An Energy-Aware Cluster Head Selection and Optimal Route Selection Algorithm for Maximizing Network Lifetime in MANETs

A MANET is a network in which no two nodes have the same physical configuration. The routers travel about at will and arrange themselves in a disorganized fashion [1]. MANET has several potential applications, including group communication, emergency response, and even combat. Since MANET has no fixed infrastructure, malicious users can readily exploit its flexibility by connecting and disconnecting nodes at will [2]. Because of how difficult these threats are to identify; it is challenging to create defenses against them. Due to its versatility and speed in network construction, MANET is employed in many different kinds of applications [3].

If every node in the network trusts and collaborates with every other node, the networks will run smoothly. Due to their extremely dynamic nature and multiple communication breaks generated by mobile nodes [4, 5], routing is challenging, and security issues are regularly highlighted. As a result, the routing protocols in a MANET need to incorporate security measures to lessen the impact of potential attacks [5]. The openness, unpredictability, lack of privacy, and lack of authentication that characterize MANETs necessitate a comprehensive security solution [6].

There are two main aspects to MANET security: functional routing security and data broadcast security. In order to avoid tampering with and replaying of routing data transmission, security-based routing systems use a variety of encryption methods [7]. There are a number of different security routing methods [8, 9], including the most fundamental routing. In order for routers in a network to regulate the best path to take between any two points in the system, they need to be able to interconnect and share relevant information [10]. The primary function of a routing system is to determine the most efficient path for sending data from one device to another. Proactive, multicast, and reactive routing methods [11] are the three main types of MANET routing protocols. While there are benefits to using a variety of routing techniques, there are also a number of technical challenges that must be overcome, including issues related to energy [12].

The LEACH protocol has seen extensive application in MANETs as a means of addressing the difficulties associated with energy efficiency in CH selection. By taking turns as CH, LEACH attempts to reduce energy usage at individual nodes and extend the lifespan of the network [13]. However, there may be barriers to obtaining maximum energy efficiency using the standard LEACH method [14]. Optimizing difficulties in many different fields has seen a rise in interest in bio-inspired algorithms in recent years [15]. These algorithms take their cues for solving problems from the workings of biological or natural systems. Bio-inspired procedures can be used to improve MANETs' energy efficiency and performance in the context of CH selection [16].

As a result, this study analyzes the results, advantages, and significance of developing and utilizing a binary WWPA method for cluster head selection. Although the method has not been employed to address the problem since its creation, its performance prompted this binarization. In order to deal with the bounded-continuity character of the issue, we present a binary version of the WWPA.

Here is a list of the important contributions:

·In this research, we describe routing strategies that both prioritize safety and low power usage.

·We present a strategy for selecting cluster heads that balances load efficiently.

·A path-choosing algorithm taking into account trustworthiness, residual energy, and shortest possible hop count is provided.

·The paper recommends a new binary algorithm inspired by the waterwheel plant algorithm (BWPA) for resolving the cluster head problem.

·The effectiveness of the projected BWPA algorithm for node selection is evaluated using a battery of computational analysis measures.

The paper is structured as follows: Section 1 delivers an impression of MANETs; Section 2 contains a literature review on the topic of security in MANETs; Section 3 details the proposed work; Section 4 details the experimental results and evaluation; and Section 5 provides a summary and final thoughts.

The projected routing scheme is alienated in two stages:

·Originally cluster heads are designated.

·After discovery paths, an optimal route is selected.

This study suggests a novel method that makes use of the BWPA to bring together mobility consciousness and energy conservation. Improved energy efficiency, and overall performance are all goals of the BWPA algorithm, which is designed to optimise CH selection in cluster-based MANETs.

3.1 Mobility concept

The success of routing algorithms and the reliability of clusters in a MANET are both affected by the nodes' capacity to move throughout the network. Poor cluster stability is the outcome of increased mobility, which causes frequent CH re- updates in cluster links. Our method uses mobility data to set up a stable path among less mobile entities, which reduces control overhead and improves stability. Using this strategy, we want to lessen MANETs' sensitivity to node mobility, improve cluster stability, and cut down on control messages that aren't strictly essential.

