Decentralized Smart Grid System:A Survey On Machine Learning-Based Intrusion Detection Approaches

Abstract

Smart grid is a two-way communication technology power system that sends information between the control server and consumer. It consists of different IoTs connected to a smart meter, creating a network known as the HAN home area network, and collections of these smart meters form a NAN neighbor area network. This data has been transferred to a WAN-wide area network, where the control server will share and analyze the information. The information shared in all layers has been secured in order to maintain this infrastructure. Traditional systems like firewalls’ general cryptographic techniques can detect anomalies for known attacks, but they sometimes fail to provide efficient security for unknown or real-time attacks. There should be a complete framework to detect real-time intruders and attacks. Here, NIDS using machine learning approach has been discussed in this survey report. Most ML techniques are able to detect real-time attacks with less time overhead and higher accuracy. On the basis of accuracy, detection rate, and F1 score, ten different types of datasets were evaluated and analyzed.