Comparison of Routing Algorithm Based on Ant Colony

Optimization in Wireless Sensor Networks

Namitha Sandeep Manya Singh

Student, Dept. of CSE Student, Dept. of CSE
CMR Institution of Technology CMR Institution of Technology
Bengaluru, India Bengaluru, India
[email protected] [email protected]

Abstract: A wireless sensor network refers to a network of dispersed nodes that can monitor and record
the physical conditions of the environment around them and it forwards collected data to a central
location. Sensor nodes are attached in the WSN with the onboard processor. These sensors monitor and
manage the environment in a particular area. These nodes are distributed in an ad hoc manner. WSNs are
primarily used to monitor an event remotely such as in the case of agriculture where the environment
conditions are continuously monitored. The ant colony optimization algorithm (ACO) is a probabilistic
technique for solving problems using the knowledge gained from the methods used by ants to locate the
shortest route from source to destination. This algorithm is categorized under ”swarm intelligence”
methods. The aim of this paper is to explore a more efficient algorithm for routing packets in Wireless
Sensor Networks inspired by the movements of ant colonies.

Keywords: Ant Colony Optimization Algorithm,Routing Algorithm, Wireless Sensor Networks

 INTRODUCTION

Wireless sensor networks are spatially dispersed and dedicated servers that monitor the surrounding
environment and send the data they collect to a central hub. Seeking to improve wireless communication
can help contribute to a wide range of different applications such as health care monitoring, disaster
prevention, monitoring the quality of water and other underwater applications [1]. WSN’s comprise of
the most important part when it comes to smart city environments built with IoT environments by using
intelligent devices for sending and sharing data over a region [2].

The Particle Swarm Optimization (PSO) algorithm draws inspiration from the way birds move in flocks.
In this method, particles, which represent possible solutions, travel through a search space, adjusting
their positions based on their own best-found solutions (personal best) and the best solution discovered
by the entire group (global best). This allows the particles to work together, efficiently exploring the
problem space and gradually converging toward optimal or nearly optimal solutions over time​. [3] The
Low Energy Adaptive Clustering Hierarchy (LEACH) algorithm is one of the most well-known and
influential distributed clustering techniques in sensor networks. In LEACH, cluster heads (CHs) are
selected in a rotating manner from the available sensor nodes. This rotation is designed to distribute
energy consumption evenly across the network, preventing any single node from depleting its energy too
quickly. Attempts have been made to combine LEACH with other protocols such as AHP for better
energy efficiency.[4]

In this paper we aim to compare the ACO algorithm with the LEACH algorithm and the PSO algorithm
(that comes from the same particle swarm category as ACO). A comparison of this helps us identify the
parameters that each algorithm is more efficient in and hence form better combinations or better choices
for future WSN routing algorithms.
 LITERATURE SURVEY

Sl Year Author(s) Title Methodology Key Findings Research Gaps

No

1 2022 Chen, C., D2CRP: A 1-hop neighbor nodes 1.Increased network 1.To evaluate the
Wang, L. Novel are used as relays to lifetime compared to performance when it is
C., & Yu, Distributed transfer data from LEACH or PEGASIS in dynamic
C. M. 2-Hop nodes within a 2-hop 2.Lowest energy environments, we need
Cluster distance. This helps consumption, highest to explore with mobile
Routing increase energy number of transmitted nodes
Protocol for efficiency. packets that reach base 2.This study also does
Wireless Based on the station not pay attention to
Sensor calculation for an various security issues
Networks : optimal cluster number
The performance is
based on the lifetime of
the network, energy
consumption, and
packets transmitted to
the base station

2 2023 Bharany, Energy-effici 1.Fitness of a node is Has the shortest The practical
S., ent calculated with respect clustering time, utilizes implementation of this
Sharma, clustering to distance from base lower energy and has a has not been done
S., protocol for station and residual higher packet delivery
Alsharabi, underwater energy rate
N., Tag wireless
Eldin, E., sensor 2.Protocol is used to
& Ghamry, networks increase the lifetime of
N. A using a network
optimized
glowworm
swarm
optimization

