6.1 INTRODUCTION

Neural networks, also called artificial neural networks (ANNs), are information processing systems with their design inspired by the studies of the ability of the human brain to learn from observations and to generalize by abstraction. In recent years, ANNs have been recognized as a useful tool for modeling and design optimization problems in RF and microwave computer-aided design [1]. Neural networks can be trained using measured or simulated microwave data, such as EM data for passive components, or device physics data for active devices. The trained neural networks become models of microwave devices/circuits and can be used in place of CPU-intensive EM/physics simulations to significantly speed up circuit/system design, while maintaining EM/physics-level accuracies. ANN-based approaches have been applied to a variety of microwave CAD problems such as modeling and optimization of high-speed VLSI interconnects, EM based modeling and optimization, device modeling, yield optimization, and circuit synthesis and more [2 9].

The motivation for using ANNs for behavioral modeling comes from their learning ability, the universal approximation property, and computational speed. ANNs can be trained to learn the nonlinear behavior of a microwave circuit from measured or simulated input-output data, avoiding otherwise manual effort of creating equivalent circuit topology. Similarly ANN avoids the need of availability of original circuit equations during model building. The universal approximation property of ANN provides a theoretical basis for the model in representing the full analog solutions of the circuit, overcoming the accuracy...