Research And FPGA Implementation Of DOA Estimation For Coherent Sources Based On Nonuniform Array

In recent decades,a lot of achievements have been made in direction of arrival(DOA)research,which have been widely used in engineering.In practical application,the non-uniform array has larger aperture and more flexible array spacing setting,so it has gradually become a research hotspot.However,the existing subspace DOA estimation algorithms are not suitable for coherent sources.Since these algorithms have a lot of non-linear operation and high implementation complexity,they are difficult to meet the needs of actual engineering.Therefore,this paper mainly studies the DOA estimation algorithm with its implementation scheme at field programmable gate array(FPGA)hardware.The main contents are as follows.Firstly,the signal model of non-uniform array is given,four kinds of non-uniform array structures are introduced,and the problem of angle ambiguity in non-uniform array is analyzed.Then,the symmetric non-uniform array with limited minimum element spacing is studied.There are multiple sub-arrays with the same structure in the array.The spatial smoothing method can be used for decoherence processing,and then the multiple signal classification(MUSIC)algorithm is used to implement the DOA estimation.The effectiveness and performance of the algorithm are verified by simulation experiments.Then,an array expansion algorithm based on Khatri-Rao(KR)product is studied for the asymmetric non-uniform array.The DOA estimation performance of KR-MUSIC algorithm for incoherent sources is verified by simulation.A matrix reconstruction algorithm for decoherence is proposed to solve the problem in which KR-MUSIC algorithm is invalid when the coherent sources received.The virtual array translation method is used to obtain the array received data under different reference array elements,and then the rank of covariance matrix is restored by matrix reconstruction.Finally,the KR-MUSIC algorithm is used to obtain the DOA estimation.The simulation results show the algorithm can effectively achieve DOA estimation of coherent sources with non-uniform array.Finally,the FPGA implementation of SS-MUSIC algorithm for symmetric non-uniform linear array is studied.According to the algorithm steps,the full rank covariance matrix calculation modules,eigenvalue decomposition modules,noise subspace estimation modules and spectral peak search modules are designed respectively.The full-rank covariance matrix module realizes the calculation of the covariance matrix,spatial smooth decoherence processing and real number processing.The eigenvalue decomposition module is implemented by parallel Jacobi rotation,and a 3-level cleaning mode is used to reduce the time consumption of the module.The noise subspace module uses thresholds to determine the number of sources.The noise subspace is determined according to the number of sources.According to the input source frequency,the guide vector is obtained,the pseudo spectrum is constructed,and the DOA estimation angle is obtained by peak search.The time consuming is reduced by dividing the search angle range.The final test shows that the entire algorithm can be implemented in 5.64μs.