Hierarchical Reinforcement Learning-Based Delay-Sensitive Video Delivery in Vehicular Networks

초록

This paper jointly optimizes node scheduling and the delivery of video chunks in delay-sensitive dynamic video streaming using a hierarchical reinforcement learning algorithm. Specifically, we presented an algorithm capable of adjusting node scheduling and the transmission of video chunks at two different slow and fast timescales, respectively. When nodes caching content are randomly distributed, mobile users dynamically select the node from which to receive video and control the number and quality of video chunks from the selected node, depending on the channel conditions with nearby nodes, the quality of cached content, and the user’s queue status.

키워드