International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.

4, July 2023

IOT RESOURCE ALLOCATION AND OPTIMIZATION

USING IMPROVED REPTILE SEARCH ALGORITHM

Prabhakar Narasappa Kota1, Pravin Balaso Chopade1, Bhagvat D. Jadhav2, Pravin

Marotrao Ghate2 and Shankar Dattatray Chavan3

Department of Electronics &Telecommunications, MES’s College of

1

Engineering, Pune, India

2
Department of Electronics and Telecommunication, JSPM's Rajarshi Shahu College of
Engineering, Pune, India
3
Department of Electronics and Telecommunication Engineering, Dr. D. Y. Patil Institute
of Technology, Pimpri

ABSTRACT
The Internet of Things (IoT) is a dispersed network system that connects the world through the Internet.
The architecture of IoT consists of more gateways and resources which cannot be allocated in a manual
process. The allocation of resources in IoT is a challenging process due to the higher consumption of
energy and high latency rate. To overcome the challenges in existing works, this research introduced an
Improved Reptile Search Algorithm (IRSA) to solve the optimization problem which occurs during the time
of allocation resources among IoT networks. IRSA employs the methodology of levy flight and cross-over
to update the candidate position and enhance the search speed in a single iteration. The proposed method
consumes less energy and has low latency during data transmission from User equipment (UE) to the base
station.IRSA has been compared with the existing Scalable Resource Allocation Framework (SRAF) and
Improved Chaotic Firefly Algorithm (ICFA). The obtained experimental results show that the proposed
IRSA attained better performance with an allocation rate of 96.40% which is comparatively higher than
SRAF and ICFA with 92.40% and 91.67% respectively.

KEYWORDS
Crossover, Improved Reptile Search Algorithm, Internet of Things, Levy flight, Resource Allocation

1. INTRODUCTION
In recent days, the Internet of Things is developed as an essential part of the daily life of an
individual. The usage of IoT in real-time applications includes medical, automation, smart
building, etc [1-4]. The networks of IoT are multi-hop ad hoc networks that are created using
more sensor nodes deployed in a supervised region. IoT networks are utilized in highly functional
applications through the cooperation of nodes [5]. During resource allocation, IoT device needs
better Quality of Service (QoS) while implementing it in the node with energy-efficient
configurations [6]. The resource allocation leads the physical IoT networks to different types of
virtual networks which is considered an advantageous process to ensure better QoS [7]. The lack
of efficient resource allocation leads to a major effect on the performance of IoT networks [8].
The resource allocation technique can be accessed using the cost and efficiency of different
processing techniques. The resource allocation can be performed based on two methodologies
such as scheduling and random access. [9]. The process of scheduling takes place by creating an
interaction among the nodes with varying geographical locations and allocating suitable resources
to tasks [10].
DOI: 10.5121/ijcnc.2023.15403 39
 International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.4, July 2023
Energy consumption is a major drawback in the uplink and downlink of IoT networks. This can
be reduced by the allocation of channels and sub-channel's bandwidth [11,12]. Moreover, the
resource allocation process serves as an energy saver and helps to improvise the capability of IoT
networks [13]. One of the most important study areas is the IoT Resource Allocation Problem
(IRAP), which considers how to lower the costs of traffic among gateways, appliances, etc. The
performance of wireless communication and foundation devices can both be improved using an
efficient resource allocation mechanism [14]. There are numerous optimization issues have
aroused while solving the problems related to allocation of resources. These problems can be
overwhelmed using evolutionary optimization techniques which make an effort to maximize the
probability to obtain solution to challenging problems [15,16,17,18]. This research is motivated
by finding an optimal solution for the resource allocation problem with minimum energy
consumption and delay. To overcome the fore mentioned problems, this research used an
effective heuristics algorithm to enhance the convergence and aids in solving the allocation
problem. The major objective of this research is to provide an optimal heuristic algorithm for
optimal resource allocation in IoT networks.

The main contributions of this research are listed as follows:

1. The problems in resource allocation of IoT networks are resolved using IRSA which
optimized the allocation issues based on latency and energy consumption.
2. Levy and crossover methodologies are employed in the proposed IRSA which updates
the position and enhance the search speed in a single iteration.

The remaining paper is organized as follows, Section 2 represents the related works of the paper.
The proposed method is discussed in Section.3. The results and discussion are provided in
Section.4. Finally, Section.5 represents the overall conclusion of the paper.

2. RELATED WORKS
Jian Jiao et al. [19] have introduced a joint network stability control and Non-Orthogonal
Multiple Access (NOMA) based power allocation in the downlink system of Satellite Internet of
Things (S-IoT). The two virtual methods such as data queuing and power expenditure were
established to evaluate the efficiency of the introduced method. An optimal solution was
provided by the particle swarm optimization algorithm for the issues related to joint allocation.
The NOMA combined with Karush-Kuhn-Tucker (KKT) to identify a sub-optimal solution using
the NOMA-KKT method. However, the NOMA-KKT method consumed more space when the
number of user equipment becomes higher.

