Wideband direction of arrival estimation using a backward nonnegative sparse method

In the field of direction-of-arrival (DOA) estimation for wideband signals, traditional methods often face a significant challenge as a result of spatial aliasing. These methods, which treat signals at different frequencies separately, may fail to detect weak signals due to the interference caused by spatial aliasing. To overcome this challenge, a wideband DOA estimation method, named the backward nonnegative sparse (BNNS) method, is introduced, which processes signals jointly across multiple frequencies. BNNS leverages a characteristic that the direction of the mainlobe remains constant across frequencies, and the grating lobe varies with frequency. By selectively retaining only those entries corresponding to the mainlobe in dictionaries, BNNS effectively reduces interference from grating lobes. Simulation results demonstrate the capability of BNNS to suppress grating lobes and detect weak signals under conditions of low signal-to-noise ratio and signal-to-interference ratio. Additionally, its ability to enhance peak-background contrast is demonstrated by the Swellex96 experimental data.