Securing Smart Meter Communication with an Ensemble Fingerprinting Framework

Advanced Metering Infrastructure (AMI) comprises several smart meters (SM) that use wireless technologies such as ZigBee to exchange data and commands between each other and the backend systems. The wireless broadcast nature and smart meters' physical vulnerability make them prone to cyberattacks like spoofing, masquerading, and man-in-the-middle. This paper addresses the secret-free smart meter identification challenge in AMI systems, focusing on the widely used Zigbee communication standard. Specifically, we introduce an ensemble device fingerprinting framework, integrating the physical and the medium access control (MAC) layer with multiple machine learning models (Convolutional Neural Network, Logistic Regression, and Decision Tree). Our analysis shows that the ensemble framework outperforms individual fingerprinting models, with a mean accuracy of 83.33 %.