Sparse Representation And Estimation Of Non-WSSUS Channel In OFDM System

Orthogonal Frequency Division Multiplexing(OFDM)technology has the advantages of high spectrum efficiency,good bandwidth scalability,and strong anti-multipath interference capability,which has wide range of applications.However,OFDM techniques require accurate estimation of channel state.Compared with the widely studied Wide Sense Stationary Uncorrelated Scattering(WSSUS)fast fading channel with a single statistical characteristic,the channel in high-speed mobile environment exhibits a Non-WSSUS characteristic with abrupt change.Most existing methods for estimating fast fading channels are based on WSSUS and cannot be applied to high-speed mobile communication environments.Therefore,there is an urgent need to develop channel estimation theory and techniques applicable to Non-WSSUS channel estimation.This dissertation will study the characteristics of the Non-WSSUS channel in the OFDM system under high-speed mobile environment,establish the sparse representation of the Non-WSSUS channel,and develop effective estimation algorithms for it.The main contributions of our work are summarized as follows:(1)Based on the sparse representation of the wireless channel of OFDM system and the characteristics of Non-WSSUS channel,the sparse representation of the impulse response of Non-WSSUS channel in OFDM system is proposed.Using the Compressed Sensing framework,the sparse representation of the Non-WSSUS channel of the OFDM system is estimated.(2)The recovery error bound of the sparse signal recovery algorithm using Majorization-Minimization(MM)-based Non-convex Optimization(NcO)are given.On the basis of analyzing this error bound,a modified sparse signal recovery algorithm based on the MM-NcO is proposed.The proposed algorithm is able to exploit the prior information of sparse signal to formulate the objective function to reduce the error bound and improve the performance of sparse signal recovery algorithms.By exploiting the relationship between the proposed algorithm and Sparse Bayesian Learning(SBL),we further modify the proposed algorithm to be adaptive to the environment noise.(3)An adaptive weighted L,minimization algorithm is proposed to estimate the sparse Non-WSSUS channels with correlated non-stationary support changes.By using the correlation of the support of channel sparse representation,the algorithm introduces the prediction-estimation-re-estimation process to get the adaptive weight.The simulation results show that the proposed algorithm improves the accuracy and robustness of the Non-WSSUS channel estimation.The proposed weighted L1 minimization algorithm is essentially a method of using the prior information of sparse signal to construct the objective function to reduce the recovery error bound and improve the performance of the sparse signal recovery.Hence,the proposed weighted L1 algorithm can be regarded as a special case of the modified MM-NcO sparse signal recovery algorithm.(4)A Particle Filter aided modified MM-NcO algorithm is proposed and applied to estimate the sparse representation of Non-WSSUS channel.The algorithm first uses the Gauss-Markov process to model the sparse representation of Non-WSSUS channel.Then,the preliminary prediction of the sparse representation of Non-WSSUS channel is obtained by using the Particle Filter(PF)technique under the Dynamic Compressive Sensing(DCS)framework.Finally,regarding this preliminary prediction as the prior information of the sparse representation of Non-WSSUS channel,we use the modified MM-NcO algorithm to achieve accurate estimation of Non-WSSUS channel.On the basis of the PF aided modified MM-NcO algorithm,an estimation algorithm for the Non-WSSUS channel with abrupt change is proposed.The proposed algorithm introduces a threshold to decide whether an abrupt change occurs or not.By exploiting the information that the abrupt change has occurred,we re-construct the prior information of the sparse representation of Non-WSSUS channel and the objective function.Then,re-estimation is introduced to achieve accurate estimation of the sparse representation of Non-WSSUS channel with abrupt change.