Role of Roadside Units in Cluster Head Election and Coverage Maximization for Vehicle Emergency Services

Abstract

Efficient clustering algorithms are critical for enabling the timely dissemination of emergency messages across maximum coverage areas in vehicular networks. While existing clustering approaches demonstrate stability and scalability, there has been a limited amount of work focused on leveraging roadside units (RSUs) for cluster head selection. This research proposes a novel framework that utilizes RSUs to facilitate cluster head election, mitigating the cluster head selection process, clustering overhead, and broadcast storm problem. The proposed scheme mandates selecting an optimal number of cluster heads to maximize information coverage and prevent traffic congestion, thereby enhancing the quality of service through improved cluster head duration, reduced cluster formation time, expanded coverage area, and decreased overhead. The framework comprises three key components: (I) an acknowledgment-based system for legitimate vehicle entry into the RSU for cluster head selection; (II) an authoritative node behavior mechanism for choosing cluster heads from received notifications; and (III) the role of bridge nodes in maximizing the coverage of the established network. The comparative analysis evaluates the clustering framework’s performance under uniform and non-uniform vehicle speed scenarios for time-barrier-based emergency message dissemination in vehicular ad hoc networks. The results demonstrate that the proposed model’s effectiveness for uniform highway speed scenarios is 100% whereas for non-uniform scenarios 99.55% information coverage is obtained. Furthermore, the clustering process accelerates by over 50%, decreasing overhead and reducing cluster head election time using RSUs. The proposed approach outperforms existing methods for the number of cluster heads, cluster head election time, total cluster formation time, and maximum information coverage across varying vehicle densities.

1. Introduction

VANETs are a type of Mobile Ad Hoc Network (MANET) where vehicles fitted with On-Board Units (OBUs) are in communication with each other (V2V) and with roadside units (V2I or V2RSU) without relying on any fixed base station. VANETs facilitate various applications, such as traffic safety, congestion control, and emergency message broadcasting, by providing real-time communication and data exchange between mobile road nodes.

Wireless communication between the mobile nodes and RSUs, or within the vehicles, will have a platform on the road [1,2]. The United States Department Of Transportation (US-DOT) is actively working on the deployment of VANET by utilizing its 5.9 GHz of Dedicated Short-Range Communication (DSRC) bandwidth [3]. The Society of Automotive Engineers (SAE) DSRC International is developing standards for supporting research, industry, and academia to optimize V2X communication [4]. In [5], the researchers used the standard SAE J2375 to describe the application-specific data dictionary intended to use the VANET safety application. This will enable the implementation of various VANET safety-specific applications, such as the convenience, comfort, traffic congestion avoidance, assistance for road jams, prior information on road events, and safety of drivers and passengers [6,7,8,9]. The application of safety measures requires the timely delivery of emergency information using multi-hops that are further away, which raises scalability issues due to the distributed nature of VANET. Thus, it is necessary to utilize the network structure’s hierarchical paradigm rather than the balanced approach. Generally, grouped vehicles form a cluster [10] to increase network performance. Clustering algorithms were initially proposed in 1997 [11,12]. In the VANET research area, some eminent clustering algorithms have been introduced that include [10,12,13,14,15]. However, some studies (e.g., [16,17,18,19,20]) found that in the timely delivery of emergency packets, mobility metrics play a vital role. Thus, in selecting relevant clustering metrics for effective CHs, there is a need to improve cluster connectivity and maximize coverage. According to [10], a cluster’s performance depends upon the connectivity and the coverage area of a cluster while transferring emergency information. Figure 1 illustrates the interaction between vehicles and RSUs in a V2V and V2RSU hybrid communication model, which forms the foundation for the proposed clustering mechanism.

1.1. Main Contribution

The main limitation of the existing research is in the clustering procedure where mobile nodes are responsible for CH selection [10,17,18]. However, using RSUs in the proposed scheme provides new directions for upcoming researchers. With RSUs, the load of vehicles can be decreased while moving on the road, making communication faster and more efficient. In addition, adding the clustering process enhances connectivity in the middle of the road. The broadcast storm, overhead, maximum coverage area, and inefficient bandwidth usage are the key metrics that need improvement [2,21,22]. Considering the above facts, this paper proposed the clustering framework to maximize the information coverage. To undertake this task, RSU identified mobility metrics to elect the CH, thereby maximizing information coverage and reducing clustering process time. The following are the main contributions of the paper:

