Low-Altitude Drone Monitoring Using Millimeter-Wave Cellular Base Stations and Covariance-Aware Multiradar Fusion

Abstract

Low-altitude drones pose significant challenges for airspace safety and navigation, often requiring dedicated radar systems for monitoring. Commercial millimeter-wave cellular base stations provide dense urban infrastructure, large instantaneous bandwidths, directive antenna arrays, and software-defined processing chains that can support radar-like sensing. In particular, fifth-generation (5G) New Radio (NR) frequency range 2 (FR2) hardware provides bandwidths up to 400 MHz and directive massive multiple-input–multiple-output (MIMO) apertures that could enable submeter range resolution and multibeam tracking using existing sites. Here, we investigate the feasibility of reusing commercial millimeter-wave cellular base stations as a distributed radar network for low-altitude uncrewed aerial vehicle (UAV) monitoring. We derive power signal-to-noise-ratio (SNR) bounds, quantify spatiotemporal resolutions, and discuss clutter and Doppler phenomenology for dense urban scenes. We develop a covariance-aware information-space fusion algorithm that combines bearing, elevation, and range observations from multiple stations. The proposed method rotates local measurement covariances into a global Cartesian frame and accumulates information using SNR gating, variance floors, and Huber reweighting for outlier suppression. Monte Carlo trials compare fused multiradar positioning against the Global Positioning System (GPS) under both global navigation satellite system (GNSS)-available and GNSS-denied conditions. Our results show that, in typical line-of-sight (LoS) FR2 deployment settings, fused millimeter-wave cellular sensing can achieve error distributions competitive with the GPS comparator and can outperform it under GNSS-degraded conditions. We then outline a 5G NR-specific implementation path in which the baseband/field-programmable gate array (FPGA) stack schedules sensing bursts in time-division duplexing (TDD) idle slots and returns fused positions to cooperative UAVs. This approach presents a promising avenue for real-time drone monitoring using existing millimeter-wave cellular infrastructure while minimizing additional hardware deployment. We conclude by discussing the implementation challenges and next steps for field validation.