3.2 Model for energy

The success of a network depends on the routing protocol's ability to do more than merely find the most direct route from one set of nodes to another. With ME routing, the path chosen for packet transmission uses the least amount of energy overall, whereas with max-min routing, the path chosen uses the most energy within the constraints of the nodes. At regular intervals, this energy hub is supplied by:

$E_{\text {energy }}\left(t_j, \Delta t\right)=E_{\text {residual }}\left(t_j, l_0\right)-E_{\text {residual }}\left(t_j, l_1\right)$             (1)

The energy node here is characterised by Eresidual at times 0 and 1, correspondingly.

3.3 MANET clustering perfect

The node learned about other nodes by sending out a packet indicating their relative weight. Both the node's degree and its throughput of data contribute to the value of the node. It is the system's mobility and energy that serve as the basic criteria for determining the cluster and CH. Increased CH selection and interface refreshment due to mobility destabilises clusters.

In order to form a cluster, the CHs will broadcast a request for a packet to any collection of sensor nodes within radio range. In the single-node mode, data is transmitted unambiguously to the CH, whereas in the node relays its data to its neighbours. CHs are assigned to rounds in the setup phase in a completely random order. This permits the CH duty to be shared throughout the network's nodes and maintain consistent energy use. Once the CHs have been picked, they announce their function to the surrounding nodes, which subsequently select their own CHs based on the advertised roles. Network scalability is increased, and communication cost is decreased, thanks to this clustering process.

During the steady state, sensor nodes provide data to their associated CHs, which then send the data in bulk to the base station. In order to reduce the quantity of data being transferred and hence the amount of energy being used, the CHs apply data aggregation algorithms. To evenly spread the energy burden and increase the lifespan of the network, CHs are swapped out at the beginning of each new round.

3.4 CH selection

To facilitate communication between nodes in various clusters and supply radio transmitters to the participants, a CH has been set up. At the start of each round, each node chooses whether or not to be a CH with the probability qs_j(t) set so that D is the expected number of CHs. In this light, it follows that if the network has nodes:

$D=\sum_{j=1}^N q s_j(t)^* 1$          (2)

here, N is the total quantity of network nodes, while N-l represents ordinary nodes.

$q s_j(t)= \frac{Anticipated\,\, number\,\, of\,\, cluster\,\, heads}{Expected\,\, number\,\, of\,\, nodes\,\, are\,\, not\,\, cluster\,\, heads\,\, in\,\, most\,\, recent\,\, rounds}$              (3)

At least one of N-k's nodes will be CH. The likelihood of transformation into a CH is the same for every node in a cluster. This ensures that the energy at each node is about equal after each round. Nodes with more energy tend to cluster more readily than anodes with lower energy. Cluster nodes are expected to number:

$\begin{gathered}\text { Energy }[\mathrm{CH}]=D \Longrightarrow(N-D *(\operatorname{rmod} N / D)) * N / D-D *(\operatorname{rmod} N / D)\end{gathered}$              (4)

Although key administration approaches facilitate the distribution of keys to cluster nodes, the process becomes inefficient and introduces key overhead if keys must be produced for each node in the cluster. Different nodes are shown in different colours to highlight the many components of the overall structure. Each cluster's CH is the node with the fewest edges.

3.5 Optimum CH selection models

The purpose of the BWPA-based clustering approach is to fortify CHs. For MANET cluster stability to be attained, this is a prerequisite. In networks where CHs are used to relay data between nodes, it is important to remember that this does not guarantee efficient clustering merely by choosing the best possible CHs. This section first describes how to set up WWPA [26], using a model behaviour, and then on to define throughout.

3.5.1 Initialization

In WWPA, several people work together to solve a problem by iteratively searching for the best option. Due to the dispersed nature of the waterwheels throughout the search area, the WWPA population exhibits a wide range of values for the issue variables. Each waterwheel represents a unique vector that may be used to solve the problem graphically. The WWPA population, including all possible waterwheel variants, might be represented using a matrix. Waterwheel placement in the search area is decided randomly as the initial step of a WWPA deployment.