• The maximum information coverage scheme is proposed where all vehicles are Legitimate Vehicles (LV) and the RSU will initialize the communication.
• To reduce the computation and complexity for selecting efficient CH, $\mathcal{O}\left({log}_{2}N\right)$ complexity is used as an acknowledgment-based approach.
• RSU acts as the backbone communication system and the vehicles use a choice-based mechanism to become CM by considering the mobility metrics. However, the studies [10,19,23] used a neighbor table for CH selection, therefore reducing cluster overlapping.
• The new bridge node (BN) selection algorithm is proposed for Cluster-to-Cluster connectivity. The solution proposed two BNs for one CH. If no BN is available, the CH itself serves the role of BN and sends an update to RSU. In [24,25,26,27], cluster connectivity is achieved through broadcasting or using gateway nodes. However, the number of gateway nodes was not fixed. This leads to redundant information, residual energy wastage, and high bandwidth utilization. However, in the paper, the flagship request was served using the bridge ID. This scheme drastically decreased the broadcast message count and reduced the broadcast storm problem [28].
• Clustering metrics are divided into microscopic and macroscopic levels according to their relevance and usage. A comparative analysis of the proposed scheme, using the time barrier-based emergency message scheme, has been made. The proposed model is divided into two traffic scenarios based on uniform and non-uniform vehicle speeds. The counter model is integrated into the proposed scheme.

1.2. Organisation

The rest of the sections of the paper are arranged as follows. Section 2 discusses the related work. Section 3 introduces the proposed model’s details. Section 4 presents the analysis of the proposed scheme’s simulation work compared to the selected models. Section 6 concludes the paper and provides future directions for study.

2. Related Work

Clustering is a crucial process in VANETs, enabling efficient message dissemination and resource utilization. The performance of clustering algorithms depends on various factors, such as application requirements and mobility metrics. This section reviews the prominent clustering approaches proposed in the literature.

Various methods were proposed in [15,29,30] to optimize message dissemination between V2V and Vehicle to Road Side Unit (V2RSU). The choice of CH is optimized because it adheres to the responsibility of cluster formation and management. The node selected as a CH possessed responsibilities to perform operations such as data forwarding, efficient data aggregation, and inter- and intra-cluster transmissions. CH broadcasting to the cluster nodes, other CHs, and RSU improves the message passing reliability and reduces the chances of link breakage. For diminishing the effect of link breakage and hidden terminal problems, vehicle-to-RSU communication holds more remarkable results than V2V communication [10]. An effective solution is presented in [31] for organizing the network and urban sensing applications. Localization data, speed, direction, and distance are the key parameters. These parameters describing moving vehicles are sent to a central server to reduce overhead. Ref. [32] highlighted the use of hypergraph modeling, the improved clustering algorithm (iTTM), the CH selection process based on multiple vehicle attributes, and the multi-decision CRITIC approach. Additionally, it emphasized the simulated results in terms of CH stability, throughput, packet delay, and switching frequency compared to existing algorithms.

Moreover, the RSU is selected as the centralized system for communication during clustering in [33]. RSU works fast in handing over road details to another RSU, thus improving network joining delay, media access collisions, and packet transmission rate compared to other studies. The unified framework of the clustering approach [28] proposed a neighbor sampling scheme to filter out stable nodes and backoff CH selection schemes to maintain link connectivity. However, this scheme lacks connectivity for information coverage. Other research works reported the challenges faced in multicast communication in VANETs, the importance of addressing factors such as energy balancing, load balancing, connectivity, and coverage, and the proposed clustering-based multicast routing approach that employs cluster head selection [26], whereas the time barrier scheme [14] defined the time window for more stable nodes selection based on calculated CH eligibility score.

To address the broadcast storm problem and reduce the impact of flooding in urban VANET scenarios, the study in [34] proposed a path-aware hierarchical structure. This approach mitigates excessive message rebroadcasts by leveraging vehicle mobility patterns, thereby reducing transmission delay, improving packet delivery ratio, and enhancing throughput and dissemination efficiency. However, ref. [35] conducted a highway study focused on CH selection and achieved more stable clusters. Thus, we can conclude that the CH selection is very important and decreases drastically the clustering time. In fact, in the [23] review article, the authors mentioned that in maximum articles, the CH and suitable CMs had a predetermined size of clusters that led to worsening the outcomes. They suggested evaluating CH selection time, cluster construction delay, and information transmission delay. Additionally, for cluster maintenance, loss of information is there due to CM leaving; a timer for the contention window is another solution provided.

