Detection of the False Data Injection Attack in Home Area Networks using ANN

The combination of high penetration of distributed energy resources with fusion of unprotected data from several sources such as digital sensors (e.g., synchrophasors, smart meters, digital relays etc.) and controllers of systems of systems (SoS) that are connected via internet networks invites more vulnerability and thus become possible vectors for potential security risks. These cyber-risks could compromise the confidentiality (C), integrity (I), and availability (A) of the system, and eventually can lead to physical tampering of devices and its systems resulting in severe economic loss. In this paper, an Artificial Neural Network (ANN) based approach is proposed to detect the false data injection attack in the Home Area Networks. An early and accurate detection of false injected measurements are essential to undertake appropriate countermeasures to avert potential risks. The implemented ANN model is trained and tested extensively on a data set containing energy profile of 200 U.S. households that models FDI attack using two attack scenarios (e.g., on-peak and off-peak hours) with Sigmoid and Trapezoidal representations, and three activation functions. Then, ANN model is compared against other machine learning methods for evaluation. Several performance metrics such as accuracy (a) and probability of detection (Pd) are used. The preliminary results of ANN show promising results with Pd values reaching 99.4% on over support vector machine (Pd =80.5%) and random forest (Pd =88.2%) methods. Thus, deploying such detection algorithms at home area network levels will help identify and reduce these false data injection attacks.