Optimum Cluster Head Selection with a Combination of Multi-Objective Grasshopper Optimization Algorithm and Harmony Search in Wireless Sensor Networks

Subject Areas : International Journal of Smart Electrical Engineering

Seyed Reza Nabavi ¹ , Mehdi Najafi ²

1 - Young Researchers and Elite Club, Arak Branch, Islamic Azad University, Arak, Iran
2 - Department of Computer Engineering, Technical and Vocational University (TVU), Tehran, Iran

Received: 2022-01-05 Accepted : 2022-01-26 Published : 2022-06-01

Keywords: Harmony search, routing, Wireless Sensor Networks, grasshopper optimization algorithm,

Abstract :

Wireless sensor networks have become extensively applied in various fields with their advance. They may be formed freely and simply in many areas with no infrastructure. Also, they gather information about environmental phenomena for decent efficiency and event analysis and send it to base stations. The absence of infrastructure in such networks, on the other hand, limits the sources; therefore, the nodes are powered by a battery with inadequate energy. As a result, preserving energy in such networks is a critical task. Clustering the nodes and picking the cluster head based on the available transmission factors is an intriguing way for reducing energy consumption in these networks, as the average energy consumption of the nodes is lowered and the network lifespan is increased. By combining the multi-objective grasshopper optimization algorithm and the harmony search, this study provides a novel optimization strategy for wireless sensor network clustering. The cluster head is chosen using the multi-objective grasshopper optimization algorithm, and information is communicated between the cluster head's nodes and the sink node using nearly optimum routing based on the harmony search. The simulation outcomes indicate that when the functionality of the multi-objective grasshopper optimization algorithm and the harmony search are taken into account, the suggested technique outperforms the previous methods in terms of data delivery rate, energy consumption, efficiency, and information packet transmission.

Full-Text:



148                                     International Journal of Smart Electrical Engineering, Vol.9, No.4, Fall 2020                   ISSN: 2251-9246        

EISSN: 2345-6221

 

pp. 143:148

 

Optimum Cluster Head Selection with a Combination of Multi-Objective Grasshopper Optimization Algorithm and Harmony Search in Wireless Sensor Networks

Seyed Reza Nabavi 1*, Mehdi Najafi 2

 

1 Young Researchers and Elite Club, Arak Branch, Islamic Azad University, Arak, Iran, [e-mail] 

2 Department of Computer Engineering, Technical and Vocational University (TVU), Tehran, Iran, [e-mail]

      

Abstract

Wireless sensor networks have become extensively applied in various fields with their advance. They may be formed freely and simply in many areas with no infrastructure. Also, they gather information about environmental phenomena for decent efficiency and event analysis and send it to base stations. The absence of infrastructure in such networks, on the other hand, limits the sources; therefore, the nodes are powered by a battery with inadequate energy. As a result, preserving energy in such networks is a critical task. Clustering the nodes and picking the cluster head based on the available transmission factors is an intriguing way for reducing energy consumption in these networks, as the average energy consumption of the nodes is lowered and the network lifespan is increased. By combining the multi-objective grasshopper optimization algorithm and the harmony search, this study provides a novel optimization strategy for wireless sensor network clustering. The cluster head is chosen using the multi-objective grasshopper optimization algorithm, and information is communicated between the cluster head's nodes and the sink node using nearly optimum routing based on the harmony search. The simulation outcomes indicate that when the functionality of the multi-objective grasshopper optimization algorithm and the harmony search are taken into account, the suggested technique outperforms the previous methods in terms of data delivery rate, energy consumption, efficiency, and information packet transmission.

Keywords: Wireless Sensor Networks, Routing, Grasshopper Optimization Algorithm, Harmony Search

Article history: Received 5-Jan-2022; Revised 20-Jan-2022; Accepted 10-Feb-2022. 

© 2022 IAUCTB-IJSEE Science. All rights reserved

 

1.       Introduction

 

WSNs are a collection of distributed microdevices that monitor the environment and relay information to end users. This technology was initially introduced over 20 years ago, and several plans have since been completed. In 2008, green calculations [1] were offered to maximize energy efficacy over a system lifespan while utilizing limited resources. WSNs are often composed of many sensor nodes with low energy resources, nonetheless, they should be able to work for a lengthy period of time without the need for charging or replacing battery. Clustering strategies were offered to ensure an effective association amongst the sensor nodes to extend the network lifespan and decrease the energy use of the network's sensor nodes [2].

Clustering procedures combine the sensor nodes to form small distinct clusters. Every cluster has a defined leader known as the cluster head (CH), and the others are referred to as member nodes (MNs). The topic of this study is the selection of the CH, which is a fundamental difficulty. The sensor nodes collect data about the environment and send it to the appropriate CH. After data aggregation and duplicate data elimination, the CH nodes collect data from all sensor nodes in the cluster and deliver it to a base station. As a result, the CH node must arrange the network, gather data, and transfer it from sensor nodes to sink node and a base station; it consumes more energy than others [3]–[5].

