Research On Key Techniques Of Direction Of Arrival Estimation For Passive Detection System

The direction of arrival (DOA) estimation of target signal is a crucial technique inpassive detection systems, which has a significant influence on the system performance andthe consequent combat dictations in the electronic warfare. The spatial spectrum estimationtechnique is employed in this dissertation to improve the DOA estimation performance inorder to achieve the high-precision of passive detection system and distinguish multipletargets arriving simultaneously. The dissertation focuses on the important problems of spatialspectrum estimation technique in practical applications, which involve the estimation ofsource number, the improvement of the angular accuracy and resolution of DOA estimation,the DOA estimation under non-uniform noises and the DOA estimation for non-circularsignals.The detailed works are described as below:The accurate estimation of source number is a prerequisite to the accurate directionfinding for spatial spectrum estimation algorithm. The source number is difficult to estimateaccurately in practical system, because of the effect of colored noises. Therefore, a newmethod for the estimation of source number based on the eigen subspace projection principleis proposed to solve this problem. For the signal subspace is orthogonal to the noise subspace,the value which the covariance matrix projects onto the post-processed eigen subspace isobtained, the source number can be estimated from the signal and noise’s respectivecontribution to the projection that is expressed by the projection’s variance. The effectivenessand the applicability of this method are verified by the simulation calculation andexperimental measurements.To improve the resolution and angular accuracy of spatial spectrum estimationalgorithms, two DOA estimation methods based on delay correlation function are proposed.Non-zero delay correlation function of received data contains the DOA information of signals,and has a restraining influence on the noise. Considering this principle, the rotationalinvariance of delay correlation functions among different antennas of the uniform linear array(ULA) has been found. An ESPRIT (Estimation of Signal Parameters via RotationalInvariance Techniques) algorithm based on delay correlation processing is proposed. For anarbitrary array, a MUSIC (Multiple Signal Classification) algorithm based on delaycorrelation processing is developed. To improve the real-time performance of the algorithm,the mutative scale chaos optimization algorithm is used to simplify the construction oftwo-dimensional spectral function and the spectral peak searching. The performances of these two algorithms are tested with the computer simulation.In actual direction finding system, the performance of classical MUSIC algorithm maybe deteriorated due to the effect of non-uniform noise received by the array. To solve thisproblem, a new method for DOA estimation is proposed under the condition of non-uniformnoise. Firstly, by processing the covariance matrix with three transformation matrixes, twoparts of data without and with noise are obtained respectively. Then, the noise covariancematrix is estimated with the two parts of data. Finally, the non-uniform noise is changed towhite noise with equivalent value by standardizing the received data with the estimated noisecovariance matrix. Tthe non-uniform noise background for DOA estimation turns into thewhite noise background, which could minimize the effect of non-uniform noise on theperformance of algorithm. Simulation results testify to the performance of proposed algorithmunder non-uniform noise background and experimental results in actual direction findingsystem prove the engineering practicability of the proposed algorithm.The non-circular characteristics of signals can be used to improve the performance ofdirection finding. According to the characteristics of non-circular signals, two DOAestimation algorithms are designed. The first method is the ESPRIT algorithm of non-circularsignals based on the matrix regrouping. Sub-matrixes that have the same amount of elementsin array with rotational invariance relationship are constructed through theconjugatereconstruction of received data. Then the influence of Gaussian white noise on thealgorithm is restrained by delaying correlation processing, the simulation results show that theperformance of the proposed algorithm are better than the ESPRIT algorithm. The second oneis a robust non-circular signal DOA estimation method on the basis of fourth-order cumulant.Fourth-order cumulant could extend array aperture and suppress Gaussian noise. In view ofthese characteristics, two fourth-order cumulant matrixes are constructed, and the DOAestimation is completed on the basis of the rotational invariance of the two matrixes. Then, therobustness of this algorithm with the channel amplitude and phase error model is analyzedand the conclusions are as follows: if only any two of the reception channels arefairlyconsistent, the accurate DOA estimation could be acquired without error correction. Theangular accuracy and the resolution are improved obviously from predictions, and theproposed algorithm is robust to amplitude-phase error.