Decentralized Cooperative Localization for UAV Swarms in GNSS-Denied Environments Using PDR Inversion

Reliable positioning is a critical requirement for autonomous Unmanned Aerial Vehicles (UAVs); however, Global Navigation Satellite System (GNSS) signals often suffer from natural degradation or intentional interference in tactical and complex environments. This paper introduces a cooperative positioning system for UAV swarms that reuses standard inter-UAV communication links as opportunistic sensors. Stationary GNSS-enabled anchor UAVs broadcast their coordinates, and a leader node in a GNSS-denied area estimates its own position onboard, without a central controller. Distances are derived from the Packet Delivery Ratio (PDR) by inverting an empirical 5 GHz propagation model and resolved through least-squares multilateration, followed by a median-seeded plausibility filter and a Kalman filter. Simulations in ArduSim, with 20 runs per configuration, show that the method provides coarse localization: mean errors among successful runs range from 221 to 275 m, and mission completion decreases from 100% to 20% as the anchor separation grows from 500 to 900 m. The plausibility filter reduces the mean error by up to 30%, whereas the Kalman filter reduces trajectory roughness by up to 93% without significantly reducing the absolute error. A 10 km mission was completed with 21 anchors, demonstrating that PDR inversion can provide coarse cooperative localization without dedicated ranging hardware, although it relies on GNSS-enabled anchors; real-flight validation remains future work.