Qoe Enhancement in HTTP Based Adaptive Video Streaming

Abstract

In the last decade, there has been an exponential increase in the video traffic over the internet. Social Medias are becoming one of the main source of live and on-demand video streaming content. With ever-increasing popularity of online different video streaming services on heterogeneous platforms, new research challenges are arising day by day. Few of the main challenges that online video streaming services face are high latency of the video, instability of the video, unfairness among the clients, inefficiency of the algorithm to adapt to the changes in the network and the start-up delay of the video. Most of the existing algorithms fail to maintain a balance between stability and efficiency of the algorithm in unstable network conditions. We have proposed SHANZ rate adaptation algorithm for which address these challenges. We have developed two versions of the algorithm. SHANZ-I algorithm works on HTTP1.1 protocol. It is a dynamic rate adaptation algorithm with feedback control mechanism and adaptive step up function, which acts as an explicit knob to maintain a balance between stability and efficiency of the algorithm, even in drastic network conditions. Moreover, it introduces randomized download delay for the clients to overcome bandwidth overestimation problem occurred in multiple clients. The second version we have proposed is SHANZ-II rate adaptation algorithm, which is based on HTTP/2 protocol. It utilizes HTTP/2 features like server-push, streams multiplexing and header compression for the enhancement of quality of experience. It minimizes the latency and start-up delay of the video, which are the main challenges for live video streaming. The algorithm defines an intelligent control mechanism for server-push, which maximizes the utility function. We have simulated our algorithm using ns-3 and compared our results with FESTIVE, PANDA and AAASH algorithms by using multiple test cases. The results demonstrate that our proposed algorithm outperforms other algorithms by addressing the key issues and by achieving higher Quality of Experience.