Adaptive squirrel coyote optimization-based secured energy efficient routing technique for large scale WSN with multiple sink nodes

Abstract

In general, Wireless Sensor Networks (WSNs) require secure routing approaches for delivering the data packets to their sinks or destinations. Most of the WSNs identify particular events in their explicit platforms. However, several WSNs may examine multiple events using numerous sensors in a similar place. Multi-sink and multi-hop WSNs include the ability to offer network efficiency by securing effective data exchanges. The group of nodes in the multi-sink scenario is described through a distance vector. Though, the efficiency of multi-sink WSNs is considerably impacted by the routing of data packets and sink node placement in the cluster. In addition, many WSNs for diverse reasons existed in the similar geographical region. Hence, in this task, a secured energy-efficient routing technique is designed for a Wireless sensor network with Large-scale and multiple sink nodes. Here, the concept of an improved meta-heuristic algorithm termed Adaptive Squirrel Coyote Search Optimization (ASCSO) is implemented for selecting the accurate selection of cluster head. The fitness function regarding residual distance, security risk, energy, delay, trust, and Quality of Service (QoS) is used for rating the optimal solutions. The consumption of energy can be reduced by measuring the mean length along with the cluster head and multiple sink nodes. The latest two heuristic algorithms such as Coyote Optimization Algorithm (COA) and Squirrel Search Algorithm (SSA) are integrated for suggesting a new hybrid heuristic technique. Finally, the offered work is validated and evaluated by comparing it with several optimization algorithms regarding different evaluation metrics between the sensor and sink node.

Keywords

Large scale WSN with multiple sink nodes secured energy efficient routing technique optimal cluster head selection adaptive squirrel coyote search optimization distance residual energy delay security risk trust quality of service

1. Introduction

WSN is described as the wireless sensor and actuator network or control network by researchers. Characterization of WSN is done through sensor nodes or small-sized devices. The sensor nodes in the environment are randomly distributed and also battery-powered to monitor the data of the specific scenario communicated “wirelessly in a multi-hop communication” [1, 2]. As WSN includes several sensor nodes, they must be communicated with each other for forming the network. In recent studies, the routing paths of several WSNs are jointly independent, where the packets are forwarded with different WSN’s routes via their sensors and routing paths [3]. If a packet is received from another WSN, then the sensors drop those packets without depending on related sinks. Generally, the routing efficiency will be improved when the sensors forward the received packets to other WSNs. The shortened distance must be observed between two neighbor sensors as more sensors transmit the packets in this heterogeneous WSN, and thus, there is a need to increase the sensor density of the corresponding monitoring platform [4]. It suffers from more redundant packets in the network [5]. The baseline routing protocols for WSNs aim especially either at security, QoS, or energy efficiency problems. On the other hand, there is a need for holistic WSN architecture since most of the applications need both security and QoS, which ensures the extension of the network lifetime. Hence, there is a necessity of considering the tradeoff among security, QoS, and network lifetime owing to the restricted energy capability of sensor nodes. Moreover, one of the complicated problems of WSN is security since unreliable channels and unattended operation is vulnerable to attacks.

To get lower latency and minimize the relay workload in intermediate nodes, several sinks are located in the network and utilized together for collecting the sensed data from the sensors in a restricted number of hops [6]. Similarly, if the hop count is smaller, then the network lifetime can be increased and energy consumption can be minimized. In addition, the energy dissipation at every node can be reduced in large-scale networks by deploying several sink nodes and sensor nodes for minimizing congestion control and increasing the manageability of the network [7]. Moreover, the probability of isolated groups of nodes with multiple sinks is minimized in a sensor area, which cannot forward their data due to the adverse signal propagation scenario. In addition, multi-sink WSNs are more scalable [8]. On the other hand, the conventional routing approaches in utilizing multiple sink nodes face some complications such as delay factors, congestion control schemes, location details for every device or sensor, and a lower number of sink nodes. The network efficiency is also dependent on the route via which the data packets must be forwarded along with the sink node placement [9]. Numerous routing protocols are implemented for getting superior performance with lesser packet delay. Similarly, a huge range of spanning-tree orientations has been explored via routing algorithms [10]. Though, these techniques are only energy efficient when disappointing the packet delay in communication [11].

Many routing protocols have been designed to reduce energy consumption or maximize the energy efficiency in a WSN [12]. Though, these approaches cannot solve the issue of overusing some of the sensors, particularly the nodes who closer to the sink node in a “static network with a single sink node”. However, it results in network collapse due to the overconsumption of energy whereas the remaining nodes in the network still have a huge amount of energy [13]. Various methodologies have been implemented in WSN, where performance can be measured regarding several factors like cost, energy, and network lifetime. Some of the new clustering protocols are Simulated Annealing (SA), Optics inspired optimization (OIO), Harmony search algorithm (HSA), Gravitation Search Algorithm (GSA), “Aware Cluster-Based Routing (ACBR) protocol”, and “Low Energy Adaptive Clustering Hierarchy (LEACH)” protocol, k-means, fuzzy, etc are suggested here for enhancing the sink node deficiency [14]. Though, these techniques suffer from a few limitations like node selection, reliability, path selection, bad network performance, limited energy in sensor nodes, packet delay, cost, energy level preservation of node, etc. [15]. Therefore, the aforementioned downsides induce the investigators to concentrate on adopting a protocol for routing in WSN. In recent studies, “nature-inspired optimization algorithms” have been utilized in developing a new routing protocol, which is “Particle Swarm Optimization (PSO), Artificial Bee Colony (ABC), Ant Colony Optimization (ACO),” etc. to explore further optimization [16]. Many optimization algorithms have been broadly utilized in WSNs to validate the solutions [17]. Conversely, various balancing of energy solutions does not confirm the QoS constraints of several practical and “mission-critical applications”. Thus, this task focuses on developing the latest nature-inspired algorithm for employing routing in WSN.

The major highlights of this study are elaborated here.

•
To plan the latest WSN method to choose the accurate cluster heads using a new heuristic algorithm by solving the “multi-objective function” with the aim of optimal routing.
•
To design an ASCSO algorithm for selecting the accurate cluster heads through deriving the fitness function concerning the reduction of delay and distance, and also the improvement of residual energy, security, trust, and QoS.
•
To examine the efficiency of efficient clustering energy in the WSN model by reviewing advanced algorithms concerning different performance measures.

The residual segments of this given offered method are provided below. Module II specifies the related approaches. Module III describes the “energy efficient routing for WSN with multiple sinks”. Module IV derives the multi-objective function derived or security-aware energy-efficient routing in WSN. Module V implements the ASCSO algorithm for routing in WSN. Module VI validates the outcomes. Module VII completes this task.
2. Literature survey

2.1 Related works

In 2021, Kaseem et al. [18] designed a new WSN system for enhancing their connectivity and extending the network lifetime by suggesting an optimized energy-efficient path planning technique. This model consisted of four processes firstly the sensing area was split into similar regions according to the count of installed mobile sinks, which handled the energy-hole issue. The “Minimum Weighted Vertex Cover Problem (MWVCP)” was used with the sojourn location determination technique for finding the optimal solutions and gathering the data as of cluster heads. At last, the “Multi-Objective Evolutionary Algorithms (MOEAs)” were used for evaluating the optimized mobile sinks’ trajectories along with three optimization techniques. The empirical outcomes were conducted by the cost function of the advanced approaches, the carrying out of the time of the sinks, network lifetime, and cluster heads number.

In 2020, Hung et al. [19] suggested Energy-Efficient Cooperative Routing (EERH) scheme to save energy in heterogeneous WSNs, where numerous WSNs are installed in a similar geographical platform. Here, routing paths were determined dynamically according to the balance energy of installed sensors along with their neighbors, and the transmission direction of event packets. Additionally, the delivery energy was saved by aggregating the packets forwarded in a similar direction through a similar sensor. They have considered the network constraints of EERH in terms of the transmission distance of a sensor and event packet, which has satisfied the real-time environment requirements.

In 2020, Singha and Nagaraju [20] implemented three diverse techniques for enhancing the efficiency of a WSN regarding sink node placement through optimization and route planning by nature-inspired algorithms. Moreover, the number of transmissions has been reduced by applying opportunistic coding at possible relays. Thus, the designed model placed the sink node using PSO, and then, the ABC algorithm was used for optimizing them. Moreover, before transmitting to neighbors, the opportunistic packet amalgamation was conducted.

In 2021, Maheshwari et al. [21] suggested a new Butterfly Optimization Algorithm (BOA) for maximizing the network lifetime and minimizing the overall energy consumption in WSNs. Here, BOA was applied for choosing the cluster heads from a set of nodes. Finally, the performance evaluation was conducted. In 2020, Mehta and Saxena [22] designed a “Fuzzy Multi-criteria Clustering and Bio-inspired Energy-efficient Routing (FMCB-ER)” for maximizing the operational time and network lifetime of WSN. The reliable clusters have been produced by a grid-based clustering method. Furthermore, the selection of an accurate cluster head was done through “An adaptive Fuzzy Multi-Criteria Decision-Making (AF-MCDM)” integrating with TOPSIS and Fuzzy-AHP. This model has processed three constraints like node position, QoS, and energy. The designed model was compared and evaluated with several conventional routing techniques.

