Improved supervision of NN's AOA determination with limited 1-D phased-array training data

A sixteen-channel linear phased array radar is calibrated by means of Artificial Neural Network (ANN). Limited data are used to train the various layers of the network, for Angle of Arrival (AOA) determinacy. This is designed to increase speed of operation and to prevent over fitting. The designed simulator generates time-series data sets representing continuous span of AOA. The generated datasets are used in multilayer feedforward NNs to find the phase, which is then used as inputs to Spiking Neural Network (SNN) o determine AOA. SNNs are used for hardware implementation feasibility. The designed simulator improved the accuracy of AOA determination, and opens the door for a wide variety of radar classification applications.