1 Introduction

Autonomous vehicles and Intelligent Transportation Systems (ITS) in Vehicular Ad Hoc Networks (VANETs) have drawn researchers’ attention due to their ability to offer a wide range of services, such as passenger safety, traffic management, and road condition monitoring [1]. VANETs consist of units installed in vehicles, called On-Board Units (OBUs), and devices located along the roads, known as Road Side Units (RSUs). These units make use of short and long-range technologies, such as IEEE 802.11p and Dedicated Short-Range Communication (DSRC), respectively [2], to facilitate communication between Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) devices. V2V plays a crucial role as it allows vehicles to exchange real-time information about traffic conditions, hazard alerts, and other relevant data, helping to prevent accidents and reduce road congestion [3]. Additionally, V2I connectivity enables more effective coordination with road infrastructure, such as traffic light optimization and signage, to improve flow and safety on roads [4]. However, it is essential to implement a comprehensive approach to address the challenges arising from the growing number of vehicles in urban environments, traffic congestion, vehicle mobility, the proliferation of applications for various purposes, channel interference, and limitations in bandwidth and coverage range [5], Moreover, new architectures for vehicular communications are being developed that can incorporate techniques such as multicast in the frequency bands assigned to vehicles. These architectures are designed to optimize bandwidth usage in urban environments [6]. These factors are critical as they negatively affect the performance of communications in VANETs, hindering their effectiveness and reliability in providing safety and traffic management services. The implementation of Wireless Sensor Networks (WSNs) with VANETs to create a hybrid VANET-WSN network could address scenarios related to Intelligent Transportation Systems (ITS) [7], through the deployment of techniques, protocols, and technologies such as Low Power Wide Area Network (LPWAN), cellular networks, and smart devices. WSNs collect, process, and efficiently distribute information among the nodes that make up the network, providing each node with extensive coverage and effective data management. This is achieved through the utilization of technologies such as LPWAN, 3G, 4G, 5G, Bluetooth Low Energy (BLE), Wireless Fidelity (Wi-Fi), and ZigBee [8,9,10]. Additionally, these technologies are capable of establishing communication with microcontrollers equipped with hardware accelerators, multicore processors, and high-speed RAM [11, 12], enabling real-time tasks, which are essential for making quick decisions and adapting to changes in driving environments [13]. However, the technologies used for WSNs alone cannot solve the mentioned challenges of VANETs, as each has specific advantages and disadvantages regarding coverage range, energy consumption, data transfer speed, and interference channels [14]. Therefore, developers and scientists are looking for and implementing new techniques, algorithms, and protocols to optimize the joint operation of these technologies, enabling the successful integration of the VANET-WSN hybrid network. This approach will help reduce congestion and data delay on the 802.11p channel, thereby contributing to decreasing the response time for critical messages such as warnings, emergencies, notifications, and errors among vehicles, crucial for protecting passengers’ lives in hazardous situations and optimizing traffic management [15, 16]. Furthermore, this integration will also facilitate the development of new applications for various purposes, such as monitoring environmental and weather conditions, processing data offline, collecting sensor information in areas with no coverage or difficult access, issuing security alerts, providing traffic updates, offering the most convenient route, providing entertainment applications, and even acting as temporary support infrastructure for some sensors to send their information to the network in emergencies.

Despite its wide acceptance in modern smart city applications, there are several challenges in the research and deployment of VANET. The sensors periodically sense, process, and transmit data to the base stations. Data collection is the systematic gathering of data sensed by multiple heterogenous sensors to transmit to the base station via a cost-effective and energy-efficient route in a hybrid network for further processing. The frequency of data acquisition, as well as the number of sensors utilized in the process, is determined by the specific application. Along with clustering in VANET, it also necessitates an effective and efficient routing on mobility scheme to overcome scalability issues within its architecture and enable seamless data dissemination [17]. This review guides researchers, practitioners, and system designers in selecting the appropriate energy-efficient clustering mechanism for VANET in their research for monitoring applications.

2 Main Contribution and Structure of the Article

The main contribution of this article is to provide researchers with a comprehensive overview of the latest methods and technologies that are increasingly being employed in VANET environments and are becoming more prevalent. These technologies leverage the IoT to optimize open-access frequency channels and enable developers to create various applications that positively impact society. These applications aim to improve people’s quality of life, monitor environmental parameters, and enhance vehicle travel experience. This, in turn, promotes a more sustainable, efficient, and reliable environment.

The document is organized as follows: In Sect. 3, a detailed analysis is conducted on the VANET architecture as a foundation for monitoring applications. We provide a clear understanding of how vehicles communicate with each other and with the infrastructure. Section 4 delves into the network’s protocols, mechanisms, strategies, and communications challenges, offering an in-depth perspective on this landscape. Section 5 focuses on sensor technologies in VANET environments, examining the types of sensors used, their potential applications, and their significant contributions to data collection and environmental monitoring in VANETs. Finally, the last section, Sect. 7, dives deep into the critical aspects of data forwarding in VANETs. Here, we explore the mechanisms and protocols governing efficient data forwarding between vehicles and infrastructure components, highlighting strategies that ensure timely and reliable information dissemination. Furthermore, this section starts with the significance of data forwarding and routing methods in VANET and then justifies the importance of clustering. After that, the following subsection provides a novel taxonomy based on the most effective and commonly utilized features of clustering schemes in monitoring and safety applications. In the following subsection, we assimilate the energy-efficient cluster-based routing schemes for VANET. It describes the anatomy of the clustering architecture in VANET and discusses the importance of having energy-efficient cluster-based routing schemes. Finally, in Sect. 8, we conclude the article by prescribing future opportunities to investigate different aspects of data acquisition in mobility, such as sensing, wireless communication over VANET, and multi-gateway interface management in heterogeneous networks. Moreover, it highlights the potential research gaps in data forwarding in VANET through cluster-based techniques for monitoring and safety applications. It also discusses incorporating artificial intelligence (AI) and software-defined networks (SDN) to improve data acquisition.

3 VANET Architecture for Monitoring Applications

In VANETs, vehicular communication and interaction with Roadside Units (RSUs) are facilitated through Wireless Access for Vehicular Environment (WAVE) [18] technology, which is crucial for passenger safety, managing traffic flow, and collecting information about the traffic and potential emergencies. Furthermore, using VANET networks, environmental monitoring applications could be built on public transport, electric vehicles, and bicycles, promoting sustainable transportation [19]. This aids in traffic optimization, especially in areas such as hospitals, schools, etc. Further, in addressing safety and emergencies, VANET implementation is ideal for highways and adverse situations (natural hazards, vehicle accidents) [20]. VANETs have three key components: the OBU, the RSU, and the Trusted Authority (TA).

