QoE-Aware Joint Segment-based Video Caching and User Association Optimization

Abstract

Mobile edge caching has been proven effective in improving the network utility and users’ quality of experience (QoE). Due to the exponential growth of mobile video traffic and the limited cache capacity of the base station (BS), how to pick appropriate video contents to cache is an important challenge. Different from existing users’ cache hit ratio maximization, this paper aims at improving both the users’ QoE and network utility for the cache-enabled network. To achieve this goal, we investigate an optimization problem of QoE-aware joint segment-based video caching and user association (JSVCUA). Then we propose an iterative framework to decouple the original non-convex problem into the cache placement problem and user association problem. For the cache placement problem, we propose utilizing the neural collaborative filtering (NCF) scheme to accurately predict the higher-level and non-linear preferences between users and video contents. Besides, the segment-based caching strategy is presented to further improve the efficiency of caching. Then the caching decision is made by maximizing the QoE weighted content transmission rate. For the user association problem, convex optimization is applied, and the performance gain of caching is squeezed. Simulation results verify that the proposed algorithm can significantly improve both the network utility and QoE on a real dataset. Specifically, the cache hit ratio increase by 10% on average.