3. 2022 Ghawy, M. An Effective 1.The PSO algorithm is It shows higher Packet 1.The energy
Z., Amran, Wireless used which helps to Delivery Ratio than consumption can be
 G. A., Sensor determine optimal AODV and DSDV improved
AlSalman, Network paths using the with better throughput,
H., Routing distance between nodes less delay and lower 2.Protocol has not
Ghaleb, E., Protocol and their energy levels. packet loss. been studied in detail
Khan, J., Based on for larger and more
Al-Bakhra Particle 2.The evaluation is complex networks
ni, A. A., Swarm based on:
... & Ullah, Optimization I.Packet Delivery Ratio
S. S. Algorithm II.Average end to end
delay
III.Normalized Routing
Load
IV.Throughput
V.Packet Loss

4 2022 Sixu, L., Particle 1.Combination of two 1.There is a reduction 1.Again , results are
Muqing, swarm algorithms- Particle in traversal path length simulation based
W., & Min, optimization Swarm and Artificial and an improvement in
Z. and artificial Bee Colony Network lifetime. 2.Does not deal with
bee colony dynamic events and a
algorithm for 2.Cluster heads larger network
clustering selected considering
and mobile energy, distance and
based cluster size.
software-defi
ned wireless
sensor
networks

5. 2020 Hung, L. Energy-Effic 1.The routes are 1.This protocol 1.These experiments
L., Leu, F. ient dynamically chosen by outperforms LEACH, were conducted in
Y., Tsai, K. Cooperative evaluating the energy and since the packets simulated
L., & Ko, Routing of the sensor nodes. are sent in groups, environments and are
C. Y. Scheme for 2.The sensors were energy is saved shown to work only on
Heterogeneo observed in both a 2.The distance that smaller networks.
us Wireless regular grid and data is being
Sensor random deployments. transmitted affects the
 Networks energy efficiency

6. 2021 Yun, W. Q-Learning- 1.A combination of 1.Has lower energy 1.Study is done in
K., & Yoo, Based reinforcement learning consumption that is simulations
S. J. Data-Aggreg and data aggregation balanced across nodes. 2.Effectiveness over a
ation-Aware 2.Initially the energy large network is not
Energy-Effic 2.To calculate optimal consumption is higher, considered
ient Routing path the residual but it reduces after the
Protocol for energy, aggregation learning process
Wireless rewards and
Sensor communication costs
Networks are considered

7. 2020 Abdulham Dragonfly 1.Three methods used 1.Lifespan of the 1.Study was done in a
eed, S. I., Algorithm for distributing the network was improved static environment
& for node : random, firefly
Aliesawi, Enhancing algorithm and 2.There is reduction in 2.Algorithm has not
S. A. PEGASIS PEGASIS delay due to reduced been used in large
Protocols in communication costs scale networks
Wireless 2.The network area is and better energy
Sensor split into parts to savings.
Networks balance out the data
transmissions
3.Works by exploiting
exploration and
exploiting phases

8 2024 Gangal, V., A 1.This is a combination 1.The network lifespan 1.Study focuses on
Cinemre, Distributed of two algorithms – is double that of stationary scenarios
I., & Leach-AHP LEACH and AHP traditional LEACH and has not been tested
Hacioglu, Routing for 2.The probability of 2.There is an increase in real world
G. Wireless forming cluster is in throughput deployment
Sensor determined by the
Networks node’s energy status 2.There was no
mention of any
possible security
concerns.
9 2020 Tianyi Routing The paper proposes a 1.Shows improved May not be adaptable
Zhang, Clustering clustering multi hop network lifetime to dynamic or mobile
Geng Protocol for routing for 3DWSNs compared WSNs
Chen, 3D Wireless using fragile collection 2.Better load balancing
Qingtian Sensor ant colony algorithm. It using load transfer
Zeng, Ge Networks includes clustering nodes and the fragile
Song, Based on nodes based on collection ant colony
Chao Li, Fragile residual energy and algorithm.
Hua Duan Collection lesser distances, also 3.Energy consumption
Ant Colony introducing new node for clustering and
Algorithm “roles” to balance routing was reduced
energy consumption 4.Shows stable
and prevent path performance
failing. Performance is
evaluated through
simulations