Xiaolan Liu et al. [20] have introduced an Efficient Channel Allocation Algorithm (ECAA)
which has less complexity while grouping the number of users in uplink communication of IoT
networks. The ECAA utilized Markov Decision Process (MDP) to predict the energy arrival and
the condition of uncertainty for each user. The MDP methodology attained better power
allocation for every individual user and improvise the throughput of the network in the specified
network frame. The ECCA attained nearest-optimization performance in the allocation of
channels with minimal complexity. However, the performance of the ECCA algorithm was
recessive when it comes to the multiple numbers of user equipment.

Zafer Al-Makhadmeh and Amr Tolba [21] have introduced Scalable Resource Allocation
Framework (SRAF) in user-based IoT utilizing deep learning techniques. The SARF performs
allocation by balancing the availed sources and balancing the processed requests. The joint
process of SARF was used to decrease the processing time and the probability of failure in
allocation. The SARF allocates the resource by solving the complexities in replication and issues
40
 International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.4, July 2023
in overloading. The SRAF combines the requests from the user and available resources to fulfill
the needs of the user and provide support for long-range access. However, the SARF framework
took more time to allocate the resource when implemented with the meta-heuristic approaches.

Zhiyong Wang et al. [22] have introduced a resource allocation scheme in IoT networks using an
improved chaotic firefly algorithm. This approach creates a multi-objective optimization based
on interference of cognitive radio's operational environment. Protecting the primary user's usual
activities was intended to increase the secondary system's throughput and the number of users
that the secondary base station can serve. The firefly algorithm incorporates the chaos algorithm
to increase convergence speed and decrease the likelihood of local optimization. The algorithm
can effectively find the best answer while making the issues easier. The algorithm doesn’t focus
on the problems related to real-time applications and it was not sensible for the response time.

James Adu Ansere et al. [23] have developed an energy-efficient iterative algorithm using the
convex approximation method and Lagrangian dual decomposition technique for optimal
resource allocation. A Jointly optimized power allocation, the count of active base stations, and a
user selection algorithm were utilized to improvise the energy efficiency in the networks of IoT.
The proposed algorithm initiates the base station count and provides better Quality of Service
(QoS) in user equipment present in the IoT. The convex approximation method and Lagrangian
dual decomposition technique used in the algorithm provide assured convergence with minimal
computational complexity. However, the delay time considerably get increased when the user
equipment gets increased.

3. OVERVIEW OF IRSA
The issue raised in resource allocation must be defined before describing the proposed algorithm.
There are two types of nodes in the IoT network considered in this research, the first one is
resources and the second is gateways connected with the networks of IoT. The various parts in
the IoT network are connected by gateways and these gateways are responsible to regulate the
traffic in resources. Moreover, these resources can be connected with various gateways, and
communication cost among them is scheduled. The cost computation is an essential process to be
carried out among the gateways and when the computational cost seems to be higher, it leads to
connection issues among the gateways. The communication cost gets varied by changing the
communication model. The load balancing is the process where the resources get separated
among the gateways. Thus, this research focuses on the cost of communication and load
balancing in the resource allocation of IoT networks.

3.1. Objective Function

All the resource node presents in the IoT networks communicate with one another and the total
cost must be computed for individual solutions of resource allocation. Each resource present in
the IoT networks sends messages to other resources present in the network. The cost for the
messages is computed which is measured as an objective function. The total cost function is
represented as and it can be minimized using the Improved Reptile Search Algorithm (IRSA).
The total cost is computed using equation (1) as follows:

|𝑉 |
𝑔
∑𝑗=1 (𝑑𝑗𝑟 ×𝑑𝑔 )
𝑇𝑐 = 𝑝
(1)

Where the total number of the gateway is represented and the cost for transferring the data among
the 𝑗th gateway and other resources is denoted as 𝑑𝑗𝑟 . The cost of communication in a gateway

41
 International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.4, July 2023
has an exponential effect on the cost of communication, so 𝑑𝑗𝑟 × 𝑑 𝑔 is considered a better
equation while computing the maximum cost. The multiplication of is effective while computing
the cost involved in communication. The communication cost among the gateway sources can be
computed using equation (2) as follows:

|𝑉 |
𝑔 𝑔 |𝑉 |
𝑑𝑔 = ∑𝑖=1 ∑𝑗=1 𝑙𝑖𝑗 (2)

The cost involved in communication among the gateways 𝑖 and 𝑗 is denoted as 𝑙𝑖𝑗 and 𝑑𝑗𝑟 is
computed by utilizing the equation (3) as follows:
𝑗
|𝑣𝑔 |
𝑑𝑗𝑟 = ∑𝑘=1 𝜀𝑗𝑘 (3)

The communication cost among the 𝑗th gateway is denoted as 𝜀𝑗𝑘 and the connected gateway
𝑗
resources are represented as |𝑣𝑔 |.

