PCCP: Proactive Video Chunks Caching and Processing in edge networks

Abstract

Mobile Edge Computing (MEC) networks have been proposed to extend the cloud services and bring the cloud computing capabilities near the end-users at the Mobile Base Stations (MBS). To improve the efficiency of pushing the cloud features to the edge, different MEC servers assist each others to effectively select videos to cache and transcode. In this work, we adopt a joint caching and processing model for Video On Demand (VOD) in MEC networks. Our goal is to proactively cache only the chunks of videos to be watched and instead of caching the whole video content in one edge server (as performed in most of the previous works), neighboring MBSs will collaborate to store different video chunks to optimize the storage resources usage. Then, by coping with the Adaptive BitRate streaming technology (ABR), different representations of each chunk can be generated on the fly and cached in multiple MEC servers. To maximize the caching efficiency, we study the videos viewing pattern and design a Proactive caching Policy (PcP) and a Caching replacement Policy (CrP) to cache only highest probability video chunks. Servers performing caching and transcoding tasks should be thoroughly selected to optimize the storage and computing resources usage. Hence, we formulate this collaborative problem as a NP-hard Integer Linear Program (ILP). In addition to the CrP and PcP policies, we also propose a sub-optimal relaxation and an online heuristic, which are adequate for real-time chunks fetching. The simulation results prove that our model and policies perform more than 20% better than other edge caching approaches in terms of cost, average delay and cache hit ratio for different network configurations. 2019