Application of neural networks and genetic algorithms to RF and wireless packaging

Conventional approaches to design of RF and wireless packages, which require both simulation and optimization of a large number of passive and active circuits, is a very time consuming process; hence, the CPU time required to carry out virtual prototyping of these circuits can be unacceptably long. This paper presents a neural-network-based approach for design of these packages in a numerically efficient manner.