LoRa reactive jamming degrades LoRa's communication stability and increases maintenance costs by reducing node lifetime. Existing countermeasures, which are Signal Strength-based, fail to detect attacks that transmit at a power level similar to legiti...
LoRa reactive jamming degrades LoRa's communication stability and increases maintenance costs by reducing node lifetime. Existing countermeasures, which are Signal Strength-based, fail to detect attacks that transmit at a power level similar to legitimate packets.
This paper proposes a detection and recovery method for attacks that can transmit jamming symbols at the signal strength of legitimate packets. A Convolutional Neural Network(CNN) is used to detect jammed packets. followed by recovery process that utilizes the unique Frequency Offset(FO) of the legitimate packet. Detection model achieved an accuracy of 0.998. Furthermore, the recovery method improved the Packet Reception Rate (PRR) from below 10% to over 63%, resulting in a throughput increase of more than 10.4 times.