LoRa-Based Decision Support for Rapid Victim Localization in Avalanche Scenarios

Search and Rescue (SaR) operations in avalanchescenarios require rapid victim localization, as survival probability drops rapidly after burial. Traditional avalanche beacons are Radio-Frequency devices used to localize victims operating at 457 kHz, estimating the distance and the target’s direction. Nevertheless, such devices have limited radio coverage, making it difficult to catch the first signal when the operator is far from the target. This work explores the LoRa technology as an extended-range complementary solution within the LoRa-SNOW project. We propose LS1, a CNN-based algorithm that analyses RSS, SNR, transmitter ID, and operator position to estimate the correct search direction. Real-world data were collected in 2025 in the Dolomite Mountains. Results show that the LS1 algorithm reaches up to 96.8% correct angular detection in the simplest setting and up to 72% in the most challenging setting, demonstrating the feasibility of LoRa-assisted avalanche localization far beyond the range of current SaR solutions.