Fig. 1
figure 1

Advanced VANET environment integrating heterogeneous systems with IoT technology for monitoring applications

Full size image

RSUs are installed along the road and typically communicate with OBU devices installed on vehicles. The OBU devices collect valuable information from the vehicle, such as position, speed, acceleration, and fuel consumption. Once the data is collected, they send all this information to the other nodes in the mobile ad-hoc network using V2V communications. Furthermore, technological advancements focus on maximizing OBU applications in remote monitoring, such as monitoring vehicle mechanical status, traffic conditions, air quality, environmental conditions, weather conditions, and more, contributing to the development of Smart Cities. Therefore, to expand the applications of the OBU, it is necessary to leverage the benefits of the Internet of Things (IoT), which introduces new low-power technologies that operate in unlicensed or restricted frequency bands [8]. However, in the implementation of IoT systems in VANET, various challenges arise due to the constant mobility of vehicles on the roads, such as high speed, rapid topological changes, intermittence of established connections, latency, bandwidth, and security [21, 22], which must be addressed for effective implementation.

Indeed, in Fig. 1, a VANET system is depicted that incorporates various IoT technologies known for their energy efficiency and long- and short-distance transmission capabilities, aiming to provide V2I or V2V connectivity. These technologies gather valuable information from vehicles and the surrounding environment [9, 23]. This approach gives rise to a paradigm known as the Internet of Vehicles (IoV) [24], involving the integration of smart sensors and IoT devices within vehicles to collect and transmit data to other components of the VANET. This integration improves transportation efficiency, safety, and comfort, mainly playing a crucial role in developing autonomous vehicles [25].

4 VANET Communications

VANETs constitute wireless systems facilitating V2V and V2I interconnection, employing protocols such as IEEE 802.11p. Designed to enhance ambient monitoring, traffic safety, and efficiency, they enable transmitting critical information, including air pollution control, safety alerts, and position data, significantly contributing to advanced traffic management and accident prevention.

4.1 Communication V2V in VANET

In a VANET, V2V communication constitutes a decentralized network, where vehicles interact directly with each other without relying on a centralized infrastructure for the communication, see Fig. 1. Protocols such as IEEE 802.11p and DSRC can be used in V2V communications. IEEE 802.11p is an extension of WiFi, designed explicitly for dedicated vehicle communications, operating in the 5.9 GHz band. It supports relatively low data rates but is suitable for traffic safety and emergency applications. Its main objective is to provide quick and efficient responses crucial for improving road safety and enabling advancements in automotive applications such as traffic flow efficiency [2]. V2V communications allow vehicles to anticipate traffic situations, such as sudden braking or lane changes, allowing quicker responses and collision prevention using Machine Learning (ML) [26]. ML algorithms analyze complex patterns and massive datasets in this context, enhancing real-time decision-making [27]. This enables the implementation of predictive systems that can anticipate driving behaviours, identify traffic patterns, and adapt to changing road conditions. The synergistic combination of V2V communications and ML promises to radically transform road safety and transportation efficiency by facilitating intelligent interconnection between vehicles [28]. However, V2V communication still has disadvantages, such as scalability since the number of vehicles equipped with V2V increases, network management may be compromised due to the rising volume of transmitted messages, leading to network saturation and overall performance degradation [29].

4.2 Communication V2I in VANET

Currently, V2I and V2X communications are becoming essential to enhance road safety, optimize traffic efficiency, and monitor various types of sensors distributed throughout cities. However, several challenges must be addressed to improve the efficiency of these communications in VANETs. These challenges include delays in intra-system communication, management of heterogeneous systems [30], the high mobility of vehicles causing constant changes in network links which directly affect the quality of service (QoS) [31] and issues related to interference [32], managing channel congestion in urban environments or heavy traffic situations, adaptability to vehicle mobility, high latency, coverage of the communication technology used, energy consumption, passenger privacy, and other factors. To address these issues, new techniques, algorithms, and protocols are being explored to facilitate the interconnection of vehicles with traffic signals, traffic lights, and other road infrastructure [33]. Figure 2 illustrates one of the architectures used in VANETs for remote monitoring using a Mobile Multi Interface Gateway (M2GW) that can be equipped with different technologies that involve different aspects such as LPWAN, IoT, or Cellular networks. The implementation of the M2GW can be crucial to address some of the challenges above since it can gather data from different sensors, utilize various technologies for bidirectional communication, store data, and perform local processing of the collected information before transferring it to other more robust processing sources, such as Edge Computing [34] It has shown better latency and a lower packet loss rate in end-to-end packet reception compared to cloud services [6], which provides quick and efficient responses by minimizing latency. Another alternative could be Cloud Computing [35], which offers a centralized environment for storing and processing data on remote servers, providing robust scalability and accessibility from any location with an Internet connection.

Fig. 2
figure 2

Architecture for air quality monitoring on ITS using M2GW

Full size image

4.2.1 The Role of M2GW for Enhancing V2I Communication in VANET

Public transport vehicles (Buses, Trams, etc.) typically have extensive coverage throughout the city or town. Hence, collecting data from the sensors attached to the OBUs poses a significant challenge in VANETs. One of the problems that arises is the congestion of data traffic in the 802.11p protocol in urban areas or areas with high vehicle density. This protocol helps to propagate emergency messages, warnings, and road status with high, medium, and low delivery priorities. These messages are crucial for preventing accidents, ensuring the safety of individuals, reducing fuel consumption, and improving the travel experience. Therefore, including additional data on the 802.11p frequency channel for remote monitoring of other sensors is inefficient, as it would compromise the communication channel and, consequently, the safety of individuals. To alleviate data congestion in the V2I communication channel, an alternative is to expand the frequency spectrum using IoT technologies that operate in unlicensed or licensed bands, such as Long Range (LoRa), Sigfox, Narrow-band Internet of Things (NB-IoT), WiFi, BLE, and Zigbee, as indicated in [36]. This allows using different communication channels for sending lower-priority data, such as sensor data and even control or emergency messages, without compromising the citizen’s privacy [4].

The underlying reason for integrating these technologies into a single gateway, referred to as a multi-interface mobile gateway (M2GW), is to overcome the incompatibility among devices, thus enabling the management and facilitation of bidirectional communication between them, regardless of the technology used by each device. Moreover, the Gateway Mobile device can perform prepossessing for other applications, such as verifying that a software update is error-free before installing it in vehicles. This reduces response time and ensures people’s safety [37]. This integration effectively addresses the challenge of device heterogeneity at the physical level [38], but managing multiple interfaces becomes a new challenge. At the application level, heterogeneity does not pose a significant challenge, as it is possible to establish protocols, such as Constrained Application Protocol (CoAP) or Message Queuing Telemetry Transport (MQTT), through software, as discussed in [39].

