A novel approach to the recovery of aperture distribution of phased arrays with single RF channel using neural networks

A novel approach to the problem of the recovery of complex aperture distribution of the phased array antenna with singe RF channel is presented in this paper. The computation of the amplitude and phase of each array element is developed as a mapping problem which can be modeled using a three-layer radial basis function neural network (RBFNN) trained with input/output pairs. RBFNN was used because it is characteristic of accurate approximation and good generalization, as well as robustness against interference and noise. The proposed approach has near optimal performance under various noise environments in terms of the error in amplitude and phase of recovered signals relative to other recovery methods.