In 2021, Rathee et al. [23] designed an ACO-based “QoS aware Energy Balancing Secure Routing (QEBSR) algorithm for WSNs.” This model has computed the “trust factor of the nodes on the routing path and the end-to-end delay of transmission” by enhanced heuristics. The designed method was reviewed over existing algorithms and showed superior performance. In 2020, Wang et al. [24] suggested an improved ABC technique for selecting cluster heads in WSNs. The fuzzy C-means clustering was developed for discovering the optimal clustering approach using the enhanced ABC technique in the initial phase with a similar energy level of entire nodes. The efficiency of the designed protocol was estimated regarding different cases and showed superior performance than other conventional techniques.

In 2021, Krishnan et al. [25] suggested a new “energy-efficient cluster head selection scheme” for reducing energy consumption along with BrainStorm Optimization (BSO) algorithm. Moreover, BSO was integrated with the “Modified Teacher–Learner Optimized (MTLBO) algorithm” for increasing efficiency. This model has also enhanced the performance regarding routing overhead, death of sensor nodes, etc. The development of the offered method was evaluated with several existing algorithms and given superior performance.

In 2021, Michailidis et al. [40] initiated a three-dimensional (3-D) geometrical presentation of the Mobile Edge Computing (MEC)-enabled network, and also the author has developed an optimization approach. It was utilized to minimize the communication-aided and computation-aided Weighted Average Energy Consumption (WTEC) of vehicles. Finally, the simulation outcomes verified the effectiveness of the MIMO transmission and also provided useful engineering insights.

In 2021, Michailidis et al. [41] developed a new computation offloading approach for the Internet of Vehicles (IoV) and also presented an Unmanned Aerial Vehicle (UAV)-aided network structure. Especially, the author has a new dual-RIS configuration. An optimization approach was proposed with the purpose of minimizing the WTEC of the vehicles.

Table 1
Superiorities and downsides of traditional routing protocol with efficient energy methods in WSN

Author [citation]	Framework	Superiorities	Downsides
Kaseem et al. [18]	SEA	• The traveling distance as well as the time is reduced. • The network lifespan is prolonged.	• It has not deployed in real-world benefits.
Hung et al. [19]	EERH	• The parameters can be adjusted to fulfilling the monitoring environment needs. • The monitoring lifetime associated with the networks is lengthened.	• It does not explore security.
Singha and Nagaraju [20]	Opportunistic coding	• It guarantees enhanced evaluation metrics. • The accuracy and the energy are optimized in several applications.	• The solution is not validated on a real-time sensor network.
Maheshwari et al. [21]	BOA and ACO	• High network performance is returned. • It can enlarge the network lifespan.	• The development is not enhanced by novel optimization algorithms.
Mehta and Saxena [22]	FMCB-ER	• It provides accurate cluster heads for achieving efficient energy. • The expenditure of energy is minimized.	• It does not apply to mobile WSN applications.
Rathee et al. [23]	ACO	• The delay is minimized. • The lifetime of the network is prolonged.	• It cannot perform without weight vectors.
Wang et al. [24]	ABC	• Better performance is attained by balancing the expenditure of energy. • The energy is saved and the network throughput is enhanced.	• It cannot be considered for WSNs to mobile networks.
Krishnan et al. [25]	BSO-TLBO	• The network lifetime is increased. • The node failures and the packet loss are avoided.	• Distinct fitness functions are not considered.

2.2 Problem statement

Table 1 provides the advanced features and critical issues of traditional “energy-efficient routing protocol” methods in WSN. SEA [18] prolongs the network lifetime and also reduces the traveling distance as well as the time. But, it cannot be deployed in real-world applications. EERH [19] lengthens the monitoring lifetime associated with the networks and also adjusts the parameter for fulfilling the monitoring environment needs. Still, security is not explored. Opportunistic coding [20] optimizes the accuracy and the energy in several applications and also guarantees enhanced evaluation metrics. Yet, it does not validate the solution on a real-time sensor network. BOA and ACO [21] return high network performance and the lifetime of the network is more. But, it does not enhance the performance of novel optimization algorithms. FMCB-ER [22] “minimizes the consumption of energy and also enhances the lifetime of the network”. Still, it does not apply to mobile WSN applications. ACO [23] minimizes the delay and also prolongs the lifetime of the network. Yet, it cannot perform without weight vectors. ABC [24] saves energy, enhances the network throughput, and also achieves better performance in balancing the consumption of energy. But, it cannot be considered for WSNs to mobile networks. BSO-TLBO [25] avoids node failures and packet loss and also increases the network lifetime. Still, it does not consider distinct fitness functions. Thus, it is necessary to develop novel methods.

3. Energy-efficient routing for WSN with multiple sinks

3.1 Model of the system

The WSN is comprised of diverse sensor nodes and also sinks. The proposed model includes multi-sinks in WSN rather than a single WSN. The multi-sink WSN architecture is given in Fig. 1.

Figure 1.

The diagrammatic representation of multi-sink-based wireless sensor networks.

The multi-sink WSN has several features when compared to the single-sink WSN. If any node fails in the network, data can be forwarded to other sink nodes or base stations [26]. The utilization of several sink nodes ignores the stopped working of the interrelationship between sensor networks and the Internet. The unbalanced expenditure of energy between sensor nodes can be solved by utilizing the multiple sink nodes [38, 39]. While considering the single sink WSNs, the entire data will be forwarded to the other kinds of sensor nodes closer to the sink. Hence, the energy can be drained among these nodes. Conversely, the burden of data transmission is distributed between all the sinks in sensor networks with multi-sink. However, the selection of sink nodes by sensor node is complicated and also plays a major role in WSN in influencing the network efficiency and delivery latency [27]. While selecting the sink node, the optimal node nearer to the sink is chosen by the sensor node and then, the data has been forwarded. The multiple sinks or relay nodes can be installed for collecting data to offer better performance. In general, the conventional single-sink framework faces two main issues latency and energy-hole problems. These issues become poor when the network diameter is huge. There is a complexity in reducing the hop distance for reaching the sink and prolonging the network lifetime [28]. Multiple sinks can be deployed for reducing the hop distance. The latency can be minimized and also, and the relay workload can be decreased.

3.2 Proposed architecture

WSNs include several counts of spatially distributed sensor nodes that are linked via a wireless medium. The network lifetime can be affected as the nodes of WSN are battery-powered [29]. The sensors existing in WSN consume more energy due to the higher data transmission to a destination via the nodes and sensing of environmental constraints [34, 35]. During the data transmission, high energy consumption is observed when compared to data processing and data sense. However, node failure also fails in the network. To enlarge the development of WSN and enlarge the lifespan of WSN, there is a requirement for optimal energy usage. Thus, the clustering of sensor nodes into groups is used for increasing network scalability and minimizing network energy consumption [36, 37]. Every cluster consists of one cluster head, which communicates with other cluster heads of the network [30]. A routing protocol is utilized in the clustered WSNs. Therefore, this paper considers the multi-objective for creating the “energy-efficient WSN” owing to the network performance improvement. The offered energy-efficient WSN architecture is shown in Fig. 2.

Figure 2.

Energy efficient routing for WSN with multiple sinks.

This task develops a new secured “energy efficient routing” technique for large-scale WSNs with multiple sink nodes. Here, the ASCSO algorithm is used for selecting the accurate cluster heads. Here, the fitness functions regarding residual energy, delay, distance, security risk, trust, and QoS for getting the optimal solutions. This fitness function-derived selection of cluster heads is utilized for minimizing energy consumption. Thus, the multiple sinks with optimal cluster head selection aim on decreasing the expenditure of energy and enlarging the network lifespan.

4. Multi-objective function derived or security-aware routing with efficient energy in WSN

4.1 Fitness function of the designed method

The offered approach considers the most important aim as the accurate selection of cluster heads by the ASCSO algorithm. Here, the optimal solution is obtained by deriving the fitness function concerning constraints like distance, packet loss ratio, residual energy, delay, security risk, trust, and QoS. The main fitness function is given in Eq. (1).

$\displaystyle F=\,\mathop{\arg\min}\limits_{\left\{{\textit{CH}_{i}}\right\}}% \left({O_{5}}\right)$ (1)

Here, the selected cluster heads using the ASCSO algorithm are called as $\textit{CH}_{i}$ , where $i=1,2,\cdots,I$ and the variable $I$ denotes the average of cluster heads in the sensor network. The limit of selection of cluster heads $\textit{CH}_{i}$ is given among [1, $N$ ] and the average of sensor nodes in the WSN is given as $N$ , which also considers multiple sinks in the network. The fitness function $O_{6}$ concerning diverse variables is equated here.

