The Research On Direction Of Arrival Estimation Of Wideband Signals

DOA(Direction of Arrival) estimation is a significant problem in the signal processing field. It has been applied on many fields, such as radar, sonar, wireless communication. Compared with the narrowband signals, wideband signals carry more information and have the advantage of anti-interference, small confused background correlation, high resolution and so on. Hence it has major significance in research on wideband signal DOA estimation algorithm. The main research work in this thesis includes:Firstly, the basic theory of spatial spectrum estimation has been described. It mainly involves the simplified array, signal model, wideband signal definitions, mathematical models, the performance parameters and important influence factors in DOA estimation.Secondly, the comparative study about wideband source DOA estimation method based on signal subspace fusion has been introduced, including incoherent signal subspace method(ISSM) and coherent signal subspace method(CSSM). Here, several principles of constructing focusing matrix in CSSM algorithm have been depicteded in detail. At last, the performance of these focusing methods is compared via computer simulation.Finally, we study the sparse signal representation based algorithms(l1-SVD and JLZA algorithm) and the performance of these two algorithms is analyzed via simulations. l1-SVD could not jointly use all the subband information to get DOA and the JLZA only utilize the energy in all frequency with regularization paremater resulting in a degration in performance. So we study the subband information fusion(SIF) method to solve the above problems. Furthermore, in order to solve the high computational complexity and inaccurate estimation of the signal altitude which appears in the SIF, the SIF algorithm based on the aperture expansion is proposed. This new algorithm improves the estimate accuracy of signal energy and increases the number of sources for estimated. Moreover, it decreases the computational complexity and sustain in low SNR. Last, the simulation results prove the above conclusions.