Performance evaluation of cluster-based schemes for message dissemination in a vehicle-to-vehicle communication in urban environment

Abstract

Effective message dissemination in vehicular ad hoc networks requires proper management of broadcast storms for effective bandwidth usage, fast and reliable delivery of messages. Studies suggest clustering of vehicles as a means of solving broadcast storm problems due to reduction of nodes that rebroadcast emergency messages. However, clustering in an urban environment faces challenges in terms of bandwidth utilization and high End-to-End (E2E) delay due to high density of vehicles, and the number of intersections and buildings. Several cluster-based dissemination schemes have been proposed however, their model evaluations lack typical urban features, like intersections and buildings, which can adversely affect their actual performances. Thus, it is still not clear which scheme is truly the best for real-world urban scenarios. To address this challenge, a fair and thorough performance evaluation of the state-of-the-art schemes in a detailed urban setup, specifically at crossroads with buildings nearby, is urgently needed. Results deduce that, in terms of E2E delay, the Time-Barrier Emergency Message Dissemination Scheme (TBEMDS) outperforms Effective Emergency Message Dissemination Scheme (EEMDS) and Position-based Emergency Message Dissemination Scheme (PBEMDS) by 75% and 50% respectively at low traffic density. However, it is outperformed by PBEMDS by around 9% in high density. For Packet Delivery Ratio (PDR), from low to high density, EEMDS has the least percentage decrease in PDR with a 10% and 40% higher than TBEMDS and PBEMDS respectively, making it the most robust scheme in maintaining PDR as traffic density.

Keywords:

• Broadcast storm
• clustering
• message dissemination
• VANET

REVIEWING EDITOR:

• Jenhui Chen, Chang Gung University, Taiwan

SUBJECTS:

• Communication Technology
• Technology

1. Introduction

In urban environments, alleviating traffic congestion is a critical concern for traffic control (Cherkaoui et al., 2019). Abnormal events such as traffic accidents often lead to severe congestion, making it essential to address such incidents promptly to minimize the impact on traffic flow (Islam, 2019). It is important to inform nearby vehicles about the occurrence of such events as quickly as possible to effectively mitigate congestion. Therefore, it is necessary to develop an efficient mechanism for disseminating warning messages in the traffic network to meet practical application requirements and alleviate traffic congestion.

Recent advances in wireless communication technologies have triggered significant research interest in the field of vehicular ad hoc networks (VANET). The VANET is a special class of mobile ad hoc network (MANET) that involves vehicles following a predefined route (Prakaulya et al., 2017). VANET is a promising technology to realize intelligent transportation systems (ITS) by enabling seamless exchange of information in real time that can be used in managing traffic flow (El Zorkany et al., 2020). In VANET, participating vehicles need to be equipped with a set of wireless sensors and On-Board Units (OBUs) so as to allow wireless communication between vehicles and their environment (Wagh et al., 2021). Dedicated Short Range Communication (DSRC) and Cellular Vehicle-to-Everything (C-V2X) are considered to be the promising wireless communication technology that support vehicular communication (Jeong et al., 2021). The aforementioned devices enable vehicles to form short-range ad-hoc networks for exchange of kinematic data in vehicular networks or with transportation authority/agencies to process and use data to improve traffic efficiency and safety on the roadways (Eze et al., 2016).

In VANET, the most common approach to disseminate information is flooding, whereby every node receiving the message and forwards the same to other nodes (Galaviz-Mosqueda et al., 2017). Flooding approach is associated with waste of bandwidth and frequent collisions which make it an inefficient method especially in the case of a dense network (Ghazi et al., 2020). In dense networks, such as urban environments, the flooding approach leads to a broadcast storm which affects transmission performance in terms of End-to-End (E2E) delay and Packet Delivery Ratio (PDR) (AlQahtani & Sheldon, 2021; Urmonov & Kim, 2020). Recently, in order to reduce the transmission of storms in urban environment, most studies employed a clustering-based approach to disseminate traffic incident messages (Banikhalaf & Khder, 2020; Benkerdagh & Duvallet, 2019; G. Liu et al., 2020; Tseng et al., 2019; Ullah et al., 2021a; Ullah et al., 2021b). Literature suggests that the clustering-based approach is the most effective approach to disseminate Emergency Message (EM) (Kaur et al., 2021). However, clustered VANETs face significant challenges due to the nature of urban road conditions, characterized by variability of vehicle density, intersections and building obstructions (Ren et al., 2023). This complexity is further exacerbated during peak hours and emergency traffic situations when the number of vehicles on the road surges. Such conditions contribute to an increase in overhead, a low PDR, and increase in E2E delay (Kaur et al., 2021).