Compared to traditional methods, data collecting on the basis of clustering-based algorithms have several advantages. First, gathering data in a cluster from multiple sensor nodes reduces the quantity of data supplied to a base station by removing duplicate ones on the basis of the CH analysis [6]. Second, each cluster's sensor nodes can transmit data to the sink nodes directly. However, because data transfer at lengthy intervals consumes more energy, direct data transfer is evaded. As an alternative, delivering data to CHs in close proximity to the member sensor nodes uses a lesser amount of energy; as a result, the overall network's energy need for data transmission is lowered [7]. Third, rotating the CHs helps provide balanced energy use in the network, preventing specific nodes from being hungry due to energy scarcity. Selecting a suitable CH with optimal capabilities while maintaining a balanced energy use percent and network efficacy in WSNs, on the other hand, is a well-known NP-hard task [8].

NP-hard challenges may not be solved utilizing polynomial or linear approaches and must be solved utilizing artificial or swarm intelligence and metaheuristic techniques. In the domains of sensor node clustering and CH selection in WSNs, heuristic and metaheuristic algorithms, which aim to improve the network's contradictory aims at the same time, have recently sparked interest [9].

Given that energy shortage is a major issue in WSNs, energy usage determines the effectiveness of most WSN applications. As a result, the main goal of this study is to lower the energy consumption of WS nodes. Because the energy necessary to detect data in the environment and receive packets from other sensor nodes in WSNs is constant, sending packets consumes the majority of the energy. The more energy is required to transport data across a larger distance between the current node and the next hop. As a result, the closest next hop with the greatest long-lasting energy and the shortest distance to sink node can conserve energy and improve QoS parameters. Alternatively, while finding the optimum path in WSNs has been offered as an NP-hard challenge, using metaheuristic algorithms is the best option for finding the optimum solution. Metaheuristic algorithms may identify local optimizations on the basis of local search, but they may encounter weaknesses or stalemates while trying to find optimum global solutions. As a result, using global search algorithms or a combination of metaheuristic algorithms to find the best global solutions may be possible.

As a result, using the multi-objective grasshopper optimization algorithm (MOGOA) and the harmony search, this research proposes a novel optimization technique for clustering WSNs (HS). In this paper, the multi-objective grasshopper optimization method is employed to select the CH, and a nearly optimum routing based on the HS is used to convey information between the CH nodes and the sink node. Taking into account the capabilities of the MOGOA and the HS, the proposed method outperformed prior approaches in terms of energy consumption, efficacy, data delivery rate, and information packet transfer rate. This study's major motivation is the use of the multi-objective grasshopper optimization method in conjunction with HS to determine the optimum local clusters and the optimum global path in a WSN.

The goal of this article is to find the optimal local clusters and global path in a wireless sensor network using a mix of the multi-objective grasshopper optimization approach and harmony search.

The remainder of this work is structured as follows. Section 2 is a literature review. Section 3 goes through the specifics of the suggested technique. Section 4 discusses the suggested method's implementation and evaluation. Section 5 brings the paper to a close.

2.       Literature Review

This part of the study will address prior research because metaheuristic optimization approaches are widely employed in the field of routing in WSNs.

In [10], one of the innovative clustering techniques is presented, which uses the grey wolf optimization process to select the cluster head. The solutions would be graded on the basis of the expected energy usage and residual energy flow of all nodes in order to pick the CHs. This new protocol uses a similar clustering in numerous consecutive cycles to improve energy efficiency, allowing this protocol to conserve the energy required to adjust the clustering. A new two-hop routing algorithm is also developed to send data between remote clusters from the base station. While remaining communication nodes, this strategy has been shown to ensure the least and most balanced energy use. They only make one hop.

For WSN clustering, Nabavi et al. developed a novel optimization technique based on the multi-objective evolutionary algorithm and the gravitational search algorithm. The CHs in this work are selected using a multi-objective evolutionary algorithm targeted at minimizing intra-cluster distances and MN energy consumption, and information is transferred between the CHs and the sink node using nearly optimum gravitational search routing [11].

Ali et al. devised ARSH-FATI-CHS in conjunction with a heuristic termed NRC for effective cluster creation in order to boost the overall LT of the network. ARSH-FATI-CHS is a simple yet effective clustering approach that alternates between exploitative and explorative search modes to get the best performance-to-cost ratio [12].

Nabavi et al. suggested a novel greedy method for determining the shortest random path in WSNs. According to the study, the greedy local search strategy is faster than other methods in locating the best solution to a variety of issues. The node energy consumption is roughly symmetric in this strategy, and hence network lifespan decreases with a modest slope [13].

Huamei et al. suggest a reshuffle leapfrog strategy based on increased non-uniform CRP. The non-uniform technique is utilized to partition sensor nodes into clusters, and the modified hybrid leapfrog algorithm is employed in clustering to select the suitable cluster head, improve node energy efficiency, regulate energy consumption in WSNs, and decrease the likelihood of energy hole incidence. To iterate node efficiency and pick the best cluster head, the protocol employs a fitness function and a sub-group elite individual updating technique [14].

Baradaran et al. presented a technique (HQCA) to generate high-quality clusters that uses a measure to determine cluster quality and may lower clustering error rates [15].

