Study On Superposition Coding For Dirty-Paper Channels

We investigate the theory and practice of dirty-paper coding (DPC) in the thesis. DPC is based upon the framework of channel coding with side information, with the interference known as side information non-causally at the encoder. Through proper coding, the interference incurs no loss of capacity compared with the standard interference-free channel.The main body of the thesis concerns the design of practical schemes of DPC. First presented are the principles of DPC, followed by detailed analysis of design of channel codes, quantization codes and corresponding joint iterative decoding algorithms. Finally, we present a realizable point-to-point DPC scheme. To summarize, the positive results obtained in the thesis are as follows:We discuss the construction methods and decoding algorithms of Low-Density Parity-Check (LDPC) codes. Simulations are respectively performed for irregular LDPC codes constructed through progressive edge-growth (PEG) algorithm and a structured LDPC code in STIMI standard. Then, we show the performance comparison and analysis of the two classes of codes.We investigate principles of encoding and decoding of trellis-coded quantization (TCQ), with detailed description of Viterbi algorithm for encoding and extended symbol BCJR algorithm for decoding. Simulations of one-dimensional and two-dimensional TCQ are conducted from perspectives of quantization performance and decoding performance.DPC scheme based on binning technique is then explored systematically, with its feasibility proved by superposition coding. Detailed analysis is given to the design of channel codes, quantization codes and joint iterative decoding algorithm. Finally, we propose a scheme based on aforementioned analysis and simulation results.We develop two approaches to reduce quantization loss using one-dimensional TCQ. The first one brings a considerable improvement of performance by optimizing power constraints of the channel code. An apparent improvement is obtained by the second approach, which employs two-dimensional TCQ as quantization code.