| Carrier frequency offset (CFO) shifts the subcarrier frequencies, destroys orthogonality between the subcarriers and consequently induces inter-carrier interference (ICI). A further channel estimation error is found if we have no accurate CFO estimation. For multiple input multiple output (MIMO) structures, inaccurate channel estimation further introduces inter-spatial stream interference (ISSI). Therefore, CFO degrades the performance significantly.;In this thesis, three receivers will be presented: (1) Kalman filter (KF), (2) expectation maximization (EM), and (3) hybrid KF and EM receiver. Comparing to minimum mean square error (MMSE) receiver, these three receivers have better performance, especially for hybrid receiver.;A lot of research has been done on KF receiver performing channel estimation and CFO estimation. Compared to the research, KF receiver performing channel estimation, CFO estimation, and Doppler estimation simultaneously is proposed in this thesis. Since KF is sensitive to residual CFO, it is difficult to find good channel estimation. However, compared to the EM receiver, the better channel estimation can be found if CFO is not present. All the above properties will be used to discuss the hybrid receiver.;EM is a very powerful technique which provides channel and CFO estimation and data decision simultaneously for MIMO OFDM systems. However, the conventional EM receiver has the high computational complexity issue. A low complexity EM receiver is developed. Using the proposed approach, the EM receiver can be performed on each subcarrier separately and hence the overall complexity is greatly reduced. Numerical results show no performance degradation.;It is shown that the EM approach is complementary to the KF approach where the latter provides poor CFO estimation but good channel estimation and the former provides poor channel estimation but good CFO estimation. A hybrid combination of EM and KF approaches is proposed to take advantages of the strengths and avoid the weaknesses of both approaches. Numerical results have shown the hybrid combination performs far better. |
