Improved bitstream extraction in dynamic adaptive streaming in http using hybrid models

Abstract

The adaptive bitrate streaming works on the principle of rapidly changing bitrate based on available resources, which is usually a bandwidth, resource limit and user query. The selection of bitrates are controlled usually by client and not by server. The adaptive HTTP video streaming is similar to progressive download streaming. In adaptive HTTP video streaming, the multimedia segments are split into same length segments. On other hand, the segments are split into video GoP that initiates with key frame. The initiation of key frame in video segments represents the past or future dependencies. Finally, the entire video segments are encoded and it is then hosted over a HTTP streaming server. Recently, the online video streaming traffic is growing massively. It is estimated that traffic in online video streaming services accounts for more than 80% of traffic in internet by 2020. Hence, the design of next generation video streaming services is driven by internet traffic. Among various HTTP video streaming methods, adaptive HTTP video streaming is considered as a dominant technique due to its flexible rate adaptation and better network utilization in delivering the video contents across its users. On contrary with conventional UDP or RTP streaming, HTTP dynamic adaptive streaming with rate adaptation mechanism based on its video quality and bitrate of present network. Further, the adaptation includes buffered video amount, available bandwidth and screen resolutions. The main aim of adaptation algorithm is to increase the maximum utilization of network while maintaining high quality video streaming. The algorithm intends to maintain the trade-off between the network utilization and video quality, which are not easy to achieve newline