无线VR多视角3D视频中基于深度强化学习的主动3C资源分配  

Proactive 3C resource allocation in wireless VR multi-view 3D video based on deep reinforcement learning

作　　者：陈雷[1] CHEN Lei(School of Public Security Information Technology and Intelligence,Criminal Investigation Police University of China,Shenyang 110854,China)

机构地区：[1]中国刑事警察学院公安信息技术与情报学院,辽宁沈阳110854

出　　处：《武汉大学学报(工学版)》2025年第2期334-342,共9页Engineering Journal of Wuhan University

基　　金：辽宁省自然科学基金项目(编号:20180550046);辽宁省教育厅科学研究项目(编号:ZGXJ2020005);中国刑事警察学院重大培育项目(编号:D2023002);中国刑事警察学院校级项目(编号:D2022045);沈阳市社会治理科技专项项目(编号:22-322-3-35)。

摘　　要：在基于移动边缘计算(mobile edge computing,MEC)的第5代移动通信网络(5th generation mobile communication network,5G)中,针对无线虚拟现实(virtual reality,VR)用户在小小区间频繁切换而降低业务体验质量(quality of experience,QoE)的问题。考虑建立多视角3维(3-dimensional,3D)视频的主动缓存、计算和通信(caching,computing and communication,3C)资源分配数学模型,并采用深度强化学习(deep reinforcement learning,DRL)算法进行求解。将多视角3D视频的主动资源分配系统建模为联合视角选择和小基站3C资源分配的马尔科夫决策过程(Markov decision process,MDP),提出了一种基于深度确定性策略梯度(deep deterministic policy gradient,DDPG)算法来寻找有效解。仿真结果表明,与另外2种算法相比,所提算法可以为无线VR用户在小小区间移动时提供更好的业务体验。In view of the problem that mobile virtual reality(VR)users frequently switch in a small range and reduce the quality of experience(QoE)in the fifth generation(5G)mobile communication networks based on mobile edge computing(MEC),a resource allocation mathematical model of multi-perspective 3-dimensional(3D)video proactive caching,computing and communication(3C)is established,and the deep reinforcement learning(DRL)algorithm is used to solve the model.First,the multi-view 3D video proactive resource allocation system is modeled as a Markov decision process(MDP)of jointing view selection and small base station(BS)3C resource allocation.Second,an algorithm based on the deep deterministic policy gradient(DDPG)is proposed to find the effective solution.Finally,the simulation results demonstrate that compared with other 2 algorithms the proposed solution can provide better business experience for wireless VR users when moving in a small range.

关 键 词：无线虚拟现实 多视角3D视频 3C资源分配 深度强化学习 

分 类 号：TN929.5[电子电信—通信与信息系统]