$\displaystyle O_{1}=\left({\alpha\times D}\right)+\left({1-\alpha}\right)% \times\frac{1}{E}$ (2) $\displaystyle O_{2}=\left({\beta\times O_{1}}\right)+\left({1-\beta}\right)% \times\textit{De}$ (3) $\displaystyle O_{3}=\left({\gamma\times O_{2}}\right)+\left({1-\gamma}\right)% \times\frac{1}{S}$ (4) $\displaystyle O_{4}=\left({\delta\times O_{3}}\right)+\left({1-\delta}\right)% \times\frac{1}{\textit{QoS}}$ (5) $\displaystyle O_{5}=\left({\varepsilon\times O_{4}}\right)+\left({1-% \varepsilon}\right)\times\frac{1}{\textit{Tr}}$ (6)

The fitness function concerning diverse variables is included in the section, where the values of as $\alpha,\beta,\gamma$ , $\delta$ , and $\varepsilon$ are assumed as 0.2. The “solution encoding of the designed cluster head selection” using ASCSO is given in Fig. 3.

Figure 3.

Optimization of the designed cluster head selection using ASCSO.

4.2 Depiction of fitness constraints

The fitness function concerning variables is analyzed here, where the length is termed as $D$ . “Euclidean distance $D$ among two nodes” is explained as “the length of a line segment between the two nodes” as equated in Eq. (7).

$\displaystyle D=\sqrt{\left({a_{u}-b_{v}}\right)+\left({a_{u}-b_{v}}\right)}$ (7)

In Eq. (7), the term $a$ and $b$ is represented as cluster heads taken for communication, and the “corresponding coordinates to these nodes” are represented as $u$ , and $v$ . The “distance between two sensor nodes in the same cluster or other clusters” is computed for recommending accurate routing.

The residual energy is called $E$ and described as the “average remaining energy of the active sensor nodes at the end of each simulation experiment” as given in Eq. (8).

$\displaystyle E=E_{n}-\left({e_{n}^{cs}+e_{n}^{sh}}\right)$ (8)

The energy consumption by collecting the count of the data unit is specified as $e_{n}^{\textit{cs}}$ the energy consumption by transmitting the count of data units is formulated $e_{n}^{\textit{sh}}$ , and the balancing energy of any other kind of node $N_{n}$ is equated as $E_{n}$ .

Delay is determined during the communication of packets, which is according to the circulation delay and communication delay. The latency time is formulated in Eq. (9).

$\displaystyle De=\frac{\max\sum\nolimits_{i=1}^{I}{\textit{CH}_{i}}}{N}$ (9)

In Eq. (9), the “data transmission from cluster head to base station” is derived $\max\sum\nolimits_{i=1}^{I}{\textit{CH}_{i}}$ and the ratio of nodes in the sensor system is termed as $N$ .

“Security constraint $S$ is considered with three security modes such as, security mode, risky mode, and $\gamma$ -risky mode”. These modes are described below. The corresponding models are defined as follows.

Security mode: “This model selects the cluster head by satisfying the security demand. Here, the security rank and the security demand are formulated as $S_{r}$ and $S_{d}$ , respectively are correlated with the cluster head selection. When the node accomplishes the constraint $S_{d}\leqslant S_{r}$ , it requires the cluster head. Further, a secure mode is considered as the cluster head selection from the huge range of nodes seems to be a conservative technique”.

Risky mode: This mode selects any of the conventional cluster heads to enable precise selection of cluster heads, and thus, it takes the entire probability of issues. This method is also classified as “the aggressive mode in the procedure of cluster head selection”.

$\gamma$ -risky mode: These “types of risks are considered in the process of cluster head selection. Here, $\gamma$ is considered as the probability measure having extremes as $\gamma=1$ and $\gamma=0$ that is equivalent to the security and risky mode, respectively”.

The secure mode is taken as very complicated. Here, $\gamma$ -the risky mode is the rapidly used mode. The security of the variable is derived in Eq. (10).

$\displaystyle S_{\textit{risk}}=\left\{\begin{array}[]{lll}1&{\textit{if}}&\ {% 2<S_{d}-S_{r}\leqslant 5}\\ {1-e^{\frac{3\left({S_{d}-S_{r}}\right)}{2}}}&{\textit{if}}&\ {1<S_{d}-S_{r}% \leqslant}\\ {1-e^{\frac{\left({S_{d}-S_{r}}\right)}{2}}}&{\textit{if}}&\ {0<S_{d}-S_{r}% \leqslant 1}\\ 0&{\textit{if}}&\ {S_{d}-S_{r}\leqslant 0}\end{array}\right.$ (10)

In this mode, the fault rate must be in the form of less than 50%, while the preferred cluster head comes to the statement $S_{d}>S_{r}$ . The choice process is executed when the condition $0<S_{d}-S_{r}\leqslant 1$ is satisfied; the process of selection gets delayed while the condition $1<S_{d}-S_{r}\leqslant 2$ is attained. However, the consequent operative must be repetitive to reach the statement $2<S_{d}-S_{r}\leqslant 5$ when the procedure of CH selection cannot be finished, which represents the condition that the deadline has passed.

Trust is described as, “the combined characteristics model for providing the security, reliability, privacy concerning the mobility”, which is represented by $T r$ . The rate of the trusted node is measured as shown in Eq. (11), which is considered time-dependent.

$\displaystyle\textit{Tr}=\sum\limits_{i=1}^{I}{\left({\sum\limits_{i=1}^{I}{% \textit{DTr}_{\textit{CH}_{i}}}\left({\textit{CH}_{i}}\right)}\right)}+\sum% \limits_{i=1}^{I}{\left({\sum\limits_{i=1}^{I}{\textit{ITr}_{\textit{CH}_{i}}}% \left({\textit{CH}_{i}}\right)}\right)}$ (11)

In Eq. (11), the average trust of a node is shown by Tr, the “indirect trust of neighbor nodes” $N_{i}$ on $\textit{CH}_{i}$ is derived by $\textit{ITr}_{N_{i}}\left({\textit{CH}_{i}}\right)$ , and nearest nodes on $\textit{CH}_{i}$ is formulated as $\textit{DTr}_{\textit{CH}_{i}}\left({\textit{CH}_{i}}\right)$ . The “trust value related to a sensor node is measured for recognizing whether the sensor node is a benign or malicious one”.

QoS is depicted as, “the optimum number of sensors that should be sending information at any given time”. These QoS are provided as “the network users with consideration of the agreement between the service providers and the network users” as in Eq. (12).

$\displaystyle\textit{QoS }=\sum\limits_{i=1}^{I}{\textit{QoS}\left({\textit{CH% }_{i}}\right)}$ (12)

Here, the “cluster head is shown” by $\textit{CH}_{i}$ . Finally, the fitness function is formulated for selecting the accurate cluster heads in multi-sink WSN.

5. Adaptive squirrel coyote search optimization for routing in WSN

5.1 Proposed ASCSO-based routing

In this proposed model, the combination of SSA [31] and COA [32] has been utilized for proposing a new ASCSO algorithm for performing the routing in WSN. This ASCSO algorithm helps in selecting the optimal cluster heads in WSN, which assists in decreasing the expenditure of energy and enlarging the network lifespan. COA has several advantages and thus, this algorithm is chosen here for performing the routing. COA offers better solutions when estimating with other traditional techniques. It can provide globally accurate solutions owing to its improvement of efficiency and unique algorithmic structural setup. Though the given recommended algorithm is caused by premature convergence. Therefore, there is essential for offering a new heuristic algorithm using the SSA algorithm. Because it has the ability for producing accurate solutions for solving high-dimension development issues in a limited period. Hence, the integration of COA and SSA is carried out in this proposed model. ASCSO is suggested here by introducing three random numbers like $c, k, l$ using worst and best fitness solutions. Utilization of worst and best solutions helps in solving the premature convergence problem and so, this algorithm offers superior routing in this suggested model.

$\displaystyle c=\textit{abs}\left({F_{\textit{worst}}-F\left(i\right)}\right)$ (13) $\displaystyle k=\textit{abs}\left({F\left(i\right)-F_{\textit{best}}}\right)$ (14) $\displaystyle l=a*\left({\left[{\frac{c}{k}}\right]}\right)_{2}+k*\left({\left% [{\frac{k}{c}}\right]}\right)_{2}$ (15)

The bad fitness result is noted as $F_{\textit{worst}}$ , the greatest fitness result is noted as $F_{\textit{best}}$ , the recent fitness solution is mentioned as $F\left(i\right)$ , where if $l==c$ , then the positions are upgraded by COA or else the given obtained results are enhanced by SSA.

COA is the inspired swarm intelligence approach, which is developed using the swarm behavior of coyotes and the social existence characteristics of the canis latrans species in North America. Initially, the COA population is assigned and it splits into the same count of coyotes (solutions) per group, where each group represents the social condition and each result location is derived as a probable number of solutions. At first, the locations of the solution are assigned arbitrarily as derived in Eq. (16), where the average results per group are limited to 14 in COA for attaining the accurate exploration ability of the algorithm.

$\displaystyle Z_{i}=lb_{i}+rs_{i}\times\left({ub_{i}-lb_{i}}\right)$ (16)

In Eq. (16), the “lower bound and upper bound” of the search space is indicated as $lb_{i}$ and $ub_{i}$ , and the location of a solution at $i^{th}$ dimension is represented as $Z_{i}$ . The better coyote is considered according to the best coyote recommended in the environment with the minimum cost function achieved from the reduction issue. These packs are maintained based on the management of solutions that helps in allocating the social situations. Additionally, the social disposition of the group is formulated in Eq. (17).

