IWHO-Based Cluster Head Selection for Vehicle-to-Vehicle Communication in Intelligent Transport System

For millions of individuals worldwide, their primary mode of mobility is an automobile. As this mode of transportation becomes more and more popular, it will be crucial to permit vehicle-to-vehicle communication in order to provide passengers with more security and entertainment alternatives. The proliferation of personal automobiles, along with worsening traffic conditions and other risk factors, has raised the probability of accidents [1]. Software that aids the driver in making decisions and guarantees the safety of all passengers was developed in response to these issues. Improving fuel economy, reducing pollution, and helping the vehicle avoid traffic jams are further advantages [2]. In the entertainment sector, setting up a communication system between automobiles might have numerous advantages, such as the ability to exchange media (such as music and movies), make it easier for passengers to engage with other cars, other vehicles, and information kiosks at the roadside, and many more. One possible solution to the problem of inter-vehicle communication is the implementation of VANETs [3]. Disconnection from preexisting infrastructure is a defining characteristic of ad hoc networks [4]. The similar idea underpins vehicular ad hoc networks, which link tiny, battery-operated devices using radio waves.

Academic interest in vehicular ad hoc networks has been on the rise in the last few years. Reason being, VANET is crucial for fixing traffic and safety problems, boosting system performance generally, and ITS entertainment features [5]. Quick advancements in communication, electricity, and computer have enabled VANET to flourish. Transforming vehicles into data-gathering, -transmitting, and -displaying smart transport systems is the ultimate aim of Vehicle-to-V2V networks [6]. In order for a network node to establish a connection with another node in the same vehicle, both nodes must be physically close by. A vehicle node's usual communication range may be exceeded if it is required to connect with another vehicle node [7]. Data may be transported from one network node to another, independent of the position of the vehicle (the "source node"), by associating with intermediate nodes within transmission ranges and finding a way to connect with both the source and destination nodes [8]. Figure 1 shows the basic layout of the smart vehicle system.

1.png

Figure 1. Smart vehicle structure

Some problems and drawbacks with VANET include multipath fading and road obstructions, traffic jams, fluctuations in vehicle speed and mobility, road topography, traffic diversion models, drivers' erratic conduct on the road, and so on [9]. In order to establish dependable network services, it is necessary to address the numerous issues with VANET. Consequently, a key challenge in VANET is ensuring stable and dependable routeing. Consequently, it is necessary to use correct approaches in actual traffic scenarios as well [10]. Even more difficult is routeing in VANET due to the vehicle's dynamic nature and high mobility speed. Therefore, the most difficult part is choosing a VANET routeing protocol, which falls into one of five distinct groups defined by their unique routeing characteristics.

1.1 Bioinspired algorithms in VANET networks

The present ITS has been fortified and performance has been enhanced by VANETs and their many uses and capacities [11]. The use of VANET technology has, however, brought forth a slew of other problems. A great deal of research in this area has concentrated on space management, safety, and routing, the three cornerstones of vehicle networks. Clustering has been shown in several studies [12] to be an effective way to fix VANET's problems and make it more dependable and scalable in city settings. The process of sorting automobiles into fictitious groups called clusters is known as clustering [13]. Because of this, Cluster can improve the utilisation of network resources and provide safer, more reliable routing [14].

Current ITS frameworks have recently included bioinspired strategies to enhance them. The purpose of this research is to examine whether or not VANET networks may benefit from evolutionary approaches in order to address the following issues:

(1) Because animal behaviours in foraging for food and other needs are comparable to optimal route finding in VANET networks, algorithms that draw inspiration from nature operate better in these networks [15]. Optimising routes with little to no human involvement is also possible using algorithms inspired by nature. Algorithms developed from natural occurrences also make VANET networks more resilient and guarantee effective routing in many network conditions.

(2) Algorithms inspired by nature are well-suited to handle the large range of topological topologies observed in VANET networks due to their intrinsic flexibility and capacity for self-organization [16].

(3) They offer more accurate sensing of the network's damaged nodes since they integrate the greatest level of naturally inspired technology. By reducing the frequency of security attacks, it enhances the security of the network.

(4) Low complexity bioinspired techniques are advantageous for addressing VANET network computing problems.