10 2020 Zongshan Energy The paper proposes a 1.Improved energy Protocol was not
Wang, Efficient routing protocol using consumption and analyzed with highly
Hongwei Routing an improved ABC better network lifetime varying or mobile
Ding, Bo Protocol algorithm (artificial bee 2.More network WSNs
Li, Liyong Based on colony) to enhance throughput
Bao, Improved energy efficiency and 3.Intra-cluster
Zhijun Artificial network lifetime for communication
Yang Bee Colony wsn’s. The improved reduced idle energy
Algorithm ABC algorithm is used wastage
for Wireless to optimize fuzzy 4.Multi-hop routing
Sensor clustering for cluster removes unnecessary
Networks head selection and energy consumption by
reduces unnecessary cluster heads far from
energy use. Multi hop base station of network
routing is done using
ant colony optimization
algorithm for
intercluster
communication to
improve energy
consumption

11 2024 Tadele A. Optimized The paper proposes 1.IMPRESEL The study does not
 Abose, Cluster two improved routing increased network address changing
Venumadh Routing protocols, IMP RES lifespan by 92% and environmental variable
av Protocol EL and EE SEP, for transmitted 44% more issues in larger and in
Tekulapall With homogeneous and data to base stations complex networks
y, Diriba Energy-Susta heterogeneous wsn. It 2.EESEP extended
C. Kejela, inable improves clustering network lifetime by
Ketema T. Mechanisms and cluster head 65%-70% better than
Megersa, for Wireless selection using residual SEP
Samuel T. Sensor energy and clustering 3.The methods show
Daka, and Networks probabilities. better performance in
Kehali A. Simulations were used packet delivery and
Jember to test the method energy efficiency

12 2023 Maha Dual-Tier The paper proposes a 1.DTCBR outperforms 1.The study does not
Ebrahim Cluster-Base novel dual tier cluster existing routing consider variables
Al-Sadoon d Routing in based routing protocol protocols like leach such as environmental
, Ahmed Mobile (DTCBR) for mobile 2.Showed up to 37% factors or security
Jedidi, and Wireless wireless sensor improvement in issues
Hamed Sensor networks (MWSN) The network life
Al-Rawesh Network for network is divided into 3. Show more energy
idy IoT main connectivity efficiency and
Applications zones and candidate reliability in large
cluster zones to scale IOT network
improve the selection
of cluster heads based
on several factors.
Matlab simulations
were conducted to
compare its
performance

13 2024 Kalavagun Optimized The paper talks about 1.Was better than 1.Study does not
ta Aravind Fuzzy Logic an energy efficient existing models in consider
and Based geographic (EEG) energy efficiency, environmental factors
Praveen Energy-Effic routing protocol using network lifetime, like temperature and
Kumar ient fuzzy logic and packet delivery, and time
Reddy Geographica optimized membership convergence rate
Maddikunt l Data functions through the 2 For group 2
 a Routing in Harris Hawk scenarios, it shows a
Internet of optimization algorithm performance increase
Things (HHO). It finds an of upto 25.6%
optimal route based on
the parameters: energy,
distance, delay,
overhead, trust, and
quality of service.
MATLAB simulations
are conducted to test its
efficiency

14 2022 Lina Research on The paper focuses on 1.The algorithm 1.May not handle real
Wang, Smooth Path improving the classic provides smoother world conditions
Hejing Planning ant colony algorithm paths where we have
Wang, Xin Method by using the Floyd 2.It provides lesser frequently changing
Yang, Based on algorithm to improve energy consumption environment
Yanfeng Improved the pheromone 3.Shows better
Gao, Ant Colony distribution and performance in more
Xiaohong Algorithm guiding paths, hence complex
Cui, Binrui Optimized improving path environments, thus
Wang by Floyd finding. It also has showing more optimal
Algorithm “fallback strategies” to path quality
handle deadlocks and
improving heuristic
functions (H(n)). The
algorithm uses a B
spline quadratic curve
technique to reduce
energy consumption

