Optimising Wi-Fi installations using a multi-objective evolutionary algorithm

Abstract

We describe a multi-objective evolutionary algorithm that designs multi-deviceWi-Fi installations optimised for three criteria: minimised 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 performed 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 models used in the simulator allow us to derive results that are more realistic than do previous methods.