Intrusion Detection System With Updated Structure and Feature Selection Algorithm for Man-in-the-Middle Attacks in the Smart Grid

With the rapid development of the smart grid relying on communication and computer science technology, its vulnerabilities to intentional cyber-attacks are also exposed to cyber-attackers. Among all types of cyber-attacks, the Man-In-The-Middle (MITM) attack stands out and poses a challenge for online detection due to its stealthiness and the severity of the potential outcome. The common countermeasure of a cyber-attack is the Intrusion Detection System (IDS). However, current IDS, based on their algorithm, are limited to running speed or accuracy. This paper provides an IDS with an updated architecture, combining the advantages of both types of IDS by adding a machine learning block to the architecture and accelerating the running speed by using a newly developed feature selection algorithm: Random Disturbance Algorithm (RDA). RDA is developed based on the idea that a more important feature is more sensitive to stochastic disturbance. To better validate the IDS proposed in this paper, the real-time testbed generates two cyber-physical datasets based on the IEEE-39 Bus system and the IEEE-118 Bus system. The performance of the architecture, RDA, and the combined IDS are thus validated.