Visible to the public Machine Learning Based Physical-Layer Intrusion Detection and Location for the Smart Grid

TitleMachine Learning Based Physical-Layer Intrusion Detection and Location for the Smart Grid
Publication TypeConference Paper
Year of Publication2019
AuthorsPrasad, G., Huo, Y., Lampe, L., Leung, V. C. M.
Conference Name2019 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (SmartGridComm)
Date PublishedOct. 2019
ISBN Number978-1-5386-8099-5
Keywordsactive intruder, carrier transmission on power lines, Collaboration, communication purposes, composability, continuous bidirectional information exchange, cryptographic techniques, Human Behavior, ideal intruder detection rates, Impedance, Intrusion detection, learning (artificial intelligence), machine learning, Metrics, modems, network stack, PHY design, physical layer methods, physical-layer intrusion detection, policy-based governance, Power cables, power engineering computing, power system security, privacy, pubcrawl, radio networks, resilience, Resiliency, Scalability, security of data, smart grid communication data, smart grid consumer privacy, smart grid data security, Smart grids, smart power grids, stand-alone solution, Task Analysis, upper layer techniques
AbstractSecurity and privacy of smart grid communication data is crucial given the nature of the continuous bidirectional information exchange between the consumer and the utilities. Data security has conventionally been ensured using cryptographic techniques implemented at the upper layers of the network stack. However, it has been shown that security can be further enhanced using physical layer (PHY) methods. To aid and/or complement such PHY and upper layer techniques, in this paper, we propose a PHY design that can detect and locate not only an active intruder but also a passive eavesdropper in the network. Our method can either be used as a stand-alone solution or together with existing techniques to achieve improved smart grid data security. Our machine learning based solution intelligently and automatically detects and locates a possible intruder in the network by reusing power line transmission modems installed in the grid for communication purposes. Simulation results show that our cost-efficient design provides near ideal intruder detection rates and also estimates its location with a high degree of accuracy.
Citation Keyprasad_machine_2019