The Research Of Multimedia Communication Over Wireless Network

Recently, the development of broadband wireless networks has stimulated greatinterest in real-time video communications over wireless network. However,delivering quality video over wireless networks in real time is still a challenging taskdue to the burst errors nature of wireless channels and the error propagation propertyof compressed video. Suitable efficient schemes are necessary to be added in thecompressed bitstreams in order to ensure reliable transmissions.This dissertation addresses some key issues in wireless communication systems.These problems are of great interest to communication engineers involved in planningand optimized design of wireless communication systems, including video qualityevaluation, error capacity, error resilience, channel coding, joint source and channelcoding, etc. Some new algorithms and schemes are proposed. Then maincontributions can be summarized as follows:1,A new effective and reliable objective video evaluation model is proposed basedon fuzzy synthetic judgment. Firstly, the essential design of the model is describedsystemically. Then the fuzzy membership functions of all factors which affectcompressed video quality are determined. And the factor weights are also estimatedby approximate algorithm. This model highlights a comprehensive evaluation bytaking account of multiple properties of the compressed video which affect the wholevideo sequence, such as quality, fluency and motion info. On the other side, thismodel has well capability of compatibility with other evaluation model or standard.Simulation results demonstrate the advantages of the fuzzy objective video qualityevaluation model especially in comprehensive judgment of the whole video.2,An optimal rate allocation algorithm in joint source-channel coding for videotransmission is proposed. We deem that error concealment techniques have beenemployed at the decoder to conceal the effect of transmission errors, which will exertindirect influence on traditional JSCC schemes. The highlight improvement of thisproblem is to add the error resilience term of decoder to rate-error evaluation function.Therefore, there is a tradeoff among source, channel, and error resilience capability.The error concealment schemes adopted in different coding standards could becombined with JSCC. Performance results show both the effectiveness of the schemeand utilization of the source distortion rate characteristics and the BER statistics of theavailable channel codes. 3,A crossed UEP packetization algorithm based on data partitioning and combinederror control scheme is presents to improve the resource utilization and the dataservice performance under fading channel condition with high error rates. Adaptivelength packetization and HARQ scheme are also adopted to optimize our combinederror control scheme. Simulation results demonstrate that our proposed schemesoutperform some other approaches especially UEP achieved by applying differentReed-Solomon coding.4,A novel error concealment scheme for intra frames of h. 264 is proposed. In thisalgorithm, the error pixel value of the concealed intra frame is copied andre-interpolation from the corresponding pixel of the previously decoded frame and theadjacent blocks received correctly. At the same time, we propose extending the JMsoftware to support the intra error concealment for bitstreams containing missingcoded picture. The proposed error concealment algorithms for intra frames weretested using the packet loss model of the video ad-hoc group of SA4 within 3GPP,which is intended to provide common conditions and valuable toolkits for testingvideo transmitted over 3GPP networks. Simulation results demonstrate that ourmethod outperforms the other method by up to 4dB.5,A novel approach for the design of rate-compatible punctured low-densityparity-check (RCPL) is proposed, which is obtained without restriction on regular orirregular codes. First, we apply density evolution method to obtain optimal degreedistributions on desirable code rate. Then we give a procedure for determining theoptimized puncturing by getting max number of survived variable nodes for everycheck node in each degree distribution. The puncturing patterns are relative to thedegree distribution of the variable nodes. It is shown the punctured nodes arerecovered more reliably. Simulation results show that rate-compatible LDPC codesconstructed in this manner perform better than other techniques.