The Improvement Of Resolution Of Spatial Spectrum Estimation Via Virtual Aperture Expansion

The spatial spectrum estimation algorithm is an important area in array signal processing, and it's applied to many fields, such as radar, communication and seismic prospecting. Though the theory and technology of the spatial spectrum estimation is well developed at the present, there are still many problems left to be solved, such as the resolution, the efficiency and the cost to implementation of the algorithm. This thesis addresses above problems to improve the performance of spatial spectrum estimation.This thesis start with the introduction of the basic principles of some widely used spatial spectrum estimation algorithm. The similarities and difference among them are analyzed and compared through simulation. The similarities of MUSIC algorithm and ESPRIT algorithm are that they are both belong to the subspace algorithm, and they realize super-resolution; the difference is that MUSIC algorithm uses the orthogonal property of noise subspace of covariance matrix, while ESPIRT algorithm uses rotation invariant of signal subspace of covariance matrix.Based on the above introduction, a spatial smoothing technique of coherent sources is proposed, which is suitable for uniform linear array antenna. The spatial smoothing technique improves the de-coherence ability at the expense of sacrificing the aperture of linear array antenna. In addition, the application of Array Signal Processing in mobile communication is introduced. The thesis also introduces the principle of SDMA and a beam forming algorithm of CDMA system, which is different from matrix factorization algorithm. The simulation results illustrate that new algorithm has low computation complexity and is easy to be implemented.The main contribution of this thesis is the resolution improve algorithm of spatial spectrum estimation, which is an important parameter-based estimation algorithm. The resolution of improving algorithm focuses on aperture of the array antenna, SNR and snapshots. New algorithm based on virtual extension is proposed in this thesis. Rectangular window is applied to the data series to eliminate the false peaks effect which is caused by virtual arrays. The flowchart of the algorithm is also given for understanding. Simulation results verify the effectiveness of the proposed algorithm.