An original theoretical approach for antenna optimization and design i
s introduced in this paper. This method consists of coupling between t
he FDTD method applied to the study of microstrip patch antennas and t
he genetic algorithm (GA) as an optimization tool. A neural network al
lows the computing time to be deceased by acting as a database, thus a
voiding multiple calls to the FDTD code. The principle of this approac
h is illustrated. (C) 1997 John Wiley & Sons, Inc.