Name:Automatic Antenna De
Description:A software suite to speed up the design of wire antennas
Abstract:Genetic optimization (GO) has been proposed in electromagnetism, and has found many successful applications, in the last decade. However GO is inherently a dimensioning approach, and as such it has quite a limited use in antenna design. The full power of evolution-inspired methods has been unchained in antenna design only when Genetic Programming (GP) approaches have been proposed. In fact, GP is able to find a suitable solution of the problem without any initial assumption about its size or shape. In this work we successfully applied the system to design small wire antennas with a significant gain and with a bandwidth considerably larger than the standard one (Yagi antennas). Key point of the design are the requirements of reduced size and robustness w.r.t. tolerances and atmospheric agents.
A wire antenna can be mathematically modelled as a set of cylinder displaced in 3D space. The space of all feasible solutions, the search space, is the set of all possible positions and orientations for N segments (with N arbitrary).
GP allow to do an effective search of a near optimum solution in such a search space. The software is build on the basis of two other open source software: ECJ (Evolutionary Computation in Java), a framework developed at George Mason University that provide the basic functionalities to deal with GP, and NEC2++ (Numerical Electromagnetic Code), a C++ porting of the original software developed at Lawrence Livermore Laboratory for the analysis, by a Method of Moments, of electromagnetic properties of wire structures .