Another parameter in resource allocation is load balancing which is denoted a penalty𝑃. The
IRSA minimizes the and the function’s denominator is taken which denotes the penalty and
denotes the normal state of the gateway. The resource with a gateway higher than the value of is
utilized to improve value. The value of the penalty𝑃 is computed using the equation (4) as
follows:
|𝑉𝑔 |
𝑃 = 1 + ∑𝑖=1 𝑃𝑖 (4)

Where the total number of the gateway is denoted as |𝑉𝑔 | and the penalty 𝑃𝑖 for each gateway is
computed using equation (5) as follows:

|𝑉 |
1 𝑖𝑓 𝑔𝑖𝑡 ≤ 𝜀 |𝑉𝑟 |
𝑔
𝑃𝑖 = { |𝑉 |
(5)
0 𝑖𝑓 𝑔𝑖𝑡 ≤ 𝜀 |𝑉𝑟 |
𝑔

The total number of a resource based on the gateway is denoted as the total number of resources
and the constant value is denoted as and respectively. The illustration of the network system is
represented in Figure 1 as follows:

Figure 1. Representation of network system

42
 International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.4, July 2023
3.2. Reptile Search Algorithm (RSA)

The RSA algorithm [24] is a type of metaheuristic algorithm developed by Abuligah et al. in the
year 2021. RSA imitates how crocodiles might act while hunting in their native environment. In
general, crocodiles are members of the "Crocodylinae" family and prefer to live in areas with
access to food and water. They come from the class of animals known as amphibians, which can
hunt both in and out of water. The following examples show how crocodiles behave in their daily
lives. The RSA is based on the natural social behavior and attacking strategy of the crocodiles.
The RSA undergoes two phases as exploration and exploitation phase, the exploration phase is
based on the mechanism of encirclement,and the exploitation stage is based on hunting behavior.
The metaheuristic algorithm is implemented with the candidate solution before beginning the
iterations. The candidate solutions can be created based on equation (6) mentioned as follows:

𝑍𝑖,𝑗 = 𝑟𝑎𝑛𝑑 × (𝑢𝑝 − 𝑙𝑜𝑤) + 𝑙𝑜𝑤 (6)

Where the stochastic value is denoted as,the maximal and minimal limits of the problems in
optimization are denoted as.

The distinctive feature of RSA's worldwide search is encircling behavior. Two behaviors such as
aerial and abdominal walks make up the process. Crocodiles are frequently prevented from
approaching food by these two acts. The crocodiles perform a search at a global level to get food
after numerous tries. Make sure that the meantime of this development stage is adjusted
continuously to obtain better results. The crocodile’s encircling behavior is represented using
equation (7) as follows:

𝑇𝑚𝑎𝑥
𝑏𝑒𝑠𝑡𝑗 (𝑡) × (−𝜂𝑖,𝑗 (𝑡)) × 𝛽 − 𝑅𝑖,𝑗 (𝑡) × 𝑟, 𝑡 ≤ 4
𝑧𝑖,𝑗 (𝑡 + 1) = { 𝑇𝑚𝑎𝑥 2×𝑇𝑚𝑎𝑥
(7)
𝑏𝑒𝑠𝑡𝑗 (𝑡) × 𝑧𝑟1,𝑗 × 𝐸𝑆(𝑡) × 𝑟, 4
≤𝑡< 4

Where the best position of the crocodile at iteration is denoted as and the stochastic value from 0
to 1 is denoted as 𝑟. The maximum number of iterations is represented as is crocodile’s
dimension at the state 𝑟1 and the stochastic value between 0 and 1 is denoted as. The operator of
the crocodile 𝑖 in dimension 𝑗 denoted as 𝜂𝑖,𝑗 and it is evaluated using equation (8) as follows:

𝜂𝑖,𝑗 = 𝑏𝑒𝑠𝑡𝑖,𝑗 (𝑡) × 𝑃𝑖,𝑗 (8)

Where 𝑃𝑖,𝑗 is represented as the percent of the variance between the best position and the current
state of the crocodile.
The reduced function 𝑅𝑖,𝑗 is utilized to lower the explored region is computed using the equation
(9) which is represented below,
𝑏𝑒𝑠𝑡𝑗 (𝑡)−𝑥𝑟2,𝑗
𝑅𝑖,𝑗 = (9)
𝑏𝑒𝑠𝑡𝑗 (𝑡)+𝜀

Where the random value among 1 and 𝑁 is represented as 𝑟2 and the smaller value is denoted as
𝜀.
The reducing probability of the arbitrary value is denoted as 𝐸𝑆 which lies in the ratio of 2 and -
2, it is represented in equation (10) as follows:
1
𝐸𝑆 = 2 × 𝑟3 × (1 − ) (10)
𝑇

43
 International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.4, July 2023
The integer value lies between -1 to 1 is denoted as 𝑟3 .
The best crocodile’s position 𝑃𝑖,𝑗 is reorganized using equation (11) as follows:
𝑧𝑖,𝑗−𝑀(𝑍𝑖)
𝑃𝑖,𝑗 = 𝛼 + 𝑏𝑒𝑠𝑡 ′ (11)
𝑗(𝑡)×(𝑢𝑝𝑗 −𝑙𝑜𝑤𝑗 )+𝜀

Where the maximal and minimal limits of 𝑗 th dimension are denoted as 𝑢𝑝𝑗 and 𝑙𝑜𝑤𝑗 , the
parameter utilized to regulate the accuracy during the search process is denoted as

The crocodile’s average position is denoted as 𝑀(𝑍𝑖 )which is represented in equation (12) as
follows:
1
𝑀(𝑍𝑖 ) = 𝑛 ∑𝑛𝑗=1 𝑧𝑖,𝑗 (12)

