I- Introduction

II- Proposed Method

III- Experimental Results

IV- Conclusion and Future Work

References

Abstract

Video streaming service, considering both live as well as video-on-demand contents, has completely changed the internet world. However, buffering remains the biggest concern, which severely degrades the quality of experience. In particular, the amount of time spent in video buffering phase has the worst impact on the user engagement. This buffering phase becomes more visible while streaming in fluctuating networks, which is a common scenario when user watches streamed video while travelling or during weather aberration. In this paper, we propose an intelligent network aware adaptive streaming method which estimates the past network trend, optimizes the video queue caching mechanism and enforces video quality in client device. By doing so, the algorithm is able to reduce buffering events by average 40% and quality switches by almost 45%, providing an almost seamless video streaming playback experience.

INTRODUCTION

As per research made with BBC iPlayer usage [1] with data taken for over nine months (1.9 billion sessions of 32M monthly users), it was observed that mobile handset users often split their content consumption across different sessions. Such sessions are either first starts on fixed-line broadband and continues while on the move (53%), or starts on a cellular connection and continues on a fixed-line connection (47%). In summary, media consumption trend clearly shows major viewership is seen during day-to-day commutating or travelling. Also, in Q1 of 2017, Mux commissioned an independent survey that asked 1,035 U.S. consumers about their viewing experience with online video [2]. As per the report shown in Fig. 1, re-buffering [5] i.e. stalling of streaming media during ongoing playback due to bad network, is the most important factor impacting the viewer’s QoE. The survey wanted to evaluate the effect the buffering events on length of user’s viewing session which is shown in Fig. 2.