Capacity Analysis of UAV Communications: Cases of Random Trajectories

This paper analyzes the link capacity between autonomous unmanned aerial vehicles (UAVs) with random three-dimensional (3-D) trajectories. This is distinctively different from existing works typically under the assumption of either two-dimensional (2-D) or deterministic trajectories, and particularly interesting to applications such as surveillance and air combat. The key idea is that we geometrically derive the probability distributions of the UAV-to-UAV (U2U) distances which, by exploiting the Jensen's inequality, can be translated to the closed-form bounds for the capacity between UAVs, and between UAVs and ground stations. Another important aspect is that we extrapolate the idea to dense UAV networks, and analyze the impact of network densification, imperfect channel state information, and interference from ground transmitters on the capacity. Corroborated by simulations, our analysis shows that a U2U link with random 2-D trajectories is superior in terms of capacity due to its short average link distance. It is also revealed that a UAV-to-ground link can incur substantially lower capacity than a U2U link even in the case the 3-D coverage of the UAVs is the same, as the result of its longer average link length.