First Results on UAV-aided User Localization Using ToA and OpenAirInterface in 5G NR

TL;DR

This work tackles localizing ground users with measurements collected by a mobile UAV when the UAV pose is imperfect. It introduces a least-squares SLAM framework that fuses ToA estimates $\hat{\tau}_{0,k}[n]$ with GPS UAV poses $\hat{\mathbf{x}}[n]$ to jointly estimate user locations $\mathbf{u}_k$ and UAV trajectory $\mathbf{x}[n]$ in a 5G NR environment, implemented via OpenAirInterface. The approach formulates a nonconvex optimization as a Graph-SLAM problem and solves it through iterative linearization, validated on real hardware with a 5G UAV base station and ground control platform. Findings show robust performance in LoS and NLoS conditions, with ToA clock drift producing a sawtooth pattern that can be modeled within the error terms. The work demonstrates the practicality of UAV-aided, ToA-based localization in 5G NR, enabling rapid deployment for emergency response or coverage in infrastructure-poor scenarios.

Abstract

This paper considers the challenge of localizing ground users with the help of a radio-equipped unmanned aerial vehicle (UAV) that collects measurements from users. We utilize time-of-arrival (ToA) measurements estimated from the radio signals received from users collected by a UAV at different locations. Since the UAV's location might not be perfectly known, the problem becomes about simultaneously localizing the users and tracking the UAV's position. To solve this problem, we employed a least-squares simultaneous localization and mapping (SLAM) framework to fuse ToA data and the estimate of UAV location available from global positioning system (GPS). We verified the performance of the developed algorithm through real-world experimentation.

Figures (8)

• Figure 1: UAV-aided ground user localization system.
• Figure 2: Signaling mechanism using PDCCH order
• Figure 3: A screenshot of our connected robotics software platform, visualizing the 3D map, the current UAV and the estimated user locations, the UAV next way-point, and the true user location.
• Figure 4: An example of our experiment setting where a single user is localized by a UAV 5G BS (a.k.a gNB).
• Figure 5: Effect of clock drift on ToA in commercial user.