Title: Deep learning for active detection of FDIAs to defend distributed demand response in smart grid

Authors: Aschalew Tirulo; Siddhartha Chauhan

Addresses: Department of Computer Science and Engineering, NIT Hamirpur, Hamirpur, Himachal Pradesh, India ' Department of Computer Science and Engineering, NIT Hamirpur, Hamirpur, Himachal Pradesh, India

Abstract: Integrating Cyber-Physical Systems (CPS) with smart grids increases susceptibility to False Data Injection Attacks (FDIAs), compromising grid stability and home automation. Current anomaly detection methods falter due to the diverse nature of smart grid data. This paper introduces a Convolutional Neural Network (CNN)-based supervised anomaly detection framework designed specifically for detecting FDIAs in Demand Response (DR) systems of smart grids. Utilising labelled real-world energy consumption data from Austin, Texas, our CNN model demonstrates superior performance over traditional methods, excelling in accuracy, precision, recall, F1 score, False Positive Rate and AUC-ROC and Precision-Recall Curves. The results affirm the model's effectiveness and potential to enhance DR mechanisms' security against FDIAs, suggesting its practical implementation in real-world scenarios.

Keywords: anomaly detection; convolutional neural networks; demand response; false data injection attacks; smart grid.

DOI: 10.1504/IJGUC.2024.142746

International Journal of Grid and Utility Computing, 2024 Vol.15 No.6, pp.572 - 587

Received: 05 Aug 2023
Accepted: 28 Apr 2024
Published online: 20 Nov 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article