Research On Underwater Acoustic Channel Estimation Method Based On Compressed Sensing Model

In view of the problems of unknown channel sparsity,strong noise interference and multipath expansion in the actual underwater acoustic channel,this paper uses the inherent sparsity of underwater acoustic channel to establish a compressed sensing model for channel estimation,and studies the sparse channel estimation algorithm and pilot position optimization in OFDM system.Firstly,the smooth L0 algorithm(SL0)based on the compressed sensing theory can reconstruct the channel parameter information better when the prior information of the sparse degree of the underwater acoustic channel is unknown.For the actual noisy underwater acoustic channel,the reconstruction performance of the traditional SL0 algorithm is significantly reduced.In this paper,an improved algorithm(MARe SL0)is proposed,which can adaptively generate reliable regularization factors and balance the errors based on the sparsity and residual of the solution in the objective function in the iterative process.In order to ensure the accurate convergence of the iterative process to the best advantage,the Nesterov Gradient Acceleration method is used as the initial value of Newton's method to mix the values after the internal iterative cycle Optimize the reconstruction algorithm.Numerical simulation shows that the proposed algorithm(MARe SL0)has better noise robustness than other classical algorithms,and improves the performance of channel state information estimation effectively.Secondly,an improved sparsity adaptive weak selection matching pursuit algorithm(MSASWOMP)is proposed to solve the problem of a priori sparsity and pilot resources for the actual underwater acoustic channel.First by the size of the sparse degree of initial estimates as the initial support set,then to weak threshold selection of atoms,the atoms support set as back to screen the candidate set,with initial support set size is value back to the initial conditions for the secondary screening,the last stage of using variable step length method for accurate estimates,the sparse degree gradually and in iterative estimation value under the condition of adaptive update back to start,eliminate the I atoms.The simulation experiment analyzed the influence of threshold parameter,sparsity estimation step size and pilot number on MSASWOMP algorithm,and the results showed that the algorithm could getmore accurate channel estimation value with less pilot number,and its estimation performance was better than traditional algorithm while saving pilot resources.Thirdly,in the sparse channel estimation of OFDM system,the selection of pilot position can effectively improve the accuracy of channel estimation and shorten the use of pilot number.Generally,the channel estimation algorithm based on the compressed sensing model uses the randomly distributed pilot position to estimate the channel.However,the pilot generated by the random distribution has great randomness.Considering this problem,the reciprocal of the cross-correlation value of the minimum measurement matrix is used as the fitness function for optimization,and the pilot position is optimized by the adaptive genetic algorithm.In the process of location optimization,the mutation probability and crossover probability are adaptively obtained to update the pilot sequence group and calculate the corresponding fitness function,and the minimum value of the set fitness function and its corresponding optimal pilot location are screened out.The simulation results show that the optimized pilot position can improve the performance of channel estimation.