In [16], a novel design strategy for optimum routing in WSNs on the basis of the multi-objective grey wolf optimization algorithm is offered. According to this study, the multi-objective grey wolf optimization method was used to pick the next node on the basis of lowering interstitial distances between nodes, reducing energy consumption in network nodes, and near-optimum routing based on the multi-objective fitness function.

Based on the typical ant colony method, Wang merged the integrated trust value, path average energy, node residual energy, and minimum energy into the pheromone update to build a cheap, high-security, high-efficiency, and high-stability wireless routing algorithm. The method's superiority in this study is demonstrated by a wide number of investigations comprising the experimental verification of life cycle, node survival time, node energy consumption, and stability indicators [17].

In [18], a routing protocol on the basis of ant colony optimization and uneven clustering is offered, which enhances the algorithm's convergence speed while avoiding local optimization. Nevertheless, the factors employed in such routing algorithms are all fixed, and there is no way to dynamically alter the algorithm factors while the process is underway.

Varshney et al. suggested an IWSN energy-saving lion optimizing routing method on the basis of lightning. The algorithm selects the best sensor placement technique on the basis of the sensor's throughput, delay, lifespan, and coverage area [19].

[20] proposes a new strategy for reducing energy use while retaining data quality. Devices may select the data rate from a minimum to a maximum frequency in this manner to monitor physical changes. Furthermore, this technique aids the reduction of a variety of relevant factors such as bandwidth resources, energy consumption, and data transfer. The results show that the suggested technique may decrease energy consumption while maintaining data accuracy.

Nayyar and colleagues proposed an energy-efficient routing approach for improving WSN real-time performance. The protocol may provide an optimal solution in terms of packet delivery and throughput, based on ant colony optimization [21].

3.       The Suggested Technique

As previously stated, this study provides a novel WSN clustering optimization technique based on the multi-objective GOA and the HS. In this study, the multi-objective GOA is used to identify the CHs by minimizing the MNs' intra-cluster distance and energy consumption, and the nearly optimum routing based on the HS is used to find the optimal route and transmit data between the CH nodes and the sink node. The goal of this research is to lower sensor node energy consumption, raise network throughput, increase data delivery rate, and decrease information packet transmission latency. The remainder of this section delves deeper into the approaches stated previously.

In cluster-based protocols, the cluster head combines the information on the received information and sends it to the sink node. As shown in figure 1, cluster head sends data to the destination through other cluster heads to transfer information to the sink node.

 

 

Fig. 1.       Multi-hop transfer from cluster head to the sink node.

A.       Energy Model

The energy model employed in [22] is used in this study. The network overall energy consumption () in this model is attributable to the energy dissipation of a transmitter () and a receiver () as shown below:

 

                                               (1)

 

The energy used to run the power amplifier and radio circuits is represented by .  also shows how much energy is used to run the radio equipment. The energy consumed by the transmitter to transmit  bits for each sensor node in the network is calculated as follows:

 

  (2)

 

Where  is the energy used per bit to drive the transmitter or receiver circuit,  and  are the amplification energies for free space and multipath models, respectively, and  is the threshold transmission distance, which is generally calculated as follows:

 

                                                             (3)

 

The quantity of energy spent by the transmitter and the amount of data to be conveyed are determined by the distance factors d. If the data transmission distance is within the threshold range, transmittance energy is proportional to ; if not, this relationship rises to . Distance and workload, therefore, have a key part in clustering the sensor nodes to improve network life.

For the receiver to receive  bit of data, the energy dissipation is obtained as below:

 

                                                 (4)

 

Where  denotes that the signal relies on parameters like digital coding, modulation, and signal propagation.

B.       Grasshopper Optimization Algorithm (GOA)

Saremi et al. introduced the Grasshopper Optimization Algorithm. For tackling optimization issues, the suggested approach theoretically modeled and emulated the swarming behavior of grasshoppers in nature [23]. Insects are grasshoppers. Because of the damage they do to crop production and agriculture, they are categorized as a pest. The grasshopper life cycle is depicted in figure 2. Grasshoppers produce one of the world's largest swarms, despite being seen singly in nature [24].

 

 

Fig. 2.       Grasshopper life cycle.

The grasshoppers' slow movement and tiny steps are a critical feature of the larval swarm. Adult swarms, on the other hand, are known for their long-distance and sudden migration. The search for food is another important aspect of grasshopper swarming. As stated in the introduction, nature-inspired algorithms logically divide the search process into two tendencies: exploration and exploitation. During exploration, the search agents are driven to move quickly, whereas during exploitation, they tend to move slowly. Grasshoppers are naturally adept at both of these tasks, as well as target hunting. As a result, we can design a new nature-inspired algorithm if we can mathematically characterize this behavior.

The grasshopper swarming behavior is simulated using the following mathematical model [23]:

 

                                                  (5)

 

Where  shows the th grasshopper position,  indicates the social interaction,  shows the gravity force on the th grasshopper, and  indicates the wind advection. Note that to obtain random behavior, the Eq. may be written as , where ,  and  represent random numbers in .