Several works have been suggested to enhance message dissemination in clustered VANETs. However, there is a need to provide a fair comparison in realistic urban setting, particularly in areas with complex features such as intersections surrounded by buildings. Such analysis is crucial to identify the most effective approach for urban environments, thereby establishing a benchmark for future research. Therefore, contribution of this study lies in the performance evaluation of the proposed schemes, setting the groundwork for their extension and application in typical urban environments.

The remainder of the paper is organized as follows: Section 2 provides an overview of clustering; Section 3 presents the material and methods; Section 4 presents the related works; Results and discussions are presented in Sections 5 and 6 and Section 7 concludes the paper.

2. Clustering in VANET

The preliminary work on clustering (EphremiDes et al., 1987) focused on the autonomous formation of subnets within MANET to enable the distribution of network resources. This work was further implemented by Gerla and Tsai (1995), who proposed the popular Lowest ID and Highest Degree (LID/HD) clustering algorithms for MANETs aiming at increasing communication efficiency among mobile nodes. Since then, many clustering methods have been proposed in the literature to satisfy the requirements of a wide range of applications, including VANET. Alghamdi (2020) described clustering as an approach of network management whereby two or more nearby vehicles with some common features join in a group resulting in efficient message dissemination. There are various forms of clustering based on algorithms, however, the key roles of vehicles within a cluster include cluster head (CH), cluster member (CM), and cluster gateway (CG). The CH coordinates data communication between members and other CHs. It is responsible for sending information to other members within a group so, its transmission range determines the coverage of a cluster. Other than CH and CM, CG enables inter-communication between clusters (Mukhtaruzzaman and Atiquzzaman, 2020).

Clustering process begins with each vehicle periodically exchanging ‘HELLO’ messages to gather information about their neighbors (Cooper et al., 2017; Kaur et al., 2021; Mukhtaruzzaman & Atiquzzaman, 2020; Senouci et al., 2020). The ‘HELLO’ message typically contains vehicle position, velocity, node ID, and other control information (Hussain et al., 2016). After gathering the data, each vehicle computes predefined metric values to assess its suitability as a CH. The vehicle with the highest metric value is then chosen as the CH. Following this selection, all nearby vehicles are grouped with the CH and become CMs. Subsequently, CGs are chosen from vehicles in the transmission range of more than one CH or cluster, a step taken to optimize network performance. Then, to ensure the uninterrupted connectivity between messages and optimal operation within the network, the cluster must be monitored on a regular basis. This monitoring ensures the continued eligibility and proper functioning of the CH, CM, and CG within the cluster (Cooper et al., 2017). This monitoring entails evaluating the CH's performance and replacing it with a new CH if it falls below a certain threshold. The monitoring also ensures that all CMs and CGs remain operational and connected.

The performance evaluation of clustering algorithms in VANETs is mostly based on network simulators because of the limitation of testing scales in the real traffic environments. E2E and PDR parameters indicate the effectiveness of the scheme through measuring timeliness and successful arrival of messages. The study by Pal et al. (2018) formulated a mathematical model for performance evaluation of clustered VANETs in terms of PDR, throughput, and E2E delay. The performance metrics were derived from the IEEE 802.11p/DSRC protocol media access control mechanism and the characteristics of clustering. The mathematical model indicates that a key issue that affects the performance parameters is vehicles trying to transmit at the same time both inter-cluster and intra-cluster and hence causing collision. Therefore, improving performance parameters requires stable clustering scheme, improved channel access scheme and optimal packets transmission, as also narrated in Jeevitha and Bhuvaneswari (2022). This implies that the frequent re-clustering of vehicles should be reduced.

