A Machine Learning Approach to Characterizing and Detecting Electronic Devices

Abstract

The unauthorized use of computers and electronic devices by foreign intelligence and disgruntled insiders poses a threat to the government. However, the consumption of energy by electronic devices induces measurable signatures back into the local power system. The detection and monitoring of these signatures provides a method of identifying unauthorized equipment use and potential counterintelligence problems. Previous works in this field focus on distinguishing between devices with a wide diversity in loads. We advance this research by focusing on nearly identical devices in the same class. By using device startup transient features and machine learning algorithms, we show that it is possible to identify unique devices in an electrically noisy office setting. This advances the state of the art and enables a more robust detection algorithm that is well-suited for counterintelligence efforts.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Oct 22, 2018
Accession Number
AD1120362

Entities

People

  • Cassie Seubert
  • David Daigle

Tags

Communities of Interest

  • Advanced Electronics
  • Autonomy
  • Energy and Power Technologies
  • Sensors

DTIC Thesaurus Topics

  • Accuracy
  • Algorithms
  • Artificial Intelligence Software
  • Computers
  • Detection
  • Detectors
  • Dimensionality Reduction
  • Energy Consumption
  • Frequency
  • Information Science
  • Information Systems
  • Load Monitoring
  • Machine Learning
  • Neural Networks
  • Power Supplies
  • Steady State
  • Supervised Machine Learning

Fields of Study

  • Computer science

Readers

  • Electrical Engineering
  • Geospatial Intelligence and Artificial Intelligence Analytics
  • Image Processing and Computer Vision.

Technology Areas

  • AI & ML
  • AI & ML - DoD AI Strategy
  • Microelectronics