$\displaystyle X_{i}^{g,t}=\left\{{{\begin{array}[]{ll}{H_{\frac{\left({\textit% {Np}_{h}+1}\right)}{2}}^{g,t},}\hfill&{\textit{Np}_{s}\,\textit{is odd}}\hfill% \\ {\frac{H_{\left({\frac{\textit{Np}_{h}}{2}+1}\right),i}^{g,t}}{2}}\hfill&{% \textit{or}\,\,\textit{else}}\hfill\\ \end{array}}}\right.$ (17)

In Eq. (17), the term $X_{i}^{g,t}$ specifies the social disposition of the implicated solution in $g^{\textit{th}}$ the pack at a time $t$ in the range of [1, $D i m$ ], the grade basis of coyotes is noted as $H$ , and the average of solutions is represented as $Np_{h}$ . In addition, in Eq. (18), the birth of a novel solution is equated by assuming the death and birth of solutions.

$\displaystyle Y_{i}^{g,t}=\left\{{\begin{array}[]{ll}{Z_{rs1,i}^{g,t},}\hfill&% {\textit{rs}_{i}\geqslant\rho_{\textit{sp}}+\rho_{\textit{ap}}\,\,\textit{or}% \,i=i_{1}}\hfill\\ {Z_{rs2,i}^{g,t},}\hfill&{\textit{rs}_{i}<\rho_{\textit{sp}}\,\textit{or}\,i=i% _{2}}\hfill\\ {\textit{RS}_{i},}\hfill&{\textit{otherwise}}\end{array}}\right.$ (18)

Here, the association probabilities and the scatter distribution are indicated by $\Re_{\textit{ap}}$ and $\Re_{\textit{sp}}$ that are formulated below, and the two solutions are arbitrarily selected from $g^{\textit{th}}$ the pack, and are indicated as $\textit{rs}1$ and $\textit{rs}2$ , respectively.

$\displaystyle\Re_{\textit{sp}}=\frac{1}{\textit{Dim}}$ (19) $\displaystyle\Re_{\textit{ap}}=\frac{\left({1-\Re_{\textit{sp}}}\right)}{2}$ (20)

The term $g^{\textit{th}}$ pack in every iterative is upgraded as shown in Eq. (21).

$\displaystyle Z_{H}^{g,t+1}=\left\{{{\begin{array}[]{ll}{Z_{H}^{g,t}+\textit{% rs}1\times\kappa_{1}+\textit{rs}2\times\kappa_{2},}\hfill&{F_{H}^{g,t+1}<F_{H}% ^{g,t}}\\ {Z_{H}^{g,t}}\hfill&{\textit{or}\,\,\textit{else}}\\ \end{array}}}\right.$ (21)

Especially, the alpha induces and the group induces are called as $\kappa_{2}$ and $\kappa_{1}$ are formulated in Eqs (22) and (23).

$\displaystyle\kappa_{1}=\textit{alpha}^{g,t}-Z_{\textit{rs}1}^{g,t}$ (22) $\displaystyle\kappa_{2}=X^{g,t}-Z_{rs2}^{g,t}$ (23)

In Eq. (22), the alpha solution is denoted as $\textit{alpha}^{g,t}$ , and the social circumstance cost is mentioned as $F_{H}^{g,t+1}$ and given in Eq. (24).

$\displaystyle F_{H}^{g,t+1}=f\left({Z_{H}^{g,t}}\right)$ (24)

If the $l==c$ statement is not fulfilled, then the positions are improved through the SSA.

SSA is inspired by taking the jumping and gliding schemes of the solutions. The whole improvement procedure is conducted in the summer and winter phases. At first, the population count is indicated as $P o$ and the lower and upper bound of the search space is indicated as $Z_{\textit{Up}}$ and $Z_{\textit{Lw}}$ . Entire solitary in the population is produced with the help of Eq. (25).

$\displaystyle Z_{i}=Z_{\textit{Lw}}+rn\left({1,ds}\right)\times\left({Z_{% \textit{Up}}-Z_{\textit{Lw}}}\right)$ (25)

In the afore-mentioned equation, the arbitrary number with the bounding range of 0 to 1 is denoted as the term rn, where the $i^{\textit{th}}$ solitary is noted as $Z_{i}$ and the proportion of the search space is given as ds. Here, the required solutions are classified according to the cost function of the population with the lowest fitness, and the solitaries are positioned at acorn trees $F_{\textit{ac}}$ and normal trees $F_{\textit{nt}}$ . Moreover, the foraging characteristic is equated here. The spot of the “flying squirrels in the acorn nut trees” $F_{\textit{ac}}$ hickory to jumps nut tree. Here, the novel location of the solutions is derived in Eq. (26), where the present iteration is precise as $i$ .

$\displaystyle Z_{\textit{ac}}^{i+1}=\left\{{{\begin{array}[]{ll}{Z_{\textit{ac% }}^{i}+\textit{rj}\cdot\textit{gh}\left({Z_{\textit{hk}}^{i}-Z_{\textit{ac}}^{% i}}\right)}\hfill&{\textit{if}\,q_{1}\geqslant\textit{PG}_{\textit{pr}}}\hfill% \\ {\textit{random}\,\textit{position}}\hfill&{\textit{others}}\hfill\\ \end{array}}}\right.$ (26)

Here, the term gh is denoted as the gliding constant and the arbitrary function $q_{1}$ is determined from the interval of $\left[{0,1}\right]$ . The arbitrary gliding distance is shown by rj and equated in Eq. (27).

$\displaystyle\textit{rd}_{G}=\frac{hs_{G}}{\tan\left(\varphi\right)\times cr}$ (27)

Terms cr and $\textit{hs}_{G}$ are considered as the stable values and the variable $\tan\left(\varphi\right)$ gliding angle is calculated in Eq. (28).

$\displaystyle\tan\left(\varphi\right)=\frac{\textit{Fr}}{\textit{Lf}}$ (28)

The afore-mentioned equation, the term Fr, is indicated as the drag force, and the variable Lf is described as the lift force and these two forces are formulated below.

$\displaystyle\textit{Fr}=\frac{1}{2\rho v^{2}\textit{jrCn}_{\textit{ds}}}$ (29) $\displaystyle\textit{Lf}=\frac{1}{2\rho v^{2}\textit{jrCn}_{\textit{lt}}}$ (30)

Here, the variables $\rho$ , $v$ , $j r$ , $\textit{Cn}_{\textit{lt}}$ and $\textit{Cn}_{\textit{ds}}$ are shown as the constant values. Various counts of outcomes are located on “the normal trees”, which are repositioned to the “acorn nut trees” for accessing the food needs, here the novel location is specified by Eq. (31).

$\displaystyle Z_{\textit{nt}}^{i+1}=\left\{{{\begin{array}[]{ll}{Z_{\textit{nt% }}^{i}+rj\cdot\textit{gh}\left({Z_{\textit{ac}}^{i}-Z_{\textit{nt}}^{i}}\right% )}\hfill&{\textit{if}\,q_{2}\geqslant\textit{PG}_{\textit{pr}}}\hfill\\ {\textit{random}\,\textit{position}}\hfill&{\textit{others}}\hfill\\ \end{array}}}\right.$ (31)

Here, the arbitrary function that is set from the limit $\left[{0,1}\right]$ is given as $q_{2}$ . Likewise, a few solutions in the regular tree are relocated its location to the hickory nut tree, where the latest location of the location is calculated in Eq. (32).

$\displaystyle Z_{\textit{nt}}^{i+1}=\left\{{{\begin{array}[]{ll}{Z_{\textit{nt% }}^{i}+rj\cdot\textit{gh}\left({Z_{\textit{hk}}^{i}-Z_{\textit{nt}}^{i}}\right% )}\hfill&{\textit{if}\,q_{3}\geqslant\textit{PG}_{\textit{pr}}}\hfill\\ {\textit{random}\,\textit{position}}\hfill&{\textit{others}}\hfill\\ \end{array}}}\right.$ (32)

The variable $q_{3}$ specifies the arbitrary function in the interval of $\left[{0,1}\right]$ .

Seasonal monitoring situations estimation: This process helps to ignore the algorithm for resolving the cost function issue. Here, the term SC seasonal constant and the corresponding lowest rates are computed in Eqs (33) and (34).

$\displaystyle\textit{SC}_{\textit{cn}}^{i}=\sqrt{\sum\limits_{ks=1}^{\textit{% ds}}{\left({Z_{\textit{ac,ks}}^{i}-Z_{\textit{ht,ks}}^{i}}\right)}^{2}}$ (33) $\displaystyle\textit{SC}_{\textit{cn}\min}=\frac{10e^{-6}}{365^{\frac{i}{\frac% {\left({i_{\max}}\right)}{2.5}}}}$ (34)

Here, the average iteration is specified as $i$ , and the highest count of iteration is given by $i_{\max}$ . The season is computed via the statement $\textit{SC}_{\textit{cn}}^{i}<\textit{SC}_{\textit{cn}\min}$ . When $\textit{SC}_{\textit{cn}}^{i}<\textit{SC}_{\textit{cn}\min}$ is fulfilled, then the winter season is started, or else the season will be unstable. When the summer season came, all solitaries upgrade their location to $F_{\textit{hk}}$ that is given in Eq. (35).