 

                                               (6)

 

Where  denotes the distance between the th and th grasshopper, s denotes a function to explain the strength of social pressures, as specified in Eq. (7), and  denotes a unit vector from the th to the th grasshopper.

The  function that describes the social forces is obtained as below:

 

                                                (7)

 

Where  shows the attraction intensity, and  represents the scale of attractive length.

Using function , figure 3 displays a conceptual model of grasshopper interactions with the comfort zone. It should be emphasized that, in a simplified version, social interaction was the driving factor behind several early locust swarming models.

 

 

Fig. 3.       Individual grasshoppers in a swarm exhibit primitive corrective patterns.

Despite the fact that the function  can divide the space between two grasshoppers into comfort, repulsion, and attraction zones, for distances greater than , it returns values close to zero. As a result, this function cannot apply high forces between grasshoppers separated by huge distances. To address this issue, we mapped grasshopper distances in the interval . In Eq. (5), the  element is computed as follows:

 

                                                             (8)

 

Where  represents the gravitational constant, and  indicates a unity vector towards the earth center.

The  element in Eq. (5) is computed as below:

 

                                                               (9)

 

Where  represents a constant drift, and  indicates a unity vector in the wind direction.

Nymph grasshoppers are wingless; thus, they move associated with wind direction.

Substituting ,  and  in Eq. (5), this Eq. may be generalized as below:

         (10)

 

Where , and  indicates the number of grasshoppers.

The ground location of nymph grasshoppers should not fall below a certain level. This Eq., however, will not be used in the swarm simulation and optimization approach because it inhibits the software from exploring and exploiting the search space surrounding a solution. Indeed, the swarm model is based on free space. As a result, Eq. (10) is utilized to model the interaction of grasshoppers in a swarm.

However, because grasshoppers quickly find their comfort zone and the swarm does not converge to a precise point, this mathematical model cannot be directly applied to optimization problems. To tackle optimization difficulties, the following modified version of this Eq. is suggested:

 

     (11)

 

Where  represents the upper bound in the th dimension,  indicates the lower bound in the th dimension ,  denotes the value of the th dimension in the target (best answer produced thus far), and  represents a decreasing coefficient to shrink the repulsion zone, comfort zone, and attraction zone.  is almost identical to the  element in Eq. (5). However, we ignore gravity (no  element) and assume that the wind direction ( element) is always directed toward a target ().

As indicated in Eq (11), a grasshopper's next location is determined by its current position, the target's position, and the positions of all other grasshoppers (11). It is worth noting that the first element of this equation takes into account the present grasshopper's location in relation to other grasshoppers. Indeed, we took into account the state of all grasshoppers while determining the placement of search agents surrounding the target.

Finally, the sum term in Eq. (11) takes into account the location of other grasshoppers, simulating grasshopper interaction in nature. The second term, , likewise denotes a proclivity to migrate towards a food source. Furthermore, factor  represents the slowing that grasshoppers experience as they approach and subsequently consume a food source. Both halves of Eq. (11) could be multiplied with random values to make the behavior more unpredictable. Individual words could also be multiplied by random numbers to provide unpredictable grasshopper contact or food source choosing behavior.

The mathematical formulas presented here can be used to explore and exploit the search space. Nonetheless, a mechanism should be in place to force search agents to modify their level of exploration in order to optimize exploitation. Because grasshopper larvae lack wings, they must travel and search for food in the wild. They will be able to freely move through the air and explore a much larger area. Exploration, on the other hand, comes first in stochastic optimization methods due to the necessity to locate attractive parts of the search space. Exploitation drives search agents to seek an accurate approximation of the global optimum locally after discovering suitable locations.

The variable  must be adjusted in proportion to the number of repetitions to balance exploration and exploitation. This method facilitates exploitation as the number of iterations increases. The following formula is used to calculate the coefficient , which is proportional to the number of iterations:

 

                                     (12)

 

Where  denotes the highest value,  denotes the lowest value,  is the current iteration, and  denotes the maximum number of iterations. The values for  and  in this investigation are  and , respectively.

The preceding sections demonstrate that the mathematical model requires grasshoppers to gradually approach a goal over the course of iterations. Nonetheless, there is no goal in a genuine search space because we don't know where the global optimum, or main purpose, is located. As a result, we must choose a grasshopper goal for each optimization phase. During GOA optimization, the fittest grasshopper (the one with the best objective value) is meant to be the aim. GOA will be able to save the most promising target in the search space and direct grasshoppers to it in each iteration as a result of this. This is done to identify a more accurate and better target in the search space to serve as the best approximation for the true global optimum [23].

C.       Harmony Search

The harmony search (HS) method [25] is based on the fundamentals of artists' harmony improvisation (see figure 4). The HS distinguishes itself from other search engines by its algorithm simplicity and speed. It has recently been successfully applied to a variety of areas, including function optimization, mechanical structure design, data categorization system optimization, and pipe network optimization.

 

 

Fig. 4.       Harmony in music.

As we all know, once a musician composes harmony, they normally experiment with a wide range of possible music pitch combinations stored in their memory. In reality, the process of discovering the best answers to engineering issues is very similar to the quest of perfect harmony. The HS technique is mostly inspired by the working principles of harmonic improvisation. There are four essential steps that must be completed.