$P=\left[\begin{array}{c}P_1 \\ \vdots \\ P_2 \\ \vdots \\ P_3\end{array}\right]=\left[\begin{array}{ccccc}p_{1,1} & \cdots & p_{1, j} & \cdots & p_{1, m} \\ \vdots & \ddots & \vdots & \ddots & \vdots \\ p_{i, 1} & \cdots & p_{i, j} & \cdots & p_{i, m} \\ \vdots & \ddots & \vdots & \ddots & \vdots \\ p_{N, 1} & \cdots & p_{N, j} & \cdots & p_{N, M}\end{array}\right]$                     (5)

$\begin{aligned} & p_{i, j}=l b_j+r_{i, j} \cdot\left(u b_j-l b_j\right), \\ & i=1,2, \ldots, N, j=1,2, \ldots, m\end{aligned}$             (6)

where, N is the total number of waterwheels and m is the total sum of independent variables, r_i,j is an independent random variable in the range [0, 1], and Lower and upper bounds for the j^th problem variable are denoted by lb_j and ub_j, respectively; P is the populace matrix of waterwheel sites; P_i is the i^th waterwheel (a potential solution) and p_i,j is the j^th issue variable. The goal function for each waterwheel may be calculated separately since they each represent a different way of thinking about the problem. The objective function of the issue may be conveniently written as a vector of values using Eq. (7).

$F=\left[\begin{array}{c}F_1 \\ \vdots \\ F_i \\ \vdots \\ F_N\end{array}\right]=\left[\begin{array}{c}F\left(X_1\right) \\ \vdots \\ F\left(X_i\right) \\ \vdots \\ F\left(X_N\right)\end{array}\right]$             (7)

where, F is a vector of all the values of the objective value anticipated for the i^th waterwheel. In order to determine which options are the most advantageous, we mostly rely on objective function evaluations. This implies that the solution with the best member) is the optimal one. The poorest solution candidate (or worst member) is represented by the lowest value. Due to the waterwheels' random drive over the search area at each iteration, the current optimum will shift over time.

Phase 1: Position identification and hunting of insects

Because of their superior sense of smell, waterwheels become formidable predators. Any bug that comes into the waterwheel's range will immediately be attacked. Once it has located its victim, it immediately begins attacking and pursuing it. WWPA simulates this waterwheel action as the first step in its population update model. We may be able to enhance WWPA's exploration capabilities in discovering leads to huge fluctuations in the waterwheel's location in the search space. The waterwheel's new position is calculated by using Eq. (8) and a simulation of its approach to the bug. If moving the waterwheel to the location below increases function, the former location will be abandoned.

$W=r_1 \cdot(P(t)+2 K)$        (8)

$P(t+1)=P(t)+W \cdot\left(2 K+r_2\right)$        (9)

where, r₁ and r₂ are independent variables that might take on any value between zero and two. K is an exponential variable having values between zero and one, and W is a vector representing the diameter of the search circle in which plant will hunt for sites. If after three repetitions the answer does not change, then the waterwheel's position can be modified using Eq. (10).

$P(t+1)=\operatorname{Gaussian}\left(\mu_P, \sigma\right)+r_1\left(\frac{P(t)+2 K}{W}\right)$          (10)

Phase 2: Carrying the insect in the suitable tube

A waterwheel draws in phase of WWPA's population update is informed by this simulation of waterwheel operation. The WWPA's power is improved during the local search as a result of tube, which creates small waterwheel in the search space, and already been discovered. By assigning each waterwheel in the population a new random site as an "excellent position for devouring insects", the creators of WWPA mimic the waterwheel population's natural behaviour. If the value of the target function is larger than the initial location, as shown by the subsequent equations, the moves to the new site.

$W=r_3 \cdot\left(K P_{\text {best }}(t)+r_3 P(t)\right)$         (11)

$P(t+1)=P(t)+K W$        (12)

where, r₃ is a random mutable with values in the variety [0, 2], P(t) is the current key at repetition t, and P_best is the finest key.

Similar to the exploration stage, the following mutation is used if the solution does not improve after three rounds in order to prevent getting stuck in a specific local minimum.

$P(t+1)=\left(r_1+K\right) \sin \left(\frac{F}{C} \theta\right)$        (13)

where, [5, 5] is a range for the values of the random variables F and C. Eq. (14) also shows that the value of K falls off exponentially.

$K=\left(1+\frac{2 * t^2}{T_{\max }}+F\right)$          (14)

3.5.2 Search space binarization

The people who make up the BWPA search space are characterised as binary vectors. Each person in the search space is represented by a binary number. The ability to easily identify good and bad qualities is facilitated by this portrayal. First, the populace size dataset dimension D determine how many people will be involved in the search. The WWPA procedure, when practical to the search space, is meant to provide optimum solutions with a novel internal representation.

