An Approach for Higher-Order Cyclostationarity based DOA Estimation of Mult ipath Signals without Eigendecomposition and Spatial Smoothing

H. Jiang, S.X. Wang, and H.J. Lu (PRC)

Keywords

Smart antennas, direction-of-arrival estimation, higher order cyclostationarity, multipath, subspace

Abstract

However, most of the subspace-based methods require either the eigenvalue decomposition (EVD) or the singular value decomposition (SVD) of a data matrix to estimate the signal or noise subspace. Moreover, the spatial smoothing [11] is needed when the signals are coherent. Those computations are cumbersome in the presence of moving sources because the system requires incessantly update the signal or noise subspace in order to acquire new data and delete old data in time [6][7]. In this paper, we propose a higher order cyclostationarity based direction of arrival (DOA) estimation approach for multipath signals without eigendecomposition and spatial smoothing. In the approach, by using a combination of higher order cyclic statistics (HOCS) and a forward backward subarray scheme, the coherency of multiple signals of interest (SOIs) are de-correlated and the performance of signal selectivity and interference suppression is improved. To reduce the computation load in order to adapt to tracking moving sources, the need for computation of the eigenvalue decomposition (EVD) or the singular value decomposition (SVD) is avoided by linear operations based on fourth-order cyclic cumulant matrix. Base on above analysis, and considering the applicability in mobile communications with multipath environment, we modify the SOCS-based method in [7] and propose a new higher order cyclostationarity based DOA estimation approach for multipath signals without eigendecomposition and spatial smoothing. In the approach, by using a combination of HOCS and a forward-backward subarray scheme, the coherency of multiple SOIs is de-correlated and the performance of signal selectivity and noise/interference suppression is improved compared with both the higher order statistic (HOS) method and the SOCS method in [7]. To reduce the computation load in order to adapt to tracking moving sources, the need for computation of the EVD or SVD is avoided by estimating the noise subspace from the fourth order cyclic cumulant matrix and based on linear independence between rows of array response matrix.

Important Links:



Go Back