The hunting process is based on two techniques such as coordination and collaboration. The
encirclement system has almost locked down the location of the intended prey for the crocodiles,
who may now approach the target more easily thanks to their hunting tactics. After multiple
rounds, the development phase will frequently discover the nearly ideal candidate solution. The
mathematical representation of the developed hunting behavior of the crocodile is represented in
equation (13) as follows:

2×𝑇𝑚𝑎𝑥 3×𝑇𝑚𝑎𝑥
𝑏𝑒𝑠𝑡𝑗 (𝑡) × 𝑃𝑖,𝑗 (𝑡) × 𝑟, ≤𝑡<
4 4
𝑧𝑖,𝑗 (𝑡 + 1) = { 3×𝑇𝑚𝑎𝑥 4×𝑇𝑚𝑎𝑥 (13)
𝑏𝑒𝑠𝑡𝑗 (𝑡) − 𝜂𝑖,𝑗 (𝑡) × 𝜀 − 𝑅𝑖,𝑗 (𝑡) × 𝑟, ≤𝑡<
4 4

Where the crocodile’s optimal position is defined as an operator for crocodile at 𝑗th dimension is
denoted as 𝜂𝑖,𝑗 .

3.3. Improved Reptile Search Algorithm (IRSA)

Less latency and energy usage are two fundamental network components for IoT administration.
Therefore, this study has concentrated on the RA problem in IoT to attain these goals. As was
previously mentioned, the key goals of RA are load balancing and lowering communication
costs. Less communication cost results in lower network latency. By load balancing, blockages
from the network are neglected and increase the performance of the network. If these limitations
are managed, there will be more opportunities for using this technology. This paper proposed a
heuristic technique known as IRSA to tackle the issues in RA in IoT networks. The leap out and
the enhancement of precision of the candidate solution are improved using the Levy flight
method and the technique based on interaction crossover improvises the ability of IRSA.

3.3.1. Levy Flight Method

The following properties of the random numbers produced by the Levy flight method are
employed in this paper:

(1) The generated random numbers are frequently small, with larger ones irregularly combined.
(2) The probability density function of the steps has broad tails. With the help of this random
number, oscillations may be induced, the area can be quickly searched within one iteration,
and a candidate solution can be helped to emerge from the local optimum.

44
 International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.4, July 2023
The levy distribution function is represented in equation (14) as follows:

𝐿𝑒𝑣𝑦(𝜔) = 𝑡 −1−𝜔 , 0 < 𝜔 ≤ 2 (14)

The length of a step in levy flight is computed using equation (15),

𝑢
𝑠 = |𝑣|1/𝜔 (15)

Where the Gaussian distribution function is represented as, the value of 𝑢 and 𝑣 is represented in
equation (16) as follows:

𝑢~𝑁(0, 𝜎𝑢2 ) and 𝑉~𝑁(0, 𝜎𝑉2 ) (16)

Where the value of 𝜎𝑢 𝑎𝑛𝑑𝜎𝑉 is represented in equations (17) and (18) respectively.

1/𝛽
𝛽
𝜙(1+𝛽).𝑠𝑖𝑛(𝜋. )
2
𝜎𝑢 = ( (𝛽−1) ) (17)
𝛽
𝜙((1+ ).𝛽.2 2 )
2

𝜎𝑣 = 1 (18)

Where the gamma function is represented as ϕ and the parameter is denoted as β with fixed 1.5

The encircling mechanism involves high walk and belly walk which takes place using the Levy
flight method is used to efficiently increase the search area. To maximize the tractability of the
ideal phase of exploitation, the Levy flying approach is used for phases of coordination and
collaboration in crocodile behavior for hunting. The method of encircling is based on levy flight
and is represented in equation (19) as follows:

𝑇𝑚𝑎𝑥
𝑏𝑒𝑠𝑡𝑗 (𝑡) × (−𝜂𝑖,𝑗 (𝑡)) × 𝛽 − 𝑅𝑖,𝑗 (𝑡) × 𝐿𝑒𝑣𝑦, 𝑡 ≤
4
𝑧𝑖,𝑗 (𝑡 + 1) = { 𝑇𝑚𝑎𝑥 2×𝑇𝑚𝑎𝑥
(19)
𝑏𝑒𝑠𝑡𝑗 (𝑡) × 𝑧𝑟1,𝑗 × 𝐸𝑆(𝑡) × 𝜆 × 𝐿𝑒𝑣𝑦, 4 ≤ 𝑡 < 4

Where 𝜆 is the parameter with a value of 0.1

The hunting of the crocodile based on coordination and cooperation using the levy flight method
is represented in equation (20):

2×𝑇𝑚𝑎𝑥 3×𝑇𝑚𝑎𝑥
𝑏𝑒𝑠𝑡𝑗 (𝑡) × 𝑃𝑖,𝑗 (𝑡) × 𝜆 × 𝐿𝑒𝑣𝑦, 4 4
≤𝑡<
𝑧𝑖,𝑗 (𝑡 + 1) = { 3×𝑇𝑚𝑎𝑥 4×𝑇 (20)
( ) ( ) ( )
𝑏𝑒𝑠𝑡𝑗 𝑡 − 𝜂𝑖,𝑗 𝑡 × 𝜀 − 𝑅𝑖,𝑗 𝑡 × 𝜆 × 𝐿𝑒𝑣𝑦, 4 ≤ 𝑡 < 4𝑚𝑎𝑥

