Iterative Decoding And Iterative Equalization Algorithms

Low-Density Parity-Check codes (LDPC) are a class of channel decoding codes published in 1962. They were forgotten during decades mainly because of the implementation complexity of the decoder, until recently, when deep submicron technologies allowed the implementation of far more complex algorithms than standard forward error correction codes: if combined with the Belief Propagation algorithm, LDPC codes can reach performance near the theoretical Shannon limit.Due to the LDPC codes adopted by various communication systems are all long codes, the conventional LDPC decoding algorithm proposed by Gallager requires large amount of storage units and needs large number of iterations to converge. I.e. the conventional LDPC decoder is hardware expensive and power consumptive. Propelled by the Turbo decoding algorithm which was introduced in 1993, we induce the Turbo decoding theory into LDPC decoding which is so called Turbo like decoding and reduce the storage units by 50% and decrease half number of the convergence iterations.Besides the storage and convergence speed, the computational complexity is another critical problem. In the thesis, we propose several complexity reducing algorithms such as column based Turbo SPA, deepest descent SPA, Jacobian SPA,normal min sum SPA andλmin sum SPA which not only greatly reduce the computational complexity but also supply various schemes for various application requirements.Since Turbo codes were introduced in 1993, numerous works have focused on the iterative detection. Perfect Channel State Information (CSI) is not available in many practical situations. The Channel has one or more unknown,and possibly time varying, parameters in its model. The soft information, so called extrinsic information, is exchanged between Soft Input Soft Output (SISO) modules which can be refined during each iteration.Message Passing (Belief Propagation) algorithm in graphical models (Bayesian networks, Factor graphs, Tanner graphs) has been used to explain the success of iterative detection, especially for Turbo codes. In the thesis, we apply the Tanner graphs theory to describe the channel response and propose a Tanner graphs based equalization algorithm whose performance approaches the performance of the BCJR equalization. As we all know, soft decision algorithms are always superior to the hard decision algorithms. In the thesis, we also propose a blind soft decision feedback (SDF) MMSE channel estimation algorithm and which greatly improves the precision of the channel estimation, especially for the practical implementation. We also induce a blind adaptive soft decision feedback LMS channel estimation algorithm.The Tanner graphs based equalization algorithm and the adaptive soft decision feedback channel estimation algorithm combined with the forward error correction (FEC) codes consist the perfect Turbo equalization loop. In the final of the thesis, we put the equalization loop into DVB-S2 system and obtain an excellent performance.