A wideband MUSIC algorithm using an improved empirical wavelet transform

Abstract The traditional multiple signal classification (MUSIC) algorithm is only suitable for narrowband array signals, however for wideband signals, sub-band division is required, most of which are based on preset filters or time–frequency analysis. These methods usually require manual selection of the parameters of the filters, basis functions, etc., according to different circumstances, which leads to reduced applicability. This paper proposes a wideband MUSIC algorithm using an improved empirical wavelet transform (EWT), which resolves the sub-band division issue by adaptively constructing a series of empirical wavelet functions to decompose the array received signals. In order to adapt EWT to array signal processing, the spectral division method of EWT is improved by utilizing the spectral mean value and order statistical filter of the array element signals, enabling EWT to uniformly decompose the received signal of each array element. Theoretical research and simulation analysis demonstrate that the wideband MUSIC algorithm using the improved EWT enables effective estimation of arrival angles of wideband signals. Due to the utilization of a more refined adaptive uniform sub-band division technique for array signal, the proposed algorithm has the lowest angle estimation error compared to the MUSIC algorithms based on short-time Fourier transform, auditory filter, and continuous wavelet transform.