Research On Direction Of Arrival Algorithms In Coherent Signals

Direction of arrival(DOA) estimation in array signal processing has been a popular research direction. Currently, most DOA estimation algorithms are based on the assumption that the spatial signals are independent or uncorrelated. However, in practical applications, the coherent signals are widespread. As the array utilization rate and poor performance in low signal to noise ratio of conventional coherent signal DOA algorithm, we presents some algorithms based on subspace theory for coherent signals.In recent years, the compressed sensing and sparse reconstruction theory have attracted wide attention from domestic and foreign scholars and they have been applied to DOA estimation algorithm. Because of its natural advantages in dealing with coherent signals,the thesis gives some more superior algorithms based on sparse reconstruction theory.The full text of work can be summarized as follows:(1) This paper presents a new method which can estimate DOAs of independent signals and each coherent group separately. Based on a conclusion that linear combination of steering vectors of uniform linear array does not produce a new steering vector, multiple signal classification(MUSIC) algorithm is applied to estimate DOAs of independent sources directly. Because of the characteristics of Toeplitz in the array covariance of independent signals and noise, differencing operation is applied to remove the influence of independent signals and noise. Then a virtual array manifold matrix which does not have cross-term effect is constructed. Each column of the matrix contains the information of DOA of one coherent group. At last, an improved vector SVD method is adopted to estimate DOAs of each coherent group.(2) When the number of coherent groups is substantial large, the previous algorithm require multiple SVD operator. On the other hand, all operations are based on the complex field. Thus the amount of calculation is quite large. By considering each column of the virtual array manifold matrix as a single snapshot virtual data, we use an aperture improved ESPRIT method based on real value to reduce calculation. Due to the differencing approach has removed the pollution of noise, the proposed method overcomes the instability in low SNR, which is the drawbacks of most real value algorithms.(3) As the drawback that traditional subspace algorithms need signal numbers asprior number, this thesis gives a new method which utilizes orthogonal matching pursuit(OMP) algorithm to estimate DOAs of each coherent group. Because it does not take advantage of the orthogonality of the subspace, the proposed method can improve the array aperture furtherly. Because of the differencing approach and operators in eliminating cross-term effects, proposed algorithm overcomes the disadvantage that OMP algorithm performs astatically in low SNR.(4) Firstly, the basic theory of DOA algorithm which based on the beam space or on the array element space is discussed. The idea of sparse reconstruction is extended to beam space. We present a novel sparse reconstruction DOA estimation method based on covariance in beam space.By setting a reasonable beamforming matrix, the received data of array element space in higher dimension is mapped to a beamspace. Finally, the basis pursuit(BP) algorithm is adopted to estimate DOAs.