The autopolyploidy enhanced evolution of large-N fractal-random arrays

This paper introduces a novel approach for the optimization of large-N fractal random antenna arrays using genetic algorithms. Genetic algorithms often become overwhelmed by large numbers of input parameters, leading to an inefficient optimization processes. Fractal-random geometries lend themselves well to genetic algorithm optimization through the ability to describe their complex array structures with only a small number of input parameters. In addition, the recursive properties of fractal-random arrays allow for the rapid calculation of the array factor, which can be exploited to significantly speed up the convergence of the GA. This paper describes a method that increases the number of generators used to construct the array as the optimization progresses, maximizing the benefit of using fractal-random geometries to describe large-N arrays. Several optimized solutions are discussed, the largest being a 1650 element linear fractal-random array with a -23.57 dB side-lobe level and a beamwidth of 0.05°.