One approach to reduce the frequent re-clustering is by proper selection of clustering parameters for a specific scenario to ensure longevity of connections. However, reducing frequent re-clustering results in CHs managing a large number of cluster members which deteriorates the performance of the entire network (Ayyub et al., 2022). A large cluster may result in more collisions and retransmissions, leading to prolonged delay. Therefore, there should be a tradeoff between the CHL and cluster size to optimize the performance.

3. Materials and methods

This research adopted a systematic review approach for analysis of recently published studies. The databases that were used to search for relevant related literature include IEEE Xplore, Science Direct, Scopus and Wiley Online Library. The search involved the most recent journals and conference articles published in English. The filtering criterion was based on the clustering schemes in urban VANET. The following keywords were used to search for relevant literature appropriate to this study: Broadcast Storm, Clustering, Message Dissemination, VANET whereby papers obtained after omitting duplicates were 352.

The screening was performed based on the relevance and qualities of abstracts whereby 105 papers were included in the study and 227 papers were discarded after reading abstracts. The 105 papers were further filtered to obtain papers focusing on the state-of-the-art ideas that were published between the years 2019 and 2023. Out of 105 papers, 30 papers were validated on highway scenarios, 35 papers validated in the urban scenario but focusing only on cluster stability and 27 involved in infrastructure only without clustering using Road Side Unit (RSU) and Base stations (BS)) in message dissemination, were discarded. Therefore, 13 papers remained for comprehensive analysis, as summarized by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart in Figure 1. The comprehensive analysis based on papers focusing on clustering schemes to suppress broadcast storms while achieving a minimum E2E delay and a high PDR in transferring information in urban environment.

Figure 1. PRISMA flow chart elaborating how literature screening was conducted.

4. Clustered emergency message dissemination schemes for urban scenario

Disseminating EMs is a challenging task in VANETs due to node mobility which causes frequent changes in network topology, resulting in excessive transmission delay and packet loss (Gasmi & Harous, 2022). Intersections in urban areas present a significant challenge to clustering in VANETs which can change traffic patterns abruptly, resulting in rapid changes in vehicle density in various areas. This can result in frequent cluster head changes, resulting in increased overhead and decreased efficiency (Hande & Muddana, 2019). Furthermore, intersections can create areas of high contention as vehicles from different directions converge at the same location (Zhao et al., 2016). This can result in more collisions and retransmissions, adding to the delay and lowering overall performance. The shadowing effect caused by buildings significantly impacts communication links, resulting in decreased connectivity performance due to the attenuation of the radio signal (Bilal et al., 2023; Ortega et al., 2018).

Figure 2 depicts a network model with clustered vehicles in a typical urban scenario at intersection. As a result, clustering in urban areas with intersections, buildings and varying density requires careful consideration and optimization to address these challenges. The approaches used to disseminate EMs in urban areas are mainly divided into two groups: pure Vehicle to Vehicle (V2V) where the infrastructure is difficult to deploy; and Vehicle to Infrastructure (V2I) which includes RSUs or C-V2X. Both approaches improve E2E delay and PDR.

Figure 2. Network model with clustered vehicles in a typical urban scenario.

To assess the performance of clustering-based emergency message dissemination schemes in the urban area, thirteen (13) related works were examined.

Durga and Varma (2016) suggested an adaptive clustering approach for data dissemination in an urban setting in an effort to reduce the number of EM retransmissions. The network was divided based on density to find the highest number of cars in a cluster, and CH was selected as the vehicle with the most energy. Clustering based on density can increase communication efficiency and decrease collisions and interference among nodes. However, clustering based on density may not be able to effectively capture the network’s topology in an urban area, resulting in inefficient clustering and potentially increased delay. In addition, picking the CH with the highest energy output may not necessarily result in the most effective use of resources. For instance, the selected node may not be at the best position to service the remainder of the cluster. This strategy may therefore be subject to node failure or movement, as the selected CH may not be able to hold its position as the node with the highest energy for a long time.