Software Techniques for the Interface Management of M2GW Managing diverse technologies integrated into the M2GW has become possible thanks to advances in developing increasingly powerful microcontrollers, as evidenced in [11, 12, 38]. These works highlight the use of development boards such as Raspberry Pi, RK3399, and STM32, which enable the execution of multiple tasks, such as reading serial ports and handling data, including data fusion and aggregation [40] [41]. These functions are achieved by applying multiprocessing, FreeRTOS [42], and Direct Memory Access (DMA) [43], as illustrated in Fig. 3.

Fig. 3
figure 3

a Flowchart with multiprocessing b Flowchart with FreeRTOS c Flowchart with DMA

Full size image

The Raspberry Pi supports multiple processes in parallel, utilizing its 3 or 4 cores. However, excessive parallel process creation may lead to resource competition, impacting performance due to shared CPU time and system resources. FreeRTOS, an open-source real-time operating system, is widely employed in STM32 microcontrollers, ESP32, and Arduino platforms. In STM32, it efficiently manages concurrent tasks with its single or multiple-core architecture. In ESP32, it serves as the underlying OS, enabling multitasking on the dual-core chip, while in Arduino, it facilitates multitasking and complex system development. FreeRTOS’s adaptability and flexibility make it a favored choice for embedded systems developers. Additionally, DMA in STM32 microcontrollers optimizes data transfer without CPU intervention, enhancing system efficiency by reducing CPU load through burst mode data transfer between peripherals and memory.

Tackling Interference Issues with IOT and WSN Technologies in M2GW Integrating different IoT technologies in an M2GW requires meticulously analyzing coexisting protocols in the same frequency band, such as Wi-Fi, ZigBee, and Bluetooth (see Fig. 4). Despite ZigBee having minimal impact on Wi-Fi, in [38], it is emphasized that configuring Wi-Fi power to 30 dBm can reduce interference. However, as a complementary approach, [32] proposes enhancing Clear Channel Assessment (CCA) to reduce continuous interference in environments where Wi-Fi and ZigBee networks coexist. Additionally, the use of CSMA/CA, a non-collaborative method inherent to the IEEE 802.15.4 physical layer, is suggested. In contrast, ZigBee can interfere with three channels when using Bluetooth Low Energy (BLE). Still, it is essential to note that BLE provides 37 general-purpose and 3 broadcast channels, offering flexibility in interference management.

Fig. 4
figure 4

WIFI, BLE, and ZIGBEE channels

Full size image

4.2.2 Access to the Cloud Using Cellular Technology

The M2GW can integrate with cellular technologies such as 3G, 4G, and 5G, enabling internet connectivity for the gateway when sending an emergency packet or when the maximum time limit to update sensor information has been exceeded. This approach addresses challenges associated with intermittent network connectivity of nodes (vehicles). Networks experiencing inconsistent connectivity and non-consecutive end-to-end paths utilize a ”store-carry-forward” mechanism [44], buffering messages until a connection opportunity arises, typical of Delay Tolerant Networks (DTNs) [45]. Best-effort protocols could be an appropriate alternative for environmental monitoring applications as they transmit the optimal number of data packets within time constraints. However, determining ideal data retention periods for resource efficiency is difficult in this protocol. Real-time protocols face challenges with data degradation over time, which is crucial in VANETs for decision-making, enhancing safety, traffic flow, and autonomous driving support [46]. Moreover, the vehicles in metropolitan areas require better support from infrastructure-based overlapped vehicular networks. This setup effectively transmits safety messages from numerous vehicles on high-density roads with multiple lanes and nearby intersections. For this reason, the use of 5G technology is a potential future solution, as it requires the infrastructure to be fully installed [47]. Clustering algorithms play an essential role in both the Best-effort and Real-time protocols. Therefore, various techniques and approaches have been proposed to use clustering-based data forwarding methods for VANET-based monitoring applications. In [48] where two techniques for internet access using 4G were compared: clustering and non-clustering techniques for gateway selection are provided by the au, the authors provided clustering and non-clustering techniques for gateway selection network. The first reduces collisions and increases network overhead, while the latter decreases control but poses gateway overloading risks. On the other hand, [49] proposes combining VANET 802.11p and 3G/UMTS for vehicle-to-base station communication. While effective, it relies on periodic beacons, increasing signalling time. In [50], a gateway selection approach with passive clustering minimizes message flooding by considering link stability with the LTE base station is proposed. In [51], although 5G cellular technology is presented as a promising solution, it also introduces new challenges, such as exposure to distributed denial-of-service (DDoS) attacks. These attacks flood the network with unexpected control information, leading to a decrease in network performance. To address this issue, solutions based on reinforcement learning are proposed. On the other hand, [52] proposes a 5G-type traffic generator that uses refactored open data of vehicles and pedestrians from Montreal to simulate various network traffic scenarios. This tool addresses the challenge of sparse 5G traffic data by creating diverse traffic patterns relevant to 5G applications, validating machine learning algorithms for traffic forecasting and resource management.

5 VANET Sensing

Wireless Sensor Networks (WSNs) and IoT can play a crucial role in the context of VANETs as they provide advanced sensing capabilities. This approach enables sensor systems integrated into vehicles and road infrastructure to collect real-time data essential for enhancing safety and efficiency in mobility [53]. It also allows for optimizing traffic signal operations, accident detection, environmental monitoring, disaster alert generation, and pedestrian safety. Below, we review potential applications, uses, and challenges for implementing M2GW.

5.1 Potential Applications of the M2GW

So far, we have observed that M2GW devices play a crucial role in supporting IoT systems in VANET environments, as highlighted in [54] for wireless healthcare monitoring. This analysis addresses the urgent need to transfer information quickly, accurately, efficiently, and promptly, especially in critical situations requiring early medical attention in urban areas. However, these applications face challenges such as interference caused by obstacles, buildings, and vehicular traffic. In this context, the implementation of routing protocols, as suggested in [55, 56], could be efficiently integrated within M2GW devices as a viable solution to prevent overflow in the communication channel of the 802.11p standard.

In the context of air pollution in VANET, the increasing traffic congestion in urban areas poses significant challenges, as vehicle emissions impact air quality and negatively impact human health. The inhalation of atmospheric pollutants can affect respiratory and cardiovascular systems, with long-term consequences on overall health [57]. These issues are addressed by implementing gateways installed on vehicles to gather information about the vehicle’s condition [58], along with data from various environmental monitoring sensors detailed in Table 1. Subsequently, this information is processed at the Edge and/or Cloud to evaluate the degree of air pollution and environmental quality to which individuals are exposed when using public transportation or navigating areas with high traffic congestion. This represents another practical case for using an M2GW, given the diversity of technologies deployed by sensors in the city for environmental monitoring.

