Thesis Title:Research On MPEG-DASH Based Dynamic Adaptive Video Streaming Algorithm Over HTTP

With the rapid development of Internet and the prevalence of personal mobile internet equipment,internet video streaming service is becoming more and more popular.The most important issue in video streaming is that the network how to adapt the varied bandwidth in order to improve the Quality of Experience(QoE)of users.Since existing video streaming transmission standards cannot deal with dynamic bandwidth easily and are difficult in establishment and mintenance,dynamic adaptive streaming over HTTP(DASH)are researched extensively these years.Dynamic adaptive streaming over HTTP made by MPEG(MPEG-DASH)has emerged as one of the efficient technologies for adaptive video-streaming and has been used by many streaming media service providers(YouTube,Netflix,etc.).As the MPEG-DASH standard is based on the HTTP protocol,it can make use of the current network structure,without special equipment support and is firewall friendly.The standard is also robust to network fluctuations.The existing web server which supports the HTTP protocol can be easily upgraded to support the MPEG-DASH standard by software.Firstly,the development of the MPEG-DASH standard and the video coding logic of the MPEG-DASH standard are introduced in detail.Based on the MPEG-DASH standard,a dynamic adaptive streaming system that supports on-demand services was designed and implemented.The system is composed of a web server that supports the MPEG-DASH standard,a MPEG-DASH player(including a computer-end player and a web-end player),and a proxy module.Then,we study the dynamic adaptive video bit rate switching algorithm in the system.By summarizing the existing video bit rate switching algorithms,we found that the most of the existing research on MPEG-DASH mainly focuses on MPEG-DASH system with only one server and one user issue.However,in real applications,the most common issue is that multiple users competing together for the limited bandwidth of a single server.In order to improve the QoE fairness of the single-server to multi-users DASH system,a cooperative game based bandwidth allocation algorithm was designed by using the proxy module so as to ensure the fairness among users.In the proposed algorithm,,the bandwidth allocation problem is first modeled as a cooperative bargaining game by taking QoE i.e.,buffer lengths and received video qualities of users into account.The Nash Bargaining Solution(NBS)which is the solution of the bandwidth allocation problem is obtained by convex optimization Experimental results demonstrates that the buffer length and the received video bit rate of each user are more stable,the delay frequency and delay duration of each user are smaller and the overall QoE is higher when comparing with the existing methods.