Shift Auto-Correlation Frequency Offset Correction Of MIMO-OFDM System In Frequency Domain And LDPC New Parameter Decoding

The past decade has seen great improvements in wireless personal communicationtechnologies. However, to provide new services over the wireless channel of strictlylimited bandwidth is still very challenging for researchers and engineers. To improvethe system capacity, the next generation wireless communication system adopts MIMO,OFDM, Turbo codes, LDPC codes, multi-user detection, and adaptive modulation tech-nologies. We focus on MIMO, OFDM and LDPC technologies in this thesis, and pro-pose several solutions to solve some specific problems in such technologies.A frequency offset estimation algorithm using the shift auto-correlation methodis proposed for OFDM systems. The key idea is, when an OFDM signal is perfectlysynchronized in both frequency and time domains, the demodulated signal of every sub-carrier is un-correlated with each other;when frequency offset exists, the demodulatedsignal of every sub-carrier contains the information from neighboring sub-carriers, sothe demodulated signals are not independent with each other. Moreover, the degreeof correlation between demodulated signals on different sub-carriers is a function ofthe sub-carrier frequency offset. Therefore, frequency offsets can be estimated by con-structing an estimation function with the shift auto-correlation method. In addition, thisalgorithm can be also applied in MIMO-OFDM, PCC-OFDM and DWMT systems.The contribution of this thesis in LDPC decoding is tri-fold. First, a new param-eter is presented for LDPC decoding, which describes the reliability performance ofeach bit in the bit ?ipping algorithm. The bit ?ipping decoding performance can belargely improved with the help of this new parameter. Second, a token based decodingalgorithm is proposed to solved the problem that the "optimal Taboo length" must becalculated in the Taboo based algorithm. With the help of the new parameter, the tokenbased algorithm provides nearly equivalent decoding performance with lower compu-tational complexity than UMP-BP. Third, in BP decoding, the new parameter can beused to evaluate every bit's reliability before each step of iteration. The bit and con-strain information with high reliability need not be updated by the decoder, and, as aresult, the computational load can be significantly reduced.Another important contribution of the thesis is constructing a spatial-time-frequency LDPC encoder structure in the MIMO-OFDM system, which allows everyLDPC code to exploit the capacity of diversity as much as possible. A scheme for SoftInterference Cancellation based on Parameter Decision (SIC/PD) is presented for thisLDPC-MIMO-OFDM structure. Simulation results show that satisfactory improvementcould be achieved under moderate SNR level.