Research On DFT-Based Multi Source DOA Estimation In Uniform Linear Array

Direction-of-arrival(DOA)estimation technology of incident signal based on sensor array is a popular research field in array signal processing.Various algorithms for accurately estimating DOA have been proposed,including parametric or nonparametric methods.The nonparametric methods require no prior knowledge of the signal,but the methods may not provide the optimum estimation performance.The parametric methods rely on the array signal model and require the number of targets to be known,thus increasing the computational cost,but this method can estimate DOA with high resolution and provide the optimum estimation performance,so it is widely used.Parametric methods based on discrete Fourier transform(DFT)can be easily computed by fast Fourier transform(FFT),which alleviates the computational complexity of DOA estimation problems.However,due to the problems of spectral leakage and fence effect in DFT,when the directions of arrival of adjacent signals are close,the estimation error will be large.Therefore,this paper studies the DOA estimation technology based on DFT under the uniform linear array(ULA)model,aiming to eliminate the influence of the fence effect and spectral leakage,and improve the estimation accuracy and anti-noise performance of the algorithm.Firstly,a direction-of-arrival(DOA)estimation method for multiple targets based on Ip DFT transform in the background of additive white Gaussian noise is proposed.In this algorithm,each DFT transform spectrum is rewritten as a polynomial multiplication form.Based on this,a multivariate linear equation system about the incident angle is established by using each DFT spectral line,so as to calculate the actual direction-of-arrival.Moreover,the complex-valued least squares framework(CLS)is adopted to extend the proposed method for multiple snapshots.The algorithm uses spectral leakage and side lobe interference as useful information,and establishes a linear equation system,so the DOA can be accurately estimated.Theoretical derivation and simulation results show that the algorithm has strong anti-noise interference ability and high estimation accuracy,and the mean square error of the estimation results is close to CRLB.In order to further improve the performance of the DOA estimation algorithm,an improved multi-target DOA estimation algorithm based on padding-zero DFT in uniform linear array(ULA)scene is proposed.For the received signal containing H incident target sources,the algorithm uses 3H padding-zero DFT bins to design a linear equation system,and then calculates the actual direction-of-arrival(DOA).In the estimation process,the algorithm considers the influence of all the target components,and then estimates each azimuth angle accurately and unbiasedly.For a uniform linear array of N sensors,DOA estimation of N/ 3 targets can be supported by the algorithm at the same time.Theoretical derivation and simulation analysis show that the algorithm can effectively resist the interference of white noise and provide high-precision angle of arrival estimation results.Compared with the previously proposed estimation method based on Ip DFT,this improved algorithm can improve about 5d B estimation performance.As a result,the improved algorithm has higher estimation accuracy and better anti-noise performance.