Table 1 Features of sensors used by different authors for air monitoring
Full size table

It is also feasible for the M2GW to retrieve vehicle information such as GPS data, vehicle status, acceleration, speed, gear changes, RPM, deceleration, and distance, among others, as it has the hardware and software capability to implement M2GW for local data tracking and analysis. These data can determine how driving habits, frequency of driving, safety awareness, speed, driver age, engine power, and vehicle displacement influence air pollution in urban areas [59, 60]. This way, authorities can identify sources of pollution, establish emission limits, and implement corrective measures when necessary, aiming to ensure compliance with environmental standards and government regulations.

This variety of applications, see Table 2, supported by various models, algorithms, and techniques such as crowdsensing in the field of remote monitoring within VANETs [61] or [40] where the authors focus their study on calibrating these sensors, aiming to obtain reliable readings through low-cost technology, continue to contribute to the concept of smart cities. This is reflected in the advancement of semi-automatic or fully automatic vehicles [62], aiming to mitigate air pollution, reduce fuel consumption, and assess future health risks.

According to the reviewed literature, M2GW can integrate heterogeneous networks [63]. Their operational robustness in handling data from diverse sources contributes to integrating large-scale systems. In addition to being a device dedicated to managing IoT sensor traffic, it is cost-effective and has reduced energy consumption. It can also perform multihop forwarding techniques [64] and prepossessing tasks, such as data filtering or aggregation, before sending them to the cloud environment [40, 65]. To improve accuracy and reduce the massive volume of data sent to the cloud, minimizing the presence of redundant, unnecessary, and erroneous measurements collected by the sensors.

Table 2 Comparison of different Gateways implemented in for Monitoring
Full size table

5.2 Implementation and Management of ML Models Through the Use of M2GW

The M2GW allows developers to implement ML models through programming, allowing for local calculations with sensor data. In a specific example mentioned in [9], the M2GW performs a correction of factors affecting concentrations of particulate matter (PM) using linear regression and multiple linear regression operations. Furthermore, the M2GW can implement models to calculate emissions and instant fuel consumption, as described in [69]. This is achieved by using the vehicle’s speed and acceleration at a sampling frequency of one second. However, the precision of this regression model decreases without having a comprehensive and detailed vehicle categorization, as evidenced in [60]. This highlights the need to include crucial parameters, such as vehicle power and driving habits, to enhance the model’s accuracy. Still, different types of sensors help understand the vehicle’s characteristics and the surrounding environment. Furthermore, in [40], the challenges of sensor accuracy, mobility, vehicular speed, and temperature fluctuations are also addressed by implementing ML algorithms. These studies show that the performance of the M2GW when using techniques and models is effective.

5.3 Challenges in Ensuring the Sustainability of M2GW Infrastructure in Vehicles

Integrating an IoT and WSN system in a VANET presents persistent challenges, primarily due to the costs associated with installing and maintaining sensors, connectivity, and servers. Managing hardware and software and addressing potential failures demands significant financial and logistical resources. Possible technological obsolescence may necessitate periodic investments if not planned with a long-term perspective. An approach discussed in [70] focuses on monitoring air pollution for Connected Vehicles (CVs), emphasizing the importance of balancing costs and latency in data collection through a heuristic algorithm to optimize the coverage of RSUs.

In contrast, another study in [71] uses GPS to analyze the distribution of buses without relying on RSU, employing a community graph-based approach. The efficiency and performance of the public transportation system are evaluated by considering various strategies and comparing them with existing routing schemes. Under specific conditions, the proposed model achieves a data transfer of 6.7 MB with a latency of less than 15 min, surpassing alternatives like BLER [72], R2R [73], GeoMob [74], and ZOOM [75]. These results highlight the feasibility and effectiveness of the proposed approach in VANET environments.

6 Key Aspects for Implementing M2GW in VANET Environments

To implement a system using an M2GW device, it is essential to conduct preliminary studies before its installation in the field. These studies should assess the system’s impact, performance, and efficiency. For this purpose, the use of simulators is crucial, as they allow for the recreation of complete scenarios that include environments, sensors, communication signals, vehicles, and obstacles. This approach helps researchers translate real-world situations into a virtual environment, enabling precise modeling of vehicle behavior, drivers, pedestrians, and traffic signage within a city. Thus, researchers can evaluate and optimize techniques to enhance system performance, avoiding time and financial resource losses. Below are some traffic simulators specialized in modeling vehicular mobility. These simulators can communicate via APIs with network simulators, thereby reducing the processing power required for simulation.

6.1 Vehicle Simulators with Real-Time Communication via APIs

Some of the commonly used software for modeling vehicular environments are presented in Table 3. Among them, SUMO (Simulation of Urban MObility) stands out for being free and open-source, making it accessible to academic researchers, developers, and urban planners. This software can handle large-scale simulations with high precision in urban traffic modeling, addressing complex and detailed scenarios without incurring significant costs [76]. Additionally, they offer real-time communication interfaces via APIs. In the case of SUMO, TraCI (Traffic Control Interface) is the communication interface provided by SUMO that facilitates integration with other systems and dynamic customization during simulation. This means SUMO can provide information about each vehicle in the simulation, such as speed, acceleration, vehicle type, Co2, So2, No2 emissions, Noise etc. [77], and it can also receive data to update simulation variables and change the established scenario, making the simulation dependent on the newly acquired data. These simulators are compatible with various input and output formats and can be extended through scripts and plugins, providing considerable flexibility. For many of these simulators, the active user and developer community ensures continuous support, as well as the availability of resources and documentation. This makes them an attractive option for research, development, and planning projects, where flexibility, time, and cost are key factors.

Table 3 Features of Vehicular Simulation Software
Full size table

6.2 Network Simulators that Integrate Other Software During Simulation

On the other hand, there are also network simulators as in Table 4, these can obtain data from other software like SUMO during simulations via APIs. This approach separates the processing tasks, where the network simulation software focuses solely on communications, while the second software handles modeling mobility, the environment, or other scenario characteristics [78, 79]. OMNeT++ is a highly versatile discrete event simulator that implements a module called INET, which covers a wide range of TCP-IP model protocols from the physical layer to the application layer, for both wired and wireless networks. It also includes mobility models and allows for the customization or creation of sensor modules, routing protocols, and optimization techniques [80]. NS-3 is an open-source simulator commonly used for packet-level and Internet protocol simulations. Although implemented in C++, it can be integrated with Python, enabling simulation scripting in both languages. It features extensive and detailed documentation, including tutorials, user guides, and API references [81]. Mininet-WiFi is ideal for emulating mobile networks and can be integrated with Python. However, it only simulates the WiFi and WAVE protocols and lacks models for mobility and interference. GNS3 combines virtual and real devices, CORE creates virtual environments for networks, Estinet simulates both packet-level and event-level scenarios, and Cooja is best suited for sensor networks.

