Deep Reinforcement Learning Based Computing Offloading for MEC-assisted Heterogeneous Vehicular Networks

Abstract

Mobile edge computing (MEC) is a promising technology that sinks cloud computing down to the edge of the cellular network, so as to reduce latency and network congestion. In this paper, we investigate computing offloading problem for MEC-assisted heterogeneous vehicular networks where both vehicles and road unit sides (RSUs) can provide computation services. With the goal of maximizing the total computation rate, we design two computing offloading schemes, namely serial and parallel offloading schemes, and formulate the optimization problem as a Markov decision process (MDP) problem. To solve it, we propose a deep reinforcement learning (DRL) based vehicular edge computing (VEC) offloading solution. Simulation results show that the proposed parallel offloading scheme outperforms other solutions in terms of the total computation rate at the cost of high computation delay.