3.3.2. Interaction crossover method

The information is interchanged among two candidate solutions and the candidate is in an ideal
position, the interaction crossover technique aids the candidate in their current position in making
adjustments. To enhance the search ability of the candidate solutions, the new position gathers
data on the ideal solution and other potential alternatives. The parameter utilized to control the
crocodile’s activity for the number of iterations is defined as 𝐶𝐹 which is represented in equation
(21) as follows:

45
 International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.4, July 2023
𝑡
𝑡 2×
𝑇𝑚𝑎𝑥
𝐶𝐹 = (1 − ) (21)
𝑇𝑚𝑎𝑥

Where the current iteration is represented as and the maximum number of iterations is defined as.
The crocodile population is classified into two categories as. The upgraded state of the crocodile
is represented in equations (22) and (23) as follows:

𝑧𝑘1,𝑗 (𝑡 + 1) = 𝑧𝑘1,𝑗 (𝑡) + 𝐶𝐹 × (𝑏𝑒𝑠𝑡𝑗 (𝑡) − 𝑧𝑘1,𝑗 (𝑡)) + 𝑐1 × (𝑧𝑘1,𝑗 (𝑡) − 𝑧𝑘2,𝑗 (𝑡)) (22)
𝑧𝑘2,𝑗 (𝑡 + 1) = 𝑧𝑘2,𝑗 (𝑡) + 𝐶𝐹 × (𝑏𝑒𝑠𝑡𝑗 (𝑡) − 𝑧𝑘2,𝑗 (𝑡)) + 𝑐2 × (𝑧𝑘2,𝑗 (𝑡) − 𝑧𝑘1,𝑗 (𝑡)) (23)

Where the best position of the crocodile is represented as and the stochastic values lie in the
interval 0 to 1 denoted as and. The crocodile lies in the position and is denoted as respectively.
After the process of crossover, the crocodiles with poor capabilities are neglected based on the
equation (24) represented as follows:

𝑧𝑖,𝑗 (𝑡), 𝑖𝑓 𝑓 (𝑧𝑖,𝑗 (𝑡)) < 𝑓 (𝑧𝑖,𝑗 (𝑡 + 1))

𝑧𝑖,𝑗 (𝑡 + 1) = { (24)
𝑧𝑖,𝑗 (𝑡 + 1), 𝑒𝑙𝑠𝑒 𝑖𝑓 𝑓 (𝑧𝑖,𝑗 (𝑡 + 1)) < 𝑖𝑓 𝑓 (𝑧𝑖,𝑗 (𝑡))

4. RESULTS AND ANALYSIS

This section provides the results and analysis of this research. The result portion is classified into
performance analysis and comparative analysis which are represented in the following sections.
The simulation of the proposed IRSA is implemented with four gateways and seven resources,
the test data is created based on three sizes of the data such as small, medium, and large.
Moreover, the total count of gateway gets varied from 4 to 100 and the a total number of
resources varies from 100 to 800. The proposed method is designed and implemented for total of
eight test samples. Moreover, the proposed method is implemented in MATLAB software with
system specifications such as i7 processor at 2.24 GHz and 8GB Random Access Memory
(RAM) in Windows 10 operating system.

4.1. Performance Analysis

In this research, the optimization problem is rectified using the IRS optimization algorithm. IRSA
is an advancement of RSA where the levy and cross-over methodology is utilized. Here, the
performance of IRSA optimization with existing optimization algorithms such as Particle Swarm
Optimization (PSO), Ant Colony Optimization (ACO), Fruit fly Optimization algorithm (FOA),
RSA, and IRSA are evaluated based on parameters like allocation rate, average delay and
consumption of energy. Table 1 represented below provides the performance of various
optimization algorithms with the proposed IRSA algorithm.

4.1.1. Allocation rate

Resource allocation is defined as the rate of assigning resources to the network to minimize
latency and power. The allocation rate of PSO, ACO, RSA, and IRSA is 87.66%, 92.35%,
94.96%, and 96.40% respectively. The results from the simulation show that IRSA achieved a
maximum allocation rate of 96.40% when compared with existing algorithms. Table 1 shows the
allocation rate of the proposed IRSA with the existing algorithms.

46
 International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.4, July 2023
Table 1. Performance based on the allocation of resources

Algorithms Allocation rate (%)

PSO 87.66
ACO 92.35
RSA 94.96
IRSA 96.40

The graphical representation of allocation rate of optimization algorithm is represented in figure 2

as follows:

Figure 2. Allocation rate

From Table 1 and Figure 2, it is concluded that the proposed IRSA achieved a better allocation
rate of 96.40% which is comparatively higher than the existing algorithms.

4.1.2. Average delay

The delay time is represented as the time between the change of component measured at the
source point and the destination point of the system. The average delay of PSO, ACO, RSA, and
IRSA is 0.008 ms, 0.005 ms, 0,007 ms, and 0.003 ms respectively. Table 2 shows the average
delay of the proposed IRSA with existing algorithms.

