Improving Message Delivery in UAV-based Delay Tolerant Networks

Delay Tolerant Networks (DTNs) are sparse networks where complete direct end-to-end paths between source and destination can seldom be established. Routing mechanisms in DTN rely on the nodes' mobility to connect disconnected nodes, carrying messages around the network to overcome path disconnection. The proactive DTN approach consists of introducing dedicated nodes whose only purpose is to establish communication between ordinary nodes, and relieve them from energy-consuming work, such as message routing and forwarding. This paper addresses the use of Unmanned Aerial Vehicles (UAVs) as DTN relays, introducing a proactive scheme called Deadline Triggered Pigeon with Travelling Salesman Problem with Deadlines (DTP-TSP-D). We envision a DTN where ground nodes can only communicate by flying UAVs with the capacity of carrying messages from one location to another. Each UAV either belongs to one node or to a cluster of nodes, and its role is to hover over their home-ground node (or ferrying around their home cluster) until they are triggered to deliver messages directed to other ground nodes. The triggering criterium is based on the deadlines of the messages present in the UAVs buffer, evaluating its ability to deliver all of them in time. It uses a developed TSP Genetic Algorithm to compute the route that achieves the most (timely) deliveries. The performance of DTP-TSP-D has been compared to dedicated node protocols found in the literature, namely SIRA and MRT-Grid. The performance metrics used were delivery ratio and average delay. The results show that DTP-TSP-D achieves higher delivery rates than its competitors, while keeping a consistent average delay.