Proactive Caching for Vehicular Ad hoc Networks Using The City Model

Abstract

Caching at Roadside Units (RSUs) is a promising technique to alleviate the load on network backhaul in Vehicular AdHoc Networks (VANETs). It also allows us to minimize communications latency between RSUs and connected vehicles, which are interested in massive multimedia contents. This work proposes novel proactive caching schemes at RSUs for the city mobility environment. We adopt the Manhattan city model, where streets are arranged in an organized manner and movable nodes are allowed to traverse along the grid of horizontal and vertical streets. Exploiting the information about vehicles demand and mobility patterns increases the caching capability and enables us to proactively cache desired files before its actual demand. The main objective of this work is to study how to minimize the communication latency for vehicles to receive their desired data items. We study non-clustered caching scheme where the optimization problem is formulated to find an optimal caching decision at each RSU. We extend our work by considering clustered caching scheme where RSUs in each cluster cooperate together to serve connected vehicles with minimum latency. Due to the complexity of these problems, we propose a sub-optimal caching policy for each scheme. We compare the performance of the optimal caching policy to that of the sub-optimal caching policy. Numerical results reveal that proactive caching has a great performance gain when compared to the reactive baseline scheme. Furthermore, results manifest that the clustered caching scheme has a significant performance over the non-clustered scheme.