Table 2. Performance based on average delay

Algorithms Average delay (ms)

PSO 0.008
ACO 0.005
RSA 0.007
IRSA 0.003

The graphical representation of average delay of optimization algorithm is represented in figure 3

as follows:

47
 International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.4, July 2023

Figure 3. Average delay

From Table 2 and Figure 3, it is concluded that the proposed IRSA achieved a minimum delay of
0.003 ms which is comparatively higher than the existing algorithms.

4.1.3. Energy consumption

The consumption of energy in the network is defined as the quantity of consumed energy during
the time of receiving and broadcasting packets that contain data. The energy consumption of
PSO, ACO, RSA, and IRSA is 1.1 mJ, 1.3 mJ, 0.9 mJ, and 0.7 mJ respectively. From Table 3 and
Figure 4, it is concluded that the proposed IRSA consumed minimal energy of 0.7 ms which is
comparatively higher than the existing algorithms

Table 3. Performance based on energy consumption

Algorithms Energy consumption (mJ)

PSO 1.1
ACO 1.3
RSA 0.9
IRSA 0.7

The graphical representation of energy consumption is represented in figure 4:

Figure 4. Energy consumption

48
 International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.4, July 2023
From table 1 – table 3, it is shown that the performance of the IRS optimization algorithm is
better when compared with other optimization algorithms. This better performance is due to
rectifying IoT Resource Allocation Problem (IRAP) effectively using levy and crossover
methodology. In the Levy method, a random number will be assisted in updating the state to
induce oscillations, perform a quick search in one iteration, and assist in the candidate solution
leap out of local optimum and cross - over method to enhance the searchability of the candidate
solutions, the new position gathers data on the ideal solution and other potential alternatives.
IRSA achieved a better allocation of 96.40 % which is higher than other optimization algorithms.

4.2. Scalability Analysis

The scalability of the simulation depends on the number of iterations and fitness value. In this
section, the scalability of the proposed method is sensed based on varying the iteration from 100
to 1000. The figure 5 presents the scalability of the proposed IRSA in an IoT network by varying
the number of iteration.

Figure 5. Graphical representation for evaluation of fitness

The results from Figure 5 show that the proposed IRSA obtained better fitness value when
compared with existing techniques. The better result of the proposed method is due to the levy
flight cross - over technique involved in IRSA. The Levy flight outperforms quick search in a
single iteration which helps to minimize the cost and the cross - over method improvises the
ability to search the ideal position for effective allocation of resources. Figure 6 mentioned below
shows the efficiency of the proposed IRSA based on scaling the allocation rate and energy
consumption. Moreover, the scalability of the proposed method is compared with the existing
optimization techniques. At the time of evaluating the scalability, the proposed IRSA has a higher
allocation rate of 96.40% and the energy consumption is 0.7 Mj. This better result is due to
incorporating the Levy flight and crossover method in IRSA which provides stability and
robustness to the algorithm. The Levy flight method enhances the convergence in a single
iteration and the cross - over search ability for an effective allocation of resources.

49
 International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.4, July 2023

Figure 6. Graphical representation for evaluating the scalability by means of energy consumption and
allocation rate

4.3. Comparative analysis

The comparative analysis of the proposed IRSA with existing methodologies such as Scalable
Resource Allocation Framework (SRAF) [21] and Improved Chaotic Firefly Algorithm (ICFA)
[22] for the problems related to resources allocation in the environment of IoT is discussed in this
section. The comparative table of the proposed IRSA with the existing SRAF [21] and ICFA [22]
is represented in Table 4.

Table 4. Comparative table

Methods Allocation rate(%) Average delay (ms) Energy consumption (mJ)

SRAF [21] 92.40 0.004 0.8
ICFA [22] 91.67 0.006 0.10
IRSA 96.40 0.003 0.7

From the above comparative table 2, it is shown that the proposed IRSA achieved better
performance measures when compared with the existing SRAF [21] and ICFA [22] for
optimizing the problems of resource allocation in the networks of IoT. IRSA utilized Levy and
cross-over method which updates the position and enhances the search speed in a single iteration.
Thus, IRSA performs to rectify the issues in IoT resource allocation.

5. CONCLUSION
The problems related to resource allocation are a challenging task in the environment of IoT.
Resource allocation is a process of allocating an optimum resource to the tasks. The inappropriate
utilization of resources leads to the consumption of energy and affects the IoT environment. So,
resource allocation in a large network system like IoT is an important topic. In this research, the
IRSA algorithm is proposed to minimize energy consumption and reduce latency in the network
IoT. As the name indicates, IRSA is an advancement in the RSA where Levy and crossover
methodologies are utilized. In the Levy method, the random number will be assisted in updating
position to induce oscillations, make a quick search in the area during one iteration, and assist in
the candidate solution leap out of the local optimum, and the cross-over method enhances the
searchability of the candidate solutions, the new position gathers data on the ideal solution and
other potential alternatives. Thus, the IRSA is utilized in solving the optimization problem.
Experimental research shows that the proposed IRSA attained a better allocation rate of 96.40%
which is comparatively higher than SRAF (92.40 %) and ICFA (91.67 %). In the future,

50
 International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.4, July 2023
hybridization can be implemented in optimization techniques to improve the performance of
resource allocation.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

