Cooperative Computation Offloading and Resource Management for Vehicle Platoon: A Deep Reinforcement Learning Approach

On the Internet of Vehicles, the computation capability of individual vehicle is restricted to finish upcoming vehicular computational tasks. Fortunately, vehicular edge computing (VEC) has been proposed to meet the pressing resource demands of vehicular services. In this paper, a cooperative computation offloading and resource management approach is proposed, in which the generated tasks can be offloaded to vehicles and roadside units that are relatively close to the computing vehicle through vehicle-to-vehicle communication (V2V) and vehicle-to-roadside communication (V2R). Taking into account the mobility of vehicles and the time-variety of computation resources for available nodes in the vehicular network, the system model, task delay, task energy consumption and optimization objectives are analysed respectively. Moreover, this optimization problem is solved by a deep reinforcement learning algorithm (DRL), namely Asynchronous Advantage Actor-Critic (A3C). Validated through simulations, the proposed approach shows superiority to other offloading strategies in terms of the calculated reward, task delay, and energy consumption.