Multi-agent nonlinear negotiation for Wi-Fi channel assignment

Abstract

Optimizing resource use in complex networks with self-interested participants (e.g. transportation networks, electric grids, Internet systems) is a challenging and increasingly critical real-world problem. We propose an approach for solving this problem based on multi-agent nonlinear negotiation, and demonstrate it in the context of Wi-Fi channel assignment. We compare the performance of our proposed approaches with a complete information optimizer based on particle swarms, together with the \emph{de facto} heuristic technique based on using the least congested channel. We have evaluated all these techniques in a wide range of settings, including randomly generated scenarios and real-world ones. Our experiments show that our approach outperforms the rest of techniques in terms of social welfare. The particle swarm optimizer is the only technique whose performance is close to ours, but its computation cost is much higher. Finally, we also study the effect of some graphs metrics on the gain that our approach can achieve.