Joint CFO And CIR Estimation For MIMO-OFDM Systems

Orthogonal frequency division multiplexing (OFDM) technique offers increased robustness against multipath fading and has high spectral efficiency, so it has become the best modulation and multiple access solution for broadband mobile communication system. Multiple-input multiple-output (MIMO) technique provides significant increases in data throughout (multiplexing gain) and link reliability (diversity gain), so it’s an effective way to improve the spectral efficiency. By combining OFDM and MIMO together, MIMO-OFDM enjoys the benefit of both techniques and has become a standard configuration of popular mobile communication system.As a multicarrier system, OFDM is very sensitive to carrier frequency offset (CFO). CFO induced by the Doppler shifts and mismatches between the local oscillators in the transmitter and receiver destroys the orthogonality among subcarriers and produces the interchannel interference (ICI), which may result in a significant performance degradation. Moreover, in order to achieve coherent detection, channel impulse response (CIR) must be estimated. In a MIMO-OFDM, CFO and CIR usually coexist and couple together. Hence, to design a joint CFO and CIR estimator is a critical issue.In the first part, we investigate the MIMO-OFDM systems suffered from time-invariant (TI) frequency selective fading channel. Based on previous research, we present a "multi-antenna, multi-path" model. Then we establish a maximum likelihood (ML) estimation and derive the Cramer-Rao lower bound (CRLB) for both CFO and CIR. To reduce the complexity of estimation, we resort to EM algorithm to decouple the multi-dimensional optimization problem into a number of one-dimensional optimization problems. Simulation results show that the performance of EM scheme reaches the CRLB approximately. At last, Subspace pursuit (SP) method is used to address the issue of sparse channel estimation and modify the joint estimation scheme. Simulation results show that the modified scheme has better performance and less computational complexity.In the second part, we investigate the MIMO-OFDM systems suffered from time-variant (TV) frequency selective fading channel. First we analysis the statistical properties of TV channel and present an effective way called basis expansion model (BEM) to describe it. Based on previous research, we establish two joint estimation methods in frequency domain and time domain for SISO-OFDM system. At last, we present a "multi-antenna, multi-path, time-variant" model. Then we expand the time domain method to two schemes:maximum likelihood (ML) estimation scheme and maximum a posteriori (MAP) estimation scheme. Simulation results show that the MAP scheme based on channel correlation properties has better performance.Finally, we summary the research methods and achievements of this thesis and list some possible improvement ideas.