Hysteretic Boundary Conditions for PSO of Antenna Array Pattern Synthesis

Abstract

Boundary conditions in particle swarm optimizers have been, quite convincingly, shown as an important algorithmic part which restricts the particles within the solution space thereby increasing their scope of participation in solution finding process. As a common practice, velocities of the particles are regulated to bring them back to solution space; however, a different fundamental approach of regulating the position of particles is taken here, particles which go out of solution place are relocated to a position where it was in its past (i.e. hysteretic position) iterative cycle inside the boundary. A six element linear antenna array is taken as the optimization target for the reduction of side lobes in the radiation pattern. Results show better performance of this boundary condition over other established ones in terms of quicker convergence and obtaining better optimization solution.