Antenna design by genetic algorithms

In many military and civilian situations, the placement of an antenna is restricted by available space and surrounding structures. Additionally, antenna design engineers are continuously considering various methods that can optimize an antenna design for criteria, e.g. radiation efficiency, directionality, or S parameters. Once the performance criteria are specified, it becomes necessary to find the best solution(s) that satisfy the design constraints. For example, one might try to place an antenna on a vehicle in such a manner as to radiate energy in one direction while causing minimal interference to other antennas nearby. Optimal antenna placement of an antenna on a platform is a challenge to current approaches to electromagnetic modeling and numerical optimization. Genetic algorithms (GAs) and other "global" search algorithms have the advantage of being able to search cost surfaces with many local minima. They can also handle a large number of variables and constraints. Some current limiting factors include the optimization mixed variable types (continuous, discrete, and categorical) and the slow speed of calculating the cost function. This proposal addresses both of these impediments through the use of a very fast and accurate commercial code and a new type of GA called a mixed integer GA.