15 2021 Debora Di A Novel Ant The paper proposes a 1.The algorithm 1.Algorithm may not
Caprio, Ali Colony fuzzy based ant colony performs better than handle very dynamic
Ebrahimne Algorithm optimization (ACOA) swarm optimization or very fuzzy data
jad, for Solving algorithm for shortest and bee colony
Hamidreza Shortest Path path problems in algorithms in terms of
Alrezaamir Problems uncertain conditions. time complexity
i, with Fuzzy The algorithm uses 2.Reduction of
Francisco Arc Weights fuzzy numbers for edge approximately 3.It is
J. weights. It uses a very effective for
 Santos-Art distance function for complex graphs.
eaga path comparison. The 4.Gives more
approach is tested on consistent solutions
various graphs and its
results are compared
with swarm
optimization, and
artificial bee colony
algorithms

16 2023 Hongzhan Wireless The paper is about an 1.IACA has lesser 1.May not be able to
g Han, Jun Sensor improved ant colony energy consumption handle more dynamic
Tang, Network algorithm (IACA) for and better network IOT environments.
Zhengjun Routing wireless sensor lifetime/lifespan 2.Faces issues for
Jing Optimization network (wsn) routing. 2.Has lower balancing energy
Based on This is done by transmission delay and consumption and
Improved analyzing pheromone better routing quality for other
Ant Colony concentrations and 3.Gives fewer dead network topologies
Algorithm in transition probabilities nodes, hence giving 3.Mentions other
the Internet of ant movements. better network lifetime advanced heuristic
of Things Simulations were 4.Has better network techniques
conducted using throughput
parameters like energy 5.Has more variety of
consumption, delay, IOT applications
and throughput
 METHODOLOGY
The problem at hand involves comparing three popular optimization algorithms—Ant Colony Optimization
(ACO), Low Energy Adaptive Clustering Hierarchy (LEACH), and Particle Swarm Optimization (PSO)—in the
context of their performance and suitability for solving optimization problems in sensor networks. By comparing
these algorithms, we aim to identify which one performs better in terms of energy efficiency, network lifetime,
and packet delivery ratio when applied to real-world optimization tasks in sensor networks.
Each of these three algorithms have their advantages and disadvantages. When placed under similar conditions we
try to identify the efficiency of Ant Colony Optimization algorithm when compared to Low Energy Adaptive
Clustering Hierarchical Algorithm and Particle Swarm Algorithm. The particle swarm algorithm belong to the
same category as Ant Colony Algorithm (Swarm intelligence methods) but in a continuous space as opposed to

ACO which deals with discrete space.

The objective of comparing these algorithms is to further enhance the knowledge available in this field and
provide guidance for future research. These algorithms are compared with respect to the field of Routing protocols
for WSN (Wireless Sensor Networks) .

Algorithm 1 :
The ACO algorithm was developed based on the foraging behavior of ants. It is a nature inspired algorithm. It is
particularly useful in solving problems such as the Travelling Salesman problem. It uses an iterative search where
the shortest path is reinforced by the trail of pheromones that each ant leaves behind. The pheromones
“evaporate” after a certain set time period.
The algorithm given below is the pseudocode used for Ant Colony Optimization in this paper. The WSN are
considered as vertices of a graph. The distance between every WSN is represented by the edges of a graph. The
edges are given weights to represent the pheromone values along them. Figure 1(a) shows the diagrammatic
representation of this algorithm.
The fitness value that is calculated in the given algorithm uses the formula:
Fitness(k)=w1​⋅Ek​1​+w2​⋅Lk​+w3​⋅Pk​
Where Ek​is the total energy consumption of the solution found by ant k.
Lk​is the network lifetime achieved by the solution found by ant k.
Pk​is the packet delivery ratio of the solution found by ant k.
w1​,w2​,w3​are weights assigned to each metric to balance their influence on the fitness value.
Algorithm Begin AntColonyOptimization:
Initializing pheromone trails;
While not termination do:
Generate the ant population;
Calculate the fitness value of each ant;
Find the best solution through selection methods;
Update the pheromone trails;
End while
End Algorithm
Algorithm 2:
The particle swarm algorithm is simulating particles moving freely in space. It is continually updating each
particle's position based on 2 factors - the particle’s best position and the best position of the swarm. This process
is done iteratively while updating particle velocities and ensuring that the particle remains within the solution
space. Figure 1(b) shows the diagrammatic representation of this algorithm.
Formula used to update velocity:
vi​(t+1)=w⋅vi​(t)+c1​⋅r1​⋅(pi_best​−xi​(t))+c2​⋅r2​⋅(g_best−xi​(t))
Where, vi​(t): Current velocity of particle i at time t.
w: Inertia weight, a parameter that controls the influence of the previous velocity.
c1​and c2: Acceleration coefficients that control the influence of the cognitive and social components.
r1​and r2​: Random numbers uniformly distributed in [0, 1].
pi_bestp: Best position of particle i (personal best).
g_best: Best position found by the swarm (global best).
xi(t): Current position of particle i at time t.