1.2 Contribution of the research work

This paper suggests an intelligent enhanced wild horse optimisation to optimise clusters in VANET networks, based on the aforementioned benefits of bioinspired algorithms. The suggested techniques begin with vehicle searches and topology creation using self-adjusted weights to reduce inaccuracy. Next, cluster heads are selected for network administration in order to improve accuracy, taking into account each horse's specific fitness function. Presented above are the key points of the paper:

(1) A new intelligent IWHO for VANET network cluster optimisation is introduced in this study.

(2) The fitness function of each vehicle is used to assign a weight to each objective in an objective clustering.

(3) The suggested algorithm clearly offers a balance between exploration and exploitation. Because it remembers various answers and searches a large region to find the global solution, this feature allows IWHO to tackle a complex optimisation problem with numerous locally optimum solutions. Last but not least, the IWHO's combination of audible fine-tuning and gentle computations makes it the best and most novel method for optimising issues.

Here is how the balance of the paper is designed: In Section 2, the pertinent literature is reviewed; in Section 3, the system model is presented together with the suggested model; in Section 4, the consequences and discussion are given; and lastly, in Section 5, the conclusion is drawn.

This research filled a knowledge gap that had been previously identified by a literature analysis and background investigation. The current VANET routeing system has issues, such as the fact that classical and meta-heuristic algorithms can't handle the routeing difficulties of today. Therefore, a system model to fix routeing problems in different types of traffic is presented in this study. Subsequent sections elaborated on the research's communication model and methods.

3.1 Communication model

In this study, he opted for the optimization of routes using communication networks from vehicle to vehicle. The problems with routing, security and node management now affect VANETs that use vehicle-to-vehicle communication. The rise of IT has opened up new ways of research on the routing, security and mobility management of VANET. Not all models of network communication are well-suited to these research difficulties. Consequently, this study introduces the suggested communication paradigm that works in VANETs with both sparse and dense traffic.

3.2 Parameter selection for cluster head

The statement asserts that utilizing an Improved Wild Horse Optimization (IWHO) algorithm for cluster head selection in Vehicular Ad-hoc Networks (VANETs) represents a novel approach that contributes new knowledge to the field, which is an innovative strategy that adds to the set of existing knowledge. Sin embargo, the veracity of this affirmation is difficult to verify in the absence of concrete tests or results of studies. In the framework of VANET research, the technique, the experimental configuration, the results and the comparison with the actual methods to determine the originality and the value of the IOMS algorithm for the selection of group heads in VANET should be examined. To add weight to the relevance of the suggested algorithm, it must be reviewed by peers and validated by the scientific community.

Several measures can be implemented to resolve this problem and improve the clarity and practicality of the methodology:

Detailed Parameter Description: Provide a comprehensive explanation of each parameter used in the methodology, including their roles, potential values, and impacts on the algorithm's performance. This ensures readers understand the significance of each parameter and how they affect the results.

Analyze the impact of changing parameter values on the performance of the algorithm by performing a sensitivity study. The objective of this study is to find the most important parameters and the best values for them in all possible situations and sets of data.

Parameter Tuning Guidelines: Offer guidelines or recommendations for parameter tuning based on experimental results or empirical observations. This could include strategies such as grid search, random search, or heuristic methods to find suitable parameter configurations.

3.2.1 Efficient energy consumption

The study uses a technique to evaluate the energy consumption efficiency of the Head of Clusters (HOCs) by calculating the total leftover energy. Whether the cumulative balance energy of the HOCs increases or decreases determines how well the approach performs overall. Every iteration, the node with the highest balancing energy is selected as the HOC node.

3.2.2 Reliability

The degree of confidence in data transmission is what the word "reliability" alludes to. Since WSN nodes run on batteries, reducing energy usage during data transfer can improve dependability. So, extending the life of the network and decreasing the cost of communication outside the cluster are both achieved by reducing energy consumption at the nodes. The sum of all costs incurred by non-cluster communication is known as the global cost of communication.

3.2.3 Energy consumption framework

Data transmission in the form of bits is the intended design of the energy consumption framework. For different network tasks involving communication with other nodes, it contains energy loss computations at each node. Both the octal-channel broadcasting method and the dual-channel broadcasting approach assess energy use during data transmission, with the latter also providing an estimate of energy loss during multi-hop data transfer. In mathematics, this framework is represented by Eq. (1).