$\displaystyle Z_{\textit{nt}}^{i+1}=Z_{\textit{Lw}}+\textit{LEVY}\left({ns}% \right)\times\left({Z_{\textit{Up}}-Z_{\textit{Lw}}}\right)$ (35)

The levy distribution is formulated in Eq. (36).

$\displaystyle\textit{LEVY}\left({\textit{sd}}\right)=0.01\times\frac{\textit{% di}_{\textit{ps}}\times\sigma}{\left|{\textit{di}_{\textit{qs}}}\right|^{1/% \beta}}$ (36)

The two diverse functions are indicated by $\textit{di}_{\textit{ps}}$ and $\textit{di}_{\textit{qs}}$ that is in the limit of $\left[{0,1}\right]$ and the stable value is denoted by $\beta$ and also the variable $\sigma$ is calculated by Eq. (37).

$\displaystyle\sigma=\left({\frac{\Gamma\left({1+\beta}\right)\times\sin\left({% \frac{\pi\beta}{2}}\right)}{\Gamma\left({\frac{\left({1+\beta}\right)}{2}}% \right)\times\beta\times 2^{\left({\frac{\left({\beta-1}\right)}{2}}\right)}}}\right)$ (37)

At last, the ASCSO algorithm ends when “the maximum count of iterations is fulfilled”. The flow diagram of the ASCSO algorithm is shown in Algorithm 1.

Algorithm 1: Designed ASCSO algorithm

The initial population and its input parameters are generated

Estimate the fitness of position of every individual

Set the arbitrary numbers

c, k, l

according to the best and worst fitness solutions

Arrange the squirrel position according to the fitness

While (stopping condition) do

l==c

The solution is upgraded according to COA

for each coyote in the pack

Set the social bias by Eq. (17)

The novel situation is validated by Eq. (21)

End for

Death and Birth by Eq. (18)

Else

The solution was upgraded according to SSA

For

i=1\,\textit{to}\,ns_{1}

q_{1}\geqslant\textit{PG}_{\textit{pr}}

The solutions are upgraded by Eq. (26)

Else

The solutions are upgraded by Eq. (26)

End if

End for

For

i=1\,to\,ns_{2}

q_{2}\geqslant PG_{pr}

The solutions are upgraded by Eq. (31)

Else

The solutions are upgraded by Eq. (31)

End if

End for

For

i=1\,to\,ns_{3}

q_{3}\geqslant\textit{PG}_{\textit{pr}}

The solutions are upgraded by Eq. (33)

Else

The solutions are upgraded by Eq. (33)

End if

End for

End while

\textit{SC}_{\textit{cn}}^{i}<\textit{SC}_{\textit{cn}\min}

Change the flying squirrel arbitrarily by Eq. (35)

End if

The lowest value

SC_{\textit{cn}\min}

is upgraded by Eq. (33)

End while

Return best solutions

Obtain a best optimal solution

Therefore, the accurate solutions concerning the selection of cluster heads are achieved by the ASCSO algorithm, and the flow diagram of the offered ASCSO algorithm is depicted in Fig. 4.

5.2 Step-by-step process for routing

The sequential process of the designed routing model is given here.

1.
Start the procedure
2.
Initialization of cluster head
3.
Utilize the ASCSO algorithm and solve the multi-objective function for selecting the cluster heads.
4.
Multi-objective fitness evaluation based on distance, packet loss ratio, residual energy, delay, security risk, trust, and QoS.
5.
If $l==c$ is satisfied.
6.
The solution is enhanced according to COA using Eq. (23)
7.
Else, the solutions are enhanced according to SSA.
8.
Find the best cluster head with minimum fitness value.
9.
If the optimal solution is reached then stop the procedure or else start step 2.

6. Calculations of results

6.1 Experimental calculation

Table 2
Experimental constraints of secured routing with energy efficient approaches for large scale wireless sensor networks with multiple sink nodes

Parameters	Values
Number of sink nodes	[3, 4, 5, 8]
Maximum number of iterations	10
Number of rounds	2000
Area	100 $\times$ 100 m
Number of sensor nodes	50

Figure 4.

Flow diagram of ASCSO algorithm for routing.

Figure 5.

Validation of the number of alive nodes of the proposed routing technique for WSN with multiple sink nodes by differing the number of sink nodes as “(a) 3, (b) Zoom in the image of (a), (c) 4, (d) Zoom in the image of (c), (e) 5, (f) Zoom in the image of (e), (g) 8, (h) Zoom in the image of (g)”.

Figure 6.

Evaluation of normalized energy of the proposed secured routing technique for largescale WSN with multiple sink nodes by varying “the number of sink nodes as (a) 3, (b) Zoom in the image of (a), (c) 4, (d) Zoom in the image of (c), (e) 5, (f) Zoom in the image of (e), (g) 8, (h) Zoom in the image of (g)”.

The suggested “energy-efficient clustering protocol in WSN” was designed in MATLAB 2020a. Here, the suggested model was evaluated by estimating with the traditional systems like the Butterfly Optimization Algorithm (BOA) [30], Jaya Algorithm (JA) [33], SSA [31], and COA [32] regarding convergence analysis, number of alive nodes, normalized energy, energy efficiency, etc. The experimental constraints are depicted in Table 2.

6.2 Evaluation of alive nodes

Evaluation of offered “energy efficient clustering protocol in WSN” was analyzed regarding the average of alive nodes by varying the count of rounds as given in Fig. 5. The better efficiency is noticed by the designed ASCSO algorithm. However, COA has superior performance while considering the number of rounds from 1800 to 2000 rounds. Similarly, SSA and BOA also overtake the performance. Hence, there is a need of improving the network efficiency.

6.3 Estimation of normalized energy

Validation of the designed “energy efficient protocol” regarding normalized energy is given in Fig. 6. Here, the higher normalized energy is attained by the proposed ASCSO algorithm while comparing it to the other algorithms from the initial number of rounds itself. Thus, the maximum efficiency is achieved by ASCSO in terms of normalized energy.

Figure 7.

Validation on the delay of the offered secured routing technique for largescale WSN by differing the number of sink nodes as (a) 3, (b) 4, (c) 5, and (d) 8.

6.4 Analysis of delay

Figure 8.

Validation of distance of the offered secured routing technique for largescale WSN by varying the number of sink nodes as “(a) 3, (b) 4, (c) 5, and (d) 8”.

Figure 9.

Validation of energy of the proposed secured routing technique for largescale WSN by differing the number of sink nodes as “(a) 3, (b) 4, (c) 5, and (d) 8”.

Figure 10.

Analysis of QoS of the proposed secured routing technique for largescale WSN by differing the number of sink nodes as “(a) 3, (b) 4, (c) 5, and (d) 8”.

Figure 11.

Analysis of security of the offered secured routing technique for largescale WSN by varying the number of sink nodes as “(a) 3, (b) 4, (c) 5, and (d) 8”.

Table 3

Statistical validation of the offered secured energyefficient routing technique concerning several dead nodes

Statistics	BOA [30]	JA [33]	SSA [31]	COA [32]	ASCSO
Number of sink nodes as 3
“Best”	8.976	5.684	6.431	7.246	6.493
“Standard deviation”	1.5013	1.5343	1.5013	1.5171	1.5343
“Median”	0	0	0	0	0
“Mean”	2.3011	2.2211	2.2711	2.2511	2.2011
“Worst”	2	1	3	1	2
Number of sink nodes as 4
“Worst”	2	2	2	2	2
“Standard deviation”	0.15817	0.15343	0.15013	0.15343	0.16292
“Mean”	2.2011	2.2011	2.2011	2.2511	2.2011
“Median”	0	0	0	0	0
“Best”	0	0	0	0	0
Number of sink nodes as 5
“Standard deviation”	1.5343	1.4676	1.4838	1.5343	1.4676
“Worst”	2	1	1	2	1
“Mean”	2.2011	2.2011	2.2011	2.2511	2.2011
“Median”	0	0	0	0	0
“Best”	0	0	0	0	0
Number of sink nodes as 8
“Mean”	2.2011	2.2011	2.2011	2.2511	2.2011
“Worst”	2	2	2	2	2
“Median”	0	0	0	0	0
“Standard deviation”	1.5171	1.5982	1.5343	1.5013	1.5665
“Best”	0	0	0	0	0

Figure 12.

Evaluation of trust of the proposed secured routing technique for largescale WSN by differing the number of sink nodes as “(a) 3, (b) 4, (c) 5, and (d) 8”.

The proposed “energy efficient protocol” is validated regarding delay by differing the count of rounds given in Fig. 7. While taking the number of sink nodes as 3 and the number of rounds as 400, the ASCSO algorithm gets a minimum delay than other heuristic algorithms, which is 62.5%, 31.8%, 61.5%, and 28.5% elevated than BOA, JA, SSA, and COA. Hence, the minimum rate in terms of delay with higher performance is observed by the ASCSO algorithm.

