LoRaLOFT-A Local Outlier Factor-based Malicious Nodes detection Method on MAC Layer for LoRaWAN

LoRaWAN is one of the network technologies that provide a long-range wireless network at low energy consumption. However, the pure Aloha MAC protocol and the duty-cycle limitation at both end devices and gateway make LoRaWAN very sensitive to malicious behaviors in the MAC layer. Moreover, this kind of sensitivity makes the false-positives problem challenging for malicious behavior detection with simple threshold methods. This study investigates two malicious behaviors - greedy and attack on the MAC layer. Furthermore, by combining the threshold method with a Local Outlier Factor (LOF) model in machine learning, LoRaLOFT is proposed. It is a centralized malicious node detection method. Analytical results show that the proposed method gives high detection accuracy while significantly reducing the false-positive rate in both behaviors.