In [36], authors described the information-centric metrics with respect to k-means clustering and without a clustering mechanism. They described cluster maintenance as required when (i) CH is changed or merged, or to track a lost link. In addition, ref. [24] proposed a handover scheme where the author mentioned an intelligent CH in the communicating range of two clusters. This method is a specific range-based method and has a limited scope of connectivity. This also increased the network load to the CH. Another drawback is that it acts as a controller for two clusters, which will halt the clustering structure if the connection breaks. Moreover, in [37], the RA-HEMB protocol proposed a target broadcasting region based on the emergency type and real-time traffic conditions. An adaptive position-based forwarding scheme was used to dynamically adjust the forwarding area based on vehicle densities and inter-vehicle ranges. This forwarding process leverages both vehicle-to-vehicle (V2V) communications as well as infrastructure support from RSUs over wired links. The goal is to balance broadcast delivery latency and reliability using the hybrid V2V/RSU approach. The simulation results showed that RA-HEMB can significantly reduce emergency notification times compared to benchmark broadcast protocols. While RA-HEMB utilizes infrastructure support, other works have focused on developing infrastructure-less solutions tailored for rural highway scenarios.

From the above research work, it has been concluded that the CH selection mobility metrics play a crucial role. There is a need for improvement in CH selection, with this being a time-consuming process during emergencies. Additionally, the information coverage enabling the consideration of a maximum number of connected nodes in the selected area is an important parameter, where having gateway nodes is one solution that can assist with this. However, the provided solutions for gateway nodes increase the broadcast storm problem that leads to flooding. Hence, the overall clustering time needs to be minimized. The pre-calculated metrics and exchange of messages increase the overhead and waste network resources.

The proposed solution enables the usage of RSU as a backbone and initiates the clustering procedure. Every upcoming node becomes a part of communication with or without clustering, wherein communication range is a crucial factor. These nodes interact with RSU and become part of the CH selection. The proposed solution releases the load of an LV using a distributed approach. RSU calculates the CH from the acknowledged LVs according to its mobility metric. This structure drastically reduces the CH selection time. RSU also sends the information of the last CH and BN, if it exists in the traffic, up to the $New\_NODE$. This will help $New\_NODE$ to either be part of CH or $AckRSU$ to re-cluster. Hence, CH is selected through RSUs and provides maximum connected nodes for information coverage, which is the main objective of this research article. New NODE refers to a vehicle that has recently entered the RSU communication range and is not yet part of any existing cluster. AckRSU denotes the acknowledgment mechanism by which a Legitimate Vehicle (LV) communicates with the RSU, confirming its presence and willingness to participate in the clustering process.

3. Proposed Solution

This article proposes a maximum coverage-based algorithm that adheres to distributed hybrid structures having V2RSU and V2V communication. All the vehicles are theoretically equipped with On-Board Units (OBUs), a Geographical Positioning System (GPS), and RSUs to a DSRC chipset [29,38]. It has been observed that the typical range of a DSRC access point is about 1000 m and for OBUs this 300 m [39]. Moreover, each RSU can calculate the relative distance between the vehicle and the RSU, whereas each vehicle is responsible for calculating the direction, relative speed, and relative distance within the communicating range. The RSU calculates the Euclidean distance between itself and each incoming vehicle using the following formula:

$${\mathrm{Dist}}_i=\sqrt{{(\mathrm{Sin}\mathrm{kX}-{\mathrm{NodeX}}_i)}^2+{(\mathrm{Sin}\mathrm{kY}-{\mathrm{NodeY}}_i)}^2}$$

(1)

where

• SinkX, SinkY represent the geo-coordinates of the RSU.
• NodeX_i, NodeY_i represent the geo-coordinates of the $i^{th}$ Legitimate Vehicle (LV).
• Dist_i denotes the calculated Euclidean distance between the RSU and the $i^{th}$ LV.

The hybrid communication clustering procedure is described in the following sections, including CH selection, cluster formation, and bridge node communication to maximize the count of connected nodes.

3.1. Vehicle States

In our proposed clustering scheme, there are five roles:

• RSU: This infrastructure is responsible for performing the CH selection only from the vehicles where acknowledgment has been received.
• LV: Those vehicles that intend to be part of communication.
• CH: The leader of the communicating group provided by RSU as a list of CHs from the acknowledged LVs. The matched $V_{id}$ of LV will change its status to CH and broadcast its message structure to the rest of the vehicles.
• CM: These are the rest of the LVs left after the CH selection and are given the right to choose the CH after receiving a message from the designated CH. They will change their status to CMs once the CH.
• BNs: The farthest nodes selected by CHs from the connected CMs to bridge the gap of communication between two clusters.

