Analysis And Application For Client Side Data Of HTTP Video Streaming In Mobile Networks

With the rapid development of mobile communication technology and the widespread of mobile intelligent terminals, the amount of data created by mobile communication services is experiencing an explosive increase. Still, the types of mobile communication services develop rapidly from voice service only to diversified data services, such as mobile video, online games, webpage browsing, mobile social network, file upload and download, etc. At the same time, user’s expectation for network are not only the insurance of indicators of QoS(Quality of Service) any more. What users care about is service’s quality that they can feel directly, which is named as QoE(Quality of Experience). Among all of these mobile services, mobile video service occupies an important position and it takes more than half of the total amount of data created by mobile services. Since the high requirement for network bandwidth, it is of a great challenge for mobile networks to fulfill QoE of video service.Although there are many video delivery solutions these days, HTTP streaming, which is based on TCP protocol, has been widely used for multimedia delivery owing to the TCP’s reliable service and other advantages such as easy traversal through network address translation and firewalls. To enhance QoE, HTTP adaptive streaming (HAS), which aims at delivering the best possible quality video to users is introduced and has become a research hotpot.In this thesis, large number of data of HTTP streaming are collected on clients for transmission characteristics analysis. Still, improvements on video transmission control are proposed based on the transmission characteristics. The main work of this thesis is listed as follows.1. Real time data collection of HTTP streaming on clientA platform is built on mobile terminals for real time data collection of different services. By applying this platform, large number of interacting data of HTTP streaming are collected under different scenarios. And these original data become the reliable basis for further analysis of service characteristics, transmission characteristics and so on.2. Deeply data analysis of HTTP streamingAfter collecting sufficient data, data analysis is conducted. The work of data analysis can be divided as three aspects:analysis of data interaction, extraction of video parameters and key indicators analysis of end to end networks. After that, the focus is put on characteristics of segment throughput of HTTP streaming in mobile networks. Analysis results show that the variation of segment throughput could be modeled by Autoregressive Process (AR).3. Improvements on transmission control of HTTP streamingBy applying AR model, segment throughput prediction is introduced. Based on that, two improvements in control of data transmission of HTTP streaming are proposed. One improvement is about buffer control in client, and another one is about video quality adaption algorithm. The performances of the improvements are examined by simulation and experimental tests, and results show these improvements can provide better experience for users.