Research and design of video adaption algorithm based on HTTP Live Streaming

• Main Authors: Sheng-Kai Hsieh, 謝勝凱
• Other Authors: Han-Chieh Chao
• Format: Others
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/6h38kf

碩士 === 國立東華大學 === 電機工程學系 === 100 === With the advance of wireless communication technology and the high market share of smartphone, the ability for enabling the smartphone user to access multimedia data on the network through wireless communication network is now available. The features for watching movies on the internet, listening music on the internet, even having a video conferencing through wireless communication network have got the major focus among mobile phone providers. However, the available bandwidth of network user in wireless communication network may change due to some factors, for example, the amount of wireless users in the same wireless base station, the signal strength of the wireless base station, and the moving velocity of wireless users. In the environment with dynamic available bandwidth on wireless communication network, the wireless user might need to wait for video or music content downloading when watching video or listening music on the internet through wireless communication network due to the lack of wireless bandwidth. For resolving the situation of waiting for the video or music content downloading during the multimedia information on the internet is playing. Apple Inc. have implemented an media streaming communications protocol based on HTTP Progressive Downlaoding, called HTTP Live Streaming(HLS), as part of their QuickTime X and iPhone software systems. It works by breaking the overall stream into a sequence of small HTTP-based files. As the stream is played, the client may select from a number of different alternate streams encoded at a variety of data rates. It allows the streaming session to adapt to the available data rate to avoid waiting for download when the stream is played. Apple has documented HTTP Live Streaming as an Internet-Draft and submitted it to the IETF as an Informational Standard. However, it is still not submitted as standardization by IETF. Moreover, in the specification of HTTP Live Streaming draft document, there is no any definition for the method of detecting the current bandwidth on video client, the timing of switching the quality of video and the quality should be switched to. Therefore, in this paper, we will propose 5 kinds of video adaption algorithms, including Greedy Algorithm, Greedy Algorithm2, Quality Switching Algorithm, Quality Switching Algorithm 2 and Quality Switching Algorithm 3. These methods will allow the video client to detect available bandwidth and switch the video quality according to the available bandwidth during the stream is downloading. Moreover, we use mathematical model to analyze these video adaption algorithms, and implement an analysis program according to these video adaption algorithms by C language. We will compare the difference of the analysis result for each video adaption algorithm in the metrics of Quality of each segment, Complete percentage of video, Wait-to-Play time, Average playback quality and drop size. Furthermore, we will analyze the difference when using different detection period in each video adaption algorithm. In our analysis result of video adaption algorithms, we can observe that when the bandwidth changed with an unstable model, the HTTP Live Streaming with Quality Switching Algorithm 3 will try to adjust the threshold value of resilience time to reduce the quality of the video to avoid waiting again in the remaining video content, increase the average playback quality, and reduce the dropped video data at the same time. In our analysis result of the different bandwidth detect duration, we can observe that the length of resilience time will result in different effects of different bandwidth detect duration.