The states are described through Algorithms 1 and 2 under the proposed clustering scheme to maximize the number of connected nodes for information coverage during emergencies in VANETs.

Algorithm 1: Acknowledgement-Based Legitimate Vehicle Entry System to RSU for CH Selection

3.2. Acknowledgement-Based Legitimate Vehicles Entry System to RSU for CH Selection

To begin the clustering procedure in Algorithm 1, RSU serves the backbone system, broadcasts its unique ID in a message $<RSU\_id>$ to all the incoming vehicles within the transmission range (e.g., 1000 m), and starts its timer $T\_window$. The mobile nodes that are approaching or leaving the RSU but are in the communicating range of 300 m acknowledge the RSU with message structure $<V\_id,Dir,V_i,NodeX,NodeY>$. The RSU populates the $C{T}_i$ table whenever a new vehicle approaches after verifying its $V_{id}$. After processing Algorithm 1, the new $C{T}_{iupdated}$ table is formed, having a sorted list of vehicles according to relative speed, direction, and relative distance. The RSU broadcasts $C{H}_{id}$ to all the vehicles. The selection of CH focused on achieving the time complexity of $\mathcal{O}\left({log}_{2}N\right)$. RSU computes $Lo{g}_{2}N$ values for selecting CHs. However, $\mathrm{CHs}\subset \mathrm{LVs}$ considered the farthest distance of the node using the Euclidean distance to form the cluster.

Furthermore, the table entry is cleared, and the new process is initiated once the RSU receives a cumulative acknowledgment from CHs. This helps to avoid over-populating the table entry and potential buffer overflows. The LVs check the value of $CH\_id$. If $V\_id$ is similar to $CH\_id$, $CH\_Status$ is set to 1. The CHs broadcast the current information containing $<CH\_id,Updated\_NodeX,Updated\_NodeY,Dir,V_i>$ to all the vehicles within the range.

3.3. Authoritative Behavior of Remaining LVs for Choosing CHs from Receiving Signal of Nearby CH: Cluster Formation

The following assumptions are made for the participation of LVs after the CHs are chosen by RSUs.

1.

To maximize the information coverage, the test scenario vehicles have been maintained with specified margins. The specific margins are made considering the minimum safety distance. This helps to gain the maximum connected vehicle count.

2.

Every node is in the range of an RSU.

3.

Every node is assigned to at least one CH.

4.

Each CH nominates two bridge nodes to mitigate the broadcast storm problem.

By having such assumptions, the researchers are trying to figure out real-time environment setup solutions for the deployment in industries. Considering the defined assumptions and providing the cluster framework will help the deployment agencies to obtain accurate results for the real-time setup, since they will have access to the defined mobility metrics as considered in Algorithm 2.

Algorithm 2: Authoritative Behavior of Remaining LVs for Selecting the CHs: Cluster Formation

Each $L{V}_i$ maintains a set of data structures $<CHids,X,Y,V_i,Dir>$ and takes a self-decision for selecting appropriate CH as per the defined metrics in Algorithm 2.

$$\begin{array}{cc}{z}_i\hfill & \le Max\left({\displaystyle \frac{1}{\mathrm{\Delta }{v}_i{d}_i}}\right)\hfill \\ \mathrm{\Delta }{v}_i\hfill & =v_2-v_1\hfill \end{array}$$

(2)

where $z_i$ defines the CH and $\mathrm{\Delta }{v}_{id}$ be the relative speed of vehicles.

Moreover, V2V communication is calculated using a timestamp $T_t$. The $REQUEST\_JOIN$ message is sent to the respective CH by a CM that verifies the assumptions. Each CH maintains a table with sorted CMs. These connected CMs are used to select the farthest nodes of CHs, named the bridge nodes. These mobile nodes are responsible for connecting the two clusters to create a maximum connected vehicle topology.

Here, $L\_{NBR}_{id}$ and $R\_{NBR}_{id}$ are Left Bridge Node id and Right Bridge Node id, respectively. Furthermore, the CH message is responsible for making the following decisions:

• Broadcasting messages in the format of $<CH\_ids,X,Y,V_i,Dir>$.
• $CH\_ids$: denoted the CH ID.
• $X,Y$: current coordinates of CHs.
• $V_i$: denoted the speed of CH.
• $Dir$: denoted direction of the vehicle.
• $L\_{NBR}_{id}$ & $R\_{NBR}_{id}$ : denoted bridge nodes for connecting two clusters.