The whole optimisation process, which is anticipated to take many rounds, will be completed before the results are sent to each ind_i. The selected characteristics should be represented by cells with values of 1 s. For an arbitrary solution ind_i, the dimension of the dataset X, |F|, is generally equal to D. We count the sum of ones in dimension D for each occurrence represented by the variable ind_i in dataset X. When solving feature selection issues, binarization of WWPA might help by formalising the search space. The following is a description of the proposed BWPA strategy and how its parts interact with one another.

3.5.3 Binarization of WWPA

To optimise solutions in a discrete solution space, a new variant of the algorithm was developed by combining the capability of the original WWPA with many additional operators. In the first stage, transformation functions are defined to allow for the continuous to discrete transformation of the solution representation and optimisation process. This is crucial so that the innovative method may address issues associated with feature assortment. The second modelled process for producing the new variant of BWPA is to modify the fitness function. The best solution may be found by calculating the fitness of each choice. We provide the fitness function to account for the details of the current scenario. A schematic of the BWPA's inner workings and an analysis of its algorithmic structure are also provided. By applying the following sigmoid function to the continuous solution returned by the WWPA method, we may obtain a binary representation of the solution.

Binary $=\left\{\begin{array}{lr}1 & \text { if Sigmoid }\left(S_{\text {best }}\right) \geq 0.5 \\ 0 & \text { otherwise }\end{array}\right.$            (15)

$\operatorname{Sigmoid}\left(S_{\text {best }}\right)=\frac{1}{1+e^{-10\left(S_{\text {Best }}-0.5\right)}}$         (16)

3.5.4 Fitness and cost functions

A combination of assessing the function was necessary to find the optimal key to the feature selection problem. The subsequent equation illustrates how the solution is evaluated based on its performance using the classifier clf when the tuning parameter is applied to a subset of the dataset X[: 1(ind_i)]. When substituted for 1(ind_i) in the notation 1indi returns the total sum of 1s in the array.

fit $=\Omega *\left(1-\operatorname{clf}\left(X\left[: 1^{\text {ind }_i}\right]\right)\right)+\left((1-\Omega) \frac{|F|}{D}\right)$            (17)

The cost function is evaluated based on function by deducting the value repaid by fit from 1, as stated in the subsequent equation. The fitness and cost function values are used to visually assess and interpret each optimal key for a certain dataset.

$\cos t=1-f i t$     (18)

3.5.5 Fitness function

The function is used to measure the quality of the solutions fashioned by the suggested algorithm. The amount of characteristics utilised for organisation and the misclassification rate are the two primary factors affecting the fitness function. If a solution reduces the error rate and the sum of features used, it is considered a good solution. The effectiveness of each solution is calculated using the following equation.

$F=h_1 \frac{|s|}{|f|}+h_2 \operatorname{Error}(D)$         (19)

where, |f| and |s|, respectively, stand for the total sum of features and the sum of features that were chosen. Error(D) stands for the classification error. and the parameters h₁ and h₂ are used to control the significance of the chosen features using $h_1 \in[0,1]$ and h₂=1-h₁.

3.6 The algorithm for optimal path selection

Finding the shortest path between two sites is essential for reaching a destination. This algorithm provides a strategy grounded on the node's greatest energy and the fewest possible hops between nodes.

1. Route RREQ is produced by the basis node.

2. The RREQ header contains the RREQ ID, the terminus address, the arrangement number, the source address, the hop count, and the NmaxE.

3. If a node getting the packet has an advanced energy stays unaffected.

4. Keep checking until the RREQ reaches at its terminus.

5. The multiple RREQ gathered from various routes are used to select the best energy-efficient path.

6. If choose a communiqué route.

7. A RREP is repaid.

8. Data is directed via the afresh stated route.

If the aforementioned methods are applied, it is not required to choose a specific algorithm to select the path. This algorithm may allow us to find multiple paths with the same high energy. The alternate route must be picked if there are any crashes along the route. Thus, computation period can be shortened. Figure 1 is an illustration of finding the node with the most energy along the shortest path.

1.png

Figure 1. Communiqué path among source and destination