying Pheromone Updates

Therefore, the pheromone disintegration constant becomes

. This update interval relies on the time factor

computed using equation (14)

(14)

Where, , and are the MAC delay, queuing and transmission delay respectively.
In this case, even through the pheromone has not completely disintegrated; the routing
paths have to be replaced. The CH need not re-instigate a new ant population rather; it can select
the next best ant for pursuing transmission. A new ant population is required when the
pheromone of either of the ants is disintegrated. This type of ant routing is dependent on both
explicit and implicit factors other than pheromone concentration. Here, routing is targeted in a
precise manner by minimizing complete energy drain of a node and also routing decisions are
made from an external manner.
Performance Evaluation
The performance of the proposed AOR is evaluated through extensive simulations using
network simulator. The evaluation metrics considered are Throughput, Packet Delivery Ratio,
Delay, Energy Consumption and Number of active nodes. The proposed AOR is compared with
the existing approaches T2FL [23] and CA-ETX [24] for the above metrics. Table 1 describes the
simulation parameters and their values.
Table 1 Simulation Parameters
 Simulation Parameters Value
ParamParameter
Network Region 1000m*1000m
Number of sensor nodes 100
Transmission range (R) 250 m
Initial energy 2J
Energy for election Packet size 50x10-9 J/bit
Mobility Model Random waypoint

MAC 802.11
Traffic Source CBR
Packet Size 512 Bytes
Simulation Time 300 seconds

Throughput Comparison

Figure 5 Throughput Comparisons

Figure 5 illustrates the performance comparison of our proposed AOR with the existing T2FL
and CA-ETX methods. In the proposed routing algorithm, the path identified by the ants is
ensured to survive the transmission besides stagnation problem that is common in conventional
ACO. The stagnation is mitigated by changing links without communication interruption based
on pheromone concentration and LET. These two factors ensure consistent link availability; that
improves continuous and dense packet flow increasing network throughput. Moreover, neighbor
selection and pheromone relies on the residual energy of the nodes, which prevents unnecessary
packet drop due to node energy drain. These factors are favorable in improving packet
transmission in the network.
Packet Delivery Ratio Comparison

Figure 6 Packet Delivery Ratio Compsrisons

There are two important factors in the proposed AOR to improve packet delivery ratio:
intermediate node selection and time based transmission reinforcement. The immediate
intermediate neighbor from the CH is selected based on the delivery factor that initially confirms
selection of reliable node. This criterion is extended in selecting paths nodes to the sink, ensuring
higher packet delivery. The time based transmission support is provided by changing feeble links
with the prior information, restricting packet loss. These factors levitate packet transmission that
in turn improves packet delivery ratio of AOR (refer Figure 6).
Delay Comparison
 Figure 7 Comparisons of Delay
AOR produces lesser delay (Figure 7) as it prevents frequent path reinforcements and
retransmissions due to packet loss. The RTT and LET metrics ensure uninterrupted transmission
through series of existing ant population wherein the necessity for initializing new ant population
is nil. This minimizes delay to a considerable extent preventing recursive routing and volatile

links from participating in the routing process. In the critical analysis of

, the time requirement for updating m ants is

diminished to a single ant, requiring lesser time than the existing approaches.
Energy Consumption

Figure 8 Energy Consumption Comparisons

The comparison of energy utilized in AOR is illustrated in Figure 8. Unlike in the other methods,
the nodes are not completely overloaded for transmitting the CH information. Also the nodes are
periodically checked for their residual energy and the maximization function decides the
participation of the nodes in routing and transmission. The higher drain energy node is obviously
the CH, preventing the other nodes from draining all their initial energy. Besides, the stagnant
problem is addressed by ensuring availability of minimum nodes in the network as the ant
population is not most often initialized. These vital factors facilitate nodes to conserve energy
despite unnecessary energy drain.
Alive Nodes