It finally returns the best particle in the swarm as shown in Algorithm given below :

Algorithm ParticleSwarm:
Initialize swarm particles (xi...xn)
For i in range(max_iter) do :
For j in range(n) do:
Compute new velocity of ith particle
Compute new position of ith particle
If position is not within range then discard it
Update bestfit of particle
End for
End for
Return best particle of swarm
End Algorithm

Algorithm 3:
The main objective behind the design of low energy adaptive clustering hierarchy algorithms is to minimize
energy consumption. It does this by selecting the cluster heads in a rotational manner in a scattered region as
shown in Figure 2 . In each rotation the nodes calculate the probability of becoming a cluster head . Every node
that is not a cluster head joins the nearest cluster.
Figure 1(c) shows the diagrammatic representation of this algorithm.

Algorithm LEACH:
Initialize sensor nodes and set parameters
For i in range(max_iter):
For each node do:
Compute probability of becoming cluster head and broadcast to neighbours
For each non-cluster head node:
 Join closest cluster
For each cluster head:
Collect and aggregate data to be sent to base station
End Algorithm

Fig 1

(a) ACO (b) PSO (c) LEACH

Algorithm Algorithm Algorithm
 Fig 2 : Shows the arrangement of nodes in LEACH algorithm with the cluster heads

Based on the above algorithms we analyze their performance based on various factors a few among which are :

I. Energy Consumption - refers to the amount of energy consumed by a sensor node during computation and
communication .
II. Network Lifetime - This refers to the period of time a network can function before all the critical nodes
fail
III. Packet Delivery Ratio - measures the rate of success at which packets reach their intended destination.

The following factors are kept common for all algorithms:

1. We assume a maximum of 100 nodes

2. An area of range 100m width and height
3. Transmission energy of 50e-7 J/bit
4. Probability of node becoming a cluster (in case of LEACH) - 0.1
5. Inertia weight of 0.5 (for PSO)
6. Cognitive cost of 1.5 (for PSO)
7. Pheromone evaporation rate of 0.1 (for ACO)
The results on running these simulations are:

Fig : Results of ACO algorithm

1. ACO algorithm

● Energy Consumption : The energy consumption is seen to vary between 0.1225 and 0.1400
Joules across 100 simulation runs. This suggests a moderate instability in energy consumption.
● Network Lifetime : The network lifetime remains constant and shows no variations for any of the
three algorithms.
● Packet Delivery Ratio : The values vary between 0.65 and 0.85.

2. PSO algorithm

Fig : Results of PSO algorithm

● Energy consumption :The energy consumption fluctuates quite significantly by ranging from about 9 to
14 Joules. This shows higher instability compared to ACO.
● Network Lifetime: constant with no variations with ACO and LEACH
● Packet Delivery Ratio : The values vary between 0.70 and 0.90. Shows highest packet delivery

Efficiency.

3. LEACH algorithm

Fig : Results of LEACH algorithm

● Energy consumption : The energy consumption is relatively low and varies between 0.006 and
0.014 Joules, showing more stability compared to both ACO and PSO.
● Network Lifetime : Remains the same and shows no comparisons with ACO and PSO
● Packet Delivery Ratio : The delivery ration ranges from 0.70 to 0.875 which shows an average
 RESULTS AND DISCUSSIONS
On analyzing the performance of the three algorithms, we form a comparison study on their various properties.

In the case of Energy consumption, the ACO algorithm seems to flicker between 0.1225 and 0.1400 Joules when
tested across 100 simulation runs. The occasional spikes seem to be due to the nature of pheromone updates in the
iterative process. We see the PSO algorithm shows much more significant fluctuations in energy consumption
from 9 to 14 Joules. This can be attributed to the continuous changes of particle velocities. The LEACH algorithm
seems to show the most stable energy consumption which ranges between 0.006 and 0.014 Joules.

