Jamming Detection in Power Line Communications Leveraging Deep Learning Techniques

Power Line Communications (PLC) is a well-established technology that allows devices connected to the power line to communicate with each other. While the majority of research in this field is devoted to issues of availability, the topic of Denial of Service (DoS) attacks has not been sufficiently addressed. Typically, current solutions might detect a jammer when situated near the target devices, yet the equipment under jamming interference may face challenges in communicating an alarm. However, when these systems are placed at a significant distance from the jammer, the negligible impact of the jamming renders its detection hardly detectable. In this work, we propose a solution to identify the presence of a jammer in a PLC infrastructure even when deployed at a significant distance. We analyze the physical layer of the PLC link and adopt state-of-the-art Deep Learning techniques to detect jamming even at a distance where the jammer's effect is negligible, thus allowing the device to trigger an alarm. Considering a jammer featuring the same transmission power as legitimate devices, we prove that we can detect the presence of such a jammer with an overwhelming probability (higher than 0.99) even at a distance of 75 m from the source.