Step 1: Initialize the HS Memory (HM): The initial HM is made up of a set number of randomly generated solutions to the optimization issues in question. For an -dimension problem, an HM of size  could be written as follows:

 

                            (13)

 

Where  () represents a solution alternate. HMS is usually set between  and .

Step 2: Create a new solution s from HM: Each element of this solution, , is gained on the basis of the Harmony Memory Considering Rate (HMCR). The HMCR indicates the likelihood of selecting an element from the HM members, whereas the 1-HMCR indicates the likelihood of producing it at random. If  is derived from the HM, it is picked from a random HM member's th dimension and changed according to the Pitching Adjust Rate (PAR). The PAR determines whether an HM alternate can be changed. As can be observed, the production of offspring in Genetic Algorithms (GAs) [26] using mutation and crossover operations is comparable to  improvisation. The GA approach, on the other hand, generates new chromosomes by combining two or more existing ones (simple crossover), whereas the HS method generates new solutions by combining all HM components.

Step 3: Update the HM: Step 2's new solution is examined. If it outperforms the worst member in the HM in terms of fitness, it will take its place. Otherwise, it will be removed.

Step 4: Repeat Step 2 to Step 3: Repeat steps 2–3 until a predefined termination requirement, such as the maximum number of iterations, is fulfilled.

The HS method, like the GA and swarm intelligence algorithms [27], uses a random search methodology. Prior domain knowledge is not required, including gradient information for objective functions. Despite this, unlike population-based techniques, it only takes a single search memory to evolve. As a result, the HS method stands out for its computational simplicity.

D.       Initial Clustering

In this section, a WSN with 100 sensor nodes and one sink node is simulated. The starting factors are based on the factors that are common in comparable approaches. Ultimately, the approach is compared to other methods that are available. After the first deployment of wireless sensor nodes in the network, the sink node sends a "Hello" message to all sensors to identify the nodes and determine the position of the network's existing nodes. After receiving the "Hello" message, all sensors in the network send a routing reply (RREP) to the sink node to determine the exact position of each sensor node for initial node clustering.

Different energy consumption constants are utilized for each packet type since the amount of energy required to send and receive RREP packets and data varies. As a result, the suggested technique's energy consumption is thoroughly assessed. After receiving the RREP and defining the initial point of the wireless sensor nodes, the suggested technique clusters the nodes based on random CHs. The CHs are chosen at random, so data flow is unaffected, and early energy discharge of some sensor nodes is avoided because the initial energy of all nodes is consistent in the first steps. The wireless sensor nodes are then grouped together based on how close they are to the CH.

The distance between the sensor nodes and the distance to sink node, which is represented as follows:

 

                          (14)

                   (15)

 

Where  is the th node coordinates, and  is the sink node coordinates. The CHs are given to the MOGOA as the first grasshoppers in this stage so that the ideal position of the CH may be discovered. In the suggested MOGOA, an optimum node in the chosen cluster may not be identified as the CH. In this scenario, the cluster is dismantled, and clustering is restarted after locating an ideal CH around the cluster.

E.       Determine the Optimum CHs Using MOGOA

Instead of the random CHs chosen in the previous stage, the MOGOA is employed to find new and optimal CHs. The grasshoppers are configured in the next section, and the MOGOA fitness function for the given technique is defined.

The recommended MOGOA starts by constructing an initial population of grasshoppers, with each grasshopper representing a possible solution to the clustering problem. As a result, a grasshopper is a vector, with the integers representing the index of a CH node.

Each grasshopper is first observed as a probabilistic clustering in the WSN, which changes as the fitness function, social interaction (Si), gravity force (Gi), and wind advection operators are applied. Finally, the grasshopper with the highest fitness level is picked as the nearly ideal cluster.

Each grasshopper's fitness function is determined using the objective function, which is a combination of the CH's residual energy, mean intra-cluster distance, and distance from the sink node. The fitness function is described as the representation of the elements of interest to balance the CH's residual energy, mean intra-cluster distance, and distance from the sink node for each viable clustering of the population. Eq. (16) is used to generate the fitness function that will be utilized to assess the given grasshoppers using the suggested objective function.

 

(16)

 

Where  represents the total number of alive sensor nodes,  represents the total number of CHs,  represents the number of sensor nodes in the cluster ,  represents the estimated distance from the next hop to the destination node,  represents the distance from the source node to the next hop, and  represents the CHs current energy .

The fitness function presented in Eq. (16) is used to select the appropriate grasshoppers from the initial population, and the remaining grasshoppers are distributed to the social interaction, gravity force, and wind advection operators to diversify the population and generate new superior grasshoppers. Every grasshopper in the new offspring population is tested to see if it can fix the problem. The unlikely grasshoppers that go above the established limitations are penalized based on their fitness value, which means they have a lower chance of being selected for generation and conversion to new grasshoppers. At each iteration, the most appropriate grasshopper that represents the approximately ideal clustering solution is saved and ordered depending on its optimality. This procedure is repeated until the termination criteria are fulfilled.

