基于深度强化学习的分布式视频缓存策略研究

Research on distributed video cache strategy based on deep reinforcement learning

为了适应大规模移动缓存网络动态且多样化的请求模式,基于深度强化学习模型提出了一种新颖的分布式视频缓存策略.该策略利用多智能体之间的协作,以自适应的方式学习最佳的缓存决策,从而减少内容访问时延和流量成本.使用真实的视频请求数据集,将该策略与主流的缓存策略进行对比评估.实验结果表明,该策略能够在降低时延和流量成本的情况下,实现较高的缓存命中率.

In order to adapt to the dynamic and diversified request patterns of large-scale mobile caching networks,this paper proposes a novel distributed video caching strategy based on the deep reinforcement learning model.This strategy uses the collaboration between multiple agents to learn the best caching decision in an adaptive manner,thereby reducing content access latency and traffic costs.This article uses real video request data sets to compare and evaluate this strategy with mainstream caching strategies.Experimental results show that this strategy can achieve a higher cache hit rate while reducing latency and traffic costs.