Research And Design Of IPTV Video Quality Monitoring System

As the carrier of IPTV telecom operating network, IP network has some inherent defects: no guarantees for good service quality, less effective management means. With the ever-increasing users and applications, the network burdens would become heavier, its performance will decline and user experiences will be degraded. Ensuring good user experiences is one of the key issues to the rapid development and promotion of IPTV, so the video quality monitoring for IPTV becomes a research hotspot.This thesis studies and designs a video quality monitoring system for IPTV under the IPv6 environment. The designed prototype framework consists of video acquisition module, multiplexing module, IPv6 network, receiver module and video monitoring subsystem. The media server multiplexes several captured video programs into one transport stream and forwards it targeting a multicast address at a RTP session. The receiver clients receive the stream, then de-multiplex, select, decode and play the wanted program. Meanwhile, the Agent at the clients sends out RTCP feedback packets periodically, and the video quality-monitoring server receives them, analyzes the received feedback information, and then sends the analysis results to the media server. According to the corresponding strategy, the media server adjusts video program quality properly for the feedback.Video flow has requirements on high bandwidth, less packet loss ratio, less time delay and jitter, but the packet loss ratio, time delay and jitter cannot be foreseen in IP network whose bandwidth is variable time to time. This thesis introduces a mechanism for streaming media end-to-end transmission control to try to improve the QoS in video transmission. To improve the performance of AIMD, the traditional transmission control algorithm for real-time flow, that cannot smooth sending rate on sender side, this thesis proposes a modified AIMD algorithm LBR-AIMD. This algorithm places agent on every client side to send feedback control information periodically, introducing the exponential weighted moving average to smooth the packet loss ratio. It compares the packet loss ratio to two thresholds and classifies the network status into three categories: light-loaded, loaded and congested, then decides the whole network status according to the proportion of the receivers'amount in three statuses. When above network status changes, the media server would adjust the program code stream on the basis of the analyzed results, and make packet loss ratio remain within a normal range and improve the performance of bandwidth, delay and jitter on receiving sides. Meanwhile the RTCP control period is also adjusted.The prototype has tested and shows that the video play is clear, fluent and users have good experiences. The LBR-AIMD algorithm test results indicate that the probability of QoS oscillation becomes lower and the performance of LBR-AIMD is improved to a certain extent than traditional AIMD algorithm.This thesis has also discussed the RTCP hierarchical feedback framework, which can improve the media server's response time to the feedback information in large IPTV system. In the end, the summary to the full thesis and the further prospects for the systems development are also made.