Table 4 Features of the most used network simulators
Full size table

7 Data Forwarding Through Clustering in VANET

Typically, in smart urban scenarios, vehicles in VANET work as network nodes, exchanging data with peers and sensing and transmitting environmental parameters. The apparent dynamic behaviour and movement with a high velocity of vehicles in VANETs make storing and forwarding the real-time information to the internet-connected collector nodes (RSUs) very complex [82]. Because of this reason, VANET routing protocols are essential to the ITS, and detailed research to categorize them into different classes based on their strengths & weaknesses for different monitoring and emergency applications is also needed. The key parameters to design such protocols are delay, packet delivery ratio, bandwidth utilization, and many other factors [83]. The state-of-the-art monitoring applications of VANET clustering protocols are (i) road safety, (ii) infotainment-based, (iii) traffic monitoring, and (iv) passenger health monitoring. However, selecting a proper routing protocol could be laborious due to the dynamic topology and characteristics of VANETs [84].

Infrastructure-based VANETs enhance V2V & V2I communication through scheduled access and resource distribution. However, the infrastructure-based widespread deployment of VANETs introduces scalability issues and unstable connection challenges. On the other hand, some advantages can be realized using clustering without physical infrastructure. Clustering, a virtual arrangement of nearby vehicles into groups, effectively solves these difficulties and could play a vital role in every emergency and monitoring application [85]. For example, the GPS attached to the vehicle monitors node location within the network while obtaining data on the vehicle’s speed and the passengers’ health. This can aid in making clustering decisions. To improve the real-time data transmission of the VANET in the urban environment, packets must be sent while considering the road conditions. Vehicles generate a network topology based mainly on the road layout, and packets are relayed from the source node to the destination through the roadways [86]. VANET operations in an on-demand environment scenario have significant drawbacks, and they cause problems with their deployment in real traffic situations. Additionally, they are vulnerable to the hidden node problem, which must be addressed with limited spectral bandwidth and a highly variable channel affected by both fixed and mobile obstructions. Clustering is essential to facilitate efficient communication between VANET nodes in such scenarios. Cluster-based routing protocols offer advantages over other routing protocols for bidirectional traffic, data dissemination techniques, and QoS guarantee [87]. Various types of cluster-based routing protocols can be used to achieve efficient communication.

Further, it is found that due to the increase in fossil fuel consumption by vehicles, there is a soar in the carbon footprints of the environment [88]. As we can see, there is a gradual shift in the demand for electric vehicles in the modern era [89]; we also need to ensure the energy consumption for the IoT (sensing, communication, and forwarding) by the set of vehicles is optimum. Because of this, energy efficiency is also one of the prime concerns in VANET’s clustering protocols. The transmission of information among a large fleet of vehicles with unpredictable behaviour leads to blockages in vehicular communication, causing poor energy efficiency, dropped packets, and prolonged delays due to restricted timing and message delivery [90]. An appropriate example would be the initial stage of route establishment, which requires a high volume of routing requests (RREQ) and routing reply (RREP) packets, leading to significant energy consumption [91]. Another instance of power dissipation is when there is communication between the gateway interfaces of the RSUs and the vehicles of VANET. Usually, the more complicated the route or the longer the distance, the more power consumption [92]. Developing such energy-efficient protocols for periodic monitoring applications remains a significant challenge [90]. Since static routing protocols experience a premature end of network lifetime, transmitting packets to the base station using multiple hop clustering is better than direct transmission in large geographic areas. Therefore, the vehicles’ computing and communication tasks might be distributed between sensing vehicles, RSUs, and relay vehicles to minimize data transmission power consumption [93]. Hence, clustering must be implemented after assessing all these critical aspects to resolve the data forwarding issues.

The following subsections explain clustering techniques, including their anatomy, advantages, and limitations in monitoring and energy efficiency applications.

7.1 Architecture and Anatomy of VANET Clustering

The concept of clustering is implemented in VANET to improve network performance by addressing complex issues such as scalability, limited resources, reliability, and the hidden terminal problem within the network [94].

Fig. 5
figure 5

Cluster-based network structure

Full size image

In a VANET at a particular instance, there could be multiple clusters that may or may not overlap. In the following stages, each cluster comprises at least one cluster head (CH) that leads the group, followed by several vehicles known as cluster members (CMs).

During the Neighbourhood Discovery stage, vehicles enter the VANET, activate the infrastructure scheme and exchange periodic HELLO messages with nearby vehicles. In the Cluster Head Selection phase, vehicles calculate metrics to determine if they qualify as CHs, updating their status accordingly and progressing to announcement or affiliation stages based on neighbour data. Affiliation occurs when a node compares received announcements with previously selected CHs; if matched, it updates and joins; otherwise, it waits for additional announcements or moves to maintenance. Following the election, each CH must broadcast an announcement packet to initiate cluster formation, receiving affiliation requests from unclustered nodes before proceeding to maintenance. Maintenance involves CHs monitoring CMs through periodic messages and updating out-of-range node lists. At the same time, CMs periodically check connections to CHs and request to join clusters if unclustered.

Clusters are classified as one-hop or multi-hop based on the distance between CH and CM. One-hop clusters are formed within the transmission range of CHs, with CHs adding one-hop neighbours. CMs communicate directly with CHs or indirectly via CHs. Multi-hop clusters connect CMs indirectly. Interconnected clusters use Cluster Gateways to monitor vehicle statuses periodically [95] [96]. Figure 5 shows clusters communicating among themselves. The number of clusters formed in clustering depends on node density and range. All message dissemination between nodes will go through cluster heads instead of every node broadcast communication. For this type of communication, electing a trustworthy and appropriate cluster head is essential since the whole cluster’s communication responsibility rests solely on the cluster head. Therefore, the node’s speed, vehicle density, and similar factors should be considered while selecting the cluster head [97]. Cluster formation and cluster head selection demand consideration of crucial metrics outlined in Table 5. A top-tier clustering algorithm aligns with these metrics, ensuring efficient performance within specified limits.

Table 5 Parameters of clustering protocols in VANET for performance evaluation
Full size table

Since introducing clustering algorithms, various approaches to the problem have been suggested. Each practice focuses on distinct issues, frequently aligned with specific applications envisioned for VANET technology. One such issue is integrating LTE, 5G, DSRC, etc., in heterogeneous networks, which poses infrastructure and cost/data sharing challenges. The prompt and efficient data dissemination in a heterogeneous network using reliable and scalable V2I clustering strategies for monitoring and energy-efficient applications are some of the benefits we discuss in detail in the following subsections [98].