F.       Find the Optimal Route Using the HS

On the basis of the CHs chosen by the suggested MOGOA, the HS finds the best path. This stage specifies the information transfer routes from each cluster to the sink node, taking into account the CHs selected. As a result, the CH in each cluster is regarded as the harmony of the HS. Therefore, at each information packet transmission stage, the HS weighs the harmonies and discards those that are of insufficient quality, while other harmonies are chosen as appropriate solutions.

The weighting of the chosen harmonies is dictated by the aims. In other words, the HS assesses whether the route between the sensor nodes and the sink node that requires data transmission minimizes overall energy consumption and end-to-end delay while enhancing data delivery rate and network throughput in the proposed technique. Routes that optimize these metrics are desirable, but routes that damage even one of these metrics are abandoned. As a result, the offered HS's fitness function is as shown in Eq. (17).

 

(17)

 

Where  represents the network total energy consumption,  represents the number of the nodes in the path,  indicates the residual energy of the nodes, and  indicates the average residual energy of the nodes.

The flowchart of the presented method is shown in figure 5.

4.       Results and Discussion

The WSN was designed to simulate our method using the standardized factors. The suggested network is simulated in a 100×100 environment, and MATLAB 2021b was used to model this scenario. To develop nodes in the network, a random location function is used. A sink node would also be designed in the network center, making it easier for nodes to reach the sink node. Other metrics associated with the offered network are listed in Table 1.

The suggested approach is examined in order to assess its quality and improve the suggested method's performance on the core challenge. Different measures for evaluating WSNs have been published in the literature, which is established with the study aims described in the first section in mind. The suggested technique is evaluated in this area of the study in terms of residual energy, energy consumption, number of missed packets, message transmission latency, and network throughput. As a result, figure 6 depicts the network's average energy use and figure 7 depicts the average residual energy with increasing data transfer steps in the suggested technique.

 

 

Fig. 5.       Flowchart of the proposed method.

 

Simulation parameters

 

 

Fig. 6.       Average energy consumption in the suggested technique.

 

Fig. 7.       Average residual energy by increasing data transfer steps in the suggested technique.

The suggested balanced energy usage in WSNs by providing a linear curve is depicted in figures 6 and 7. Therefore, in the suggested technique, the energy consumption balance is established, and the energy of some nodes does not finish earlier than others, and the energy consumption is the same in all nodes. As a result, all nodes gradually lose energy, showing that the network has a lengthy lifespan.

The term network lifespan refers to the amount of time that the network is available at a time that does not interfere with network data aggregation. As a result, the network's life can be defined as the time when some nodes in the network's energy supply run out and the network is unable to continue operating with the remaining nodes. Figure 8 depicts a network lifespan diagram.

 

 

Fig. 8.       Death process of sensor nodes in the network.

As depicted in figure 8, the suggested network has a lifespan of 1490 cycles before the first node dies after 1250 repetitions. When a node dies in a WSN, all paths leading to that node become stalled and must be rerouted, and the responsibility of gathering information from the area covered by that node falls on the shoulders of its nearby nodes. This increases the energy consumption of surrounding nodes and increases their chances of dying. As depicted in figure 8, when the first node dies, the fatality rate of the other nodes rises.

Another indicator used to evaluate the proposed strategy is the number of lost packets. Fewer packets are lost during the information aggregation process the better the quality of data collecting and the precision of the routing system. One of the main causes of information packet loss could be the selection of nodes with congestion of sent packets that appear to be the network bottleneck. In the suggested technique, the lost packets are shown in figure 9.

According to figure 9, the suggested technique loses around 70 packets, resulting in a maximum of 0.02 lost packets per 15000 total transferred packets via routing. The throughput is defined as the ratio of packets transferred per unit time to the transmission medium's bandwidth. The data transmission rate per unit time is measured by network throughput because all sensor nodes in a WSN have the same capacity. Figure 10 depicts the recommended technique's network throughput ratio.

 

 

Fig. 9.       Lost packets in the suggested technique.

 

Fig. 10.       Throughput ratio in the network.

As illustrated in figure 10, the suggested technique attempts to avoid bottlenecks and build a safe channel to transfer data to the entire node optimum routing. As a result, the offered network has a high throughput rate, with an average of 99.95 percent.

The final criterion for evaluation in the suggested method is the end-to-end latency of network nodes. Given that a packet's transfer time coefficient is for fixed nodes, the distance between nodes is the primary cause of end-to-end transfer delay. Because the suggested technique involves the transfer of sensor nodes to the cluster head, the shorter distance between the CHs and other nodes means more accurate grouping and lower average distances inside the cluster, which is one of the goals. In addition, figure 11 depicts a diagram of the end-to-end delay of 100 nodes.

The offered technique for 100 nodes has an average latency of 1 millisecond each cycle, as can be seen in figure 11. This small number denotes the close proximity of the CHs to the cluster MNs.

 

 

Fig. 11.       The end-to-end delay in the network.

We compare the suggested technique to [16], [21], [28]. Figure 12 compares the proposed strategy to older methods regarding average energy consumption in the network.

 

 

Fig. 12.       Comparing the suggested technique and the earlier approaches regarding energy consumption.