The analysis of Network Lifetimes does not show much variation since it remains stable across all three
algorithms. Thus we can say that the choice of algorithms has no significant effect on the lifetime of a network.

The packet delivery ratio of ACO ranges from 0.65 to 0.85 . This indicated moderate performance hence making
ACO algorithm a reasonable choice for reliable and stable algorithm. However the ACO algorithm may also cause
loss of packets due to suboptimal path selections. The PSO algorithm exhibits the highest packet delivery ratio
which makes it effective for optimal or near optimal solutions. LEACH uses a clustering approach which makes
data transmission reliable however even LEACH can suffer from packet dropping issues.

The ACO algorithm is effective in situations requiring iterative or adaptive pathfinding with frequently changing
topologies. Its weakness is having a higher energy consumption while having a low to moderate packet delivery
ratio. Therefore we can say that ACO is suitable for applications that are dynamic and adaptable.

The PSO algorithm has the highest packet delivery ratio and is effective in search spaces that are continuous in
nature. It shows fluctuations in energy consumption therefore is considered best for applications with high packet
delivery ratio.

The LEACH algorithm shows lowest energy consumption and provides stable operation . However it exhibits
average packet delivery ratio and may therefore cause scalability issues in large networks. It is more suitable for
situations where network stability is prioritized over packet delivery ratio.

Exploring hybrid approaches that integrate ACO with other energy efficient algorithms to mitigate its higher
energy consumption while taking advantage of its adaptive capabilities. For the PSO algorithm, implementation of
various techniques to stabilize its energy consumption can make the usage of this algorithm more efficient.
LEACH’S packet delivery system can be made more efficient by including error correction techniques.

Investigating Hybrid approaches that combine the strengths of these algorithms can be done as future research on
this topic. The application of these algorithms in WSN with varying node densities , mobility patterns and
environments can be further tested.

By conducting this comparative study we aim to provide valuable insights and various guidelines for selecting and
optimizing algorithms in Wireless Sensor Networks.
 CONCLUSION
In this comparative study , we examined the performance of Ant Colony Optimization Algorithm (ACO), Low
Energy Adaptive Clustering Hierarchy (LEACH) and Particle Swarm Optimization (PSO) . These algorithms
were analysed in the context of their application as routing protocols for Wireless Sensor Networks. Each
algorithm was tested for different parameters that included energy consumption, network lifetime and packet
delivery ratio.

The algorithms were proved to be effective in different fields leading to the conclusion that they would prove
effective in different situations when paired with other algorithms. Using hybrid algorithms for routing can help
mitigate qualities in the individual algorithms that we wish to avoid while making use of the properties such as
high energy efficiency, high packet delivery ratio . This leads to better performance and results.

Finding such hybrid algorithms can be included as part of future research. This study has tried to provide detailed
analysis and valuable insights to select and optimize algorithms for Wireless Sensor Networks.
 REFERENCES:

[1] 2023, Bharany, S., Sharma, S., Alsharabi, N., Tag Eldin, E., & Ghamry, N. A, Energy-efficient clustering
protocol for underwater wireless sensor networks using optimized glowworm swarm optimization

[2] 2022, Ghawy, M. Z., Amran, G. A., AlSalman, H., Ghaleb, E., Khan, J., Al-Bakhrani, A. A., ... & Ullah, S. S.,
An Effective Wireless Sensor Network Routing Protocol Based on Particle Swarm Optimization Algorithm

[3] 2022, Sixu, L., Muqing, W., & Min, Z. , Particle swarm optimization and artificial bee colony algorithm for
clustering and mobile based software-defined wireless sensor networks

[4] 2024, Gangal, V., Cinemre, I., & Hacioglu, G. , A Distributed Leach-AHP Routing for Wireless Sensor
Networks

Web Sources:

[5] https://www.geeksforgeeks.org/introduction-to-ant-colony-optimization/

[6] https://www.geeksforgeeks.org/particle-swarm-optimization-pso-an-overview/

[7] https://www.geeksforgeeks.org/data-gathering-in-sensor-network/