Research On Two-dimensional Spatial Spectrum Estimation Algorithm For L-shaped Array And Its FPGA Implementation

Spatial spectrum estimation,also known as Direction of Arrival(DOA)estimation,is a key issue in array parameter estimation and its applications have also been extended to many military and civilian realms such as sonar and communications.L-shaped array is commonly used for two-dimensional DOA estimation and it has simple structure and is easy to implement in practice.Therefore the topic of two-dimensional DOA estimation algorithms with L-shaped array in this paper has great theoretical significance and practical value.The work of this paper is listed as follows:(1)About the issue of two-dimensional DOA estimation for non-uniform L-shaped array,three algorithms including Multiple Signal Classification(MUSIC)-like,CAPON-like and spectral peak search Propagator Method(PM)-like are proposed.In these three algorithms,a successive one-dimensional spectral peak search instead of two-dimensional peak search is carried on to realize joint two-dimensional estimation and avoid huge computational complexity,meanwhile estimation performance is very close to it with two-dimensional DOA algorithm.(2)About the issue of two-dimensional DOA estimation for coprime L-shaped array,two algorithms with rotational invariance techniques and two algorithms with spectral peak search are investigated.The former includes Estimating Signal Parameters via Rotational Invariance Techniques(ESPRIT)algorithm and rotational invariance PM algorithm,these two algorithms use coprime property to eliminate the ambiguity angles estimated with two subarrays,respectively.Compared with uniform L-shaped array,coprime L-shaped array can achieve better estimation performance.The latter includes one-dimensional MUSIC algorithm and one-dimensional CAPON algorithm,these algorithms carry on one-dimensional spectral peak search with arrays on the X-axis and on the Y-axis of coprime L-shaped array,respectively,and achieve extra pair match with the estimated angle information.The algorithms with coprime L-shaped array have the merits of lower computational complexity compared to corresponding two-dimensional DOA estimation algorithms and obtain higher DOA estimation accuracy compared to it with uniform L-shaped array.(3)Two two-dimensional DOA estimation algorithms based on one-dimensional MUSIC and one-dimensional General Estimating Signal Parameters via Rotational Invariance Techniques(GESPRIT)for adjoined coprime L-shaped array are proposed.The algorithms carry on one-dimensional spectral peak search with arrays on the X-axis and on the Y-axis of adjoined coprime L-shaped array,respectively,and achieve extra pair match with the estimated angle information.The algorithms with adjoined coprime L-shaped array have lower computational complexity compared to corresponding two-dimensional DOA estimation algorithms and achieve higher DOA estimation accuracy compared to it with uniform L-shaped array.(4)Two-dimensional DOA estimation algorithm based on CAPON-like for non-uniform L-shaped array is realized with Verilog-HDL on PFGA.The main work includes FPGA design for covariance matrix construction,matrix inversion,spectral peak search modules.The Modelsim simulation results of above modules are given,and the comparison between the simulation results on FPGA and on MATLAB are conducted.Test results verify the effectiveness of the FPGA design with two-dimensional spatial spectrum estimation algorithm based on CAPON-like for non-uniform L-shaped array.