Touil and Ghadi (2018) proposed an efficient dissemination method for VANETs based on a passive approach and dynamic clustering, with the goals of maximizing EM delivery coverage and minimizing the number of exchanged messages. Clustering of vehicles based on number of vehicles in a cluster in order to balance clusters and average speed. The vehicle with the optimal average speed is eligible to form and lead a cluster. The scheme’s demonstrated relatively good performance in terms of packet loss. However, the method did not consider geographic distribution of vehicles, which can be crucial in urban areas with intersections where traffic patterns may be more complex.

Liu et al. (2018) proposed a direction-based clustering and probabilistic forwarding scheme for node communication that is stable and reliable. Clustering was accomplished through parameters such as link connectivity and packet successful transmission probability. The likelihood of link connectivity was determined primarily by driving direction in order to ensure a stable link between nodes, which is critical in urban environments with obstacles such as buildings and other vehicles. The probabilistic forwarding scheme was designed to adapt to variable vehicle density, which can help optimize the system’s delay performance, particularly in congested urban areas where traffic density varies greatly. The scheme performed better in terms of information coverage, transmission delay, and PDR. However, in densely populated urban areas with a high number of vehicles and obstacles, the approach may have limited scalability.

Shah et al. (2019) proposed a Time-Barrier Emergency Message Dissemination Scheme (TBEMDS) for the urban environment based on V2V communication. In their study, the clustering considered the vehicle interest compatibility and packet transmission probability to improve the lifetime of the cluster. A key feature of the scheme is the adaptation of the time barrier technique to prevent the broadcast storm problem, allowing only the farthest vehicle to rebroadcast the message. However, in highly congested urban areas with many vehicles in close proximity, the use of a timer that is inversely proportional to the distance between the CM and the sender of EM may not be effective. As a result, multiple nodes with the same distance from the CH would rebroadcast at the same time, leading to an increased congestion and latency.

Ali et al. (2019) proposed a position-based Emergency Message dissemination scheme (PBEMDS) to tackle the prevalent challenges of mobility, connection lifetime, and network stability. The proposed scheme focuses on sharing information with vehicles moving in the opposite direction of the incident, aiming to enhance the dissemination performance of incident-related information such as accidents or congestion. In their study, CH was selected based on the compatibility among vehicles’ interest in a cluster and the probability of successful packet transmission. Vehicles with similar destinations were clustered to improve network stability and lifetime. EMs are only sent to CHs on the opposite side of the incident or direction, aiming to reduce the dissemination delay and to expedite the spread of the message. Utilizing CHs in the opposite direction in vehicular communication systems offers a targeted and efficient approach. By focusing on relevant recipients, it ensures timely dissemination of critical information of EMs and reduces unnecessary broadcasting, thereby preserving network bandwidth. However, effectiveness of such an approach may be compromised in dense urban environments where traffic behavior can be highly variable such as the unpredictable changes in speed, direction and traffic patterns.

Alghamdi (2020) proposed a Cellular-5G VANET architecture, which combines VANET with cellular-5G technology. In urban areas with intersections, integration with 5 G technology can provide high data transfer rates, low latency, and reliable communication. In their study, clustering was performed using the adaptive mobility aware path similarity algorithm, with CHs chosen based on future path similarity, considering speed difference, average distance, and acceleration difference. In this study, V2V communication was carried out using DSRC, and V2I communication was carried out by applying the Device to Device (D2D) concept to 5 G mobile network users. This architecture is appropriate for urban environments where efficient and reliable communication is critical to managing traffic congestion and ensuring road safety. The integration of VANET with cellular-5G technology, on the other hand, may necessitate significant infrastructure investments and upgrades, which can be costly.

A self-assessment-based clustering method for sparse and dense traffic conditions for Internet of Vehicles (IoV) was designed by Qureshi et al. (2020) to improve stability and overhead. In their method, BS performed clustering based on node density, and the front most vehicle in a cluster was chosen as CH after a self-assessment approach that included checking whether there was already an existing CH, distance, and lower velocity. This approach is appropriate for urban areas because it can adapt to both sparse and dense traffic conditions, as traffic density varies greatly, depending on the time of day or day of the week. The reliance on BS to perform clustering and CH selection, may introduce a single point of failure and increase network latency, particularly in highly dynamic and congested urban environments with intersections where network conditions can change rapidly.

