Digital-Twin-Enabled Outage Detection and Uplink Availability Optimization With UAV Assistance

Abstract

Digital Twins (DTs) offer a powerful tool for outage management in wireless networks, but their direct deployment during emergencies is limited by the high cost of near real-time ray tracing. This work develops a DT-enabled outage management framework that combines vision-based outage detection with availability-oriented uplink control. A lightweight detector (DTOD), trained on DT renders and SINR-consistent labels, produces calibrated outage probabilities in almost real time. These predictions trigger a threshold-capped (ThreshCap) uplink power allocator, which leverages UAV relays to sustain service during macro base station (MBS) failures. Using Markovian MBS outages, realistic DT mobility, and surrogate channels fitted to ray-traced propagation, we show that UAV relays substantially improve availability and that ThreshCap consistently outperforms baselines under varying load and outage intensities.