On the application of a novel Grouping Harmony Search algorithm to the Switch Location Problem

This paper presents the adaptation of a novel heuristic algorithm to the Switch Location Problem (SLP), a NP-hard problem where a set of distributed terminals, with distinct rate demands, is to be assigned to a fixed number of concentrators subject to capacity constraints. Each terminal must be assigned to one and only one concentrator while keeping the overall rate demanded from such concentrator below its maximum capacity. Related literature demonstrates that the inclusion of the so-called grouping concept into the allocation algorithm is essential when dealing with this specific kind of optimization scenarios. As such, previous studies introducing Grouped Genetic Algorithms (GGA) combined with local search and/or repair methods show that their proposed allocation procedure alleviates significantly the computational complexity required by an exhaustive search strategy for the SLP problem, while outperforming other hybrid heuristic algorithms. In this manuscript, a novel Grouped Harmony Search (GHS) algorithm designed for the SLP paradigm is hybridized with both a local search method and a technique aimed at repairing the solutions iteratively supplied by the heuristic process. Extensive Monte Carlo simulations assess that our proposal provides faster convergence rate and better statistical performance than the previously proposed GGA, specially when the size of the SLP scenario increases.