Azzaoui et al. (2021) proposed a real-time emergency message dissemination approach in IoV networks that employs a clustering strategy. Clusters were formed based on relative distance and relative velocity, which can help to balance the network load and ensure that vehicles with similar speeds and trajectories are grouped together. The CHs were selected on the basis of vehicles that have a strong signal and communication link with the Long-Term Evolution (LTE) antenna, ensuring that the network has adequate coverage and that the CHs can efficiently forward data packets. The research was carried out with the goal of reducing transmission delay and overheads while maintaining high coverage by selecting the best forwarder vehicles. In densely populated urban areas with high-rise buildings, which can obstruct the signal path and reduce link connectivity, this approach may lead to higher packet drop rates and prolonged delay. Furthermore, the use of LTE antenna may not be feasible in all areas, particularly those with limited cellular network coverages.

Ullah et al. (2021a) devised a scheme to enhance cluster stability and extended information coverage by avoiding concurrent CG rebroadcast, named Effective Emergency Message Dissemination Scheme (EEDMS). To avoid simultaneous transmission, the CG with the lowest link stability was chosen. When more than one vehicle is furthest to the CH, link stability was calculated based on communication range, velocity, and distance. Only CH and CG were permitted to re-broadcast EM in order to avoid broadcast storms. For selecting a suitable CH, mobility parameters such as moving direction, velocity, distance, and time to leave were considered. The parameters were chosen with the goal of increasing CH lifetime, lowering communication overhead, and achieving a high PDR. However, choosing a CG with lower link stability to avoid simultaneous transmission may not always be effective in reducing concurrent rebroadcast of CGs, particularly in high-density urban environments with numerous obstacles that can interfere with the signal path and reduce link stability.

The study was extended by Ullah et al. (2021b) to include categorizing EMs based on zone of interest with the goal of selectively rebroadcasting depending on the scenario, as well as allowing CMs to rebroadcast in the absence of a gateway. As a means of improving network coverage in areas with low vehicle density or when the CH or CG is unavailable, the scheme allows the CM to rebroadcast the EM. However, the scheme has demonstrated good performance in low density scenarios since allowing CMs to rebroadcast in dense scenarios may result in broadcast storms. Besides, the schemes’ performance is also not guaranteed in sparse networks and when transmission is limited due to signal blockages.

A study by AlQahtani and Sheldon (2021) focused on overcoming DSRC signal blockage caused by intersection buildings. The proposed method combined clustering to alleviate the broadcast storm problem, cluster forwarders to allow communication around obstacles, and the 5 G cellular network to extend network coverage to increase the reach of the disseminated messages. The use of 5 G cellular network can introduce additional latency and is not always reliable in all areas, especially in remote or low-coverage areas. It may also raise the cost of system deployment and necessitate additional infrastructure.

The study by Wang et al. (2023) explores the challenges and solutions related to the mixed transmission of EMs and safety messages (SMs) using V2V communication in urban environments for the IoVs. Their study employs a bus-based clustering algorithm in conjunction with a prioritized scheduling scheme for both EMs and SMs. The bus-based algorithm selects buses as CH, taking advantage of their fixed schedules to overcome the unpredictable movement of other vehicles. By considering factors such as path similarity, link reliability, average relative velocity, and average relative distance, the algorithm aims to achieve reliable and efficient data dissemination. Alongside this clustering approach, the researchers also introduced a scheduling algorithm that prioritizes messages based on their importance and urgency, particularly in scenarios where EMs and SMs are being transmitted concurrently. This scheme has proven especially useful in areas with well-defined bus routes, as the predictability of buses as CHs contributes to stability and effectiveness in data dissemination. However, the approach may have limitations in areas without consistent bus service.