Figure 9 Alive Nodes Count Comparisons

The number of alive nodes in the network post the termination of ants is illustrated in Figure 9.
The proposed AOR conserves energy of the individual nodes through clustering and residual
energy check. The number of nodes participating in active communication is less, minimizing
unnecessary energy drain. Moreover, the clustering approach harmonizes all the sensor nodes
such that the nodes requirement is serviced by the CH. Based on energy the nodes are identified
internally (within the cluster) and based on link delay, the external nodes are classified. The link

delay is estimated for nodes satisfying in equation (13). This distributes the chances of

identifying unattended nodes in the network minimizing single node overloading. Therefore,
optimal number of operations is carried out by a node such that it retains an appreciable amount
of energy, retaining higher number of active nodes.
Table 2 illustrates the Simulation values the proposed AOR with the existing T2FL and CA-ETX.
Table 2 Simulation Values of AOR, T2FL and CA-ETX

Metrics T2FL CA-ETX AOR

Throughput (Kbps) 66.89 69.37 83.52
Packet Delivery Ratio (%) 55.24 62.4 74.3
Delay (s) 0.538 0.453 0.412
Energy Consumption (mJ) 25.15 30.06 24.79
Alive nodes Count 39 48 62

Conclusion
In this manuscript, we propose an Ant based Opportunistic Routing (AOR) that
incorporates the functionalities of ant colony optimization in a selective manner. The stagnation
in ACO is rectified through a maximization function satisfied by independent metrics that
identifies active nodes in a seamless manner. This is facilitated by setting up common evaluation
metric for both opportunistic routing and clustering process. Both the internal and external
neighbor selection criteria are satisfied through residual energy metric at iterative intervals. With
the consideration of residual energy metric, the consistency of the node links are verified over
each ant update. The diverse operations are unanimously derived to identify an optimal path. The
proposed approach improves network performance in terms of Throughput, Packet Delivery
Ratio, Delay, Energy Consumption and Alive nodes count.

