A neural network approach to the design of the high frequency devices

Designing the high-frequency devices by using artificial neural networks (ANN) emphasize the ANN propensity for storing knowledge, for generalizing patterns in data and for offering computational efficiency. Two neural models are discussed in this paper. The model of a band-pass forward-coupled filter uses the ARMA coefficients, which describe the FDTD signal at the output port of the filter in time-domain. On the contrary, the second model, which was developed for a tunable planar resonator, uses directly the device response in frequency-domain.