Ullah et al. (2023) proposed a method to improve message dissemination by employing cooperative relay forwarders based on mobility and connectivity factors and selecting the best pedestrian to serve as a gateway for reliable V2I communication with low delays. The V2V and Vehicle to Pedestrian (V2P) communications were established between vehicles and pedestrians using DSRC standard and 5 G technology, respectively. In order to reduce unnecessary rebroadcasting and congestion, the communication efficiency was also improved by categorizing EM into one-hop or multi-hop broadcasts based on the severity of the emergency. Despite increasing the efficacy of EM dissemination, the method heavily depends on reliable V2I communication and connectivity between forwarder vehicles and pedestrians, which may be disrupted in certain circumstances. Their method necessitates the installation of RSUs on the roadsides which increases the overall cost.

According to the reviewed literature, the initial research on emergency message dissemination based on a pure V2V environment with all vehicles connected. Multi-hop communication is required for information to reach many vehicles, which contributes to E2E delay. Pure V2V communication is prone to link failure and hidden terminal issues, whereas, V2I communication yields more impressive results than V2V communication (Kaur et al., 2021). Recent trends in EM dissemination include the use of C-V2X, software defined network (SDN) controllers, machine learning, and edge computing (Abbas et al., 2022; AlQahtani & Sheldon, 2021; Azzaoui et al., 2021; Bhabani & Mahapatro, 2023; Bijalwan et al., 2022; Li et al., 2022; B. Liu et al., 2021; Maan & Chaba, 2021; Zhang et al., 2021). However, since not all urban areas have similar environments, a fair performance comparison is needed to establish the most effective cluster-based methods for pure V2V scenario and pave the way to more usable research outputs. This consideration is particularly vital for developing countries where infrastructure challenges prevail (Yasser et al., 2017); paving the way for ITS implementation without the need for RSUs.

5. Results and discussions

Simulation tools employed to conduct the performance evaluation include Vehicles in Network Simulation (VEINs) framework incorporating Objective Modular Network Testbed in C++ (OMNeT++) and Simulation for Urban Mobility (SUMO) linked by Traffic Control Interface (TRACI). The SUMO was utilized to depict the flow of traffic on cross road scenario as depicted in Figure 3, then it was linked to VEINS through Transmission Control Protocol (TCP) socket and the OMNeT++ was used to reflect the movement of vehicles. The VEINS encompassed implementations of the IEEE 1609.4 and IEEE 802.11p communication standards.

Figure 3. Traffic simulation scenario in SUMO (a) crossroad and blocks representing buildings in (b) Enlarged view of the road intersection.

5.1. Channel modelling

The study by Ullah et al. (2021a) used Two ray interference (TRI) for channel modelling whereas those by Ali et al. (2019) and Shah et al. (2019) used Nakagami-m model to represent effects caused multipath, reflection, and diffraction that are prevalent in common urban environments. However, it is worth noting that the Nakagami model, when used alone is not ideally suited as a path loss model for simulations. As highlighted by Hota et al. (2022), for enhanced performance results, it is beneficial to integrate the Nakagami model with other loss models. In this study, all three models were incorporated; TRI, Nakagami-m, and Shadowing models (Sommer et al., 2019). By combining these models, the deterministic path loss from distance, stochastic fluctuations from multipath fading and the large-scale variations due to obstacles can be represented. Furthermore, this combination allows for capturing the broader variations caused by obstacles, thus offering a more holistic representation of signal propagation in urban scenario.

Mathematically, combining the afore-mentioned effects, the received power $P_r$ in an urban scenario can be represented as (1)(1) $$P_r=P_t+G_t+G_r+L_p+L_{\mathit{\text{nam}}}+L_{\mathit{\text{sha}}}$$(1) : (1) $$P_r=P_t+G_t+G_r+L_p+L_{\mathit{\text{nam}}}+L_{\mathit{\text{sha}}}$$(1) where P_t is the transmitted power G_t and G_r are the gains of the transmitter and receiver antennas, respectively, L_p represents the loss due to the Two Ray Interference Model, L_nam represents the loss due to Nakagami-m fading, and L_sha represents the loss due to shadowing effects from obstacles.