Table 4

Statistical analysis of the designed secured energyefficient routing technique in terms of normalized energy

Statistics	BOA [30]	JA [33]	SSA [31]	COA [32]	ASCSO
Number of sink nodes as 3
“Best”	4	3	5	2	1
“Worst”	509	509	509	509	309
“Mean”	256	256	257	257	255
“Median”	386	394	392	395	393
“Standard deviation”	2	202	201	2	202
Number of sink nodes as 4
“Best”	0	0	0	0	0
“Worst”	52	501	52	501	52
“Mean”	258	259	258	257	256
“Median”	382	392	375	375	374
“Standard deviation”	199	195	199	194	199
Number of sink nodes as 5
“Best”	0	0	0	0	0
“Worst”	506	506	506	501	501
“Mean”	256	0.000254	256	254	254
“Median”	378	386	391	39	381
“Standard deviation”	195	196	197	197	197
Number of sink nodes as 8
“Best”	0	0	0	0	0
“Worst”	54	51	54	59	51
“Mean”	256	255	256	257	254
“Median”	392	393	39	392	386
“Standard deviation”	2	202	198	199	201

Table 5

Statistical validation of the offered secured energyefficient routing technique in terms of Distance

Statistics	BOA [30]	JA [33]	SSA [31]	COA [32]	ASCSO
Number of sink nodes as 3
“Best”	249.03	229.58	258.62	220.22	249.98
“Worst”	500.21	482.64	530.22	482.39	509.81
“Mean”	364.93	366.8	363.84	364.67	366.61
“Median”	365.59	367.75	364.68	364.66	367.43
“Standard deviation”	32.829	32.22	32.07	33.557	32.714
Number of sink nodes as 4
“Best”	227	234.69	253.86	232.89	222.26
“Worst”	522.46	484.33	505.18	463.3	479.27
“Mean”	356.86	360.99	360.5	332.45	338.24
“Median”	357.9	362.15	361.84	332.22	337.83
“Standard deviation”	34.043	33.166	33.292	32.163	34.127
Number of sink nodes as 5
“Best”	250.07	232.06	254.63	215.6	219.03
“Worst”	469.31	449.23	524.58	441.67	452.54
“Mean”	339.9	341.7	338.1	327.84	328.51
“Median”	339.27	340.46	337.45	327.53	328.76
“Standard deviation”	29.227	28.534	29.505	28.989	29.218
Number of sink nodes as 8
“Best”	243.44	241.74	228.49	247.88	236.09
“Worst”	488.72	477.13	475.53	464.96	481.16
“Mean”	359.77	352.46	355.96	359.26	352.66
“Median”	360.73	352.76	356.04	359.42	353.34
“Standard deviation”	31.134	31.189	31.915	30.856	30.867

Table 6

Statistical validation of the offered secured energyefficient routing technique in terms of the number of residual energy

Statistics	BOA [30]	JA [33]	SSA [31]	COA [32]	ASCSO
Number of sink nodes as 3
“Standard deviation”	2.29 $\times$ 10 ${}^{-23}$	2.32 $\times$ 10 ${}^{-23}$	2.31 $\times$ 10 ${}^{-23}$	2.33 $\times$ 10 ${}^{-23}$	2.42 $\times$ 10 ${}^{-23}$
“Worst”	1.00 $\times$ 10 ${}^{-22}$	1.00 $\times$ 10 ${}^{-22}$	1.00 $\times$ 10 ${}^{-22}$	1.00 $\times$ 10 ${}^{-22}$	1.3 $\times$ 10 ${}^{-22}$
“Mean”	1.39 $\times$ 10 ${}^{-22}$	1.39 $\times$ 10 ${}^{-22}$	1.39 $\times$ 10 ${}^{-22}$	1.40 $\times$ 10 ${}^{-22}$	1.60 $\times$ 10 ${}^{-22}$
“Best”	1.79 $\times$ 10 ${}^{-22}$	1.80 $\times$ 10 ${}^{-22}$	1.80 $\times$ 10 ${}^{-22}$	1.80 $\times$ 10 ${}^{-22}$	1.89 $\times$ 10 ${}^{-22}$
“Median”	1.40 $\times$ 10 ${}^{-22}$	1.39 $\times$ 10 ${}^{-22}$	1.39 $\times$ 10 ${}^{-22}$	1.41 $\times$ 10 ${}^{-22}$	1.60 $\times$ 10 ${}^{-22}$
Number of sink nodes as 4
“Mean”	1.40 $\times$ 10 ${}^{-22}$	1.39 $\times$ 10 ${}^{-22}$	1.40 $\times$ 10 ${}^{-22}$	1.39 $\times$ 10 ${}^{-22}$	1.59 $\times$ 10 ${}^{-22}$
“Worst”	1.00 $\times$ 10 ${}^{-22}$	1.00 $\times$ 10 ${}^{-22}$	1.00 $\times$ 10 ${}^{-22}$	1.00 $\times$ 10 ${}^{-22}$	1.00 $\times$ 10 ${}^{-22}$
“Standard deviation”	2.35 $\times$ 10 ${}^{-23}$	2.32 $\times$ 10 ${}^{-23}$	2.26 $\times$ 10 ${}^{-23}$	2.31 $\times$ 10 ${}^{-23}$	2.33 $\times$ 10 ${}^{-23}$
“Median”	1.39 $\times$ 10 ${}^{-22}$	1.39 $\times$ 10 ${}^{-22}$	1.39 $\times$ 10 ${}^{-22}$	1.40 $\times$ 10 ${}^{-22}$	1.70 $\times$ 10 ${}^{-22}$
“Best”	1.79 $\times$ 10 ${}^{-22}$	1.79 $\times$ 10 ${}^{-22}$	1.79 $\times$ 10 ${}^{-22}$	1.79 $\times$ 10 ${}^{-22}$	1.80 $\times$ 10 ${}^{-22}$
Number of sink nodes as 5
“Worst”	1.00 $\times$ 10 ${}^{-22}$	1.00 $\times$ 10 ${}^{-22}$	1.00 $\times$ 10 ${}^{-22}$	1.00 $\times$ 10 ${}^{-22}$	1.3 $\times$ 10 ${}^{-22}$
“Standard deviation”	2.32 $\times$ 10 ${}^{-23}$	2.29 $\times$ 10 ${}^{-23}$	2.35 $\times$ 10 ${}^{-23}$	2.26 $\times$ 10 ${}^{-23}$	2.40 $\times$ 10 ${}^{-23}$
“Mean”	1.39 $\times$ 10 ${}^{-22}$	1.41 $\times$ 10 ${}^{-22}$	1.40 $\times$ 10 ${}^{-22}$	1.40 $\times$ 10 ${}^{-22}$	1.61 $\times$ 10 ${}^{-22}$
“Median”	1.39 $\times$ 10 ${}^{-22}$	1.41 $\times$ 10 ${}^{-22}$	1.40 $\times$ 10 ${}^{-22}$	1.41 $\times$ 10 ${}^{-22}$	1.62 $\times$ 10 ${}^{-22}$
“Best”	1.80 $\times$ 10 ${}^{-22}$	1.79 $\times$ 10 ${}^{-22}$	1.79 $\times$ 10 ${}^{-22}$	1.80 $\times$ 10 ${}^{-22}$	1.89 $\times$ 10 ${}^{-22}$
Number of sink nodes as 8
“Standard deviation”	2.35 $\times$ 10 ${}^{-23}$	2.32 $\times$ 10 ${}^{-23}$	2.29 $\times$ 10 ${}^{-23}$	2.30 $\times$ 10 ${}^{-23}$	2.27 $\times$ 10 ${}^{-23}$
“Worst”	1.00 $\times$ 10 ${}^{-22}$	1.00 $\times$ 10 ${}^{-22}$	1.00 $\times$ 10 ${}^{-22}$	1.00 $\times$ 10 ${}^{-22}$	1.3 $\times$ 10 ${}^{-22}$
“Best”	1.79 $\times$ 10 ${}^{-22}$	1.80 $\times$ 10 ${}^{-22}$	1.79 $\times$ 10 ${}^{-22}$	1.80 $\times$ 10 ${}^{-22}$	1.89 $\times$ 10 ${}^{-22}$
“Mean”	1.40 $\times$ 10 ${}^{-22}$	1.39 $\times$ 10 ${}^{-22}$	1.40 $\times$ 10 ${}^{-22}$	1.41 $\times$ 10 ${}^{-22}$	1.60 $\times$ 10 ${}^{-22}$
“Median”	1.40 $\times$ 10 ${}^{-22}$	1.39 $\times$ 10 ${}^{-22}$	1.39 $\times$ 10 ${}^{-22}$	1.42 $\times$ 10 ${}^{-22}$	1.60 $\times$ 10 ${}^{-22}$

Table 7

Statistical validation of the offered secured energyefficient routing technique in terms of delay

