Task prioritization offloading scheme for vehicular edge computing networks: a game-theoretic DRL approach

With the development of Vehicular Edge Computing (VEC) computing architectures, in the study of task offloading problem, based on the differences in task delay sensitivity and the dynamic characteristics of environmental information, this paper designs a distributed DRL framework based on non-cooperative game, and introduces a memory mechanism(RNN) and a shared experience mechanism (Shared Experience Actor-Critic) in the MADDPG algorithm, which improves the learning efficiency of the algorithm as well as the convergence speed by capturing and sharing the timeseries data across time steps.