The Pilots Design For The Compressed Sensing Based Sparse Channel Estimation

The traditional channel estimation methods based on pilots need to use a large number of pilots,which leads to low-frequency spectrum utilization,and does not fully consider the channel sparsity.In broadband wireless communication system,the application of compressed sensing(CS)in pilots assisted sparse channel estimation has less pilot overhead and better performance.At present,the channel estimation methods based on CS have been widely studied,and many sparse signal recovery algorithms have been applied to channel estimation.Another key problem in the sparse channel estimation is the design of pilot position and their values of the symbols.This paper mainly studies the CS based sparse channel estimation and pilot pattern optimization in the orthogonal frequency division multiplexing(OFDM)system and the multiple input multiple output orthogonal frequency division multiplexing(MIMO-OFDM)system.The main contributions are as follows:1)In order to solve the problem of pilot pattern design in channel estimation of CS based OFDM systems,this paper compares several existing algorithms in terms of the advantages and disadvantages of their optimized pilot patterns by simulation and the conclusion is that the stochastic search schemes(SSS)is better.Then,a tree-based stochastic search schemes(TSS)algorithm is proposed,and the performance of TSS and SSS is compared by simulation.Simulation results show that,as comparsed to the SSS algorithm,the TSS algorithm costs shorter time and the system employing the pilot pattern obtained by TSS algorithm has smaller mean square error(MSE)and system bit error rate(BER).2)To solve the problem of pilot pattern design in MIMO-OFDM channel estimation based on CS,the TSS algorithm is improved to obtain the tree-based stochastic search algorithm in multiple-antennas(TSS?MA),which can obtain the pilot pattern of multiple transmit antennas in MIMO-OFDM system.As compared to the existing Stochastic Search Schemes Extension Scheme(SSS?ES),simulation results show that the TSS?MA can get smaller mutual incoherence property(MIP)of recovery matrix and a slightly better MSE performance.3)According to the set of pilot pattern selected based on extension scheme 2(ES2),it is possible that the MIP of recovery matrix corresponding to pilot pattern on some antennas is still too large.So we propose the pilot optimization criteria as follows.When the pilot pattern of other antennas is determined,the MIP of recovery matrix of an antenna to be optimized is minimized.On the basis of the new criteria,an algorithm called layer optimization search scheme(LOSS)is proposed.As ompared to the TSS?MA and SSS?ES algorithms,the complexity of LOSS is lower under low cycle times,and its mean square error of channel estimation and the bit error rate of the system are better.