Federated Learning Empowered Routing for Opportunistic Network Environments

Opportunistic network systems (OppNets), a subclass of Delay-tolerant networks (DTNs) systems, are a category of wireless networks designed for operation in demanding and constantly changing environments, where traditional network infrastructure is often unreliable, constrained, or absent. These adverse network conditions make conventional routing algorithms impractical. This paper proposes a Federated Learning-based Router (FLRouter) suitable for use in Opportunistic network environments, which harnesses the capabilities of federated learning. In this framework, the nodes opportunistically select the next-hop relay for message forwarding based on the real-time network state and nodes' local knowledge. The contribution of this work is the seamless integration of federated learning algorithms into opportunistic network systems, allowing for optimal routing decisions. Through simulations and in-depth analysis, the effectiveness of our approach in achieving efficient routing within OppNets is demonstrated.