Rate adaptation with Bayesian attractor model for MPEG-DASH
Masayoshi Iwamoto, Tatsuya Otoshi, Daichi Kominami, Masayuki Murata · 2019
Recently, over-the-top video service providers focus on the quality of experience (QoE) as an important factor when they provide video content. Today, most video streaming service providers, such as Youtube and Netflix, provide video content to users with adaptive bitrate (ABR) control techniques for increasing the user QoE. To maximize the QoE under a fluctuating network condition, in this paper, we propose an ABR algorithm using the Bayesian attractor model, which models cognition and decision making of the human brain, as the name suggests, according to the Bayesian inference. Simulation results show that our proposed method achieves a higher video bitrate with less video quality switching to improve the user QoE compared to the methods even in the situation where network available bandwidth greatly fluctuates.