文档介绍：: .
系统仿真学报和不完全信息博弈下的策略。完全信息博弈下，证明该博弈存在唯一纳什均衡解；不完全
信息博弈下，将环境建模为部分可观测的马尔可夫决策过程(Partially Observable Markov Decision
Process, POMDP)，并提出一种基于二阶段深度强化学习（Two-Stage deep reinforcement learning,
TSDRL）的最优卸载策略。仿真实验表明，该算法相较于 D-DRL 算法能减少 ％的时延及
％的能耗，有效提高用户 QoE(Quality of Experience)。
关键词：5G 混合专网；计算卸载；Stackelberg 博弈；Nash 均衡；POMDP
中图分类号： 文献标志码：A
DOI: .-1118
Computation offloading strategy based on Stackelberg game and DRL
Zhou Xianwei, Gong Qixu, Yu Songsen
(School of Software, South China Normal University, Foshan Guangdong 528225, China)
Abstract: In order to achieve optimal computation offloading strategy for different types of MEC users in
5G hybrid private network, Stackelberg game is utilized to model for two kinds of users competing for
MEC server resources. Subsequently the strategies of complete information game and partially incomplete
information game are discussed respectively. It is proved that there is a unique Nash equilibrium solution
in the complete information scenario. Whereas in the incomplete information scenario, the environment is
model as POMDP, and a two-stage deep reinforcement learning(TSDRL) is proposed to obtain optimal
computation offloading strategy. Simulation results show the effectiveness of the proposed algorithm with
a total reduction of % time delay and % energy consumption compared with the D-DRL
algorithm, and thus better user QoE(Quality of Experience) is achieved effectively.