Transmission Rate Control Techniques For Scalable Video Coding

Along with the fast development of network and video coding techniques, the network-based video service has been widely applied. There are mainly three characteristics in the current video application environment:heterogeneous networks, various receiving terminals and complexity of users’ demands. These features lead to difficulties as well as challenges for network-based video service. With the increasement of user requirements, people are less satisfied with the current video transmision methods and seek for a more convenient and effective video communication solution. The video transmission process will greatly affect the quality of the reconstructed video. In practical environment, the transmission channel is bandwidth limited and time varying and the rate constraint is an important factor affecting the video transmission. Video transmission rate control is an effective way to solve the problem of potential degrading of the video quality, both objectively and subjectively, at receiver side due to the imperfect transmission channel. Therefore, the video transmission rate control has theoretical significance and important practical usage.The dissertation studies the transmission rate control under the basis of scalable video coding (SVC). The main contents and novelties are as follows:1) This dissertation investigates the transmission characteristics with respect to the packet rate of the packet switching network.This dissertation analyzes and models the transmission process based on the packet-switched network, simplifies the model under the characteristics of video transmission and simulates the video trasnmission process. The results show that when the packet loss rate is low and the length of burst is as same as the length of buffer, the packet loss rate is directly related to the packet rate of the stream. The streams with the same packet rate will bring the same network congestion. The result gives a better performance metrics for video transmission under low packet rate and provides theory basis and standard for the performance comparison method.2) This dissertation proposes the transmission rate control methods for different scenarios.Based on the research results of the packet-switched network, this dissertation proposes the congestion aware transmission rate control method for video transmission under the high congestion wire network. The method is based on the medimum grain scalability of SVC, and provides better reconstruction video quality while not aggravating the network congestion to bring packet loss. The results show that the method can get an average0.4dB gain.This dissertation proposes the adaptive packet encapsulation method for the video transmission from wire to wireless. The method considers the relationship among the wireless bit error rate, packet length and the importance of the different parts of the video stream. The method can utilize the bandwidth of the wireless channel and gives a better reconstruction video quality. The results show that the method can get an average0.72dB gain.This dissertation focuses on the situation of serious congestion network with packet loss. This dissertation analyzes the previous FEC methods and proposes the internal packet FEC(IP-FEC). The IP-FEC provides a certain degree of error protection capability without aggravating the network congestion. The IP-FEC can be easily combined with the previous FEC methods. The results show that the IP-FEC can get an average4.7dB gain compared with no-FEC under3%packet loss rate and an average4.9dB gain under5%packet loss rate,1.1dB and1.6dB respectively compared with traditional FEC with10%parity packets. The reconstruction video quality is better than the previous FEC method with at least10%parity.3) This dissertation proposes the novel fast stream switching method.This dissertation analyzes the problems of video communication under current network conditions based on the latest DASH method. This dissertation focuses on the the problems of small buffer and buffer replacement algorithm mismatch, proposes the adaptive predict update algorithm. The proposed algorithm gives the approximate optimal replacement results from estimating the transition probabilities of the program content and channel bandwidth. The results show that the proposed algorithm can achieve about12.6%hit ratio gain under the high switching scenario and comparable performance under other scenes.