In a MIMO-OFDM system, including channel fading, time delay andfrequency offset and so on, the three key types of channel parametershave great impact on the performance for receiving end, and themain work of this dissertation can be concluded as follows:1.Considering MIMO-OFDM system is very sensitive to frequencyoffset, compared with maximum likelihood estimation algorithm andthe S&C algorithm, It can be concluded that the frequency offset of MLalgorithm which use cyclic prefix is verified limited, however, frequencyoffset range of the S&C algorithm, relatively simple and occupyingmore system resources, is more wider.2.Considering MIMO-OFDM system receiver needs to identify thechannel information for equalization, after analyzing the complexity ofthe wireless channel algorithms, LS algorithm, which is convenient butperformance comparatively poor, is certificated the least computingcapacity. The property of MMSE algorithm is in the middle whileconsidering the white noise process. SVD algorithm effect is confirmedthe same as MMSE algorithm in low SNR while the BER is deteriorated inhigh SNR. ML algorithm performance is better but the computationcomplexity is high.3.Due to the actual joint parameter estimation of matrix inversion,iteration EM-PF algorithm, which use weighted particle parametersapproximate the real-time posterior probability density function, isproposed and improved iterative dimensionality reduction EM-PFalgorithm is explored. Simulation investigates the impact on the numberof iterations after the PF algorithm added and it improves on the initial,and at last verifying the performance of each algorithm on differentsimulation platform. |