(A) study of two phase cache management scheme with multiple prefetching-based chunk popularity for mobile video streaming

Abstract

The demand for video streaming services by mobile devices worldwide is increasing rapidly. The additional delay caused by the core network and the radio access network of the wireless network between the cache server of the conventional content delivery network and the mobile user degrades the Quality-of-Experience (QoE) in the streaming of the high-quality video. Generally, in the mobile environment, video transmission is a heavy load on high-speed wireless backhaul such as LTE and LTE-A, and a solution for improving network performance is required. As a solution to this problem, in a mobile edge computing (MEC) environment, a cache server is hosted in a base station very close to a mobile user, thereby guaranteeing video streaming QoE to a mobile user and reducing network load on the core network in wireless due to video transmission. However, the cache server in the MEC environment has a limited storage space compared to the servers in the existing content delivery network, and accordingly, a new caching scheme suitable for a low capacity storage space is required to improve the efficiency of the cache server in MEC. In general, content providers such as Yotube and Netflix adopt dash streaming for video streaming. It provides streaming by dividing a video file into chunks of files which can be played for a certain period of time. In particular, with live DASH streaming, chunks requested by users change over time, therefore, cache server efficiency can be improved in MEC environment through chunked caching. This thesis proposes a new cache management scheme for live dash streaming in MEC environments. The proposal scheme is a two-phase cache management scheme that estimates the popularity of live media based on the number of requests similar with existing scheme in phase 1 and estimates the request probability of chunks in each live channel formatted DASH over time in phase 2. Based on the estimated chunk request probability, proposal scheme perform selective caching for chunks which have high popularity or high request probability and guarantee high cache performance and QoE of users who request live dash stream. As an evaluation of the proposal scheme, we configure the environment in which users receive live DASH stream from a single MEC cache server and compare with existing cache scheme. Compared with the existing cache scheme, the proposal scheme can improve the cache performance and QoE of users watching live DASH streaming by selective chunk caching and prefetching.