Neural network model for 2D DOA estimation of two coherent sources

摘要

This paper presents an artificial neural network (ANN)-based model for two-dimensional direction of arrival (DOA) estimation of two coherent sources. The proposed model is composed of two neural networks, one for each radiating source. Datasets for training and testing of the neural model are formed assuming narrowband signal model and a 4 × 4 rectangular antenna array at the receiver. Unlike MUSIC algorithm with the spatial smoothing pre-processing (SSP) scheme, the ANN model is available to separate two closely spaced coherent sources. Comparison results are presented to show advantages of the neural model both in terms of accuracy and speed of calculation.