Cooperation and joint source-channel transmission in wireless networks

摘要

In this thesis, we study the problem of cooperation and joint source-channel transmission in wireless networks, with an emphasis on some fundamental information-theoretic aspects. The majority of this thesis focuses on the analysis of fundamental performance limitations of joint source-channel transmission in wireless cooperative networks. We made three major contributions in this topic. The first contribution is a study on the end-to-end distortion of joint source-channel transmission in multi-relay cooperative systems, in terms of the distortion exponent at high signal-to-noise ratio (SNR). Building upon results from the diversity-multiplexing tradeoff (DMT) analysis, the achievable distortion exponents of multi-relay cooperative systems with layered coding and transmission strategies are obtained. We next propose to improve the achievable distortion exponent by employing limited channel state feedback in the multiple-relay system. We show that combining a simple feedback scheme with single-rate coding outperforms the best known non-feedback layered transmission strategies with only a few bits of feedback information. The third part focuses on the recently proposed two-way relaying cooperative networks, where two users communicate in both directions with the help of one relay. We introduce and analyze a new concept - achievable distortion exponent region, which characterizes the end-to-end distortions of both users and addresses the multiuser nature of the two-way communication system. In addition, we extend the DMT analysis to two-way relaying cooperative networks and obtain the DMT regions of various bidirectional cooperation protocols. This thesis also investigates the cross-layer resource allocation in wireless systems. We consider transmitting a layer-coded source over a slow fading channel using the broadcast strategy, where the channel state information is not available at the transmitter. An efficient iterative algorithm is proposed to minimize the end-to-end distortion by jointly solving the power allocation problem and the channel discretization problem at an arbitrary SNR.