Basis Study On The Application Of Super-resolution Spatial Spectrum Estimation Technique

Super-resolution spatial spectrum estimation is very important in the field of the array signal processing, and it has received considerable attention because of its better parameter estimation capability and broad application space. Spatial spectrum estimation technique has played a significant role in many fields, such as radar, communication, sonar, seismography, reconnaissance, radio astronomy, biomedicine engineering and so on. Through the research of several decades, spatial spectrum estimation technique has obtained much great progresses in the theory aspect. However, there are a few problems to be solved in the application. In this thesis, the key problems about the spatial spectrum estimation application are been researched deep. The main contributions of the thesis can be summarized as follows:In order to solve the DOAs estimation of coherent souces, a novel decorrelation method is proposed fully using the eigenvector corresponding to the maximum eigenvalue obtained by the array received data covariance matrix. This method constructs a new matrix whose rank is equal to the number of coherent sources.The presented method avoids the loss of the array aperture and the constructed matrix has Toeplitz property. Compared with the spatial smoothing algorithms and the other algorithms also using the eigenvector, the proposed method has better decorrelation capability and has better estimation performance in the presence of low SNR. The technique provides a novel idea about the decorrelation method based on the eigenvector or the technique without the loss of array aperture.According to the different reflection surfaces, the specular reflection model and the diffuse reflection model of the meter-radar height-finding is researched. Moreover, the effective reflecton region of the meter-wave radar is analyzed. The direct signal and the indirect signal have strong coherence, small spatial spacing, and can not be differentiated for low SNR situations. Therefore, we propose that the conventional array model (called model one) is modified. In the modified model (called model two), the steering vector is the linear combination of the steering vectors of the direct part and the indirect part. When the direct signal and indirect signal can not be differentiated, the model two can give the rough estimation of the direct angle, i.e, the height of the object. In the meantime, the estimation Root Mean Square (RMS) error of the model two is smaller than that of the model one in the same simulation conditions. The real data processing results further show the effectiveness of the model two in the meter-wave radar height-finding applications.Based on the spatial smoothing and the propagator operator concept, a novel computationally effective method is presented in order to further reduce the computational load of the eigen-subspace algorithms in the meter-wave radar height-finding applications. The method fully uses the information included in the received data covariance matrix to construct new matrices. Finally, the noise subspace is estimated based on the propagator operator idea. The proposed method avoids the other decorrelation processing and the eigendecomposition. Moreover, compared with the forward-backward spatial smoothing algorithm, the proposed method has less computational load in the case of the same estimation performance. Computer simulations and the real data processing results demonstrate that the proposed method is effective in the meter-wave radar height-finding applications.For the mutual coupling calibration of the L-shaped array and the cross array, the self-calibration algorithms for the L-shaped array and the cross array are proposed respectively using the properties of the mutual matrices between sensors in each subarray and those between subarrays. The proposed method calibrates the mutual coupling between subarrays besides the mutual coupling between sensors in each subarray. The method decouples the the 2-D angles and mutual coupling coefficients, and has high estimation precision. The method requires neither the multidimensional nonlinear search nor iterative computation, and then the computational load is greatly reduced. The effective proof for asymptotic equivalence is addressed. A necessary condition for DOA identifiability is discussed, and the relevant CRB is derived also.The errors calibration problem for the L-shaped array and the cross array is researched using the simple assistant array. When the channel disaccord, mutual coupling, sensor position error are all exit, the DOAs of signals and the error steering vector can be directly estimated using the assistant array. The computatioanl load is small. When the channel disaccord and the mutual coupling are both exit, a novel method is proposed. Based on the analysis of error characteristic for L-shaped array and cross array, the complex nonlinear multi-dimension search problem is transformed into simple iteration computation, and the computation burden is greatly decreased. Moreover, it can not converge at the local extremum. Finallly, the block diagram is given to summarize the calibration methods under different situations for the L-shaped array and the cross array.