On Infrastructure Design for Large UAV Networks

With increasing Unmanned Aerial Vehicle (UAV) applications, their connectivity with network infrastructure plays a critical role in their mission performance. In this paper, we propose a simple model for UAVs as network elements with two mission-critical UAV-specific traffic flows, namely, command-and-control (C2) traffic a UAV receives from its serving base station (BS), and payload traffic in the opposite direction. We analyze the delay and throughput of these flows in a network composed of a large number of such mutually interfering UAV s and their serving (ground) base stations using tools from stochastic geometry. To the best of our knowledge, this is the first such analysis. Our findings indicate that when the UAV s and their serving BSs are distributed as independent spatial Poisson processes, the mean C2 delay exhibits a hard phase transition at a critical transmission rate (i.e., it remains finite only below this rate) and also increases rapidly with the UAV's deployment height and the base station density. On the other hand, with mean payload throughput, such an increase in base station density can help offset the loss in signal strength with increasing deployment height. This disparity can be traced to the difference in the requirements of these traffic types, and to the network geometry at the receiving ends of C2 and payload packets.