Server-side Adaptive Bitrate Method Based on Perceptual Video Quality in Networks with Controllable Bandwidth

Abstract

In HTTP adaptive bitrate (ABR) streaming, the ABR algorithm is an important factor determining the quality of experience. Traditional ABR methods control the bitrate of individual clients autonomously so that the quality variation occurs even when the bandwidth of a delivery network is managed. Here, we propose a server-side ABR method achieving quality fairness among clients and suppressing a quality variation over time in networks where external applications dynamically control the quality of service. The method controls both the bitrate and delivery bandwidth based on a perceptual quality score. We evaluated the basic performance of the proposed method on an experimental network whose bandwidth could be controlled by the API. We confirmed that the proposed method could maintain the target quality within the upper-limit bandwidth and suppress the quality variation among clients and over time.