SUBBANDING ESPRIT BY DAUBECHIES WAVELETS
Yanbo Xue, Jinkuan Wang, Zhigang Liu, Z. Liu · 2007
A novel subbanding ESPRIT method by Daubechies wavelets is suggested in this paper. The proposed approach filters the signals into different subbands with reduced wavelet packet filters and then applies standard ESPRIT to the subband signals. The rotational invariance in the subarrays is proven and the mapping method from the subband to the fullband is also formulated. Experimental results show that in sense of root mean square error (RMSE) reduction and output gain enhancement, the suggested method outperforms ESPRIT in scenarios of highly correlated signals and/or low signalto-noise ratio (SNR). The RMSE and output gain versus Daubechies wavelets with different lengths also show the impacts of choice of Daubechies wavelets on the performance of the new method.