Smart cattle monitoring with UAV edge processing and resilient multi-channel alert delivery

Monitoring cattle in large grazing areas requires telemetry solutions that are energy-efficient on the animal side, responsive during field operations, and robust to intermittent connectivity. This paper presents a smart livestock-monitoring architecture that combines LoRa-enabled wearable collars, a UAV serving as both a mobile gateway and an edge monitoring node, and a farmer-side node that exposes a configurable runtime monitoring interface. The proposed system moves time-critical reasoning closer to the field through lightweight on-drone rule evaluation, while preserving richer inspection, persistence, and rule management on the ground. To improve alert availability under unstable connectivity, the architecture integrates a dependable dispatcher that prioritizes terrestrial communication and escalates critical alerts to Iridium Short Burst Data (SBD) when Wi-Fi or LTE/4G delivery is unavailable or cannot satisfy the required timeliness. Livestock-relevant conditions, including missing beacons, geofence violations, prolonged immobility, communication degradation, and delivery failures, are represented as explicit runtime rules. The paper reports a controlled, prototype-based evaluation that combines measured LoRa first-hop characterization, replayable mission scenarios, and direct timing observations along the RockBLOCK-Cloudloop-CONCERN integration path. In the considered mission profile, on-drone rule evaluation reduces detection latency by about 2 orders of magnitude compared to ground-centric processing, while the hybrid configuration with dependable dispatch delivers all critical alerts within the adopted latency budget during simulated terrestrial outages. These results indicate that combining explainable edge-side monitoring with resilient multi-channel alert delivery is a practical direction for livestock supervision in infrastructure-poor agricultural environments.