Abstract:
Vehicular edge computing (VEC) brings computational resources at the edge of vehicular networks (VANETs). In VEC, the roadside unit (RSU) across the road segment acts as an edge server. The vehicle having less computational capability offloads high computation tasks to its nearby RSU for processing. There is a significant energy consumption occurs at the RSU in computing each high computation task. To minimize the energy consumption, a caching technique is used at RSUs. The greatest challenge of caching in VEC is the mobility of vehicles. In this poster, we propose a Mobility-Aware Caching technique (MobiCache) in VEC. MobiCache uses an actor-critic deep reinforcement learning framework to find the best routes for migrating the popular cache contents of RSUs according to the mobility pattern of vehicles. Simulation results show that our proposed caching strategy reduces the energy consumption by an average of 39.54% as compared to other existing caching techniques.