$F^{S W}\{k, c\}=\prod\binom{k F^{s e l}-k F^{e t} c^{i+1},(c-1) \leftarrow c^i}{k F^{s e l}-k F^{n o} c^{i+4},(c+1) \leftarrow c^i}$     (1)

where, F^et is the loss of energy at the single hop, F^no is the loss of energy at multi-hop. F^sel is digital modulation, signalling, computerised coding, sorting, and other factors comprise the electrical energy. Eq. (2) mathematically expresses the hybrid separation as co, where c is the distance among the transmitter and receiver nodes.

$c^o=\int \frac{F^{e t}}{F^{n o}} \times \sum F^{s W}\left\{k, \sum c\right\}$    (2)

The energy spent by the vehicular sensor devices to receive the message F^QW is exactly uttered as Eq. (3).

$F^{Q W}(k)=e^{\Sigma c} k F^{s e l}+\bigvee \frac{F^{e t}}{F^{n o}}$      (3)

So, the physical and medium access control layer of WSN is constructed to contain the energy spent during (i) message transmission and (ii) message reception.

3.3 Optimal selection of HOC

3.3.1 Wild horse optimizer (WHO)

The wild horse optimizer (WHO) method exactly mimics the social behaviour of wild horses in nature to solve optimisation problems [25]. Typically, a herd of horses will consist of a stallion and many mares and foals. Mating, grazing, chasing, controlling, and commanding are just a few of the behaviours they display. The following are the five stages of the WHO algorithm:

A). Generating an initial populace and creation horse groups and selecting leaders

A large sum of subsets is first created from the basic population. The algorithm's group count is G, and the population size is N. The method uses a stallion as the head of each group, with G being the number of stallions. The remaining populace, consisting of foals and mares, is distributed groups, with (N-G) being the difference.

B). Grazing behavior

The subsequent equation was projected to fake the grazing behavior:

$\begin{gathered}X_{i, G}^j=2 Z \cos (2 \pi R Z) \times\left(\text { Stallion }^j-X_{i, G}^j\right) + \text { Stallion }^j\end{gathered}$     (4)

where, $X_{i, G}^j$  denotes the current site of member, Stallion^j is the stallion position, R is an unchanging stochastic sum from the range [-2,2], besides Z is the subsequent equation:

$\begin{gathered}P=\overrightarrow{R_1}<T D R ; I D X=(P==0) \\ Z=R_2 \Theta I D X+\vec{R}_3 \Theta(\sim I D X)\end{gathered}$      (5)

where, $P$ is a vector containing of 0 to $1, \overrightarrow{R_1}$ and $\vec{R}_3$ are a random sum from the variety [0,1], R₂ is a sum from the range [0,1]. According to the subsequent equation, TDR is an adaptive parameter that begins at 1 and declines until it grasps 0 at the conclusion of the algorithm's implementation:

$T D R=1-i t \times\left(\frac{1}{\text { maxit }}\right)$       (6)

where, it is the present repetition and maxit is the extreme sum of iterations.

C). Horse mating behavior

As part of the mating process, a foal will move from group, and a foal from group j will do the same. It was suggested that the following Crossover operator of the mean type be used to model horse mating behaviour:

$\begin{gathered}X_{G, K}^P=\operatorname{Crossover}\left(X_{G, i}^q, X_{G, j}^Z\right) i \neq j \neq k, p=q=\text { end } \\ \text { Crossover }=\text { Mean }\end{gathered}$       (7)

D). Group leadership

The WHO algorithm depicts a herd's leaders, or stallions, guiding the rest of the herd to a watering site. In order to ensure that the dominant group uses this water source first, the stallions compete with one another for it. At this stage of the procedure, the following equation was suggested:

$\overline{\text { Stalllon }}_{G_i}$$=\left\{\begin{array}{c}2 Z \cos (2 \pi R Z) \times\left(W H-\text { Stallion }_{G_i}\right)+W H \text { if } R_3>0.5 \\ 2 Z \cos (2 \pi R Z) \times\left(W H-\text { Stallion }_{G_i}\right) -W H \text { if } R_3 \leq 0.5\end{array}\right.$          (8)

where, $\overline{\text { Stalllon }}_{G_i}$ is the subsequent site of the leader. WH is the hole.