As depicted in figure 12, the suggested approach consumes less energy on average for 100 nodes than the prior methods. Figure 13 compares the suggested strategy to existing methods in terms of the longevity and the initial node’s death.

As seen in figure 13, the offered approach has a longer lifespan than the prior method, and the initial node in the offered approach dies after others, showing that the offered approach has balanced energy consumption. regarding data delivery rate, figure 14 compares the suggested method to earlier methods.

As shown in figure 14, the data delivery rate of the proposed method is higher than previous methods, demonstrating choosing the optimal route, avoiding bottlenecks, and loss of the fewest information packets.

 

Fig. 13.       Comparison of the suggested technique and the earlier approaches in terms of lifespan.

 

Fig. 14.       Comparison of the suggested technique and earlier approaches regarding data delivery rate.

5.       Conclusion

The WSN is a cutting-edge environmental monitoring and control network that collects data from the environment and aggregates it for network applications without the need for infrastructure. As a result of their popularity, a number of difficulties have arisen, one of which is uneven energy usage. Due to the sensor nodes' limited energy, unbalanced energy consumption may have an influence on all network performance metrics. Therefore, for WSN clustering, a novel optimization strategy on the basis of the MOGOA and the HS is provided in this paper. In this work, the MOGOA is used to choose CHs based on reducing intra-cluster distances and MN energy consumption, and the nearly optimal routing based on the HS is used to transfer data between the CHs and the sink node. When compared to earlier methodologies, the results of the suggested method's implementation reveal that combining the multi-objective grasshopper optimization algorithm and the HS, the provided model considerably improved average energy consumption, data delivery rate, and network lifespan. Using density-based clustering approaches, future research will seek to cluster wireless sensor nodes in the network. We believe that this method will improve the clustering of network nodes more efficiently.

References

[1]        S. Murugesan, “Harnessing green IT: Principles and practices,” IT Professional, vol. 10, no. 1, pp. 24–33, Jan. 2008, doi: 10.1109/MITP.2008.10.

[2]        L. Xu, R. Collier, and G. M. P. O’Hare, “A Survey of Clustering Techniques in WSNs and Consideration of the Challenges of Applying Such to 5G IoT Scenarios,” IEEE Internet of Things Journal, vol. 4, no. 5, pp. 1229–1249, Oct. 2017, doi: 10.1109/JIOT.2017.2726014.

[3]        H. Ouchitachen, A. Hair, and N. Idrissi, “Improved multi-objective weighted clustering algorithm in Wireless Sensor Network,” Egyptian Informatics Journal, vol. 18, no. 1, pp. 45–54, Mar. 2017, doi: 10.1016/J.EIJ.2016.06.001.

[4]        S. Misra and R. Kumar, “A literature survey on various clustering approaches in wireless sensor network,” 2nd International Conference on Communication, Control and Intelligent Systems, CCIS 2016, pp. 18–22, Mar. 2017, doi: 10.1109/CCINTELS.2016.7878192.

[5]        T. T. Truong, K. N. Brown, and C. J. Sreenan, “Multi-objective hierarchical algorithms for restoring Wireless Sensor Network connectivity in known environments,” Ad Hoc Networks, vol. 33, pp. 190–208, Oct. 2015, doi: 10.1016/J.ADHOC.2015.05.005.

[6]        S. Arjunan and S. Pothula, “A survey on unequal clustering protocols in Wireless Sensor Networks,” Journal of King Saud University - Computer and Information Sciences, vol. 31, no. 3, pp. 304–317, Jul. 2019, doi: 10.1016/J.JKSUCI.2017.03.006.

[7]        A. Kumar Wadbude and R. Shyam Panda, “A Survey on Multi Cluster Head Analysis Scheme for Wireless Sensor Network,” International Research Journal of Engineering and Technology (IRJET), vol. 6, no. 1, pp. 761–765, 2019.

[8]        P. S. Mann and S. Singh, “Improved artificial bee colony metaheuristic for energy-efficient clustering in wireless sensor networks,” Artificial Intelligence Review, vol. 51, no. 3, pp. 329–354, Mar. 2019, doi: 10.1007/s10462-017-9564-4.

[9]        F. Fanian and M. Kuchaki Rafsanjani, “Cluster-based routing protocols in wireless sensor networks: A survey based on methodology,” Journal of Network and Computer Applications, vol. 142, pp. 111–142, Sep. 2019, doi: 10.1016/J.JNCA.2019.04.021.

[10]        S. M. M. H. Daneshvar, P. Alikhah Ahari Mohajer, and S. M. Mazinani, “Energy-Efficient Routing in WSN: A Centralized Cluster-Based Approach via Grey Wolf Optimizer,” IEEE Access, vol. 7, pp. 170019–170031, 2019, doi: 10.1109/ACCESS.2019.2955993.

[11]        S. R. Nabavi, V. Ostovari Moghadam, M. Yahyaei Feriz Hendi, and A. Ghasemi, “Optimal Selection of the Cluster Head in Wireless Sensor Networks by Combining the Multi-Objective Genetic Algorithm and the Gravitational Search Algorithm,” Journal of Sensors, vol. 2021, 2021, doi: 10.1155/2021/2292580.

