VANETs can help smart cities by improving vehicle mobility and implementing an efficient system for communicating and managing warning messages. For instance, efficient traffic alerts and up-to-date traffic incident information will reduce traffic congestion, improve road safety, prevent car accidents, and enhance city driving. Additionally, real-time traffic alerting will reduce travel distances, fuel consumption, and, as a result, emissions of CO2 [4]. Furthermore, due to the increasing need for communication, computation, and storage resources, emerging vehicular applications, and exponentially growing data, Vehicular Edge Computing (VEC) has great potential to improve traffic safety and travel comfort by bringing communication, computing, and caching resources closer to vehicular users. It could also be able to meet the growing demand for low latency and bandwidth in edge devices [5].
Since people are consuming more information with their mobile devices, vehicles are equipped with edge devices and RSUs technologies in the road transport network, and the popularity of new mobility services such as ridesharing and carsharing has increased communication between vehicles, people, and everything else [6,7]. Therefore, the information gathered by them can be used to evaluate and predict real-time traffic density and compute an accurate map of road traffic density, as well as assist VANET in improving transportation efficiency, and pedestrian comfort, and provide a QoS. Cloud, fog, and edge computing techniques enable the real-time transmission and processing of terabytes of data. The cloud node has a lot of memory and processing power, but the fog and edge nodes have limited capacity. Additionally, the physical distance between the cloud data center and the fog and edge nodes influences the data transfer rate, and if it is long, it increases latency and potential packet loss. Furthermore, one of the primary goals of VANET is to provide QoS to end users, while infrastructure deployment is the most significant challenge in the traffic improvement application of VANET.
city car driving download 64 bit
Therefore, the arguably most important task to enable VANET-based mobility in smart cities is to ensure a high QoS, which is influenced by the two main forms of communication occurring in a VANET context: First, vehicles transfer information with each other in a peer-to-peer, or V2V manner. Second, vehicles can tap into a flow of data through RSUs either through a direct connection with the RSU or with a relayed connection through a V2V network path [18]. In all communication between nodes in the network, transmission follows one simple rule: Two nodes can only exchange data if they are within broadcasting range of their wireless devices [19]. In most VANET applications, accordingly, nodes greedily forward transmissions by selecting a node in the target direction [20]. Single-hop transmission occurs when the broadcasting node sends information to neighboring nodes; multi-hop transmission requires nodes to re-broadcast information [21]. To ensure an uninterrupted flow of data in multi-hop transmission scenarios, nodes can store information and only forward them once a suitable transfer node is found [22]. Over the last decade, a plethora of routing protocols have been proposed by researchers and compared in regard to their performance [23,24,25,26]. In a 2014 meta-analysis, Dua et al. [18] cluster routing protocols into five predominant groups such as topology-based, geographic, hybrid, clustering, and data fusion. All of these protocols aim to create a network that can withstand the demanding nature of smart city connected mobility. Belamri et al. [27] provide a framework of parameters in regard to which a VANET routing protocol should be optimized: Most importantly, routing quality should be assessed concerning message delay, network node distances, link reliability, hop count, and mobility of nodes. In evaluating the QoS of a network with specific routing protocols, researchers should use network metrics such as end-to-end delay (EED), packet loss, throughput and bandwidth, and packet sending rate (PSR). Following this logic, a VANET routing protocol and its technology should be sufficient to enable whichever application needs to be run in the network. VANET applications for smart city mobility can generally be clustered into one of two applications: efficiency-oriented and safety-oriented optimization [28]. Efficiency-oriented applications are mainly concerned with the overall flow of traffic in a VANET environment. A VANET infrastructure can be used to host a variety of applications such as traffic congestion detection and mitigation [29,30,31], traffic forecasting [32,33], fuel-saving vehicle routing [34,35], or secondary efficiency enabled by internet access, for example, by providing internet during traffic jams [36]. All of these applications prove to be a use case for optimization techniques. Safety-oriented services in a VANET are concerned with vehicular security. Given VANET connectivity, these services can be used to prevent collisions [37,38,39], facilitate emergency service response [40,41,42], or support safe overtaking [43,44].
VEC is a promising technology that can be used to support ITS services, smart city applications, and urban computing. Figure 3 shows the problems and methods that are used in the literature reviewed in VEC.
Qiao et al. [62] proposed a new edge caching scheme that optimizes content placement and delivery in VEC and networks with limited storage capacity and bandwidth by taking into account time-varying content popularity, dynamic network topology, and vehicle driving paths. Edge caching was modeled as a double time-scale Markov decision process (DTS-MDP). The joint content placement and the delivery problem is NP-hard long-term mixed integer linear programming (MILP). As a result, the variable participation of vehicles increases the operational complexity of the edge caching system, making it difficult to find the best solution. Thus, they proposed a deep deterministic policy gradient (DDPG) learning algorithm based on a DRL-based cooperative caching scheme to provide low-complexity decision making and adaptive resource management, and they accelerated the learning speed and improved caching performance by using mini-batch gradient descent.
Zhao et al. [74] proposed a carsharing service in VANET based on a dual SGA in order to improve the quality and robustness of carsharing services while also reducing passenger wait times and avoiding traffic congestion. After a successful match, the vehicle will respond to carpool matching requests via relay vehicles. The DSGA procedure will first calculate the geographic matching based on the driver and passenger destination correlation when the relay vehicle receives the request. The vehicle will then conduct a VSGA identity check. If the match fails, the request is forwarded by the relay vehicle. In the final match step, each relay vehicle merges and collects its traffic data with the help of nearby vehicles via the beacon package. They then distribute a certain amount of tokens to each relay vehicle to control the peripheral congestion message. The authors also proposed a layered congestion monitoring method to collect congestion information and improve matching accuracy. In this process, the relay vehicle first distributes some tokens to its neighbors, and then the neighbor vehicle that received the tokens collects the atomic congestion message in its driving region, such as acceleration, speed, and brake frequency, and then performs fuzzy clustering on the message. The fuzzy clustering method can reduce information and computation redundancy by extracting key information.
Another challenge is prediction accuracy, where knowing about potential traffic problems can aid in congestion relief and road capacity expansion. Based on collected vehicular data and the Continuous Time Markov Chain (CTMC), El Joubari et al. [85] developed traffic behavior in multi-lane roads and near intersections. In order to analyze system performance, the queuing theory was used to describe urban traffic dynamics, and CTMC in continuous time was used to forecast long-run average quantities such as congestion rates and average waiting times. Long-term estimates of traffic distribution can be obtained using this method, which employs a numerical method for solving the stationary distribution. In order to validate their model, the results were compared to a queue-based model and realistic traces. The numerical results show that the model accurately reflects real-world urban traffic behavior when historical traffic data are used. 2ff7e9595c
Comments