Alert Message Dissemination using Graph-based Markov Decision Process Model in VANETs

Vehicular Ad-hoc Networks (VANETs) have many promising applications such as improving vehicle driver's safety, and so, decreasing car deaths. In VANETs, the main communication way is broadcast. Such Broadcast needs to be done efficiently especially in congested areas to avoid the broadcast storm problem. A crashed vehicle disseminates a message about its incident to all nodes in the network. The vehicle nodes of the network have to disseminate in their turn the alert message carefully to maximize both reachability and saved rebroadcast (by reducing congestion) while minimizing the delay. This concern is demonstrated and modeled by decision hypothesis. In this paper, at first, we are keen on the Markov decision process (MDP), which is utilized to model and tackle such successive decision issues. We aim to optimize a utility for each vehicle relying upon an irregular domain and choices made by it. As the MDP formalism achieves its cutoff points when it is important to consider the cooperation between the several vehicles. We will begin utilizing the Graph-based MDP where the state and activity spaces are factorizable by factors. We then demonstrate that the transition functions and rewards are deteriorated into nearby functions, and the reliance relations between the vehicles are spoken to by a graph. To Figure the optimal strategy, we use Mean Field Approximation (MFA) for tackling GMDP issue. © 2020 IEEE.