Deep Learning-Based, Passive Fault Tolerant Control Facilitated by a Taxonomy of Cyber-Attack Effects

Abstract

There have been several cyber-attacks on the cyber-physical systems (CPS) that monitor and control critical infrastructure over the last few years. The need for increased cyberspace security for these industrial control systems (ICS) has been widely discussed extensively researched. This work presents a novel controller design that does not rely on fault or attack detection. It incorporates deep learning and ensemble learning techniques to holistically consider the state of the system under control and determine which model to use for further control signals. This work also presents a taxonomy of effects for use in designing training and testing FTC. The Taxonomy is foundational to the proposed controller.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Nov 23, 2020
Accession Number
AD1124108

Entities

People

  • Dean C. Wardell

Organizations

  • Air Force Institute of Technology

Tags

Communities of Interest

  • Cyber
  • Energy and Power Technologies
  • Sensors
  • Space

DTIC Thesaurus Topics

  • Air Force
  • Artificial Intelligence
  • Bayesian Networks
  • Computational Science
  • Computer Languages
  • Control Systems
  • Control Systems Engineering
  • Cyberattacks
  • Cybersecurity
  • Detectors
  • Industrial Control Systems
  • Information Science
  • Information Systems
  • Kalman Filters
  • Machine Learning
  • Mathematical Models
  • Model Predictive Control
  • Multiple Input Multiple Output
  • Network Science
  • Neural Networks
  • Scada
  • Systems Engineering

Fields of Study

  • Computer science

Readers

  • Cybersecurity.
  • Neural Network Machine Learning.
  • Robotics and Automation.

Technology Areas

  • AI & ML
  • AI & ML - Autonomous Systems
  • AI & ML - Neural Networks
  • Cyber