DESIGN OF FRACTAL ANTENNAS AND FREQUENCY SELECTIVE SURFACES USING BIOLOGICALLY INSPIRED COMPUTATIONAL TECHNIQUES

Abstract

The ever growing demand for antenna bandwidth in wireless communication market is continuously posing antenna designers with challenges of designing multiband and/or broadband antennas having additional characteristics like low profile, compact size, and in some cases conformal also. Over the years, number of such antenna designs has been reported with excellent characteristics. All these structures are associated with a background analysis theory. But the real problem that an antenna engineer faces, even today, is the availability of a tool for designing custom-made (multiband) antennas in one shot. The research done in this thesis is an attempt to develop a user-friendly tool for design of custom made fractal antennas, a class of antennas that is well known in antenna theory for designing multiband antennas. The same theory has also been extended to develop custom-made multiband fractal frequency selective surfaces (FSSs). Since the introduction of fractal theory in Electromagnetics, although several fractal structures have been analyzed, but the major problem with these structures are: (I) all the operational frequencies of a fractal electromagnetic structure, in its classical form, are not the user-defined frequencies, and furthermore (ii) in order to reach to a perturbed structure that can function at user required frequencies, the user has to play with a number of design parameters besides the scaling and to do several trial simulations before landing to the final optimized structure. Our aim, in this research, is to circumvent both these demerits of custom-made fractal multiband electromagnetic structure design and to provide the designer with a flexible design methodology, that is, not only accurate but at the same time will take drastically less design time. The novel design methodology has been developed with the help of two different biologically inspired computational techniques, viz, artificial neural networks (ANN or simply NN) and particle swarm optimization (PSO) technique. A conventional microwave design process, using simulators, usually starts with a rough design, whose dimensions are decided according to some closed form formulation. After observing the frequency response of the rough design, the designer changes some or all of its design dimensions in a pure trial and error way until reaching to the desired response. Sometimes it takes hundreds of simulations to reach to the final optimized design, depending on the number 1ff of design variables. In the present work we have tried to automate this conventional microwave design process. The methodology starts with the development of a trained NN connecting the design variables of the fractal electromagnetic structure with its frequency 40 response. Then the design task was approached as an optimization problem and was solved using the PSO tool whose cost function was evaluated with the help of the previously developed trained NN. Because the response of NN is very fast, so the complete optimization process takes few seconds only. We have taken two different types of fractal antennas, viz. Sierpinski gasket monopole, Koch monopole and one type of fractal frequency selective surface (FSS) to check the validity of the developed design methodology. Laboratory prototypes of all the structures were made and experimentally measured to cross validate the developed design methodology. Here it may be noted that, although the developed novel design methodology has been tested for limited number of fractal structures, it can be extended for other fractal structures also. Furthermore, this design methodology is very much suitable for design of microwave structures having more number of design variables. Even today, although number of commercial softwares and freewares are available for simulation of microwave components, if one looks from the designer point of view, a flexible design technique is still missing in Microwave Engineering area. In view of this fact, we feel that the work done in this thesis will certainly give new direction in the area of microwave design.