Estimating two-dimensional angles of arrival by neural networks

A two-layer neural network to estimate two-dimensional (2-D) angles of arrival (AOA) for multiple narrowband radiating sources is proposed in this paper. To avoid performing a 2-D search, two one-dimensional (1-D) neural estimators are applied in the first layer to perform 1-D searches using rows and columns of the rectangular array. To match the data obtained, a 2-D verification is then carried out in the second layer. Therefore, this approach saves a great number of neurons and computations in programming the network. Simulation results are given to illustrate the good performance of this method