Segment Prefetching at the Edge for Adaptive Video Streaming

Segment prefetching is a technique that transmits the next video segments in advance closer to the user to serve content with reduced latency. Due to its location and capabilities, an edge computing node is an ideal component for executing segment prefetching policies and storing/caching the prefetched segments. In this work, we study segment prefetching techniques deployed at the edge computing node for adaptive video streaming. We propose different types of segment prefetching policies and study their costs and benefits, including segment prefetching based on past segment requests, transrating, a Markov prediction model and machine learning. Besides, we analyze and discuss which segment prefetching policy is better under which circumstances and the influence of the ABR algorithm and the bitrate ladder on segment prefetching.

Automated summary

This study investigates the technique of segment prefetching at edge computing nodes for adaptive video streaming. Various segment prefetching policies are proposed and their costs and benefits are analyzed. Additionally, the influence of ABR algorithms and bitrate ladder on segment prefetching is examined.