A new neural network DOA estimation technique based on subarray beamforming

A new neural network DOA estimation technique based on subarray beamforming is proposed. The proposed technique improves previously reported modified neural multiple source tracking algorithm (MN-MUST). MN-MUST algorithm has three stages, the new technique replaces the first two stages of it with a new beamforming stage based on subarrays. The whole direction of arrival angular region is divided into subsectors as in MN-MUST. Detection and filtering stages are replaced by subarray beam forming stage. Subarray beamforming stage filters out the signals outside the sector of interest. Beamforming is not the scope of this study however, the phase differences between virtual subarrays are used in DOA estimation stage. The proposed algorithm dramatically reduces process of MN-MUST algorithm, thus improves the accuracy and speed.