Frequency Synchronization In MIMO For Wireless Communications

As an important approach to increasing the spectral efficiency and the system capacity of wireless communications, multiple input multiple output (MIMO), with emphasis on distributed MIMO deploying distributed antennas, has been attracting much research interest. Orthogonal frequency division multiplexing (OFDM) can effectively combat the multipath fading of wireless channels and has become another hot point in the area of wireless communications. Hence the combination of MIMO and OFDM is a promising technique for the new generation of wireless communications.Like that of single antenna systems, the performance of MIMO systems may seriously degrade with the presence of the carrier frequency offset (CFO) that is unavoidably present due to the possible oscillator mismatch as well as the relative motion between the transmitter and the receiver. In particular, MIMO-OFDM systems are very sensitive to the CFO. Any CFO causes a loss in the orthogonality of the subcarriers which results in inter-carrier interference (ICI) and hence performance degradation. Therefore, it is of primary importance to accurately estimate this frequency offset and compensate for it prior to performing detection. This is the task of frequency synchronization. Since there may exist multiple CFOs in distributed MIMO systems, the CFO estimation and compensation become more difficult for this case.This thesis focuses on the problem of frequency synchronization for MIMO systems. More specificly, it consists of the following three aspects: maximum likelihood (ML) CFO estimation in spatially correlated MIMO channels, CFO estimation for distributed MIMO, and CFO compensation for distributed spatial multiplexing MIMO-OFDM.Firstly, taking the spatial correlation among channels corresponding to different pairs of transmit and receive antennas into account, a data aided (DA) ML CFO estimator based on the marginal likelihood function is proposed. It can exploit spatial diversity and make use of the knowledge of spatial correlation. The Cramer-Rao bound (CRB) for the problem is also derived as a benchmark.Secondly, considering the general case where both the time delays and the CFOs are possibly different between the receiver and each transmit antenna, a general model for CFO estimation in distributed MIMO systems is presented. Then a suboptimal CFO estimator based on ML is proposed by making a suitable choice of the training sequences, which can estimate the CFO for each transmit antenna respectively and can exploit spatial diversity.Finally, the proplem of CFO compensation in distributed spatial multiplexing MIMO-OFDM systems is addressed. Firstly, considering macroscopic fading and multipath Rayleigh fading, the optimal initial CFO correction value is derived by maximizing the average signal to interference and noise ratio (SINR). Secondly, by exploiting the underlying characteristics of the zero forcing (ZF) detection and assuming low delay spread of the multipath channel, a CFO compensation method without need for computationally complex matrix inversions is proposed, which can compensate for the CFO for each spatially multiplexed substream respectively. Finally, a detection method involving two CFO correction processes is presented by combining the initial CFO correction method and the CFO compensation method presented above.The research in this thesis extends the existing researches on frequency synchronization for traditional MIMO systems, and explores the problem of frequency synchronization in distributed MIMO systems. The results have some theoretical and practical values.