A Data-Based Detection Method Against False Data Injection Attacks

Charalambos Konstantinou, Michail Maniatakos

Research output: Contribution to journalArticle

Abstract

State estimation is one of the fundamental functions in power grid. In this paper, we address the vulnerability of state estimators to false data injection attacks (FDIAs) by proposing a data-driven anomaly detection algorithm. The proposed technique applies dimensionality reduction on grid measurements along with a density-based Local Outlier Factor (LOF) analysis and a feature bagging framework of combining predictions from multiple LOF outlier detection outputs. The work also addresses the handling of critical measurements. Instead of removing the attacked measurements, which may cause the system to become unobservable, we replace them by forecasted measurements. Numerical tests on IEEE 14-bus system verify the effectiveness and performance of the proposed method.

Original languageEnglish (US)
JournalIEEE Design and Test
DOIs
StateAccepted/In press - Jan 1 2019

    Fingerprint

Keywords

  • Cybersecurity
  • dimensionality reduction
  • false data injection attacks
  • outlier detection
  • state estimation

ASJC Scopus subject areas

  • Software
  • Hardware and Architecture
  • Electrical and Electronic Engineering

Cite this