Abstract:
Replay attacks on transmission protective relays installed at modern smart grid substations are the stealthiest and most challenging to detect. Since the pre-recorded act...Show MoreNotes: This article was mistakenly omitted from the original submission to IEEE Xplore. It is now included as part of the conference record.
Metadata
Abstract:
Replay attacks on transmission protective relays installed at modern smart grid substations are the stealthiest and most challenging to detect. Since the pre-recorded actual fault events are played back at the sensor’s terminals, mimicking the real-time fault. To launch such attacks, the attackers take advantage of known vulnerabilities of protection algorithms and communication infrastructures in the smart grid. Thus, the existing information technology-based security mechanism is inadequate to counter such attacks. This work proposes a data-driven replay attack detection framework leveraging the power of mathematical morphology and unsupervised deep learning long short-term memory autoencoder. With the help of significantly crafted features of three-phase current signals provided by the mathematical morphology signal processing technique, the long-short-term memory autoencoder detects replay attacks. It coordinates with the relay to prevent the false tripping command to breakers. The result shows that the method achieved a high recall and precision rate. The performance assessment is based on the unsynchronized and partially synchronized replay injections. Alternatively, the analysis of the choice of threshold for optimal performance is also studied.
Notes: This article was mistakenly omitted from the original submission to IEEE Xplore. It is now included as part of the conference record.
Published in: 2022 IEEE International Conference on Advanced Networks and Telecommunications Systems (ANTS)
Date of Conference: 18-21 December 2022
Date Added to IEEE Xplore: 28 August 2023
ISBN Information: