A fuzzy mutual information-based intrusion detection system for enhancing cybersecurity in smart energy grids

The shift from traditional power grids to smart grids has enhanced efficiency, reliability, and flexibility through advanced digital communication, automation, and real-time control. However, this increased connectivity expands the attack surface, exposing smart grids to diverse cyber threats and legacy system vulnerabilities. Cybersecurity is, thus, essential to ensure reliable, safe, and secure energy services in smart grid scenarios. This paper proposes a novel FMI-Reduct-Based NIDS Classifier, which integrates a fuzzy approach along with Conditional Mutual Information (CMI) and the Quick Reduct Algorithm (QRA) to prioritize the most relevant features in smart grid data effectively. This methodology not only improves classification performance but also significantly boosts detection capabilities, providing a robust solution for addressing the challenges posed by evolving cyber threats in a smart grid. Extensive experiments using public and benchmark datasets such as NSL-KDD and UNSWNB15 demonstrate that the proposed FMI-Reduct-Based Classifier outperforms traditional classifiers in precision, recall, and overall detection rate. The limitations and future research works are discussed thoroughly.