E). Leader’s interchange and selection

Leaders are designated based on their fitness in the subsequent phases. According to this formula, the leader and the important member will switch places:

$\begin{aligned} & \overline{\text { Stallıon }}_{G_i} =\left\{\begin{array}{c}X_{G, i} \text { if } \cos t\left(X_{G, i}\right)<\cos t\left(\text { Stallion }_{G_i}\right) \\ \text { Stallion }_{G_i} \text { if } \cos t\left(X_{G, i}\right)>\cos t\left(\text { Stallion }_{G_i}\right)\end{array}\right.\end{aligned}$      (9)

3.3.2 Improved wild horse optimizer (IWHO)

The cuckoo search (CS) method is the foundation of the (IWHO) [26]. The following equation is used to produce new solutions throughout iterations of the proposed method, which utilises the Levy flight:

$\begin{aligned} & X_{i, G}=X_{i, G}-\gamma\left(X_{i, G}-X_g\right) \oplus \operatorname{Levy}(\lambda)=X_{i, G}+\frac{0.01 u}{|v|^{1 / \lambda}}\left(X_{i, G}-X_g\right)\end{aligned}$     (10)

where, X_i,G is the ith site, γ designates the phase scaling size, X_g designates the global best solution, the ⊕ mentions to multiplications, λ denotes to the Levy flight advocate, while u and v are well-defined as:

$u \sim N\left(0, \sigma_v^3\right), v \sim N\left(0, \sigma_v^2\right)$    (11)   

The standard deviations σ_u and σ_v are expressed as:

$\sigma_{\mathrm{u}}=\left[\frac{\sin \left(\frac{\lambda \pi}{2}\right) \cdot \Gamma(1+\lambda)}{2^{(\lambda-1)} \lambda \cdot \Gamma\left(\frac{1+\lambda}{2}\right)}\right]^{1 / \lambda}, \sigma_v=1$     (12)

Eq. (6) is used to construct the new candidate solution, where Γ is function. One major benefit of this upgrade is that the suggested method can strike a better local exploitation. Lastly, it utilises a GA-based mutation method to solve the problem of quicker convergence, which is the same idea as in the study [27-29].

3.4 Cluster creation

Creating and grouping the cars into several clusters is the second step after selecting the cluster heads using the suggested technique. As an alternative to the conventional distance-based clustering, this method clusters nodes according to their link reliability (REL). Taking REL into account is important since it reveals how well data packets may be sent with little chance of link failure. This metric measures the robustness and efficiency of the system. The reliability of the connections between cars is not shown by their geometrical distance.

All nearby vehicles can receive ELECTED Messages sent by a CH. The ELECTED Message contains the following information on the cluster head: its ID, location, speed, and direction. The time when the cluster leader sends the ELECTED Message can be determined by examining the Life Time. In response to an ELECTED Message, every node k communicates with the CH sender by sending a JOIN Message. The message contains the following information about the node: its location, velocity, direction, and cluster head ID. Node k will conduct a test if it gets more than one ELECTED Message. CHs is calculated by node k. This node has sent the JOIN Message to indicate its intention to join the CH that has a higher REL value. A node k's reliability connection to a CH may be determined by applying the following formula.

$\begin{aligned} & R E L\left(\operatorname{link}_k, C H_i\right)=\operatorname{Erf}\left(\frac{\left(\frac{2 R}{t}-\mu_{\Delta v}\right)}{\sigma_{\Delta v} \sqrt{2}}\right)  -\operatorname{Erf}\left(\frac{\left(\frac{2 R}{t}+T_p-\mu_{\Delta v}\right)}{\sigma_{\Delta v} \sqrt{2}}\right), \text { when } T_p>0\end{aligned}$        (13)

that is, the GAUSS error function is represented by Gerf. The fields of probability, statistics, and diffusion often make use of this particular integer function. Here is its definition:

$\operatorname{Erf}(x)=\frac{2}{\sqrt{\pi}} \int_0^Z e^{-t^2} d t$      (14)

where, µ and σ characterize speed correspondingly. ∆v designates the difference in speed among two vehicles.

T_p shows how long it is likely that a connection L between two cars will be accessible. Two scenarios are used to compute it:

Scenario 1: Equally vehicles transfer in the similar direction.

