Study On Doubly Selective Channel Estimation Based On Compressed Sensing In OFDM Systems

High mobility induces time and frequency selective fading in wireless communication channels within orthogonal frequency division multiplexing(OFDM)systems.To ensure the data through the fading channel can be received correctly,channel state information has to be estimated.The application of compressed sensing is considered in the estimation of doubly selective channels because of its sparsity on delay-doppler domain.This thesis studies the doubly selective channel estimation based on channel estimation.The main contents include:(1)A optimized method of pilot pattern selection is proposed to estimate double selective channels.Compressed sensing has been applied to the estimation of doubly selective channels within orthogonal frequency-division multiplexing(OFDM)systems,while the pilot pattern selected randomly leads to the unstability of recovery capability about the observation matrix and affects the channel estimation performance.A couple of coherence functions were adopted in selecting optimal pilot patterns according to the coherence of the observation matrix for better estimation of the doubly selective channels.Simulation results demonstrate that pilots selected through coherence functions could achieve better performances.(2)Two optimized recovery algorithms are proposed based on original algorithms.The complexity of ROMP recovery algorithm increases with higher sparsity,two optimized algorithms are proposed to reduce the complexity.To reduce the complexity of atom selection and solving least squares problem in ROMP,one optimized recovery algorithm with a rigorous computational bound is proposed,which identifies a fixed number of atoms to make the recovery submatrix be a nonsingular matrix.In addition,the recovery submatrix is renewed at the end of each iteration to improve the precision.The other modified recovery algorithm with QR-decomposition avoids the pseudo inverse operation in LS,which reduces the computation complexity.(3)A doubly selective channel estimation method based on compressed sensing with the existence of ICI is proposed.Main diagonal and other diagonals of a doubly selective channel coefficient matrix are nonzero when ICI exists.However,traditional channel estimation methods ignored the other part outside the main diagonal.Each diagonal of doubly selective channel coefficient matrix is estimated through compressed sensing technology due to its sparsity on delay-doppler domain.And the optimal estimation performance is obtained by adjusting the size of effective bandwidth in the channel coefficient matrix.The simulation results show that the method improves the estimation performance of double selective channels with ICI.