Adaptive Live Streaming for Multi-user Access with Fairness Guarantee

Chang Liu,Jiaoyang Yin,Yiling Xu

DOI: https://doi.org/10.1109/bmsb58369.2023.10211604

2023-01-01

Abstract:Adaptive bit rate (ABR) algorithms are widely adopted in video streaming applications to optimize user quality of experience (QoE). However, most ABR approaches place emphasis on QoE improvement for the single user, while lacking consideration for multi-user access. In the real environment, users may access network in the meantime and multiple clients under the same bottleneck bandwidth will compete for the limited network resource. It is challenging for current ABR strategies to guarantee satisfactory QoE for each user under more constraints, e.g. latency requirement, resource competition, and limited server overhead. This paper proposes an ABR system for multi-user live streaming. In this system, clients under the same bottleneck bandwidth are clustered into several groups according to specific network conditions and user preferences. Then ABR server conducts bitrate decisions for each user group to achieve overall QoE optimization. Extensive experiments show that the proposed ABR system can guarantee fairness for multiple users with satisfactory overall QoE and acceptable resource consumption.