$T_p=\left\{\begin{array}{l}\frac{2 R+L_k, C H_i}{\left|V_k+V_{C H_i}\right|} \text { if the vehicles move towards each other } \\ \frac{R-L_k C H_i}{\left|V_k+V_{C H_i}\right|} \text { if the vehicles move towards each other }\end{array}\right.$    (15)

Scenario 2: The diverse directions.

$T_p=\left\{\begin{array}{l}\frac{2 R+L_k, C H_i}{\left|V_k+V_{C H_i}\right|} \text { if the vehicles move towards each other } \\ \frac{R-L_k C H_i}{\left|V_k+V_{C H_i}\right|} \text { if the vehicles move towards each other }\end{array}\right.$      (16)

The two vehicles' speeds are denoted by V_k and $V_{C H_i}$, respectively, and R is the area of gearbox. L is the distance in Euclidean terms between vehicles k and $V_{C H_i}$.

In order to exchange and retrieve the data needed for the formula, when the network is initialised, every vehicle broadcast HELLO packets to all of its nearby cars (13). Primarily, the HELLO packet structure includes the subsequent elements: preamble, life time, location, velocity, and direction.

The main phases in creating clusters are:

1) Apiece broadcasts HELLO mails to share flexibility info.

2) The CH_i transmissions ELECTED communications to vehicles.

3) If vehicle k receives a solitary, it directly sends a JOIN message back to CH_i to join the cluster.

4) When vehicle k gets more than one ELECTED message, it uses all of the available cluster heads to determine the REL value. The JOIN message is sent by vehicle k to the cluster head that has the highest REL value in order to join the cluster.

3.5 Cluster maintenance

When new vehicles enter or leave a cluster, the cluster heads will occasionally send ELECTED Messages to all accessible cars to structure. This is how the new vehicle might become a part of the cluster—by getting the ELECTED Message. Every vehicle that gets this kind of message works with the cluster head senders to calculate the REL value. The vehicle is eligible to join the cluster due to its high REL charge [30].

If any of the cluster heads decide to step down, the process for electing new leaders can be repeated.

The cluster keep phase deliberates three conceivable cases:

1. Cluster heads permission the cluster.

2. Vehicles leave the clusters.

3. Novel vehicles access the clusters.

At regular intervals, the cluster heads will transmit ELECTED Messages to all available cars in order to refresh the cluster structure whenever new vehicles join or depart the clusters. By receiving the ELECTED Message, the new car can join the cluster in this fashion. The REL value is computed by each vehicle that receives this sort of communication in conjunction with the senders of the cluster heads. The car's high REL value qualifies it to join the cluster.

It is possible to choose new cluster leaders by repeating the head election procedure in the event that any cluster heads resign.

1) Cluster Stability Performance Metrics

Cluster stability in VANETs is shown by a number of measures. Both the constancy of cluster leaders and members are included in these measurements.

A: Cluster Lifetime

- Cluster head lifetime: The lifespan of a vehicle head is the amount of time it stays in that role. An extended lifespan indicates that the clusters are quite stable. To get it, divide the total amount of time that cluster heads have been assigned roles by the total sum of roles in the cluster.

- Cluster member lifetime: The average term of being a member of a cluster is shown by the members lifespan. Cluster stability increases as this measure becomes longer. By isolating the total time of cluster role assignments by the total number of cluster role assignments, we may get the lifetime of each cluster member.

- Change rate of cluster head: The frequency with which a vehicle a certain time frame is represented by this measure. It is calculated by separating the sum of cluster heads assigned roles by a time period, often the duration of the simulation. This value decreases as clustering efficiency increases.

The complexity and unpredictability of real-world settings may be better captured by empirical validation on actual networks than by simulations, which may not provide a realistic representation of algorithm behaviour.

The term "empirical validation" refers to the process of testing the algorithm in real-world settings, collecting data, and evaluating its performance in various situations. Researchers may test the algorithm's resilience, scalability, and practicality using this method.

The statement claims that more testing on real networks is necessary to confirm the algorithm's effectiveness and applicability to (VANETs) and related domains, regardless of how promising the simulation results may be. With this form of verification, the reliability and validity of the study results will be enhanced.