Research Of High-Efficient Spatial Spectrum Estimation Algorithm

Spatial Spectrum Estimation is an importance research filed of the array signal processing.The applications of spatial spectrum estimation occur in many military and national economy field,including radar,sonar,communications,navigate,earthquake,exploration and astronomy,etc.With extension of spatial spectrum estimation's application field and development of radar,communication,etc,a much higher accuracy,resolution and computing complexity of spatial spectrum estimation is demanded.For the sake of decreasing computing complexity,increasing computing speed and improving accuracy,we propose two algorithms: the one is real-valued beamspace root MUSIC algorithm,and another is spatial spectrum estimation algorithm based covariance matrix of modified data.The main contributions of this thesis are shown below:1.This analyzes the basic theories and classic algorithm of Spatial Spectrum Estimation.The array basic structure,mathematics model and the basic principle of measurement are first investigated.Then,two classical algorithm(including MUSIC algorithm and ESPRIT algorithm)and their improved algorithms are analyzed.Finally,simulation experiments are conducted,and performance of two kinds of algorithm for DOA Estimates are analyzed.2.In view of The computational complexity of the methods based on subspace is prohibitively expensive for real-time applications,we propose in this paper a real-valued beamspace root MUSIC(RV-B-Root MUSIC)algorithm,which exploited a subspace decomposition on either the real-part of array covariance matrix or the imaginary-part of array covariance matrix with real-valued computations.The Method used that the real-part of array covariance matrix shares the same null subspace with the imaginary-part of array covariance matrix,which collides with the intersection of the original noise subspace and its conjugate one.Furthermore,since the DOA estimates in RV-B-Root MUSIC are processing in beamspace,it produces a dimension reduction to processing in arrayspace.Therefore,the algorithm gives a further computational efficiency.Simulations illustrate that the proposed approach has a similar performance with the conventional MUSIC,Root MUSIC and Unitary Root MUSIC approaches while its complexity is lower than Root MUSIC,etc.3.Due to a small number of samples or low SNR,the methods based on estimating the signal and noise subspaces from the sample covariance matrix are exposed to performance breakdown.The performance breakdown is associated with the subspace leakage.So we propose a method that improves the performance by modifying the sample covariance matrix.The main idea of the proposed algorithm is to modify the sample data covariance matrix at the second step based on the DOA estimates obtained at the first step.The modified covariance matrix is obtained by deducting a scaled version of the estimated undesirable terms from the sample data covariance matrix.Because the proposed algorithm modified the sample data covariance matrix,reduced the amount of the subspace leakage,so the accuracy of DOA estimates has improved.Furthermore,the subspace leakage is theoretically derived and analysised.Numerical examples and simulation results shows that,for a small number of samples or low SNR,the proposed algorithm have higher of estimate accuracy to the other algorithms which are based on subspace.