[12]        H. Ali, U. U. Tariq, M. Hussain, L. Lu, J. Panneerselvam, and X. Zhai, “Arsh-fati: A novel metaheuristic for cluster head selection in wireless sensor networks,” IEEE Systems Journal, vol. 15, no. 2, pp. 2386–2397, Jun. 2021, doi: 10.1109/JSYST.2020.2986811.

[13]        S. R. Nabavi, N. Osati Eraghi, and J. Akbari Torkestani, “WSN Routing Protocol Using a Multi-Objective Greedy Approach,” Wireless Communications and Mobile Computing, vol. 2021, 2021, doi: 10.1155/2021/6664669.

[14]        Q. Huamei, L. Chubin, G. Yijiahe, X. Wangping, and J. Ying, “An energy-efficient non-uniform clustering routing protocol based on improved shuffled frog leaping algorithm for wireless sensor networks,” IET Communications, vol. 15, no. 3, pp. 374–383, Feb. 2021, doi: 10.1049/CMU2.12067.

[15]        A. A. Baradaran and K. Navi, “HQCA-WSN: High-quality clustering algorithm and optimal cluster head selection using fuzzy logic in wireless sensor networks,” Fuzzy Sets and Systems, vol. 389, pp. 114–144, Jun. 2020, doi: 10.1016/J.FSS.2019.11.015.

[16]        S. R. Nabavi, N. Osati Eraghi, and J. Akbari Torkestani, “Intelligent Optimization of QoS in Wireless Sensor Networks Using Multi-Objective Grey Wolf Optimization Algorithm,” Wireless Communications and Mobile Computing, vol. 2022, 2022, doi: 10.1155/2022/5385502.

[17]        X. Wang, “Low-Energy Secure Routing Protocol for WSNs Based on Multiobjective Ant Colony Optimization Algorithm,” Journal of Sensors, vol. 2021, pp. 1–9, Oct. 2021, doi: 10.1155/2021/7633054.

[18]        N. Moussa and A. el Belrhiti El Alaoui, “An energy-efficient cluster-based routing protocol using unequal clustering and improved ACO techniques for WSNs,” Peer-to-Peer Networking and Applications, vol. 14, no. 3, pp. 1334–1347, May 2021, doi: 10.1007/S12083-020-01056-4/TABLES/5.

[19]        S. Varshney, C. Kumar, and A. Swaroop, “Lightning-Based Lion Optimization Algorithm for Monitoring the Pipelines Using Linear Wireless Sensor Network,” Wireless Personal Communications 2020 117:3, vol. 117, no. 3, pp. 2475–2494, Nov. 2020, doi: 10.1007/S11277-020-07987-8.

[20]        X. Zhu, Q. Xu, X. Qi, and H. Liu, “A Self-adaptive Data Acquisition Technique and Its Application in Landslide Monitoring,” Advancing Culture of Living with Landslides, pp. 71–78, 2017, doi: 10.1007/978-3-319-53487-9_7.

[21]        A. Nayyar and R. Singh, “IEEMARP- a novel energy efficient multipath routing protocol based on ant Colony optimization (ACO) for dynamic sensor networks,” Multimedia Tools and Applications 2019 79:47, vol. 79, no. 47, pp. 35221–35252, May 2019, doi: 10.1007/S11042-019-7627-Z.

[22]        P. C. S. Rao, P. K. Jana, and H. Banka, “A particle swarm optimization based energy efficient cluster head selection algorithm for wireless sensor networks,” Wireless Networks, vol. 23, no. 7, pp. 2005–2020, Oct. 2017, doi: 10.1007/s11276-016-1270-7.

[23]        S. Saremi, S. Mirjalili, and A. Lewis, “Grasshopper Optimisation Algorithm: Theory and application,” Advances in Engineering Software, vol. 105, pp. 30–47, Mar. 2017, doi: 10.1016/J.ADVENGSOFT.2017.01.004.

[24]        S. J. Simpson, A. R. McCaffery, and B. F. HÄgele, “A behavioural analysis of phase change in the desert locust,” Biological Reviews, vol. 74, no. 4, pp. 461–480, Nov. 1999, doi: 10.1111/J.1469-185X.1999.TB00038.X.

[25]        Z. Woo Geem, J. Hoon Kim, and G. v Loganathan, “A New Heuristic Optimization Algorithm: Harmony Search,” Simulation, vol. 76, no. 2, pp. 60–68, 2001, doi: 10.1177/003754970107600201.

[26]        W. B. Langdon and R. Poli, “Foundations of Genetic Programming,” Foundations of Genetic Programming, 2002, doi: 10.1007/978-3-662-04726-2.

[27]        A. P. Engelbrecht, “Fundamentals of computational swarm intelligence,” p. 599, 2005.

[28]        W. Lu, H. Zhao, and H. Zhao, “Distributed energy balancing routing algorithm in wireless sensor networks,” in Lecture Notes in Electrical Engineering, 2012, vol. 127 LNEE, no. VOL. 4, pp. 227–232. doi: 10.1007/978-3-642-25769-8_33.

 

---