A fair server adaptation algorithm for HTTP adaptive streaming using video complexity

Abstract

The increasing popularity of online video content and adaptive video streaming services, especially those based on HTTP Adaptive Streaming (HAS) highlights the need for streaming optimization solutions. From a server perspective, the main drawback of HAS is that the user selects the quality of the next video segment without taking the server constraints into account. These constraints include the number of users simultaneously being served and the server’s congestion. Here, we present the Fair Server Adaptation (FSA) algorithm, which is designed to maximize user Quality of Experience (QoE) by tackling the server’s bottleneck problem. The algorithm provides the quality representation that is closest to the user’s request, subject to the server’s constraints. Simulation results show that compared to standard Dynamic Adaptive Streaming over HTTP (DASH) server, FSA increased the number of served users and decreased both the number of rebuffering events and the average rebuffering event duration. Furthermore, the average number of unserved users decreased to almost zero and Jain’s fairness index rose. It is clear that these changes increase users’ QoE.