ORTIA: An algorithm to improve quality of experience in HTTP adaptive bitrate streaming sessions

Abstract

Adaptive Bitrate (ABR) is used at large scale in online video streaming to improve viewer perception. The advanced online streaming process utilizes adaptive bitrate adaptation algorithms that works in video-players. To improve viewer’s quality of experience, ABR algorithm has to select high quality video as per the available network throughput and transmit with minimal stops and low-bitrate fluctuation. Although, current ABR algorithms suffers from stops, and low bitrate fluctuations because they don’t consider the influence of the underlying atypical value also known as outlier in statistics during throughput prediction. Here we propose a new ABR Outlier-Removal-based Throughput Improvement Algorithm (OTRIA), that provides a realistic forecast of throughput in changing network circumstances, thus minimizing stops and bitrate fluctuations. We used various statistical methods to detect outliers. We used the method of Inter-Quartile-Range to identify a range of throughput values that differs significantly from the other values in dataset and utilized this tool to maximize the precision for the predictions of throughput. Results from real-time experimentation have shown the elimination of stops during the implementation of our algorithm. In addition, a comparison is made with the latest generation ABR algorithm named DYNAMIC to exhibit that the ORTIA-algorithm is superior to the current ABR algorithm. Simply put, our algorithm delivers an excellent viewer’s Quality of Experience and aperture for adaptive transmission of real video. It is noteworthy that our algorithm beats the DYNAMIC, which is now officially the part of DASH-IF reference player and is used by video content supplier in production atmosphere.