Research On Fast Algorithms Of Two-Dimensional DOA Estimation

Compared with traditional direction finding technology, spatial spectrum estimation technology has higher direction finding accuracy and stronger lateral resolving power, it has been widely used in civil communication systems and electronic reconnaissance and other fields. Based on subspace two-dimensional DOA (Direction-of-Arrival, DOA) estimation algorithm in spatial spectrum estimation technology is a typical DOA estimation algorithm, the proposed and research of this kind of algorithm promoted the development of the spatial spectrum estimation technology, but in the two-dimensional DOA estimation or more higher dimensional DOA estiamtion, this kind of algorithm has big computation complexity and poor real-time performance, it is difficult to implement in the actual engineering application. This paper does research on the fast DOA algorithm under studying the problem of two-dimensional DOA estimation algorithms complexity, the main work is as follows:Fisrtly, according to the conventional signal subspace algorithm can not estimation the DOA of coherent signals and the heavy computation complexity of two-dimensional MUSIC algorithm, a fractional dimension reduction MUSIC algorithm based on the Forward/Backward spatial smoothing Technique was proposed. This algorithm de-correlates the coherent signals through the Forward/Backward spatial smoothing Technique, and then used the method of fractional dimension reduction obtained two-dimensional DOA of target signals, implemented the conventional signal subspace algorithm for coherent signal. This algorithm avoids two-dimensional spatial spectrum search and reduces the computational complexity.Secondly, according to the traditional ESPRIT algorithm in the two-dimensional DOA estimation algorithms was conducted in complex domain, and the fractal dimension calculation and angle matching problem, then the computation complexity is big, so a kind of unitary ESPRIT algorithm by Cyclic Maximization Dimension Reduction based on the rectangular array was proposed. By the principle of unitary transformation, constructed a mathematical model of the unitary matrix for array processing and transformed complex domain operation to the real number domain, then constructed the dimension reduction matrix by the method of Cyclic Maximization Dimension Reduction for the blocked dimension data model in the real number domain. This algorithm obtains two-dimensional DOA without angle matching and improves the real-time performance.Finally, for conventional two-dimensional MUSIC algorithm and conventional two-dimensional ESPRIT algorithm need eigenvalue decomposition, spectral peak searching and exist the problem of angle matching computation complexity, and combining the PM algorithm without eigenvalue decomposition, this paper puts forward a fast parameter matching for 2D PM algorithm. The algorithm firstly using PM algorithm to deal with parallel array data, then uses a quick matching method to obtain the Propagator, finally obtaining the two-dimensional DOA of target signals. The algorithm just does once eigenvalue decomposition and has good real-time performance.