Research Of Technology About Super-resolution DOA Estimation For Coherent Sources

Super-resolution DOA (Direction-Of-Arrival) estimation is a research hotspot of spatial spectrum estimation. It has received considerable attention because of its superior estimation performance and wide application. In the past, most of super-resolution DOA estimation techniques are performed in an ideal environment. In real environments, however, due to the presence of multipath effects and (or) undesired environmental noise, the performance of most of classic super-resolution DOA estimation algorithms declines rapidly or even turns to be invalid. In this thesis, some problems about the super-resolution DOA estimation of coherent sources (or multipath signals) have been studied. The main subjects are concluded as follows:1. The classic algorithms of the super-resolution DOA estimation for coherent sources are classified and compared, and the decorrelation principle of the spatial smoothing algorithm is interpreted from the changes of the correlation coefficient and the phase difference between coherent sources respectively. The derivation process shows:the decorrelation capability of the classic algorithms are achieved at the expense of array effective aperture or through a complex multidimensional nonlinear fitting; and in the physical sense, the spatial smoothing algorithm decorrelates the coherent sources by changing the phase difference between the received signals on each sensor element at different snapshot.2. In order to reduce the computational complexity by avoiding the eigenvalue decomposition (EVD) of sample covariance matrix, a method for DOA estimation of coherent sources, called cross-correlation vector unitary matrix projection (UPM) algorithm, is proposed based on the spatial smoothing and projection matrix concept. This method uses two cross-correlation vectors to construct a pseudo covariance matrix first, and then the signal subspace is estimated using the linear operation of the pseudo covariance matrix instead of the traditional EVD. Simulation results show that the UPM algorithm has good performance and is more suitable for the lower SNR environment. 3. In the presence of colored Gaussian noise with unknown covariance, considering that most of the existing four-order-cumulant-based DOA estimation algorithms use not all the information of the four order cumulant, a method based on the joint diagonalization of the four order cumulant matrices is proposed. This method is a modification of the EVESPA algorithm, and it improves the overall performance of the former algorithm.