Self-adapt Source Coding And Transmission System For Wireless Video

The development of mobile media service not only brings new hope to the increment of wireless service, but flings down a challenge to the current video coding and transmission technologies. The situation in wireless network is unstabler than in a wired one. Despite the fluctuation of the network bandwidth, the moving rate and locality of the mobile equipment also severely affect the transmission rate. At the same time, wireless video encounters higher error rate than in wired network, which introduces the distortion into the error-sensitive compressed data, reducing the quality of the playback video. Therefore, to ensure the end-to-end quality, wireless video needs a transmission system including both efficient source coding scheme and transmission technology against the influence of the channel error, adapting to the environment of wireless channel.Supported by the national high technology research and development program of China (No.2002AA119010) and the National Natural Science Foundation of China (No.60496315), this thesis tries to find out solutions to the problem described above.Some successful methods used in wired networks guaranteeing the video quality are no longer useful in wireless transmission. Moreover, using of ACK or retransmission is also limited in some wireless services, for the end-to-end delay has to be minimized to avoid any perceptual disturbances and to maintain synchronicity of audio and video. Therefore, we mainly depend on source coding to perform error resilience. a new multiple description coding scheme—auto-resilient multiple description coding(ARMDC)—is proposed basing on the comparison and analysis upon many self-adapt source coding methods. Experiments demonstrate that our scheme has significant gains over other schemes such as MDR when applied to in wireless error-prone channels and ARMDC achieves better results in worse environment.Scholars have put forward many R-D models for non-layered, single-description video coders in the interest of rate controlling or video quality prediction. But most of them cannot describe experimental results very accurately due to simplistic assumptions. Particular R-D model for particular coding scheme is necessary. A R-D model for ARMDC is then set up in the thesis, which characterize the coding and transmission of ARMDC streams, giving the predicting expressions of transmitted video used at the coding end. Packet path diversity is enlightened by diversity techniques in wireless communication. It is just like transmitting multiple streams over several paths. This advance has encouraged scholars to do further research. But few people consider the problem in wireless channel and all the works only use simplest and basest MDC, just for the sake of providing streams for diversity. It is time to think of finding proper MDC scheme and well combine it to the transmission strategy to develop a real self-adapt system. After ARMDC is proposed, we combined it with multi-path transmission technology. The whole system is optimized and a framework of a self-adapt transmission system for wireless video is finally completed. Both experiments and emulation have been established for it.At the end of the thesis, DSP-based realization of real-time QCIF video coding is introduced. The hardware system and code optimization according to the system for two different type of high performance DSPs are compared in detail.