An Experimental Study of Parameter Selection in Particle Swarm Optimization Using an Automated Methodology

In this work, an experimental study to evaluate the parameter vector utility brought by an automated tuning tool, so called Hybrid Automatized Tuning procedure (HATp) is given. The experimental work uses the inertia weight and number of iterations from the algorithm PSO; it compares those parameters from tuning by analogy and empirical studies. The task of PSO is to select users to exploit concurrently a channel as long as they achieve the Signal-to-Interference-Ratio (SINR) constraints to maximize throughput; however, as the number of users increases the interference also arises; making more challenging for PSO to converge or to find a solution. Results show that, HATp is not only able to provide a parameter vector that improve the search ability of PSO to find a solution but also to enhance its performance on resolving the spectrum sharing application problem than those parameters values suggested by empirical and analogical methodologies in the literature on some problem instances.