REFERENCES

[1] Mukherjee, Amrit, Pratik Goswami, Ziwei Yan, Lixia Yang & Joel JPC Rodrigues, (2019) "ADAI
and adaptive PSO-based resource allocation for wireless sensor networks", IEEE Access, Vol. 7,
pp131163-131171.
[2] Ren, Hong, Cunhua Pan, Yansha Deng, Maged Elkashlan & Arumugam Nallanathan, (2020)
"Resource allocation for secure URLLC in mission-critical IoT scenarios", IEEE Transactions on
Communications, Vol. 68, No. 9, pp5793-5807.
[3] Li, Yun, Shichao Xia, Qianying Yang, Guoyin Wang & Weiyi Zhang, (2020) "Lifetime-priority-
driven resource allocation for WNV-based internet of things", IEEE Internet of Things Journal, Vol.
8, No. 6, pp4514-4525.
[4] Kota, P.N., Chandak, A.S., Patil, B.P. (2023) “ IOT-HML: A hybrid machine learning technique for
IoT enabled industrial monitoring and control system” , Concurrency and Computation: Practice and
Experiencet,Vol 35, No. 3, pp 1-21
[5] Sun, Huaiying, Huiqun Yu, Guisheng Fan & Liqiong Chen, (2020) "Energy and time efficient task
offloading and resource allocation on the generic IoT-fog-cloud architecture", Peer-to-Peer
Networking and Applications, Vol. 13, No. 2, pp548-563.
[6] Raveendran, Neetu, Huaqing Zhang, Lingyang Song, Li-Chun Wang, Choong Seon Hong & Zhu
Han, (2022) "Pricing and resource allocation optimization for IoT fog computing and NFV: An EPEC
and matching based perspective", IEEE Transactions on Mobile Computing, Vol. 21, No. 4, pp1349-
1361.
[7] Hu, Han, Qun Wang, Rose Qingyang Hu & Hongbo Zhu, (2021) "Mobility-aware offloading and
resource allocation in a MEC-enabled IoT network with energy harvesting", IEEE Internet of Things
Journal, Vol. 8, No. 24, pp17541-17556.
[8] Mukherjee, Amrit, Pratik Goswami, Mohammad Ayoub Khan, Li Manman, Lixia Yang & Prashant
Pillai, (2021) "Energy-efficient resource allocation strategy in massive IoT for industrial 6G
applications", IEEE Internet of Things Journal, Vol. 8, No. 7, pp5194-5201.
[9] Li, Xi, Yiming Liu, Hong Ji, Heli Zhang & Victor CM Leung, (2019) "Optimizing resources
allocation for fog computing-based internet of things networks", IEEE Access, Vol. 7, pp64907-
64922.
[10] Chen, Jie, Lin Zhang, Ying-Chang Liang, Xin Kang & Rui Zhang, (2019) "Resource allocation for
wireless-powered IoT networks with short packet communication", IEEE Transactions on Wireless
Communications, Vol. 18, No. 2, pp1447-1461.
[11] Fu, Yaru, Xiaolong Yang, Peng Yang, Angus KY Wong, Zheng Shi, Hong Wang & Tony QS Quek,
(2021) "Energy-efficient offloading and resource allocation for mobile edge computing enabled
mission-critical Internet-of-Things systems", EURASIP Journal on Wireless Communications and
Networking, Vol. 2021, No. 1, p26.
[12] Wang, Jun, Daquan Feng, Shengli Zhang, An Liu & Xiang-Gen Xia, (2021) "Joint computation
offloading and resource allocation for MEC-enabled IoT systems with imperfect CSI", IEEE Internet
of Things Journal, Vol. 8, No. 5, pp3462-3475.
[13] Jain, Deepak Kumar, Sumarga Kumar Sah Tyagi, Subramani Neelakandan, Mohan Prakash &
Lakshmaiya Natrayan, (2022) "Metaheuristic optimization-based resource allocation technique for
cyber twin-driven 6G on IoE environment", IEEE Transactions on Industrial Informatics, Vol. 18,
No. 7, pp4884-4892.
[14] Hussain, Fatima, Rasheed Hussain, Alagan Anpalagan & Abderrahim Benslimane, (2020) "A new
block-based reinforcement learning approach for distributed resource allocation in clustered IoT
networks", IEEE Transactions on Vehicular Technology, Vol. 69, No. 3, pp2891-2904.
[15] Tsai, Chun-Wei, (2018) "SEIRA: An effective algorithm for IoT resource allocation problem",
Computer Communications, Vol. 119, pp156-166.

