Multimedia Communication Over Multi-Antenna Systems

Nowadays, with growing wireless communication capability and the demand for multimedia applications, wireless multimedia communication has become an increas-ingly active area of research. In the progress of wireless communication, multi-antenna technology is an important strategy to achieve high-quality and high-rate communi-cation and it is widely used in the next generation wireless networks. Compared with single-antenna systems, multi-antenna systems could take advantage of multipath prop-agation to provide diversity gain for higher reliability, or to greatly increase the trans-mission rate by achieving multiplexing gain. Considering that wireless channels are time-varying, error-prone, and bandwidth limited, how to improve the quality of the wireless multimedia communication, particularly in multi-antenna systems, is an im-portant and challenging problem.Separate source-channel coding is commonly employed in conventional wireless multimedia systems. However, due to the existence of threshold effect and level-off effect, the performance of these separate coding systems is not robust with respect to wireless channels with bad condition. In order to strike a balance between system ef-ficiency and reliability, the idea of joint source and channel code design has attaches more and more attention, and is applied in a lot of systems. Most of these systems fall into two categories:the layered source coding based digital schemes, and the analog-like or hybrid digital-analog coding schemes. However, only a few discrete levels of recovery quality can be achieved by the layered digital schemes; while the second kind of schemes are mainly designed for single-antenna systems, and they are difficult to be efficiently applied in multi-antenna systems. Considering that analog coding strate-gies have low complexity and graceful behavior, analog coding based new multimedia transmission schemes for multi-antenna systems are proposed in this thesis.We study the multimedia communication problem in three different kinds of multi-antenna systems, and we design specific analog coding based strategy for each one of them. The discussed systems include:MIMO-OFDM based unicast system, the multi-antenna broadcast system with receive antenna heterogeneity, and the low-cost virtual multi-antenna system. The main contributions are summarized as follows: (1) For the MIMO-OFDM based video unicast systems, we propose a new linear joint source-channel coding scheme. This scheme first separates the source and channel into independent components, and then matches the more important source components with higher-gain channel components in channel allocation process. After that, it im-plements the transmit power redistribution with joint consideration to the source and the channel, and finally adopts pseudo-analog modulation for transmission. The proposed system achieves fine-grained unequal error protection across source components, and ensures the recovery video quality commensurate with channel condition.(2) For the multi-antenna image broadcast network with receive antenna hetero-geneity, we propose to solve the diverse channel quality and different numbers of an-tennas across receivers by integrating compressive sensing (CS) into the transmission. The sender transmits linear combinations of the image coefficients with analog modu-lation, and the receivers adopt a CS recovery algorithm. In particular, we first study the transmission of compressible statistic signal, which has been used for modeling image coefficients. The analysis and simulation results show that the decoding for all kinds of users is feasible and multi-antenna users are able to benefit from the additional anten-nas. After the study on abstract model, we consider image broadcast in such systems and still use CS technology. We also introduce a power allocation strategy to achieve performance tradeoff between users with different antenna numbers. The experimental results demonstrate the advantages of our system compared to other reference systems.(3) For the transmission of non-uniform distributed bit stream over the virtual multi-antenna systems, we propose four decode-and-forward cooperative strategies and analyze their respective achievable rates. The main feature of these strategies is that they utilize CS as the. channel code and directly transmit multi-level CS projections through amplitude modulation. The discussed virtual multi-antenna system is formed by multi-ple single-antenna users of a relay network. The non-uniform distributed bit source is the model of the coded bits in conventional multimedia coding schemes, but remaining some statistical redundancy by skipping the binary entropy coding. The experimen-tal results show that the CS based cooperative strategies have great potential in both transmission efficiency and adaptation capability to channel variations.