References
[1] D. Gong and Y. Yang, "Low-Latency SINR-Based Data Gathering in Wireless Sensor
Networks," in IEEE Transactions on Wireless Communications, vol. 13, no. 6, pp. 3207-3221,
June 2014.
[2] K.-W. Fan, S. Liu, and P. Sinha, “Structure-Free Data Aggregation in Sensor Networks,”
IEEE Transactions on Mobile Computing, vol. 6, no. 8, pp. 929–942, 2007.
[3] S. Xiao, B. Li, and X. Yuan, “Maximizing precision for energy-efficient data aggregation in
wireless sensor networks with lossy links,” Ad Hoc Networks, vol. 26, pp. 103–113, 2015.
[4] X. Liu, “Atypical hierarchical routing protocols for wireless sensor networks: A review,”
IEEE Sensors Journal, vol. 15, no. 10, pp. 5372–5383, October 2015.
[5] S. Ehsan and B. Hamdaoui, “A survey on energy-efficient routing techniques with QoS
assurances for wireless multimedia sensor networks,” IEEE Communications Surveys &
Tutorials, vol. 14, no. 2, pp. 265–278, Second Quarter 2012.
[6] O. Younis and S. Fahmy, “HEED: a hybrid, energy-efficient, distributed clustering approach
for ad hoc sensor networks,” IEEE Transactions on Mobile Computing, vol. 3, no. 4, pp. 366–
379, December 2004.
[7] M. M. Afsar, and M.-H. Tayarani-N, “Clustering in sensor networks: A literature survey,” J.
Netw. Comput. Appl., vol. 46, pp. 198–226, Nov. 2014.
[8] Hu Gang, Xie Dongmei, Wu Yuanzhong, “Research and improvement of LEACH for
wireless sensor networks,” Chinese Journal of Sensors and Actuators, vol. 6, no. 20, pp. 1391-
1396, 2007.
[9] H. He, Z. Zhu, and E. Makinen, “A Neural Network Model to Minimize the Connected
Dominating Set for Self-Configuration of Wireless Sensor Networks,” IEEE Transactions on
Neural Networks, vol. 20, no. 6, pp. 973–982, 2009.
[10] Y. Shen and M. Wang, “Broadcast scheduling in wireless sensor networks using fuzzy
Hopfield neural network,” Expert Systems with Applications, vol. 34, no. 2, pp. 900–907, 2008.
[11] Z. Han, J. Wu, J. Zhang, L. Liu, and K. Tian, “A General Self-Organized Tree-Based
Energy-Balance Routing Protocol for Wireless Sensor Network,” IEEE Transactions on Nuclear
Science, vol. 61, no. 2, pp. 732–740, 2014.
[12] R. Xie and X. Jia, “Transmission-Efficient Clustering Method for Wireless Sensor Networks
Using Compressive Sensing,” IEEE Transactions on Parallel and Distributed Systems, vol. 25,
no. 3, pp. 806–815, 2014.
[13] Ghaffari, A., “An Energy Efficient Routing Protocol for Wireless Sensor Networks using A-
Star Algorithm,” Journal of applied research and technology, Vol. 12, No.4, pp. 815-822, 2014.
[14] E. Amiri, H. Keshavarz, M. Alizadeh, M. Zamani, and T. Khodadadi, “Energy Efficient
Routing in Wireless Sensor Networks Based on Fuzzy Ant Colony Optimization,” International
Journal of Distributed Sensor Networks, vol. 10, no. 7, p. 768936, 2014.
[15] D. A. A. Raj and P. Sumathi, “Enhanced Energy Efficient Multipath Routing Protocol for
Wireless Sensor Communication Networks Using Cuckoo Search Algorithm,” Wireless Sensor
Network, vol. 06, no. 04, pp. 49–55, 2014.
[16] J.-S. Leu, T.-H. Chiang, M.-C. Yu, and K.-W. Su, “Energy Efficient Clustering Scheme for
Prolonging the Lifetime of Wireless Sensor Network With Isolated Nodes,” IEEE
Communications Letters, vol. 19, no. 2, pp. 259–262, 2015.
[17] H. K. D. Sarma, R. Mall, and A. Kar, “E2R2: Energy-Efficient and Reliable Routing for
Mobile Wireless Sensor Networks,” IEEE Systems Journal, vol. 10, no. 2, pp. 604–616, 2016.
[18] M. Dong, K. Ota, A. Liu, and M. Guo, “Joint Optimization of Lifetime and Transport Delay
under Reliability Constraint Wireless Sensor Networks,” IEEE Transactions on Parallel and
Distributed Systems, vol. 27, no. 1, pp. 225–236, Jan. 2016.
[19] C.-H. Lee, J. Kwak, and D. Y. Eun, “Towards Distributed Optimal Movement Strategy for
Data Gathering in Wireless Sensor Networks,” IEEE Transactions on Parallel and Distributed
Systems, vol. 27, no. 2, pp. 574–584, Jan. 2016.
[20] C. Ma, W. Liang, M. Zheng, and H. Sharif, “A Connectivity-Aware Approximation
Algorithm for Relay Node Placement in Wireless Sensor Networks,” IEEE Sensors Journal, vol.
16, no. 2, pp. 515–528, 2016.
[21] P. Kar and S. Misra, “Reliable and Efficient Data Acquisition in Wireless Sensor Networks
in the Presence of Transfaulty Nodes,” IEEE Transactions on Network and Service Management,
vol. 13, no. 1, pp. 99–112, 2016.
[22] Y. Zhou, N. Wang, and W. Xiang, “Clustering Hierarchy Protocol in Wireless Sensor
Networks Using an Improved PSO Algorithm,” IEEE Access, vol. 5, pp. 2241–2253, 2017.
[23] P. Nayak and B. Vathasavai, “Energy Efficient Clustering Algorithm for Multi-Hop Wireless
Sensor Network Using Type-2 Fuzzy Logic,” IEEE Sensors Journal, vol. 17, no. 14, pp. 4492–
4499, 2017.
[24] S. Yang, U. Adeel, Y. Tahir, and J. A. Mccann, “Practical Opportunistic Data Collection in
Wireless Sensor Networks with Mobile Sinks,” IEEE Transactions on Mobile Computing, vol.
16, no. 5, pp. 1420–1433, Jan. 2017.
[25] Parvatkar, S., and Gore, D, “An ACO Approach to Maximize Lifetime of Heterogeneous
WSN,” International Journal of Computer Science and Information Technology, vol. 5, pp.
5773-5775, 2014.
[26] H. J. A. Nasir, K. R. Ku-Mahamud, and E. Kamioka, “Ant Colony Optimization Approaches
in Wireless Sensor Network: Performance Evaluation,” Journal of Computer Science, vol. 13,
no. 6, pp. 153–164, Jan. 2017.