Pilot Pattern Design In MIMO-OFDM System Based On Compressive Sensing

In order to meet the high-quality transmission requirements of massive data in future wireless communication systems,multiple-input multiple-output(Multiple-Input Multiple-Output,MIMO)technology is considered to be the key technology of next-generation communication because of its high spectrum efficiency and spatial multiplexing.For MIMO systems,beamforming and signal detection require accurate channel state information.However,with the increase in the number of base station antennas,how to use less pilot resources to accurately estimate multi-dimensional channels is an urgent problem for MIMO systems.Therefore,the research on the pilot design scheme in MIMO system is of great significance to the channel estimation performance of communication system.In the broadband communication system,the wireless channel is sparse,this is the channel delay spread is large,but the number of significant paths is small.The traditional channel estimation does not take into account the sparsity of the wireless channel,and the pilot density is determined by the two-dimensional Nyquist sampling rate,which consumes more pilot overhead.Applying compressed sensing theory(Compressed Sensing,CS)to sparse channel estimation in MIMO-OFDM systems can significantly reduce the number of pilots and have good channel estimation performance.The traditional channel estimation thinks that the pilot placement method with equal interval is optimal,but this conclusion is not true in the channel estimation based on CS.Therefore,it is of great significance to study the pilot pattern design in CS-based channel estimation.This paper focuses on the channel estimation and pilot pattern design based on CS in OFDM system and MIMO-OFDM system.The specific work is carried out as follows:(1)Under the slow fading scenario of broadband wireless communication,in OFDM system,the compressed sensing-based traditional channel estimation has the problems of low accuracy of channel estimation and low channel estimation efficiency,a variable proportion variable step size stagewise adaptive matching pursuit(VP-VSSt AMP)algorithm is proposed.Firstly,a dynamic threshold parameter is introduced in the initial stage of atom selection,and the quality of atoms is strictly controlled in different iteration stages to effectively improve the estimation accuracy.secondly,combined with the idea of??variable step size,which can quickly approximate the true sparsity of the channel.The simulation results show that,compared with the existing adaptive reconstruction algorithms,the proposed algorithm in this paper can gain 1?2dB improvement in the channel mean square error.(2)In SISO-OFDM system,in order to improve the performance of OFDM sparse channel estimation based on compressed sensing,the sum of the distance cubes between the elements of the cross-correlation matrix is ??proposed as a new pilot design criterion,which can more accurately evaluate the overall cross-correlation of the sampling matrix;In addition,in view of the slow efficiency of traditional pilot search methods,a tree-based sequential replacement pilot interval search algorithm(TSS,Tree-based Search Schemes)is proposed,which avoids pilots with too large and small intervals from being selected into the candidate set,and optimizes pilot search efficiency.The simulation results show that,compared with the existing pilot design schemes,the proposed criterion can effectively gain 4dB improvement in the channel mean square error.Furthermore,the newly proposed pilot search method has lower complexity and better convergence performance.(3)In MIMO-OFDM system,aim at the problems of low efficiency and poor estimation quality of joint design algorithm for multi-antenna pilot patterns,a pilot pattern design(SATSS-SM)method for MIMO-OFDM channel estimation based on adaptive branch tree is proposed.Firstly,the sampling matrix cross-correlation model is constructed,and the minimum cross-correlation distance cube sum criterion is proposed.secondly,combined with the shift mechanism theory,the SATSS-SM algorithm is used to search the pilot and quickly converge to the global minimum cross-correlation value to achieve low-complexity sparse channel estimation.Simulation results show that,compared with the existing pilot optimization schemes,the pilot pattern designed by the proposed algorithm can reduce the channel estimation mean square error of about 2?4dB.