PASA: Towards fair rate adaptation for http-based video streaming through server assistance

HTTP-based adaptive streaming (HAS) has become the de facto standard for video streaming over the Internet. However, when multiple clients compete for available bandwidth at a bottleneck link, traditional rate adaptive algorithms lead to unfair bandwidth allocation. This paper proposes Probe and Adapt with Server Assistance (PASA) algorithm based on PANDA. PASA uses a decentralized fairness controller that works with the assistance of server to improve fairness among clients. The main advantage of PASA is that it is still a client-driven, pull-based algorithm and it does not bring challenges as in centralized management. Besides, we use a bitrate switching logic to make a balance between response agility to bandwidth fluctuation and bitrate stability. We use ns-3 simulator to evaluate the performance of PASA. Compared with PANDA and state-of-the-art method, the results show that PASA is able to reduce the unfairness of video bitrate by over 40% and also has excellent stability and convergence rate. © 2020 Computer Society of the Republic of China. All rights reserved.