Research On Key Techniques Of LDPC-Based MIMO System

The past decade has seen great improvements in wireless personal communication technologies. However, to provide new services over the wireless channel of strictly limited bandwidth is still very challenging for researchers and engineers. For high speed information transmition with low-delay and high reliability in selective-fading channels, an irregular repeat-accumulate (IRA) codes based, threaded layered space-time-frequency system is proposed. The proposed system employs several new techniques, such as multi-input multi-output (MIMO), low-density parity-check (LDPC) codes, orthogonal frequency division multiplexing (OFDM), and turbo iterative receiver. We give some solutions to solve the practical problems that arise in such technologies, with emphasis on the design of involved signal processing algorithms. In the design of these algorithms, we not only emphasize the tradeoff between performance and complexity but also give consideration to the robustness of them.Firstly, an efficient decoder is proposed for the perfect space-time block (perfect-STBC) codes. Based on the special structure of perfect-STBC, an equivalent vertical Bell Labs layered space-time (V-BLAST) decoding model is derived. Minimum mean-square error decision feedback equalizer (MMSE-DFE) preprocessing is applied to this model and a decoder with boundary controlled Fano tree search algorithm is given. This decoder achieves almost maximum likelihood (ML) decoding performance at lower complexity than that of existed near-ML decoders. Furthermore, the decoder is robust to antenna configurations and to modulation constellationsSecondly, for space-time bit-interleaved coded modulation (ST-BICM) system, the analytical calculation of extrinsic information transfer (EXIT) chart is derived. In particular, the EXIT-band technique is introduced to the practical finite-length ST-BICM system and a detailed algorithm for the EXIT-band analysis and calculation is given. Using this algorithm, we give an analysis and comparison to several typical soft-input soft-output (SISO) MIMO detectors.Thirdly, as to the SISO MIMO detector design, we improve the List-TM detector and obtain a robust detection algorithm named List-MTM. First, the unbiased MMSE-DFE filter is used to preprocess the MIMO channel and a tree-search structure is given. Second, the survived paths are constructed by the TM algorithm. Then, some partial-length paths are augmented selectively. Finally, the soft information of coded bits is calculated according to both the survived paths and the augmented paths. The proposed List-MTM detector behaves flexible in performance-complexity tradeoff and provides a unified framework for the breadth-first tree-search based MIMO detectors. In addition, by adding one-bit complement vectors and by setting particular parameters to the List-MTM algorithm, another efficient MIMO detector named U-MMSE-ITS is derived. This detector not only provides high detection performance at very low complexity but also is robust to antenna configurations and to MIMO channel conditions. The complexity of the U-MMSE-ITS detector is almost invariant for all channel conditions, which makes the hardware implementation conveniently.Fourthly, for general LDPC codes encoding, we give a practical algorithm that transforms the parity check matrix into an approximate lower triangular form by performing row and column permutations only. Online encoding according to the check matrix of this approximate lower triangular form needs only linear time complexity. In addition, for the short-length IRA codes construction, we proposed a constraint optimization to the progressive edge growth (PEG) algorithm based on the special structure of IRA codes. Short-length IRA codes constructed by the optimized PEG algorithm exhibit very good performance.Fifthly, based on the iterative receiver structure of the LDPC coded MIMO system, we improve the standard belief propagation (BP) decoding algorithm from the following three aspects: check-nodes information retain, message-passing schedule, and check-nodes information calculation method. The improved decoder not only converges very fast (5 iterations is sufficient) but also takes only a small amount of add operation and multiply operation to calculate the information updating. Compared to the standard BP decoder, it has almost no performance loss.Finally, based on the above research, an IRA codes based threaded layered space-time-frequency system is proposed. The simulation results demonstrate that, with low processing complexity, the proposed system can provides high data rate with high reliability over selective-fading channels. Furthermore, this system is robust to arbitrary antenna configurations and to any MIMO channels.