DeePref: Deep Reinforcement Learning For Video Prefetching In Content Delivery Networks

Content Delivery Networks carry the majority of Internet traffic, and the increasing demand for video content as a major IP traffic across the Internet highlights the importance of caching and prefetching optimization algorithms. Prefetching aims to make data available in the cache before the requester places its request to reduce access time and improve the Quality of Experience on the user side. Traditional prefetching techniques are well adapted to a particular access pattern, but fail to adapt to sudden variations or randomization in workloads. This paper explores the use of deep reinforcement learning to tackle the changes in users' access patterns and automatically adapt over time to predict future requests. We propose DeePref, an auto-aggressive video prefetcher that is sensitive to different storage capacities at edge networks. DeePref is agnostic to hardware design, operating systems, and applications in which it utilizes only the video ID to make prefetching decisions online. DeePref outperforms baseline approaches that use video content popularity as a building block to statically or dynamically make prefetching decisions. Our results show that DeePref, using a real-world dataset, achieves 17% increase in terms of prefetching accuracy and 28% increase in prefetching coverage. We also study the possibility of transfer learning of statistical models from one edge network into another, where unseen user requests from unknown distribution are observed. Our results indicate that DeePref effectively adapts to distribution shifts where the increase in prefetching accuracy and prefetching coverage are [30%, 10%], respectively.