H.264Video Transmission Quality Control Technology Research Under Multicast Network

With the rapid development of information technology, H.264video multicast has becomean important mode for exchanging information. Since the quality of video transmissiondetermines users viewing experience, video transmission quality control is always the key link topromote the competitiveness of video transmission products. Recently, video transmissionservices gradually showing the features of high-cluster and strong-sensitivity, under this newsituation，H.264-video transmission quality control in the multicast network still has manyproblems.Internet does not guarantee the quality of service, classical transmission quality controlscheme usually use Forward Error Correction (FEC) at server and RTCP feedback at client toagainst transmission errors, but since the large-scale multicast service for H.264-video is morecomplex to network and more sensitive to data, such scheme often has poor performance underthe real-time, reliable, and subjective-feeling considered transmission requirements.This dissertation explored research around the key technologies of transmission qualitycontrol, focusing on the optimization of multicast transmission architecture and the improvementof forward error correction to make them more suitable for H.264compressed video multicast.At first, we proposed a RTCP-net based transmission architecture. By using the aggregatenode,"network-neighbor" detection mechanism and OSPF-like algorithm the transmissionsystem was optimized to net-topology, so that it can meet the rapid response demand oflarge-scale multicast; based on the heart-beat and correction mechanism control scheme couldreduce the vulnerability of key nodes, thus enhanced the reliability of the system; with thesecondary forward error correction mechanism transmission system could provide specificprotection for subnet so as to solve its heterogeneous problem.Secondly, we proposed an enhanced forward error correction (S-FEC) method based onH.264feature and user experience. On the one hand, using the characteristics of H.264compression data to extract decode-importance and content-complexity as reference informationfor S-FEC, on the other hand, using the IQX hypothesis and RTCP feedback information toevaluate the user experience as reference information for S-FEC. By combining them to decidethe S-FEC parameters, the result of H.264video transmission protection could be more accurateand consistent with the human visual experience.Finally, we designed a complete video transmission quality control system based onRTCP-net and S-FEC, and implemented it under Linux environment.We verified above system under the video codec chip MG35000, experiment results showed that our system could effectively respond to many H.264multicast problems such as large-scale,sensitive data, network deterioration and lack of consideration of user experience; it couldenhance the quality of the video transmission and satisfy the requirement of real-time broadcastbusiness at the same time.