Mu - Mimo - Ofdm Carrier Frequency Offset In The Uplink Synchronization Technology Research

Beside the improvement of channel capacity, Multiple-User Multiple-Input-Multiple-Output (MU-MIMO) also brings new technical difficulties into Orthogonal Frequency Division Multiplexing (OFDM) system. Due to the sharing of the same time-frequency resources by multiple users, the Carrier Frequency Offsets (CFO) result in multiple Inter-Carrier Interferences (ICI) in the received signals, so it is very important to estimate and compensate the CFOs in MU-MIMO-OFDM. The receiver needs to estimate multiple CFOs from the received signals, which are combined by multiple signals from independent channels, so the synchronization techniques in single-user systems cannot be applied in MU-MIMO-OFDM uplink.In this dissertation, we work to improve the performance of estimation and compensation in the CFO synchronization in MU-MIMO-OFDM, including three parts:CFO acquisition, Residual CFO (RCFO), and CFO compensation.Traditional CFO acquisition algorithms search the jointly optimal solution of CFOs and channels according to some criterion like Maximum Likelihood (ML) or Minimum Mean Square Error (MMSE), needing to search in a high-dimension space. However, the posteriori of CFOs can be obtained directly by marginalizing the channels using Bayesian approach, and then Maximum A Posteriori (MAP) and MMSE solutions can be searched in a low-dimension space, reducing the complexity. Besides, bringing in the priori of the CFOs can improve the estimation accuracy of the CFOs.In order to obtain the close-form solution, traditional RCFO tracking methods using pilots in frequency domain always use linear approximation to simply the nonlinear problem. However, the approximation causes estimation biases, degrading the estimation performance. A jointly tracking of RCFO and channels based on Expectation Maximum (EM) is proposed in this dissertation. EM instead of linear approximation to solve the nonlinear problem avoids the estimation biases, and the impact of the channel estimation errors can be reduced since the error probability distribution is considered while estimating the RCFOs. Meanwhile, proper initial values are chosen in tracking phase to reduce the iteration number.The channel and CFO estimations in frequency domain always suffer from ICI, because the compensations should be accomplished after estimation in frequency domain. After analyzing the components of ICI and the performance of its compensation, a partial CFO pre-compensation mechanism is proposed to suppress the ICI power by forming a feedback structure. Optimization problem is used to describe the relationship between ICI power and pre-compensation frequencies, and the close-form of an approximated optimal solution is given at last.