3.4. Cluster Maintenance

Cluster maintenance is a crucial step after a complete clustering cycle. The cluster maintenance algorithm is given in Algorithm 3. The cluster maintenance is divided into three components, including CM joining, CH update, and RSU update. The timer $T_n$ is initiated when any new node arrives in the network. The timer is attached to calculate the total time taken by a new node to become a part of the existing CH or to initiate a new clustering process. RSU starts reconstructing the cluster only after receiving the $New\_Node$ notification because cluster maintenance relies on the one assumption and two conditions listed below.

Algorithm 3: Cluster Maintenance

1.

Assumption:

(a)

RSU already identifies the arrival of $New\_Node$.

2.

Conditions:

(a)

If there exist vehicles, RSU will send the last CH and BN details to $New\_Node$. $New\_Node$, after verifying its route id $R_{id}$, destination id $D_{id}$, and relative distance $D_{ir}$, will select the CH. Thus, CH updates its BN.

(b)

If $New\_Node$ does not have relevant matched mobility metric criteria to become the member of existing CH, it will send a $REPLY$ to RSU for re-clustering only if it is in transmission range, i.e., $<=300$ omnidirectional. Till that point, the node is considered to be a free node.

To support the above-mentioned arguments of cluster maintenance, we present the following two use cases.

3.5. When a New Node Arrives and Road Information Is Provided by RSU

Consider the $New\_Node$ coming after the delay of 2 s and receiving a broadcast from RSU in 0.01 ms. The message structure consists of $<RS{U}_{id},C{H}_{id},CHLB{N}_{id},C{H}_{XCoord},$ $C{H}_{YCoord},CHLB{N}_{XCoord},CHLB{N}_{YCoord},$ $C{H}_{Rid},C{H}_{Did}>$. We assume a two-dimensional Cartesian coordinate system for calculating the relative distance between two nodes. The coordinates $(x_1,y_1)$ and $(x_2,y_2)$ represent the geographic positions of two vehicles in meters (m). The relative distance $d_i$ between them is computed using the Euclidean distance formula as shown below:

$$d_i=\sqrt{{(x_2-x_1)}^2+{(y_2-y_1)}^2}$$

(3)

Here, $d_i$ is measured in meters (m), and it is used to determine whether a node lies within the communication range (≤150 m) for potential cluster participation.

The new node calculates the relative distance between the CH and itself, as well as between the bridge node (BN) and itself, using the above formula.

Two cases arise:

1.

If the $d_i<=150$ and $CH{r}_{id}==New\_Node$ and $CH{d}_{id}==New\_Node$ between CH and New$\_Node$, then according to the bridge node selection methodology, it has been assumed that the CH is also BN and the last node of traffic flow. The New$\_Node$ sends $Request\_Join$ to CH and CH sends the $B{N}_{id}$ to the New$\_Node$. According to the simulation results, it is found that the completion of this process increases by 1.536 ms (approximately) for the one node that is seen.

2.

Similarly, if the $d_i<=150$ and $CH{r}_{id}==New\_Node$ and $CH{d}_{id}==New\_Node$ between BN and New$\_Node$, then New$\_Node$ sends the $Request\_Join$. BN sends the updated joined request to CH. Then, CH will send the updated $B{N}_{id}$ to all the CMs and RSU.

3.6. Re-Clustering Using RSU

However, if both cases fail, the $New\_Node$ sends $Ac{k}_{R}equest$ to RSU, and RSU starts re-clustering. This process will increase the delay and require more nodes to be part of the communication. There exist two cases: (1) If there is a road fatality and traffic congestion, then $New\_Node$ will receive the RSU notification. It can re-route if the vehicle is fully automatic. However, it will follow the same route according to the driver’s behavior. But if the transmission range of the vehicle is more than 150, then the acknowledgment would not be received by RSU. Secondly, if the vehicle is in range, it is confirmed that the acknowledgment will be received and traffic congestion decreases. During such a situation, re-clustering happens. Assuming that there are 20 nodes in the vicinity, the number of CHs formed is approximately six.

