The Research On Direction Of Arrival Estimation Algorithm Of3G Smart Antenna

In recent years, as the development of global wireless communication, mobile communication occupies an increasing important position, which also has a higher requirement on communication technology requirements. Modern mobile communication systems not only require larger communication capacity, higher call quality, higher bandwidth efficiency, but also can achieve the goal in complex wireless environment and limited frequency band. In the development of wireless communication, if a huge leap wanted to be achieved, the key point is to overcome three issues:multipath fading, delay spread, multiple access interference. Smart antenna technology is introduced into the mobile communication under this background. It can format nulling point in the direction of interference signal, which can reduce interference largely, solves the three problems efficiently, become the efficient means of capacity improvement and interference reduction.For the key technology of smart antenna--direction of arrival, several classical algorithm of DOA are introduced, such as MUSIC algorithm, ESPRIT algorithm. The common characteristic of these algorithms are that received data is divided into two orthogonal subspaces by mathematical decomposition (eigenvalue decomposition, singular value decomposition, QR decomposition), one is signal subspace, the other is noise subspace. The basic idea of MUSIC algorithm is that signal subspace and noise subspace can be achieved by eigenvalue decomposition of the covariance matrix of received data, then the parameter of the signal can be get by using the orthogonal of the two subspace. ESPRIT algorithm doesn’t search peak by using the rotation invariance of covariance matrix. The experimental results show that the resolution of the classical algorithm has good performance; however, this method will fail in case of coherent signals.Because classical algorithm will fail in case of coherent sources, an improved singular-value decomposition algorithm is proposed. This algorithm reconstructed the correlation matrix by using largest eigenvectors of singular value decomposition, Lots of computer simulations prove that SVD algorithm overcome the loss of aperture, and has better stability and resolution capability in low SNR.