Development of an Experimental Platform for Analysis of Cyber Attacks on the Power Grid

James Kollmer, Robert J. Irwin, Saroj Biswas, Walid Saad, Arif I. Sarwat, Li Bai2018 · DOI

Abstract

This research focuses on electrical machines, power systems, and the control of dynamic systems through the development of a physical, hardware-in-the-loop experimental platform. The system is designed to analyze the impact of cyber-physical attacks on power grid stability, bridging the gap between theoretical simulations and real-world industrial applications.

Introduction: The Vulnerability of the Smart Grid

As the traditional power grid transitions into a "Smart Grid," it increasingly relies on bidirectional communication networks for real-time monitoring and control. While these advancements enhance efficiency and reliability, they also introduce significant cybersecurity risks. Because the power grid is a critical infrastructure, direct experimentation is impossible, and software simulations often fail to capture the complex, non-linear dynamics of physical hardware. This research addresses the need for a controllable physical testbed to study how cyberattacks can deceive controllers and destabilize grid components.

Experimental Platform and Methodology

The researchers developed a "Networked Control System" (NCS) that bridges the gap between digital control and physical power generation.

The physical hardware-in-the-loop experimental setup, including the generator, prime mover, and Arduino-based controllers.

The platform consists of several key components:

  • Physical Plant: A LabVolt EMS8001 system featuring a 200W, 4-pole synchronous generator driven by a dynamometer (prime mover).
  • Data Acquisition (DAQ): An Arduino Uno with a Wiznet w5500 TCP/IP stack samples voltage and current at 1000Hz.
  • Controller: A software-based PI (Proportional-Integral) controller running on a laptop calculates necessary adjustments based on received sensor data.
  • Actuation: A second Arduino receives control signals via Ethernet and uses Pulse Width Modulation (PWM) to drive an IGBT DC/DC chopper, regulating the generator's field current.

Networked Control Architecture

The system is modeled as a closed-loop Networked Control System where the communication network introduces potential vulnerabilities.

Conceptual diagram of the Networked Control System (NCS) showing the feedback loop between the physical system and the controller via a communication network.

A critical design choice was the use of the UDP protocol for communication. In high-speed power control loops, the delays associated with TCP's retransmission mechanisms can be more destabilizing than the loss of an occasional packet. UDP ensures that the controller always acts on the most recent data available, even if some packets are dropped.

Cyber-Physical Attack Scenarios

The platform was used to simulate four distinct attack vectors:

  1. Constant Bias: Injecting a steady offset into sensor data. This "tricks" the controller into maintaining a physical voltage that is either dangerously high or low, while the controller's internal logs report a stable, correct value.
  2. Ramp Bias: A linearly increasing bias causes the actual voltage to drift steadily, potentially triggering protective relays in a real-world grid.
  3. Random Data Injection: Adding noise to sensor packets causes erratic variations in the IGBT duty cycle, threatening the physical integrity of the generator.
  4. Denial-of-Service (DoS): By simulating high packet dropout rates (98%), the researchers found that the system could maintain marginal stability by "holding" the last known good value, provided the dropout rate did not exceed a specific threshold.

Conclusion and Future Work

The experimental results demonstrate that cyberattacks can effectively "deceive" grid controllers, leading to physical instability that is invisible to the controller's monitoring software. Future work will involve expanding the testbed to multi-generator configurations, studying phase-matching synchronization under attack, and developing control-theoretic defense algorithms to detect and mitigate these threats in real-time.

Summary updated

Abstract

in electrical machines and power systems, multimedia tutoring, and control and optimization of dynamic sys-
BibTeX
@misc{kollmer2018development,
  author = {James Kollmer and Robert J. Irwin and Saroj Biswas and Walid Saad and Arif I. Sarwat and Li Bai},
  title = {{Development of an Experimental Platform for Analysis of Cyber Attacks on the Power Grid}},
  year = {2018},
  doi = {10.18260/1-2--28167},
}
Figure