4. Performance Evaluation

In this section, the analysis of a maximum information coverage-based clustering framework for emergencies in VANETs is depicted for the highway scenario, having a uniform and non-uniform relative speed with different density vehicles. In the first part, we provide the simulation results to investigate the clustering time and CH selection time for the fast delivery of emergency messages in the proposed scheme. Additionally, for CH stability, total CH count, the average capacity of CHs, and the maximum information coverage are computed. In the second part, the proposed model is compared with the time barrier-based emergency message dissemination algorithm in vehicular ad hoc networks [14]. Since the proposed algorithm works for both uniform and non-uniform speeds of vehicles, the time barrier scheme is integrated into our simulation scenario. All the schemes are implemented in OMNET++. The simulation configuration is described as follows:

4.1. Testing Scenario

We consider two testing scenarios: the single-lane unidirectional highway with a distinct density of vehicles at either a uniform or non-uniform speed. To our knowledge, this is the first research work in which RSU performed the backbone structure to select CHs and evaluated the proposed algorithm in high-density traffic scenarios. The road scenario consists of either 50, 100, 150, 200, and 250 unidirectional vehicles. These are known as Type C highways that represent the same type of vehicle behavior in terms of infrastructure and vehicle density [4]. The vehicles move with a uniform speed of 20 m/s and non-uniform speed within the range of 25 m/s–45 m/s, with the road length set to 1km. The simulation runs for 600 s. The transmission range for vehicles is 300 m and RSUs are 1000 m.

The two testing scenarios consist of one relatively stable traffic scenario at a speed of 20 m/s and the second with a highly dynamic traffic scenario with a random speed range from 25 m/s to 45 m/s. There is only one type of vehicle with a standard vehicle size of 4.5 m. The mobility pattern of the second scenario is unpredictable as compared to the first one. In the road length of 1000 m, all vehicles are in the communicating range of one of the three RSUs. The rest of the LVs are authenticated vehicles and are directly connected to one of the CHs of their own choice. The inter-vehicle spacing, denoted as s, is calculated using the formula:

$$s={\displaystyle \frac{L}{n}}$$

(4)

where L represents the length of the road in meters (m) and n denotes the number of vehicles (vehicle density) on that road segment. For each scenario, the simulation runs for 600 s. The initialization of the clustering process is at time $T\_window$ when the RSU starts broadcasting its ID $RSU\_id$. The CHs are elected by RSUs as per the CH selection scheme, and the CM/CH connections are made by following the cluster formation scheme. For cluster formation, $T_t$ is the timer for CH/CM connections and for establishing the maximum connected nodes in the network for emergency message dissemination. The default simulation parameters are listed in

Table 1. Default simulation parameters.

Parameters	Values
Simulation Time	600 s
$T\_window$	0.1 ms
$T_t$	0.1 ms
Length of Road	1 km
Number of Nodes/Vehicle Density	50, 100, 150, 200, 250
Transmission Range for vehicles	300 m
Transmission Range for RSU	1000 m
Uniform Speed	20 m/s
Non-Uniform Speed	25 m/s–45 m/s
Size of a Vehicle	4.5 m
Mobility Model	Car—following model
Number of Iterations	30
Number of Repetition	30
Frequency/Channel Bandwidth	5.9 GHz/10 MHz
MAC Protocol	IEEE802.11p

.

Rationale for simulation parameters in Table 1: The simulation time of 600 s ensures a sufficient duration for evaluating network behavior under both sparse and dense vehicle scenarios. The time window $T_{\mathrm{window}}=0.1\mathrm{ms}$ and timer $T_t=0.1\mathrm{ms}$ were selected to simulate the rapid response times required in emergency scenarios. The road length of 1 km is a standard segment size often used in highway-based VANET research. Vehicle densities ranging from 50 to 250 allow us to test scalability and performance under different traffic loads. Transmission ranges of 300 m (vehicle) and 1000 m (RSU) are consistent with DSRC communication standards. The uniform speed of 20 m/s and non-uniform speed range of 25–45 m/s are chosen to reflect typical highway vehicle dynamics. The IEEE 802.11p MAC protocol and 5.9 GHz frequency band are widely adopted for VANET simulations, aligning with the DSRC communication specifications.

4.2. Performance Metrics

The stability of clusters mainly depends on the implementation of upper-layer applications. The comprehensive analysis, including microscopic and macroscopic levels describing the cluster performance, is listed as follows. The macroscopic performance level includes the number of clusters and cluster efficiency for cluster stability, whereas microscopic parameters depict vehicle behavior. Thus, it provides the CH counts and CM connectivity percentage.

1.

Macroscopic Performance:

• Number of Clusters defines the total count of clusters formed with different densities of vehicles on the road.
• Clustering efficiency is defined as the number of LVs that participated during the clustering procedure. The maximum number of connected vehicles determines the clustering performance.

2.