7.2 Cluster-Based Information Dissemination in Monitoring Applications

In most monitoring applications, data transmission occurs in the V2V and V2X, considering several parameters, as mentioned in Table 5. Several surveys quantified and pointed out the impacts of these parameters and categorized them according to the application viewpoint. The section discusses the different clustering challenges in VANETs, assessing the advantages and drawbacks of V2V and V2I data dissemination for safety and monitoring applications. Typically, the cluster-based routing protocols for monitoring applications, which are responsible for continuous sensing and transmitting the data, have some standard features regarding their implementation. Vehicles in VANET share their status through periodic beacon messages for safety, creating accurate proximity maps to prevent accidents and alert drivers. Due to the dynamic behaviour of the VANET clusters, it is also essential to have multi-hop and relative mobility between the intra and inter-vehicle communication, which is a crucial parameter for forming clusters. The multi-hop approach provides stability and increases the coverage area necessary for emergency message exchange. Depending on their size, the traffic flow could also be responsible for the overhead and delay in forming the clusters. Furthermore, Broadcasting of packets and Cost effective data aggregation are critical parameters for effective bandwidth utilization and collection of the data at the cluster gateways and RSUs.

In literature, application-specific algorithms are categorized based on purpose, while some aim for versatility across diverse applications. In this section, we categorized the cluster-based routing techniques, focusing on Beaconing Based Methods, Relative Mobility and Multi-Hop centric Approaches, Traffic Density Estimation, Broadcast Storm Problem, Optimal Resource Utilization, providing a taxonomy for VANET clustering algorithms and exploring specific algorithms within this framework. Figure 6 depicts the taxonomy we deduced.

Fig. 6
figure 6

Classification of clustering algorithms for monitoring applications

Full size image

7.2.1 Beaconing Based Methods

The short-range inter-vehicle safety information exchange occurs in V2V in a VANET. These safety messages carry highly time-critical data, which imposes high demands on the timing and reliability of the underlying communication protocols. Typically, the exchange of messages in realistic large-scale areas based on the IEEE 802.11p standard takes place using beacons. However, supporting certain real-time safety-critical applications under challenging communication conditions while meeting stringent QoS requirements is difficult for IEEE 802.11p [99]. Chhabra et al.in [100] proposed a context-aware hybrid beaconing system to enhance driver safety by sharing data with nearby vehicles. The system embeds driver status in beacons and uses Service Broadcast Notification (SBN) to collect driver and network status. The network congestion is avoided dynamically by adjusting transmission range, power parameters, beacon interval, and contention window using the Enhanced Distributed Channel Access (EDCA) mechanisms. In the proposed scheme of [101], Beacon-oriented Emergency Message Delivery (BEMD) improves the coverage and distribution of emergency messages with minimal delay. The objective is to use beacon-oriented communication for high-mobility traffic with low density on highways, minimizing end-to-end delay and increasing reliability.

7.2.2 Relative Mobility and Multi-hop Centric Approaches

Typically, for the message exchange in V2V, V2I, or inter-VANET communication, the vehicle communicates with multiple hops due to dynamic topology changes. Many prior studies concentrate on vehicle clustering techniques in VANET in this aspect [102,103,104,105]. These methods typically employ hierarchical network structures instead of flat ones. Within a cluster, vehicles can communicate directly with others or through a designated CH. In [106], a position-based protocol features a multi-hop greedy forwarding scheme. It organizes candidate lists from detected neighbours for emergency message carrier selection in the next hop. Density changes are managed with fixed and dynamic beacon control strategies. In [107], the author implemented a stable Clustering Algorithm for Efficient Multi-hop Vehicular Communication (CAMVC) to enhance the observation of cluster speed, acceleration, closeness centrality, and position parameters in VANETs. This approach extends cluster lifetime using a density-based clustering non-parametric algorithm (DBSCAN) for cluster configuration and Fuzzy Logic Control for CH selection. The Link Connectivity Duration (LCD) calculates the connectivity duration and stability to select the gateway to establish multi-hop communication with the CH. This clustering is perfect for implementing on highways where velocity is high. The authors in [108] presented An efficient hierarchical clustering protocol (EHCP) for multi-hop communication that reduces control message broadcasting, increases the network overhead and packet collision, and extends network lifetime. EHCP assumes vehicles connect to the Internet via Roadside Unit Gateways (RSU-G), enabling them to gather neighbour information and select CHs. In [109], the authors proposed an Advanced multi-hop clustering (AMC) algorithm for VANETs. A designated vehicle serves as a cross node at road junctions, assigning weights to road segments based on connectivity analysis to determine optimal data forwarding paths. The AMC routing protocol is recommended to disseminate traffic emergency information.

7.2.3 Traffic Flow Estimation Methods

Density estimation is one of the most crucial aspects of VANET clustering schemes. With the increase in traffic density, the possibility of collision also increases. However, the stability of the clusters also reduces frequent CH selection overhead. The authors in [110] introduce the traffic density-based-congestion control algorithm (TDCCA) method. This updated mathematical method improves the packet delivery ratio (PDR) compared to prior work using a quarter of the congestion window size. It introduces a congestion management technique that adjusts beacon message rates and considers diverse VANET scenarios from substantially saturated to sparsely dispersed networks. Further, it analyses and predicts traffic-dense paths using different vehicle ID-based back-off values and message collisions of vehicles. To counter the problem of no communication zones in traffic applications in [111], Gillani et al. proposed a Real-Time Traffic-Aware Data Gathering Protocol (TDG) that adopts dynamic partitioning of vehicular communication zones to manage communication limitations, aiming to reduce network overhead and meet real-time data collection constraints. TDG utilizes lightweight and dynamic design principles. The Real-Time Cluster Head Election (R-CHE) algorithm dynamically selects the most appropriate Cluster Head (CH) to gather data from neighbouring vehicles and share aggregated data with the Sink.

