Research On Channel Estimation Of OFDM System

The application of Orthogonal Frequency Division Multiplexing(OFDM) in wireless broadband digital communication has drawn more and more attention in recent years, because it has been proved to being good at combating the Inter-Symbol Interference(ISI)caused by time dispersion of radio channel. In OFDM system, the coherent demodulation requires the Channel Status Information (CSI) being known by the receiver through channel estimation, which makes the channel estimation a key technology in the OFDM system. First, the dissertation introduces the fundamentals of researches, and the application of channel estimation. Then the dissertation discusses the channel estimation of wireless OFDM system as follows:Firstly, the principles of cross-relation are introduced, and blind multichannel LMS method and Newton method are emphasized. The design of generalized optimal stepsize of LMS method is discussed in detailed, and a blind channel estimation simulation of SIMO-OFDM system is obtained based on CR iterativeness.Secondly, the conditions of typical subspace method of ZP-OFDM system are studied, analyzing the feasibility of estimating noise subspace by standard orthogonal iterative method. In order to improve the convergent speed and to obtain better estimation performance, two improved methods are proposed, LMS-NEWTON orthogonal iterative method and modified LMS-NEWTON orthogonal iterative method. Reduced rank orthogonal iterative method is applied to track subspace in order to track subspace fast and reduce the compute complexity. Since the inverse of autorelation matrix has the same eigenvectors with, another noise subspace reduced rank orthogonal iterative method is proposed. The simulation results show that both methods can achieve stable convergence.Then, among channel estimation methods of MIMO-OFDM systems, oversampling is carried out to estimate MIMO channel when the number of transmit antenna is greater than that of the received antenna. After oversampled, noise is not white any more, so the prewhiten method is applied. The semi-blind method with pilots or training sequences is proposed to solve the prewhiten problem. The simulation results show that reduced rank orthogonal iterative method could achieve to good the estimation of channel impulse response.Last but not least, signal subspace fitting and noise subspace fitting are analyzed. Combined with weighted subspace fitting and iterative fitting method, a OFDM iterative channel estimation method is proposed. In order to avoiding the high computation and achieve high speed to track time-variated data subspace, a OFDM channel estimation based on subspace fitting with conjugate gradient method is proposed. Since the conjugate gradient method and multistage wiener filter are equivalent, an OFDM semi-blind channel estimation method based on Krylov subspace is proposed. The simulation results show that the three methods could accomplish the similar result to classical subspace fitting method, and improve the estimation performance, while the last two methods show better performance than the first one in convergent speed, tracking effect and computation complexity.