Key Technologies Of 3G Video Services Data Recovery

Along with the development of wireless communication technology, the video service has become the core business of the mobile communication system. Compared with the traditional voice service, the video service has a larger data amount, and is much more sensitive to the time delay and error coding. Both domestic and overseas research institutions have started researches aimed at improving the video service's receiving performance,. The purpose of the research project is to improve the video service's receiving performance. Based on a deep analysis of current wireless video processing technology, the paper researches on corrupted signaling fixing, corrupted data frame extraction, video graphics error detection and error concealment.(1) With the study of the H.245 control protocol and its process of signaling, the mechanism of key parameters'configuration is analyzed, and an H.245 control signaling fixing algorithm with prior knowledge is given. According to the control channel protocol stack defined in the H.324 protocol, the fixing algorithm is introduced into the H.245 control signaling parsing solution. The experiment result indicates that this algorithm has effectively increased the success ratio of fixing corrupted control signaling.(2) The multiplexing protocol of H.223, especially the frame assembling mechanism and frame structure of composite layer and adaptive layer, is studied, An adaptive threshold selecting algorithm based on BP neural network is given in order to solve problems of recognizing frame identifier and setting threshold of MUX-PDU, which can adjust MUX-PDU frame identifier selecting threshold adaptively. Simulations show that it can improve the MUX-PDU extraction performance compared with a fixed threshold.(3) A posterior information based verification algorithm is given combined with the characteristics of MUX-PDU frame structure in order to solve the MUX-PDU extraction problem. The algorithm can identify the false alarm and omission of MUX-PDU identifier. Combined the algorithm with BP neural network based adaptive threshold selecting algorithm, an joint extraction algorithm based on the BP Neural Network and posterior information is proposed. Simulations show that the algorithm can decrease the frame losing ratio of MUX-PDU, and can improve the MUX-PDU extracting performance effectively.(4) Due to the problem of video image error detection, a priori constraint forward space-time video error detection algorithm is proposed with the consideration of the prior knowledge and the video correlation in the time domain. It fuses the information of video sequence's extensive characteristics of error detection. Simulations show that the algorithm has a much better error location precision than traditional ones. (5) With the comparison of the spatial domain error concealment algorithm and time domain concealment algorithm, a reconstruction algorithm of average motion vector based on bordary-matching is given in order to solve the video image error- concealment problem. The experiment result indicates that this algorithm further improve the error- concealment effect through improving the border smooth level between the recovery block and the neighborhood block.