Microscopic Performance:

• Number of CHs is the total count of CHs selected by RSUs.
• Total Clustering Procedure Time is defined as the complete clustering process time beginning from the selection of CHs to cluster formation.

The discussed microscopic and macroscopic performance evaluation model is presented below, with the proposed model having a uniform and non-uniform speed at different vehicle densities. The proposed scenarios are formalized using the uniform speed of 20 m/s and non-uniform speed of 25 m/s–45 m/s.

4.3. Performance Comparison with Time Barrier Mechanism and RA-HEMB with Uniform and Non-Uniform Speed Parameter

In this section, a comparative analysis of the proposed scheme has been carried out with the time barrier emergency mechanism [14] for macroscopic and microscopic parameters with uniform and non-uniform speed highway traffic scenarios. According to [40,41,42], it is found that the compared algorithm is new and reflects popularity by recent citation counts.

According to the time barrier scheme, there is a Cluster Head Eligibility Score (CHE) for the selection of CHs. The original scheme is only for different speed scenarios, whereas we optimize the uniform and non-uniform highway scenarios for dense vehicles. The [14] has been compared with the same speed of 20 m/s and random speed ranging from 25 m/s to 45 m/s. The non-uniform speed is assigned randomly to all moving vehicles; therefore, their mobility cannot be predicted. To achieve fair sample results, we set the average results in one direction in each simulation. We repeat each simulation 30 times with a random speed of 30 m/s.

We compare the average number of clusters, the average number of CMs connected to CHs, CH selection time, clustering time, and network connectivity percentage with the high-density highway scenario. The time taken by the proposed scheme to select the CH by RSU is 0.6 ms, considering $\mathcal{O}\left({log}_{2}N\right)$ complexity in both scenarios, whereas the CH selection in the time barrier scheme increased exponentially as shown in Figure 2. In Figure 3, two scenarios with 4 sub-scenarios are described as the total clustering time. It is observed that the proposed scenarios are three times faster as compared to the time barrier scheme. The timely dissemination of messages during emergencies is a crucial factor. So, there is a need to select the appropriate number of CHs in a short period, as shown in Figure 4.

In the time barrier scheme gateway, the nodes are selected to make cluster-to-cluster communication. The gateway nodes are the nodes that are in the range of two or more clusters. There can be a n number of gateway nodes. Therefore, the count of the gateway node is not predicted. In addition, there may be a use case where the clusters do not have a common gateway. This will drastically affect the network connectivity and information coverage. However, this problem can be addressed while considering protocol-defined margins and range. However, considering the bridge node concept, the gateway nodes are reduced to two per cluster. These intermediate nodes are the CMs of different clusters. The average CH capacity is described in Figure 5. It is found that the compared scheme has the peak value of CMs at 150 nodes. After 150 nodes, the CH capacity started decreasing, whereas, in the proposed scheme, there is uniformity. It is concluded that the CH capacity in the proposed scheme is lower than the compared one. This parameter drastically decreases the overhead, and hence, the excessive utilization of bandwidth is reduced. As shown in Figure 6, using the proposed approach, the simulated results doubled the information coverage as compared to the time barrier scheme.

Figure 5 depicts the average CH capacity, which is the total number of CMs associated with the CHs. In the proposed algorithm, the formation of the cluster is an independent decision. The LVs received the details of nearby CHs and selected the most suitable CH as per $z_i<=Max\left({\displaystyle \frac{1}{\mathrm{\Delta }{v}_{i}d}}\right)$. However, in the time barrier scheme, the vehicles received cluster association requests from CHs and then joined more than one CH. This is, again, a time-consuming process. Thus, the proposed method outperforms during CH selection.

From the above analysis and simulation results, the proposed scheme outperforms in terms of the average number of clusters, the average number of cluster heads, total CH selection time, and total clustering process time.

4.4. Summary of Observations

The following are the observations made from Section 4.3 and Section 4.4.

1.

The time frame computed for CH selection and clustering is much lower in the proposed scheme and has a marginal difference in the sub-scenarios.

2.

The bridge nodes of the proposed scheme are limited to at most two in number in comparison to the time barrier scheme where the gateway node number reaches n. The fixed number of $BN$ leads to providing uniformity. However, the second most $CM$ of CH and RSU is reserved as a backup if the system fails. This will provide the confirmed path for a message to travel. This illustrates the importance of new bridge node formation.

3.

