Vehicular Ad Hoc network mobility models applied for reinforcement learning routing algorithm

Vehicular ad-hoc networks (VANET) are specialized applications of mobile ad-hoc network. To analyze the complex and dynamic topologies of these scalable applications a simulation based analysis plays a vital role in realizing the effective deployment of vehicles in realistic scenarios and its corresponding effect on routing protocols. Mobility models mimic the movement of vehicles and we consider Manhattan, City section and INVENT mobility models for our analysis. We measure the performances of these mobility models with suitable mobility metrics and try to correlate its corresponding impact on the performances of routing protocols. Routing protocols play a central role in the design of these types of networks. To meet the challenging requirements of the vehicular networks we analyze the suitability of a reinforcement learning based routing algorithm. We compare the performance of a reinforcement learning algorithm with AODV which is considered as one of the robust routing protocols under varying traffic and load conditions.