The Study Of LDPC Codes And Its Feedback Iterative Equalization Techniques

The linear block code is called a binary low-density parity-check code if it is constructed based on a sparse parity-check matrix. Since it has been proved that the performance of LDPC codes is extremely close to the Shannon limits when combined with iterative BP algorithm, the discovery of LDPC codes is a great progress in channel coding field after turbo codes. In recent years, LDPC codes has drawn the world-wide attentions in channel coding community due to its great performance and potential worth in application.On the basis of comprehensive studies of the performance of LDPC codes, this thesis discusses two effective encoding methods, Gauss elimination and the one based on approximate lower triangular matrix. As for BP decoding algorithm, simulation results are obtained over an AWGN channel. Furthermore, in order to overcome the problem of inter-symbol interference (ISI), we study the equalization techniques with feedback iterations suitable for LDPC codes. Applying the principles of turbo decoding, extrinsic information is exchanged between the decoder and the equalizer at the receiving end iteratively to form turbo equalization. First, by applying MMSE criteria, we combine the BP decoder of LDPC codes with linear equalizer (LE) and decision feedback equalizer (DFE) respectively to form the non-adaptive equalization receiver. Second, we present LDPC/LMS-DFE algorithm of adaptive equalization. Finally, we focus on the performance of LDPC/CMA-DFE algorithm of blind equalization.For all three kinds of equalization techniques, we perform simulation and comparative studies. The results show that the LDPC receiver based on turbo equalization can effectively improve system performance. Compared to adaptive equalization, the non-adaptive equalization offers better BER performance and faster convergence speed. But the computational complexity and the request of given channel characteristics seriously limit its applications in practice. When the channel characteristics are unknown, blind equalization has similar performance to adaptive equalization. Because it doesn't need training sequence, it can use the frequency band more efficiently, which makes it quite promising for the future applications in mobile communications.