It is observed that the average capacity of clusters is lower whereas the total cluster formation is greater than the compared scheme, respectively. However, there is a uniform pattern in CH selection and CH average capacity, whereas the counter scheme started decreasing its value after 150 nodes. Therefore, we can conclude that our proposed scheme provides small clusters with high robustness, cluster stability, and maximum connected networks, especially under dense traffic scenarios.

5. The Framework Analysis of a New Clustering Scheme to Maximize Information Coverage

The proposed model is based on 100% information coverage with uniform and 99.5% non-uniform speed highway traffic scenarios. The role of clustering in the proposed solution is important for providing maximum connectivity, including for the vehicles that are present in the middle of the road. The range of RSU covers up to 1 km omnidirectional whereas the onboard range of vehicles is 300 m. Thus, bidirectional communication is not possible till both are in range. Thus, clustering plays a vital role in considering the connectivity of the vehicle. This is an important metric for disseminating messages and provides maximum coverage.

5.1. Overhead Analysis

Clustering overhead is defined as the total message exchange during the clustering process. The type of communication messages used are both broadcast and unicast in the proposed framework. The process of communication starts with a Beacon message between the RSU and vehicles which is a broadcast message. The reply from vehicles is the unicast message. Furthermore, $Request\_Join$ and $CH\_Details$ are the messages used for CH selection and CM selection. However, the time barrier scheme is used to broadcast and multicast different types of messages as CH association and cluster association requests.

5.2. The Performance Analysis of a New Clustering Scheme to Maximize Information Coverage

According to the above-mentioned results, it can be seen that the proposed framework has performed better in terms of CH selection and clustering time, with complete route setup. The following are the key findings of the paper.

• During CH selection, only the vehicle sends an acknowledgment to the RSU to become part of the selection procedure, thus having V2RSU communication. However, only V2V communication has been carried out on time barrier schemes.
• In the proposed scheme, two sub-scenarios are simulated, whereas the implemented time barrier scheme works with non-uniform speed. However, the original time barrier scheme provides an infinite value of CH eligibility for uniform speed scenarios. Considering the above-stated result, the uniform speed of vehicles is set to 1 in a simulated environment.
• The maximized connected network has been deduced by the proposed bridge node method where at least two nodes act as intermediate nodes for different cluster connectivity, whereas the time barrier scheme has a n number of gateway nodes that increase the delay and broadcast storm problem during emergencies.

6. Conclusions and Future Work

In this paper, we propose a new clustering scheme to maximize the information coverage for emergencies in VANETs. The proposed framework includes uniform and non-uniform speed high-density highway scenarios, a new CH selection approach using RSU, and bridge node selection for maximum information coverage. The transmission range of LVs is responsible for RSU and LVs connectivity for selecting CHs. The selection of CHs and the choice of LVs to be part of a particular CH, according to mobility metrics, reduces the clustering overhead. Moreover, the most suitable bridge node connectivity has provided maximum coverage and reduced clustering time. We compared the proposed framework with time barrier-based emergency message dissemination under two different traffic scenarios.

Furthermore, the proposed framework can further solve the cluster maintenance and re-clustering problem with the usage of RSU. The RSU is the backbone of the proposed model, and in the future, it may work for different scenarios, such as urban and traffic light scenarios. The proposed scheme forms a one-hop cluster scheme where the CHs and CMs can communicate directly instead of a pure broadcast scheme. This scheme will further aggregate different emergency messages and be destined to different segments of vehicles according to geographic positions. Here, RSUs can communicate with nearby hospitals, chemist shops, police stations, and many more to disseminate messages during emergencies. RSUs and vehicles can decide whether the provided information is useful to send further or to stop. In this manner, secure clustering can be performed.

Moreover, the deployment of the proposed scheme can be optimized based on data packet size, the partition of the road on a segment basis, and the grouping of vehicles as per the destination using route IDs, and CH selection can be optimized using different road conditions, as per application requirements.

The previous study is limited to the usage of RSU for forwarding and broadcasting. A significant amount of CH selection is on the vehicles only, but with the proposed scheme, we provide new insight to the upcoming researchers. This transforms the broadcasting of messages to multicast and unicast, which leads to bandwidth saving. Also, the maximum coverage achieved is 100% and 99.55% as per the uniform and non-uniform speed scenario, which is not the case with the compared scheme. The total clustering time is reduced to half, which leads to the fast delivery of messages. It has been observed that the proposed scheme outperforms in terms of information-based full connectivity, CH selection times, and clustering time under high dynamic traffic scenarios.

In the future, we will work on more complex traffic mobility models and extend the functionalities of data dissemination and data aggregation for various VANET applications.