Cybersecurity in Cyber-Physical Systems: Wireless Jamming Attack Detection in Noisy LoRaWAN Environment

Cybersecurity in safety-critical data communication infrastructures presents a significant and open challenge for cyber-physical systems (CPS). While extensive research exists on wireless jamming, effectively detecting attacks in noisy wireless environments remains difficult. This paper introduces a novel framework for detecting mobile jamming attacks in LoRaWAN-based CPS, designed to operate robustly in the presence of communication faults. Unlike traditional methods that rely on physical layer metrics like RSSI from end devices, our strategy uses only data upload statistics available at the backend network server. We propose a multi-stage filtering approach that first identifies anomalous upload failures and then distinguishes jamming attacks from communication faults by analyzing their distinct spatio-temporal signatures. Furthermore, the framework can reconstruct the attacker's trajectory from the identified attack events. We evaluate the proposed strategy via extensive discrete-time simulations under various network densities and attacker speeds. Results demonstrate high efficacy, achieving an F1-score of up to 1.00 for event detection and a trajectory reconstruction mean absolute error (MAE) as low as 7.07 meters in dense networks, even in the presence of faults. This work's primary contribution is a backend-centric, fault-tolerant detection methodology that enhances situational awareness without imposing additional overhead on resource-constrained end devices.