Statistics	BOA [30]	JA [33]	SSA [31]	COA [32]	ASCSO
Number of sink nodes as 3
“Best”	1.6744	1.6759	1.6709	1.6753	1.6704
“Worst”	9.8769	9.8791	9.8752	9.8736	9.8748
“Mean”	5.7321	5.7846	5.7504	5.7483	5.6912
“Median”	5.7164	5.8005	5.7548	5.7498	5.5804
“Standard deviation”	2.3604	2.3848	2.3645	2.3501	2.3712
Number of sink nodes as 4
“Best”	1.6733	1.673	1.679	1.674	1.6827
“Worst”	9.8702	9.874	9.8757	9.876	9.8733
“Mean”	5.8129	5.7531	5.8643	5.6994	5.8031
“Median”	5.8325	5.6472	5.8585	5.6319	5.8226
“Standard deviation”	2.3567	2.3536	2.3404	2.3519	2.3547
Number of sink nodes as 5
“Best”	1.6827	1.6735	1.6737	1.6792	1.679
“Worst”	9.8711	9.879	9.8709	9.8782	9.8791
“Mean”	5.7125	5.7766	5.629	5.7865	5.7983
“Median”	5.6751	5.7525	5.6414	5.8089	5.8574
“Standard deviation”	2.3807	2.3595	2.3782	2.378	2.3569
Number of sink nodes as 8
“Best”	1.6813	1.6725	1.6772	1.6729	1.6758
“Worst”	9.8737	9.8795	9.8791	9.8694	9.877
“Mean”	5.7538	5.8123	5.7201	5.7888	5.6981
“Median”	5.7812	5.8713	5.6845	5.7783	5.6648
“Standard deviation”	2.3098	2.3811	2.3683	2.371	2.3319

Table 8

Statistical validation of the offered secured energyefficient routing technique in terms of security

Statistics	BOA [30]	JA [33]	SSA [31]	COA [32]	ASCSO
Number of sink nodes as 3
“Best”	100.06	100.39	100.12	100.47	100.23
“Worst”	499.87	499.87	499.41	499.84	499.74
“Mean”	299.86	301.56	298.79	300.36	299.45
“Median”	299.89	303.36	298.44	298.68	298.11
“Standard deviation”	115.64	117.34	114.87	117.29	113.9
Number of sink nodes as 4
“Best”	100.41	100.23	100.36	100.04	100.13
“Worst”	499.99	499.76	499.86	499.96	499.91
“Mean”	299.55	300.03	300.23	293.4	297.05
“Median”	296.98	299.57	300.29	288.91	297.22
“Standard deviation”	117	117.96	116.44	115.65	115.96
Number of sink nodes as 5
“Best”	100.13	100.03	100.13	100.36	100.01
“Worst”	499.44	499.93	499.32	499.97	499.8
“Mean”	299.78	299.1	303.49	302.79	301.57
“Median”	300.11	298.37	305.45	303.83	305.12
“Standard deviation”	115.61	115.23	114.72	115.09	117.44
Number of sink nodes as 8
“Best”	100.05	100.04	100.58	100.04	100.19
“Worst”	499.71	499.99	499.4	499.97	499.66
“Mean”	302.66	301.07	299.92	299.9	302.4
“Median”	305.73	301.5	297.49	302.19	305.37
“Standard deviation”	116.27	118.39	117.57	114.74	115.91

Table 9

Statistical validation of the offered secured energyefficient routing technique in terms of QoS

Statistics	BOA [30]	JA [33]	SSA [31]	COA [32]	ASCSO
Number of sink nodes as 3
“Best”	1.4286	1.4286	1.4286	1.4286	1.4286
“Worst”	4.1667	3.5714	4.7619	4	4.3478
“Mean”	8.9147	8.9292	9.0584	8.9005	8.9979
“Median”	9.0909	9.0909	9.0909	9.0909	9.0909
“Standard deviation”	1.604	1.6072	1.5813	1.5894	1.5301
Number of sink nodes as 4
“Best”	1.4286	1.25	1.25	1.25	1.25
“Worst”	3.8462	3.7037	3.7037	4	4.3478
“Mean”	8.5155	8.4839	8.464	8.2941	8.2969
“Median”	8.3333	8.3333	8.3333	8.3333	8.3333
“Standard deviation”	1.5975	1.5335	1.5989	1.5303	1.4838
Number of sink nodes as 5
“Best”	1.25	1.25	1.25	1.25	1.25
“Worst”	4.3478	4.1667	4.1667	4.5455	4.3478
“Mean”	8.221	8.2342	8.2334	8.6138	8.677
“Median”	8.3333	8.3333	8.3333	8.3333	8.3333
“Standard deviation”	1.4227	1.3889	1.4167	1.4391	1.4308
Number of sink nodes as 8
“Best”	1.25	1.1111	1.25	1.25	1.25
“Worst”	3.5714	4.1667	4.1667	4.1667	4.1667
“Mean”	8.0414	7.9821	7.9401	8.0142	8.0423
“Median”	8.3333	7.6923	7.6923	8.3333	8.3333
“Standard deviation”	1.4564	1.3922	1.3857	1.3697	1.3527

Table 10

Statistical validation of the offered secured energyefficient routing technique in terms of Trust

Statistics	BOA [30]	JA [33]	SSA [31]	COA [32]	ASCSO
Number of sink nodes as 3
“Best”	110.06	110.03	110	110.06	110.05
“Worst”	198.92	198.97	198.96	198.97	198.95
“Mean”	153.47	155.22	154.31	154.78	154.77
“Median”	151.71	155.04	154.01	154.87	155.29
“Standard deviation”	25.053	25.863	25.877	25.4	25.58
Number of sink nodes as 4
“Best”	110.14	110.02	110.03	110.08	110.02
“Worst”	198.98	199	199	198.99	198.98
“Mean”	154.85	154.85	155.14	154.74	154.66
“Median”	155.16	155.11	155.13	154.97	155.36
“Standard deviation”	25.562	25.715	25.472	25.695	25.324
Number of sink nodes as 5
“Best”	110.18	110.11	110.12	110.06	110.08
“Worst”	198.85	198.97	198.93	199	198.96
“Mean”	155.62	153.89	155.03	153.98	154.46
“Median”	156.47	154	154.55	154.61	154.84
“Standard deviation”	25.489	25.247	25.593	25.751	26.135
Number of sink nodes as 8
“Best”	110.03	110.04	110.03	110.15	110.09
“Worst”	199	198.95	198.99	198.96	199
“Mean”	154.57	156.09	154.14	153.28	155.02
“Median”	154.91	157.13	153.97	151.94	155.31
“Standard deviation”	25.071	25.615	25.597	25.396	25.246

6.5 Evaluation of distance

Validation of the designed cluster head selection method is carried out in terms of distance as shown in Fig. 7. The elevated performance concerning distance is observed by the ASCSO algorithm.

6.6 Validation of residual energy

The designed clustering model using the ASCSO algorithm is given in Fig. 9. The better performance is reached by ASCSO-aided cluster head selection strategy in terms of higher energy and thus, it ensures the maximum performance.

6.7 Analysis of QoS

The performance of the designed “energy efficient clustering” is examined concerning QoS by differing the number of rounds as given in Fig. 10, which shows maximum QoS improvement is achieved by the ASCSO algorithm by evaluating with other baseline approaches.

6.8 Evaluation of security

Validation of the offered method is examined regarding security by differing the number of rounds as given in Fig. 11. The higher security is accomplished by the ASCSO algorithm by differing the number of sink nodes.

6.9 Validation of trust

Validation of the offered “energy clustering” method using the ASCSO algorithm concerning trust is given in Fig. 12. The maximum trust is guaranteed by designed cluster head selection by the ASCSO algorithm.

6.10 Statistical analysis

Estimation of statistical analysis of the offered “energy-efficient clustering model” for several constraints is given in Tables 3 to 9. As the “considered optimization algorithms are in stochastic nature and the experiment is executed five times”. This analysis is executed by considering the “measures like best, worst, mean and standard deviation”. “The mean is the average value of the best and worst values and the median is referred to as the center point of the best and worst values whereas the standard deviation is represented as the degree of deviation between each execution”. Finally, the statistical evaluation shows higher efficacy regarding different constraints like the number of dead nodes, trust, QoS, normalized energy, residual energy, security, distance, and delay.

7. Conclusion

This work has implemented a secured “energy-efficient routing” technique for large-scale WSNs with multiple sink nodes. Here, ASCSO was implemented for the selection of the accurate cluster head. It was carried out by solving the objective function regarding “distance, residual energy, delay, security risk, trust, and QoS”. Finally, the proposed work was evaluated by comparing it with several optimization algorithms in terms of different evaluation metrics between sensor nodes and sink nodes. From the results, the maximum performance was observed by the designed ASCSO algorithm. However, COA has superior performance while taking the number of rounds from 1800 to 2000 rounds. Similarly, SSA and BOA also overtake the performance. When considering the number of sink nodes as 4 and the number of rounds as 800, the ASCSO algorithm was 37.9%, 17.6%, 60%, and 33.3% superior to BOA, JA, SSA, and COA techniques. The efficacy improvements of this work are applicable to other research like IoT healthcare applications, target tracking, and military applications. However, it was required to deploy the Routing Protocol for Low-Power and Lossy Networks (RPL) nodes to ensure the consistent of the outcomes. Additionally, need to discover the multihomed solutions on different QoS metrics. Hence, there was a need of improving the network efficiency. This can be solved in the future by utilizing intelligent techniques. We will explore the impact of diverse constraints on the efficacy of the network and further improve its performance.

References

Sun

Zhao

Feng

Zhang

Luo

. Mobile multi-sink nodes path planning algorithm concerned with energy balance in wireless sensor networks. IEEE Access. 2019; 7: 96942-96952.

