Long-Term Adaptive Bitrate Control Mechanism

Abstract

Adaptive bitrate (ABR) video streaming is an important application on the Internet. To ensure that users enjoy high-quality services, ABR control mechanisms need to be designed that select chunks wisely on the basis of the available network throughput. To address the chunk selection problem, this paper describes an adaptive bitrate control mechanism that leverages long-term throughput information in the chunk selection process. While previous work has considered how quality should be requested on a per-chunk basis, the proposed method increases the timeframe of the analysis and allows higher quality of experience (QoE) to be reached. This is done by appropriately selecting a sequence of consecutive chunks' quality values instead of a single chunk's value. Simulation results are reported on a large variety of real-world network conditions and various throughput prediction algorithms and show the benefit of the proposed method over conventional ABR control mechanisms.