Classification and optimization of beam responses synthesized from non-uniformly spaced linear arrays

Abstract

A probability distribution for the element positions on a linear array is derived so as to match a target beam pattern that approximates a Dolph-Chebychev array. The random beam responses generated from such a configuration exhibit a mean profile in observation angle that is invariant with element size N but the variance decays as 1/N. The beam responses are classified considering the N-1 inter-element distances as features that control the deviation from the mean beam profile and the variance of the class. The inter-element distance constraints are incorporated in a computational model of the firefly algorithm (FA) designed to optimize the location of the array elements that match the target beam pattern. The FA consists of a set of fireflies, representing arrays, which move through the solution space with directed movement based on each fireflies' fitness and random movement to ensure adequate exploration and avoid local minima. The fitness function is a combination of a target beam-width, side-lobe lev...