51
 International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.4, July 2023
[16] Tyagi, Sumarga Kumar Sah, Amrit Mukherjee, Shiva Raj Pokhrel & Kamal Kant Hiran, (2021) "An
intelligent and optimal resource allocation approach in sensor networks for smart agri-IoT", IEEE
Sensors Journal, Vol. 21, No. 16, pp17439-17446.
[17] Kota, P.N., Gaikwad, A.N. (2017) “ Fireflies algorithm based optimal scrambling to reduce PAPR in
SFBC based MIMO-OFDM”, International Journal on Communications Antenna and Propagation,
Vol. 7, No. 7, pp. 626–634.
[18] Reginald Jude Sixtus J and Tamilarasi Muthu (2022) “Particle Swarm Optimization–Long Short-
Term Memory based Channel Estimation with Hybrid Beam Forming PowerTransfer in WSN-IoT
Applications”, International Journal of Computer Networks & Communications, Vol. 14, No.5, pp
65-79
[19] Jiao, Jian, Yunyu Sun, Shaohua Wu, Ye Wang, and Qinyu Zhang, (2020) "Network utility
maximization resource allocation for NOMA in satellite-based Internet of Things", IEEE Internet of
Things Journal, Vol. 7, No. 4, pp3230-3242.
[20] Liu, Xiaolan, Zhijin Qin, Yue Gao & Julie A. McCann, (2019) "Resource allocation in wireless
powered IoT networks", IEEE Internet of Things Journal, Vol. 6, No. 3, pp4935-4945.
[21] Al-Makhadmeh, Zafer & Amr Tolba, (2021) "SRAF: Scalable Resource Allocation Framework using
machine learning in user-centric internet of things", Peer-to-Peer Networking and Applications, Vol.
14, No. 4, pp2340-2350.
[22] Wang, Zhiyong, Dong Liu & Alireza Jolfaei, (2020) "Resource allocation solution for sensor
networks using improved chaotic firefly algorithm in IoT environment", Computer Communications,
Vol. 156, pp91-100.
[23] Ansere, James Adu, Guangjie Han, Li Liu, Yan Peng & Mohsin Kamal, (2020) "Optimal resource
allocation in energy-efficient Internet-of-Things networks with imperfect CSI", IEEE Internet of
Things Journal, Vol. 7, No. 6, pp5401-5411.
[24] Abualigah, Laith, Mohamed Abd Elaziz, Putra Sumari, Zong Woo Geem & Amir H Gandomi, (2022)
“Reptile search algorithm (rsa): a nature-inspired metaheuristic optimizer”, Expert Systems with
Applications, Vol. 191, p116158.

AUTHORS
Prabhakar Kota received a Ph.D. degree in Electronics and Telecommunication
Engineering from SPPU University, Pune in 2018. He is now an Associate Professor at
MES College of Engineering, Pune, India. His research interests include signal
processing and wireless Communication systems.

Pravin B Chopade is currently working as an Associate Professor of E&Tc at M. E. S.

College of Engineering, Pune-01.He holds BE in Electronics, ME in Electronics
,andPh.D. in Electronics and Telecommunications Engineering and has more than 20
plus publications in reputed national and international journals. He has published a
patent based on the superresolution of images using a hardware platform. His research
papers have been widely cited, in technical papers and reference books He is a member
of IETE and ISTE. His area of interest is image processing and signal processing.

Bhagavat D. Jadhav received the B.E. degree in Electronics engineering from Pune
University, Pune India, in 2002, M.Tech. and Ph.D. degrees from Savitribai Phule Pune
University, India, in 2008 and 2017 respectively in Electronics and Telecommunication
Engineering. Since July 2005, he has been with the Department of Electronics and
Telecommunication Engineering and currently he is working as Professor and Dean
Controller of Examinations. His current research interests include Signal and Image
processing. He is a Life Member of the Indian Society for Technical Education (ISTE). He has filed two
patents and published more than twenty papers in various IEEE, Elsevier, Springer conferences and reputed
journals.

52
 International Journal of Computer Networks & Communications (IJCNC) Vol.15, No.4, July 2023
Prof. Dr.Pravin Marotrao Ghate is a graduate in Electronics &amp;
Telecommunication Engineering with a post-graduate degree of Electronics
Engineering. For his doctorate, He did research in the area of Speech Processing with
the design and development of Speech synthesis Methods for improving the quality of
syllables in Marathi Language. Prof. Dr.Pravin Ghate is a Professor at JSPM’s
Rajarshi Shahu College of Engineering, Tathwade, Pune ,and a Professor at the
Department of E & Tc Engineering. He has 17 technical published papers in National
& International Journals and Conferences. His research papers have been widely cited, in technical papers
and reference books. He is guiding 2 Doctoral students and 15 postgraduate students and is credited with
the establishment of a Project laboratory (Electronics ). He had worked on two major projects sponsored by
Matrix Automation and SPPU. His research has led to 02 patents being registered to his credit. He is
actively associated with various prestigious Education/Research organizations like Dr. Babasaheb
Marathwada Ambedkar University Aurangabad as an External examiner and Matrix Automation in Pune
for R&D activities & member of the Purchase Committee. He is also a Member of the Indian Society of
Technical Education & Society of Automotive Engineers.

Shankar Dattatray Chavan completed his Ph.D. in Electronics &

Telecommunication Engineering from Savitribai Phule Pune University, Pune, and his
Master of Engineering in the Electronics from Shivaji University Kolhapur in 2007.
Currently working as Associate Professor of Electronics and Telecommunication
Engineering Department at Dr. D. Y. Patil Institute of Technology, Pimpri, Pune, with
an overall teaching experience of over 23 years at Undergraduate and Postgraduate
level and Industrial experience of 5 years. Guided more than 12 students towards post -
graduation and currently guiding 02 research scholars toward Ph.D. More than 34 papers published in
international conferences and reputed journals. Area of interest includes Wireless Sensor Networks and
Power Electronics

53