5.2. Performance parameters

5.2.1. E2E delay calculation

The E2E delay is calculated by subtracting the original timestamp, $T_0$_, from the reception time at the destination, $T_{\mathit{\text{CH}}}.$ This computed delay encapsulates the entire duration the message took to traverse from its originating vehicle to the CH, irrespective of whether it was a direct transmission or relayed through several vehicular nodes following a broadcast. It includes all inherent delays within the network: transmission delays from source to intermediaries, propagation delays across the medium, queuing delays at buffer points, and processing delays during routing decisions.

The E2E delay, is presented in (2)(2) $$E2E\hspace{0.17em}\mathit{\text{delay}}=T_{\mathit{\text{CH}}}-T_O$$(2) : (2) $$E2E\hspace{0.17em}\mathit{\text{delay}}=T_{\mathit{\text{CH}}}-T_O$$(2)

5.2.2. PDR

The PDR signifies the ratio of the total number of packets successfully received, $N_R,$ to the total number of packets dispatched, $N_{\mathit{\text{CH}}-\mathit{\text{CMs}}}.$ For this scenario, the number of packets sent are considered as those transmitted by the CH to its CMs. Conversely, the number of packets received corresponds to the count of vehicles that successfully obtain EMs from a distinct CH.

The PDR, is presented in (3)(3) $$\mathit{\text{PDR}}=\frac{N_R}{N_{\mathit{\text{CH}}-\mathit{\text{CMs}}}}\times 100$$(3) : (3) $$\mathit{\text{PDR}}=\frac{N_R}{N_{\mathit{\text{CH}}-\mathit{\text{CMs}}}}\times 100$$(3)

6. Results discussion

Figure 4 depicts the variation of E2E delay with the flow rate for the three message dissemination schemes in a pure V2V communication environment, using parameters given in Table 1. It is observed that at low traffic density (25 vehicles/hour), TBEMDS is the most efficient scheme with 75% lower delay compared to EEMDS, and 50% compared to PBEMDS. At high traffic density (250 vehicles/hour), PBEMDS appears to have the least delay, outperforming EEMDS and PBEMDS by 36% and 9%, respectively.

Figure 4. Variations of E2E delay with vehicles flow rate.

Table 1. Simulation parameters.

Parameter	Value
Road topology	Two way urban
Lanes	2 for each direction
Road section length	1.5 km
Maximum vehicle velocity	50 km/h
Simulation time	200 s
Transmission power	20 mW
Bit rate	6 Mbps
Min power level	−85 dBm
Noise floor	−98 dBm
Channel model	Nakagami-m
Beacon interval	0.1 s
MAC and PHY layer	IEEE802.11p

The relative performance of TBEMDS degrades the most as density increases. Although PBEMDS is not the best at low density, it scales the best with increasing traffic and ultimately outperforming the performance of TBEMDS at high density. EEMDS, on the other hand, consistently has higher delays across both low and high traffic densities compared to the other schemes. Therefore, PBEMDS appears to be the most suitable since it offers the best scalability and maintains consistent performance across a range of increasing traffic density, which is characteristic of urban environments. This is mostly contributed by utilizing vehicles moving in opposite directions to extend transmission.

For PDR, as depicted in Figure 5, it is observed that all schemes perform well with PDRs of around 100% at low traffic densities. As traffic density increases, EEMDS shows a minor decrease in PDR indicating its robustness in more congested environments. TBEMDS manages to maintain a relatively high PDR up to moderate densities but then struggles as the density gets very high. PBEMDS experiences a significant drop in performance as density increases, which suggests it is less capable of handling the broadcast storm problem that occurs in high-density scenarios. At high density (250 vehicles/hour), EEMDS outperforms TBEMDS and PBEMDS by 10% and 40%, respectively.

Figure 5. Variations of PDR with vehicles flow rate.

From low to high density, EEMDS has the least percentage decrease in PDR, suggesting that it is the most robust scheme in terms of maintaining PDR as traffic density increases. TBEMDS experiences a moderate decrease, indicating it has some resilience but is not as effective as EEMDS at maintaining PDR under high traffic conditions. PBEMDS shows the most significant decline in PDR, indicating a considerable drop in performance with increased vehicle density. Therefore, EEMDS is most suitable for dense urban areas, maintaining a high PDR and showing only a slight decrease as density increases.

