Optimising Wi-Fi Installations Using a Multi-Objective Evolutionary Algorithm


We describe a multi-objective evolutionary algorithm that designs multi-device Wi-Fi installations optimised for three criteria: min- imised cost, maximised coverage, and minimised service refusal. At the heart of the system is a detailed simulator for Wi-Fi installations, and a simple parallel evaluation scheme to allow these simulations to be per- formed in reasonable time. We show that the algorithm can derive good installations for two real-world maps requiring respectively around ten and fifty access points. The fine-grained connectivity and mobility mod- els used in the simulator allow us to derive results that are more realistic than do previous methods.