7.2.4 Broadcast Storm Problem

Collisions can occur when safety messages are improperly broadcasted and packets are transmitted simultaneously from multiple vehicles. The design of efficient broadcasting algorithms is essential as a basic service to support safety and emergency applications. Conversely, transmitting packets might cause regular disputes and collisions of packets in transmission between nearby vehicles. In [112], the Novel Segment-based Safety message broadcasting in Cluster (NSSC) for Vehicular Sensor Networks is introduced. NSSC employs the segment-based forwarder selection (SFS) scheme to combat the broadcast storm issue in disseminating safety messages. This method utilizes the Fuzzy-Vikor algorithm within the squared region to determine the optimal forwarder, thereby mitigating packet duplication. Consequently, NSSC effectively prevents broadcast storms while transmitting safety messages among vehicles. Another approach was taken in [113] cluster-based RSU-enabled message aggregation (CluRMA), a message aggregation protocol that uses clusters and RSUs to aggregate data. The protocol uses local aggregation at cluster heads and global aggregation at RSUs to address broadcast storm issues, optimizing bandwidth for diverse messages in heterogeneous networks while reducing redundancy and traffic. In their research article [114], Khan et al. introduced VP-CAST, a broadcast suppression protocol designed for VANETs; VP-CAST incorporates car position and velocity data from the Global Positioning System (GPS) in emergency messages and employs vehicles travelling in the opposite direction via store-and-carry mode. It is adaptable to sparse and dense networks, avoids emergency message collision, and improves the message delivery ratio. All these features make it useful in safety applications.

7.2.5 Cost Efficient Approaches

Although research on vehicle clustering aims to reduce the amount of LTE resources used, data rate and delay sensitivity are crucial for road traffic management applications like route planning and driver assistance. The authors in [115] introduce a Self-Adaptive clustering algorithm based on justified distribution to each vehicle, which dynamically adjusts the maximum number of hops according to vehicle density. The approach balances data aggregation over the cellular network and network congestion in V2V, enhancing packet delivery ratios and equitable communication cost distribution. As a result, it offers potential benefits in efficiently transmitting large amounts of data, particularly images and videos, in VANET. Based on the Cooperative Interest-Aware Clustering (CIAC) protocol [116], a clustering framework called destination and interest-aware clustering (DIAC) is proposed in [117] extended to VANET-LTE heterogeneous clustering. The cluster is built by keeping speed, location, and information from online traffic information services (TIS) for route planning. The framework allows the sharing of data and costs between VANET vehicles while moving on the road. Akbar et al. in [118] proposed SeAC (SDN-enabled adaptive clustering). They improved cluster stability by introducing an SDN controller and made the clustering efficient by considering physical location, social ties, and interests. SeAC minimizes communication costs and enhances cluster lifetime, minimizing delays and packet loss. In [119] Khan et al. proposed CVoEG model which optimally clusters VANET nodes using Eigen-gap heuristic, selecting cluster heads (CHs) based on maximum Eigen-centrality scores. The CVoEG model is paired with RAODV to form CEG-RAODV, enhancing network reliability, packet delivery ratio, end-to-end delay, and throughput. Simulation results show significant performance improvements in dynamic vehicular scenarios, while scalability and reduced control message overhead are key benefits.

Table 6 Comparison between different clustering algorithms
Full size table

7.2.6 Discussion

Clustering in vehicular networks enhances resource optimization over flat routing, benefiting safety and monitoring applications. Recent significant works offer a hybrid architecture combining LTE and DSRC for independent routing and forwarding optimization, ensuring low delay and high data delivery ratio. Algorithms focus on bandwidth optimization through message aggregation and compression, preventing path blockage between gateways, base stations, and vehicles. Some algorithms predict car positions, direction, cluster size, and nearest RSUs to select optimal Cluster Heads, enhancing throughput and reducing latency. Table 6 compares several monitoring algorithms in terms of their size (number of vehicles), scenario (urban/semi-urban/highways), types of applications and objectives.

7.3 Cluster-Based Energy-Efficient Routing in VANET

Efficient communication in VANETs requires energy optimization for prolonged operation. The advanced capabilities of 5G/6G networks may increase energy consumption due to higher data rates and complexity. Meeting the increasing demand for enhanced Quality of Service (QoS) involves optimizing network throughput, yet high throughput often leads to unsustainable energy consumption. The energy consumption of OBUs is another significant concern in clustering-based routing protocols, as the sensors consume a large amount of power for opportunistic V2X communication. Due to the above reasons, designing energy-efficient protocols is essential for VANETs while maintaining their performances. It will more effectively minimize energy consumption within the VANET monitoring applications and improve the QoS. These factors motivated us to document the study of recent developments of such energy-efficient algorithms for monitoring and emergency information exchange algorithms. In this section, we explain a recent set of energy-efficient routing algorithms along with their pros and cons.