For urban vehicular networks, especially those with a complex layout of intersections and buildings, the selection of a communication scheme must balance the need for timely message delivery (low E2E delay) with the reliability of message receipt (high PDR). Therefore, optimal performance in an urban setting requires a dynamic or hybrid system that can switch between EEMDS and TBEMDS based on real-time conditions. For-instance, EEMDS can be used during peak hours to benefit from its high PDR and TBEMDS during off-peak hours when the PDR is no longer a serious issue, and lower delay is more desirable. For PBEMDS, significant enhancements would be needed to improve its PDR in high-density conditions for it to be a viable option for such environments.

7. Conclusion

This study analyzed the performance of clustered-based message dissemination schemes in a pure V2V urban environment. The performance parameters of different message dissemination schemes were examined for comparative analysis. A systematic review approach was used to analyze the most recent literature published on cluster-based message dissemination schemes in urban environments. Furthermore, 13 studies were selected for further analysis based on clustering approach, CH selection parameters, and performance parameters associated with message dissemination. During simulation, the VEINs framework, OMNeT++ and SUMO were used assuming pure V2V environment. The analysis of the three schemes EEMDS, TBEMDS, and PBEMDS reveals distinct performance profiles. The EEMDS outperforms other schemes for its robust PDR across varying traffic densities, making it a reliable choice for ensuring message delivery in dense urban traffic where reliable communication is a priority. The trade-off is a higher E2E delay, which may be acceptable given the critical nature of maintaining high PDR in such environments. The TBEMDS offers an advantage in environments with lower to moderate traffic densities, where its low E2E delay enhances communication efficiency. However, its performance in terms of PDR at high densities suggests limitations for peak urban traffic conditions. The PBEMDS, on the other hand, shows predictability in E2E delay, which does not significantly worsen with increased traffic density. However, its sharply decreasing PDR in high-density conditions limits its suitability for urban areas during busy periods. Future study will involve devising a dynamic or hybrid approach that leverages the strengths of each scheme and adapts to real-time traffic.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The authors received no direct funding for this research.

Notes on contributors

Elizabeth Ngatunga

Elizabeth Ngatunga received Bsc. degree in Telecommunications Engineering from University of Dar es Salaam, Tanzania and M.Sc. degree in Information Technology and Management from the Avinashilingam Deemed University, India. Presently, she serves as an Assistant Lecturer in the Department of Electrical, Electronics, and Telecommunication Engineering at the National Institute of Transport. She is currently a dedicated PhD student in Telecommunication Engineering at the University of Dar es Salaam. Her research focuses on Intelligent Transportation System (ITS), Vehicular Ad-Hoc Network (VANET) and Vehicle-to-Vehicle (V2V).

Mussa Kissaka

Mussa M. Kissaka received B.Sc. degree in electrical engineering from the University Dare es Salaam, Dar es Salaam, Tanzania, in 1989, and Ph.D. degree in telecommunications engineering from the University of Manchester, Manchester, U.K., in 1994., Currently, he is Senior Lecturer with the Department of Telecommunications Engineering, College of Information and Communication Technologies, University Dare es Salaam. His research interest includes rural telecommunications, computer networks, wireless network and optical fiber, Dr. Kissaka is a Registered Professional Engineer with the Engineers Registration Board (ERB) of Tanzania.

Abdi T. Abdalla

Abdi T. Abdalla received the B.Sc. degree in electronic science and communication and M.Sc. degree in electronics engineering and information technology from the University of Dar es Salaam, Tanzania, in 2006 and 2010, respectively, and the Ph.D. degree in electrical engineering majoring in communication and signal processing, from King Fahd University of Petroleum and Minerals (KFUPM), Saudi Arabia, in 2016. Currently, he is Associate Professor at the Department of Electronics and Telecommunications Engineering, University of Dar es Salaam, Tanzania. His research interests include through-the-wall radar imaging, indoor target localization, sparse arrays processing, application of compressive sensing to radar signal processing, Vehicular Ad-Hoc Networks, and application of ICT in blue economy and community development.