UAVs Enhance Delay-Tolerant Network Communication with AI

Delay-tolerant networks (DTNs) are crucial for communication in environments with sparse or intermittent connectivity, relying on a store-carry-forward mechanism. However, challenges like limited contacts, finite buffers, and message time-to-live often lead to poor delivery rates and congestion. This research addresses these issues by exploring a joint optimization strategy involving decentralized opportunistic routing and controllable unmanned aerial vehicle (UAV) flight. The proposed framework, JUROR (Joint UAV flight and Opportunistic Routing), leverages the Proximal Policy Optimization (PPO) reinforcement learning algorithm. It casts the problem as a factored partially observable Markov decision process, where decentralized actors make decisions based on local observations, while a centralized critic guides training. UAVs are used to strategically create new contacts, and nodes can replicate messages, significantly improving performance over existing methods like PRoPHET and MaxProp in various traffic modes.