Fuzzy logic-based approaches leverage the principles of fuzzy set theory to enhance energy efficiency in VANETs by optimizing clustering algorithms. Some recent works use these techniques to use parameters like node distance, residual energy, and neighbour nodes to manage network performance and lifespan. In [120], the authors introduced a fuzzy-based clustering algorithm to improve energy efficiency in vehicle ad hoc networks. The algorithm uses three significant parameters: node distance, residual energy of the node, and the number of neighbour nodes. These parameters are used as input for the fuzzy management system. The system’s output provides information on the network lifetime, including the number of live nodes, expiry nodes, and node availability at each round. Data transmission can be a bottleneck in VANET, presenting a significant challenge. In [121], the authors introduced the TP-FUZZY clustering algorithm to improve energy efficiency in vehicle ad hoc networks. The algorithm uses parameters such as node distance, residual energy, and neighbour nodes as inputs to a fuzzy system. The algorithm employs a threshold mechanism that considers vehicle speed, data rate, and distance. The output determines network lifetime, identifying live nodes, expired nodes, and node availability in each round, addressing challenges in VANET data transmission. Blessy et al. in [122] proposed mechanism uses an intelligent fuzzy system, optimized by the Bald Eagle Search (BES) algorithm, to enhance the CHS process. Evaluated with MATLAB, it achieved a 13.58 ms delay, 15.5 J energy consumption, and high clustering efficiency and packet delivery rates of 94.15% and 97.65%, outperforming existing methods. Optimization approaches focus on enhancing energy efficiency in VANETs by fine-tuning resource allocation and routing protocols. These methods utilize advanced optimization techniques to minimize energy consumption while maintaining high network performance and stability. In [123], the authors aim to reduce energy consumption in edge offloading by optimizing computing tasks and file transmission costs in 5G heterogeneous networks, proposing an Energy-Efficient Computation Offloading (EECO) scheme optimizing offloading and resource allocation with latency constraints. It enables priority-based ambient and traffic monitoring applications at the User End while reducing energy consumption. This algorithm is helpful in schemes as it derives the priorities of the devices that offload their tasks to the Mobile Edge Cloud server. In [90], the Elhoseny et al. s study presents the Intelligent Energy-Aware Oppositional Chaos Game Optimization-based Clustering (IEAOCGO-C) protocol for VANETs. IEAOCGO-C efficiently selects network Cluster Heads (CHs) using oppositional-based learning (OBL) and chaos game optimization (CGO) algorithms. It constructs clusters based on these techniques to establish effective data transfer paths, taking into account distance, trust factor, and energy while also optimizing energy consumption through a fitness function. The authors in [124] developed an alternative Mobile Energy Aware Cluster Based Multi-hop (MEABCM) protocol for routing using heterogeneous WSNs in a cluster-based multi-hop network to reduce energy consumption. The nodes that are not part of any cluster or are no longer reachable are identified using a notion of sub-clustering. The results show that it outperformed the contemporary algorithms regarding stability and network lifetime and provided better energy efficiency in heterogeneous WSNs. The SDNFoG-IoT architecture in [125] fundamentally aggregates SDN-controller, FOG-controller, and cloud service tiers to control the data transmission rates. It employs the Energy-Efficient Routing Methodology (EERM) mathematical model, a static and dynamic clustering hybrid, ensuring optimal energy utilization. Pal et al. in [31] considered the minimum distance between two nodes among RSUs, destination vehicles and other vehicles to transmit the data considering the maximum throughput and least energy (threshold 0.05j) consumption. To minimize the hop count the sender vehicle sends the data to the furthest vehicle reachable. It is found to provide better QoS and is more energy efficient than EERM. Artificial Intelligence approaches leverage advanced algorithms to enhance energy efficiency and network performance in VANETs. These methods utilize clustering and optimization techniques to manage data transmission and clustering consistency effectively. A clustering-based optimization technique, the Energy Efficient Clustering Technique (EECT), was developed by the authors [126] for V2V communications using the ad hoc on-demand distance vector (AODV) protocol and machine learning (ML) based K-Medoids clustering algorithm. The method aims to cluster vehicles and identify those capable of interacting in a predetermined, secure, and reliable path. Efficient nodes are identified as CHs from each cluster to improve energy-efficient communication and network performance. The authors in [127] introduce the Energy-based Clustering Model with Data Distribution Strategy (ECM-EDT) to sustain network clustering consistency, decreasing energy consumption and extending network lifespan. The Head Vehicle retrieves data from the base station and efficiently transmits it to other Moving Vehicles through the Data Distribution Strategy. This strategy optimizes energy consumption rates during data transmission. In [128], the authors presented the performance outcomes of a self-organizing map neural network (SOMNN). This unsupervised ML algorithm identifies energy-efficient cluster head nodes from each cluster for vehicular data uploading and downloading applications. It achieves higher cluster stability in a highly dynamic vehicular environment and outperforms K-Means and Fuzzy-C Means in all performance indices. Additionally, SOMNN consumes less energy compared to K-Means and Fuzzy-C Means. In another article [92] Choksi et al. introduces a hybrid fuzzy c-means and machine learning-based (FM), dynamic clustering algorithm (DCA) for energy-efficient multi-hop routing, considering vehicle mobility and energy parameters. It outperforms k-means-based algorithms and DSR/AODV protocols in packet delivery, throughput, delay, and energy consumption.

Table 7 Comparison between different energy efficient algorithms
Full size table

7.3.1 Discussion

We witnessed that the reduction in power consumption is not directly proportional to the performance degradation. In most algorithms, the results show improvement; however, the throughput and packet loss remained static in some cases. Although the study is not exhaustive, it puts all the essential aspects of energy-efficient VANET clustering for monitoring and safety applications. Table 7 compares different aspects of the above-mentioned algorithms like energy consumption reduction, underlying clustering method and algorithm influence in QoS.

8 Conclusion and Future Works

As VANETs continue to evolve, researchers are increasingly focusing on addressing the complex challenges posed by the growing number of vehicles, urban congestion, and limitations in communication technologies. One emerging trend is the integration of WSNs with VANETs to create hybrid networks capable of addressing various scenarios in ITS. By leveraging technologies such as LPWAN, cellular networks, and smart devices, these hybrid networks aim to enhance communication reliability and coverage while optimizing data management. However, the integration of WSNs alone cannot fully address VANETs’ challenges, as each technology has strengths and limitations. Therefore, future research in this area will likely explore novel techniques, algorithms, and protocols to optimize the joint operation of VANET-WSN hybrid networks. This includes reducing congestion and data delay on communication channels, improving the response time for critical messages, and enabling the development of new applications for traffic management, environmental monitoring, and passenger safety. Additionally, researchers may explore the potential of these hybrid networks to serve as temporary support infrastructure in emergencies, further enhancing their versatility and effectiveness in real-world deployments.

Next-generation VANET clustering will likely leverage dynamic and intelligent algorithms that adapt to real-time traffic conditions, vehicle density, and speed. Machine learning integration could optimize cluster formation for scalability, network efficiency, and critical data handling (e.g., emergency broadcasts of traffic and safety data). However, several key challenges remain. Handover management is under-explored in VANET, which minimizes delays, and data loss is significant in real-time critical applications. Protocol design for VANETs should start with the monitoring application and then analyze known challenges to leverage or mitigate them with diversity techniques. Multi-level clustering, including the potential benefits of single or double cluster heads, is a wide-open research area. We also lack definitive answers on optimal CH and member selection techniques, particularly the crucial attributes for a robust leader. Finally, the effectiveness of clustering in ensuring QoS and security needs further investigation. Research into multi-homing clustering techniques, which could enhance stability, holds significant promise.

This study comprehensively reviewed and analyzed the literature on data exchange in V2V and V2I communication systems in monitoring applications. Due to ambiguous descriptions and boundaries, VANET communication is an ongoing research field. Understanding this area is crucial, and this paper contributes by surveying and categorizing relevant research efforts. It followed a systematic literature review approach, starting with the architecture and anatomy of VANET infrastructure. Furthermore, it discusses accumulating data regarding environmental, safety & emergency, and traffic monitoring metrics using OBUs and multiple sensors. It further analyses the advantages and disadvantages of implementing a multi-interface mobile gateway (M2GW) to have an efficient data collection system by transmitting data packets using heterogeneous communication modules, LPWANs, and cellular technologies. Moreover, it also analyses the implementation of software-based interface management, ML-based techniques for local data calculations, and the sustainability challenges of M2GW. It also highlights the gateway to cloud data dissemination technique using different clustering approaches.

Moreover, we proposed a novel classification of clustering frequently utilized metrics from distinct schemes and according to their implementation in traffic management, health, and safety applications. The comparative analysis of clustering schemes includes challenges like beacon-based message passing, relative mobility, dynamic cluster formation, optimal resource utilization, dynamic scheduling, traffic density estimation, and existing routing schemes. Although the survey did not categorize the energy-efficient clustering schemes for VANETs, the accumulation and comparison of these routing schemes for the monitoring application has not been done before, to our knowledge. The synthesis will also allow various users in this domain to select and deploy one of the energy-efficient routing schemes based on its merits over the others in terms of optimal energy utilisation.