Hybrid Task Offloading and Resource Optimization in Vehicular Edge Computing Networks

Abstract

In this paper, we propose a hybrid frequency division multiple access (FDMA)-non-orthogonal-multiple-access (NOMA)-assisted computation offloading framework for vehicular networks, where each vehicle offloads its computational task to vehicular edge computing (VEC) server nearby a small cell base station (SBS) via NOMA and VEC server nearby a macro base-station (MBS) via FDMA for execution. By considering the partial task offloading model, we formulate a joint task offloading, bandwidth and computational frequency allocation optimization problem to minimize the total cost in terms of overall delay and energy consumption. To address the non-convexity of the optimization problem, we transform the original problem into two sub-problems. Then, the bandwidth and computational frequency allocation subproblem can be solved with convex optimization method, while the task offloading subproblem can be solved by deep deterministic policy gradient (DDPG) method. The simulation results demonstrate that our proposed scheme outperforms other benchmark schemes.