Detection and Mitigation of Jamming Attacks in LoRaWan Using Machine Learning

The security and efficiency of low-power wide area networks (LPWANs) for connecting an ever-growing number of IoT devices, expected to exceed 40 billion by 2030, are becoming increasingly important. LoRaWAN, a leading LPWAN technology, enables long-range, low-power communications but remains susceptible to jamming attacks that degrade network performance at the physical layer. This paper introduces a robust framework for detecting and mitigating jamming in LoRaWAN networks using comprehensive threat modeling and machine learning-based countermeasures. The framework simulates two types of jamming attacks: a channel-oblivious jammer, which transmits continuously to randomly interfere with channels, and a channel-aware jammer, which selectively disrupts active transmissions. We evaluate LoRaWAN's resilience through extended simulations in the ns-3 module, adapted for jamming scenarios. Additionally, we provide a high-precision LSTM-based detection model to identify jamming patterns and a mitigation strategy to counteract the channel-oblivious jammer, including an automatic restoration process for returning devices to normal operation post-disruption. The proposed framework enhances network robustness against jamming, showing that ML-based detection significantly reduces disruptions.