Design of Ultrawideband Antennas using Genetic Algorithms and Integral Equation Techniques

The design of broadband antennas finds application in areas such as cellular communications, remote sensing, television etc. Thin-wire antennas are frequently used when the systems are required to be light and portable. The broadband characteristics of the thin-wire antennas are achieved either by loading the wires with combinations of passive loads or by an adequate selection of the geometry of the structures. In both cases, due to the great variety of parameters involved, optimization techniques such as Genetic Algorithms (GA) (1) are very appropriate tools to search for the best antenna models. GA techniques are becoming widely used to solve electromagnetic problems due to their robustness, wide range of applications and readiness in their implementation. Nevertheless they are computationally intensive because they involve the solution of hundreds of possible designs before converging to the one that best fulfils the initial requirements. Therefore it is important to optimize their implementation minimizing the computational time utilized in the whole procedure. In this communication we apply GA techniques to the design of thin-wire broadband antennas when they are implemented in two different domains, i.e, utilizing either a frequency- domain or a time-domain technique to solve for the thin-wire problem. The Electromagnetic solver is based in the solution by the Method of Moments of the Electric Field Integral Equation (EFIE) formulated either in the frequency domain (NEC code) or directly in the time domain for wires with arbitrarily connected loads. The GA optimization of wire broadband antennas in the frequency-domain (1)-(2) maximizes the bandwidth computed from the antenna response at a discreet number of frequencies in a certain frequency range. If the number of frequencies used is not too high, working in the frequency domain may be computationally advantageous but, in some cases, it is necessary to call for the electromagnetic solver at many frequencies within the band of interest to appropriately model the behaviour of the system with the consequent increase in computational time. An alternative is to achieve the broadband characteristic of the antennas through a maximization of the fidelity factor directly in the time domain providing wide-band information from a single execution of a marching-on-in-time procedure (3). The later formulation offers the advantage of treating all the frequencies in the same manner and usually reduces the computational time in cases presenting a very broad bandwidth. Results will be shown comparing the performance of GA tools in the frequency and time domains when they are applied to achieve a maximum bandwidth by either modifying the geometry of the thin-wire antennas or by selecting the appropriate values of passive linear loads, to be located at specific points along an particular thin-wire geometry. Advice will be given on the convenience of using a specific GA approach depending on the kind of problem to be optimized. REFERENCES