Multiple-source angle-of-arrival estimation using neural-network-based smart antennas

The Neural Multiple Source Tracking (N-MUST) algorithm, which is based on an architecture of a family of radial basis function neural networks (RBFNN), is investigated for multiple source tracking with neural network-based adaptive array antennas. The N-MUST algorithm consists of an initial stage with a number of RBFNN's trained to detect the presence of the sources, while a second stage of networks is trained to estimate the exact locations of the sources. The field of view of the antenna array is divided into separate angular sectors, which are in turn assigned to a different pair of RBFNN's. When a network detects one or more sources in the first stage, the corresponding second stage networks are activated to perform the direction of arrival estimation step. No prior knowledge of the number of present sources is required. Simulation results are performed and experimental data is applied to the networks to investigate the required training criteria to achieve good generalization in the detection mode, with respect to the angular separations and relative SNR of the sources. The results show substantial reduction in the computational complexity of the network training compared to the single network approach.