Reduced wavelet packet and its application to beamspace ESPRIT algorithm in correlated signals

Subbanding method is one of the most successful applications of wavelet theory in direction-of-arrival (DOA) estimation problem. This paper illustrates the use of reduced wavelet packet filters for ESPRIT method, namely RWP-ESPRIT. Though we use the modified wavelet packet filters, the rotational invariance of the filtered data matrix holds. The proposed approach filters the signals into different subbands and then applies standard ESPRIT to the subband signals. 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, RWP-ESPRIT outperforms ESPRIT in scenarios of highly correlated signals and/or low signal-to-noise ratio (SNR).