Key Technology Of Video Transmission Over Wireless Channel

With the explosive growth of the Internet and dramatic increase in wireless access, there is a tremendous demand on multimedia delivery over wireless networks. Despite a large number of standards and commercial products for applications of visual communications, providing high quality visual information over resource-limited wireless networks still remains a major challenge. This is because of the hug data size of visual information and the complicated characteristics of wireless channels. Due to the limited bandwidth of wireless channels, it is desired that visual information is compressed by highly efficient compression schemes while, due to the time-varying error prone environments of wireless channels, controlled redundancy is necessary to be added in the compressed bit streams in order to ensure reliable transmissions. Therefore, there is a tradeoff between efficiency and reliability. The objective of this research is to find the optimal tradeoff between efficiency and reliability for video transmission over wireless channels. Supported by the national high technology research and development program of China (No.2002AA119010) and the National Natural Science Foundation of China (No.60496315), in this dissertation, some key technologies for wireless video transmission are researched and video transmission system over wireless channel is developed based on robust error control and optimal bit allocation.In order to further increase the efficiency of source coding and to relieve the resource conflict between huge amounts of information of video and the limited bandwidth of wireless channel, we introduce a novel video coding algorithm based on ROI (region of interest). Content-based or ROI-based video coding can improve subjective video quality and utilize the bandwidth usage due to taking into account the characteristic of the human visual system (HVS). It is known that algorithms of object segmentation from video are essential to realize ROI-based video coding. Unfortunately, those frequently used object segmentation algorithms at present have such shortcomings like computational intensive, unsuitable to resource limited wireless environment. Therefore, we propose video object segmentation algorithm based on simple characteristics of moving blocks in this dissertation. Comparing with DCT which is commonly used in video coding, the computational complexity of the proposed algorithm is reduced at least two order of magnitudes. So the additional computational complexity is negligible. Experimental results have shown that accuracy of the proposed segmentation algorithm can satisfy the requirement to further improve video coding efficiency. Moreover, a novel global movement estimation algorithm based on the ROI is proposed. Experimental results also verify computational efficiency and accuracy of the algorithm. This algorithm also proves computational efficiency and accuracy of the proposed video object segmentation algorithm.Because of shadowing, fading and noise, wireless channel, as time varying and bandwidth limited channel, has such natures that it is error prone and has bust error pattern. It has been shown that burst errors have different effects on different communication layers, and these effects are not i.i.d. Therefore, it is not sufficient to investigate average error rate to completely describe error characteristic, higher order statistics are needed to describe the correlations between errors. In order to improve communication quality, it is necessary to design forward error correction (FEC) and automatic error request (ARQ) which are based on full knowledge of error characteristics of the underlying wireless channel. Therefore, a model of wireless channel is proposed in this dissertation. This model is divided into several different layers. We study the method of modeling and the algorithm to estimate the corresponding parameters of the model to accurately describe burst characteristic of errors. The study has also revealed the inner relationship between errors of data units on different protocol layers.Then, we utilize the model of data unit errors on different protocol layer to analysis the performance of FEC on physical layer and ARQ on link layer (include efficiency of throughput and characteristic of delay). Furthermore, we reveal the inner relationship between error characteristic and error-controlling parameters of wireless channel. These results provide a certification to optimize communication characteristic of wireless channel from theory. Especially, these results have not only made it possible to optimize communication system across different layers, but also give a foundation to build operational wireless communication system which guarantees quality of services (QoS).Finally, a joint source-channel optimization scheme for wireless communication is proposed. This scheme is based on the study results of source coding and wireless channel which are mentioned above. To minimize distortion, rate is optimally partitioned between source and channel. The video source coding is based on ROI, while rate is divided between source coding and channel coding, and different regions of video. Parameters of different error control methods are optimized based on the optimization across different layers. In order to improve accuracy of the estimation wireless channel's distortion and to increase response to time varying channel, we propose a novel algorithm to eliminate distortion of video signal which is based on feedback information and the error model of channel. Experimental results have proved that this video communication system has very good ability to adapt to the changes of channel, and still has pretty good subjective quality under poor channel conditions. Moreover, the computational efficiency is high. Since there are no more requirements about coding and channel environments, this method for communication is general.