Vijayashree

Dhas

CSG

. Energy efficient data collection with multiple mobile sink using artificial bee colony algorithm in large-scale WSN, Automatika. 2019 Nov; 60(5): 555-563.

Singh

Malik

Kumar

. Energy efficient heterogeneous DEEC protocol for enhancing lifetime in WSNs. Engineering Science and Technology an International Journal. 2017 Feb; 20(1): 345-353.

Abo-Zahhad

Ahmed

Sabor

Sasaki

. Mobile sink-based adaptive immune energy-efficient clustering protocol for improving the lifetime and stability period of wireless sensor networks. IEEE Sensors Journal. 2015 Aug; 15(8): 4576-4586.

Qing

Zhu

Wang

. Design of a distributed energy-efficient clustering algorithm for heterogeneous wireless sensor networks. Comput. Commun. 2006 Aug; 29(12): 2230-2237.

Wang

Gao

Liu

Sangaiah

Kim

H-J

. Energy efficient routing algorithm with mobile sink support for wireless sensor networks. Sensors. 2019 Mar; 19(7): 1494.

Koosheshi

Ebadi

. Optimization energy consumption with multiple mobile sinks using fuzzy logic in wireless sensor networks. Wireless Networking. 2018 Jun; 25(3): 1215-1234.

Byun

. Mobile collector-based cost balancing scheme for uniform data gathering delay and energy consumption in wireless sensor actuator networking systems. IEEE Sensors Journal. 2020 Apr; 20(8): 4260-4268.

Gharaei

Malebary

Bakar

Hashim

SZM

Butt

Sahar

. Energy-efficient mobile-sink sojourn location optimization scheme for consumer home networks. IEEE Access. 2019; (7): 112079-112086.

10.

Khan

Gansterer

Haring

. Static vs. Mobile sink: The influence of basic parameters on energy efficiency in wireless sensor networks. Computer Communications. 2013 May; 36(9): 965-978.

11.

Sharmin

Karmaker

Lambert

Alam

Shawkat

MSA

. Minimizing the energy hole problem in wireless sensor networks: A wedge merging approach. Sensors. 2020 Jan; 20(1): 277.

12.

Hung

L-L

Leu

F-Y

Tsai

K-L

C-Y

. Energy-efcient cooperative routing scheme for heterogeneous wireless sensor networks. IEEE Access. 2020; 8: 56321-56332.

13.

Mohammed

Elrahim

AGA

. Energy efficient routing protocol for heterogeneous wireless sensor networks. SUST Journal of Engineering and Computer Sciences (JECS). 2019; 20(1): 1-10.

14.

Liu

Zhao

Tie

Bai

Jin

. An improved energy efficient routing protocol for wireless sensor networks. Sensors. 2019 Oct; 19(20): 479.

15.

Israr

Awan

. Multilayer cluster based energy efficient routing protocol for wireless sensor networks. International Journal of Distributed Sensor Networks. 2008 May; 4(2): 176-193.

16.

Cheng

Gao

Zhang

Yang

. An energy-efficient two-stage cooperative routing scheme in wireless multi-hop networks. Sensors. 2019; 19(5): 1002.

17.

Liu

Wang

. Maximizing energy utilization routing scheme in wireless sensor networks based on minimum hops algorithm. Computers & Electrical Engineering. 2012; 38(3): 703-721.

18.

Al-Kaseem

Taha

Abdulmajeed

Al-Raweshidy

. Optimized Energy – Efficient Path Planning Strategy in WSN With Multiple Mobile Sinks. IEEE Access. 2021; 9: 82833-82847.

19.

Hung

Leu

Tsai

. Energy-Efficient Cooperative Routing Scheme for Heterogeneous Wireless Sensor Networks. IEEE Access. 2020; 8: 56321-56332.

20.

Amit

Nagaraju

. Low latency and energy efficient routing-aware network coding-based data transmission in multi-hop and multi-sink WSN. Ad Hoc Networks, vol. 107; 2020 October.

21.

Prachi

. Dr. Ajay

Karan

. Energy efficient cluster based routing protocol for WSN using butterfly optimization algorithm and ant colony optimization. Ad Hoc Networks, vol. 110; 2021 January.

22.

Deepak

Sharad

. Hierarchical WSN protocol with fuzzy multi-criteria clustering and bio-inspired energy-efficient routing (FMCB-ER). Multimedia Tools and Applications, 2020 September.

23.

Rathee

Kumar

Gandomi

Dilip

Balusamy

Patan

. Ant Colony Optimization Based Quality of Service Aware Energy Balancing Secure Routing Algorithm for Wireless Sensor Networks. IEEE Transactions on Engineering Management. 2021 Feb; 68(1): 170-182.

24.

Wang

Ding

Bao

Yang

. An Energy Efficient Routing Protocol Based on Improved Artificial Bee Colony Algorithm for Wireless Sensor Networks. IEEE Access. 2020; 8: 133577-133596.

25.

Krishnan

Yamini

Alenazy

Nalini

. Energy-Efficient Cluster-Based Routing Protocol for WSN Based on Hybrid BSO-TLBO Optimization Model. The Computer Journal. 2021 June; 64(10): 1477-1493.

26.

Dhurandher

Misra

Obaidat

Gupta

. An ant colony optimization approach for reputation and quality-of-service-based security in wireless sensor networks. Security and Communication Networks. 2009; 2(2): 215-224.

27.

Abdmeziem

Tandjaoui

. An end-to-end secure key management protocol for e-health applications. Computers & Electrical Engineering. 2015; 44: 184-197.

28.

Liu

Chung

. Secure user authentication scheme for wireless healthcare sensor networks. Computers & Electrical Engineering. 2017; 59: 250-261.

29.

Lee

Mitra

Kumara

. Distributed energy balancing routing in wireless sensor. Computers & Industrial Engineering. 2009; 57: 125-135.

30.

Maheshwari

Sharma

Verma

. Energy efficient cluster based routing protocol for WSN using butterfly optimization algorithm and ant colony optimization. Ad Hoc Networks, Vol. 110, 1 January 2021.

31.

Jain

Singh

Rani

. A novel nature-inspired algorithm for optimization: Squirrel search algorithm. Swarm and Evolutionary Computation. 2019 February; 44: 148-175.

32.

Yuan

Wang

Yildizbasi

. Developed Coyote Optimization Algorithm and its application to optimal parameters estimation of PEMFC model. Energy Reports. 2020 November; 6: 1106-1117.

33.

Meenalochani

Sudha

. Range-based localisation of a wireless sensor network using Jaya algorithm. Institution of Engineering and Technology. 2019 September; 13(7): 937-943.

34.

Jose

Tamilselvan

. Fault Tolerant and Energy Efficient Signal Processing on FPGA Using Evolutionary Techniques. Computational Intelligence, Cyber Security and Computational Models. 2015; 155-164.

35.

Jose

Kumar

Saravakanthan

Dheeraj

. Computer-aided partitioning for design of parallel testable VLSI systems. 2013 International Conference on Advances in Computing, Communications and Informatics (ICACCI), 2013, pp. 1363-1366.

36.

Vinotha

Jose

. VLSI Implementation of Image Encryption Using DNA Cryptography. Intelligent Communication Technologies and Virtual Mobile Networks. 2019; 33: 190-198.

37.

Roy

Dutta

. A systematic review and research perspective on recommender systems. Journal of Big Data. 2022; 9: 59.

38.

Rajkumar

Rebeiro

. Implementation of Cryptographic Primitives. International Journal of Engineering Trends and Technology. 2017; 49(4).

39.

Rajkumar

. Implementing Software Defined Load Balancer and Firewall. International Journal of Scientific Research and Engineering Development. 2022; 5(5): 300-304.

40.

Michailidis

Miridakis

Michalas

Skondras

Vergados

. Energy Optimization in Massive MIMO UAV-Aided MEC-Enabled Vehicular Networks, IEEE Access, August 2021.

41.

Michailidis

Miridakis

Michalas

Skondras

Vergados

. Energy Optimization in Dual-RIS UAV-Aided MEC-Enabled Internet of Vehicles. MDPI Sensors. 2021; 21: 1-24.

Adaptive squirrel coyote optimization-based secured energy efficient routing technique for large scale WSN with multiple sink nodes

Abstract

Keywords

1. Introduction

2.1 Related works

Table 1 Superiorities and downsides of traditional routing protocol with efficient energy methods in WSN

3. Energy-efficient routing for WSN with multiple sinks

3.1 Model of the system

4.1 Fitness function of the designed method

5.1 Proposed ASCSO-based routing

6.1 Experimental calculation

Table 2 Experimental constraints of secured routing with energy efficient approaches for large scale wireless sensor networks with multiple sink nodes

6.3 Estimation of normalized energy

6.6 Validation of residual energy

6.7 Analysis of QoS

6.8 Evaluation of security

6.9 Validation of trust

6.10 Statistical analysis

7. Conclusion

References

Table 1
Superiorities and downsides of traditional routing protocol with efficient energy methods in WSN

Table 2
Experimental constraints of secured routing with energy efficient approaches for large scale wireless sensor networks with multiple sink nodes