Research on network security technology of industrial control system

Kai Jin, Zhanji Niu, Jieping Liu, Jinxue Bai, Lei ZhangMATEC 2022 · DOI

TL;DR

As industrial control systems (ICS) move from isolated "air-gapped" environments to the interconnected Industrial Internet, they face unprecedented security risks. This paper explores penetration testing as a proactive defense strategy, detailing methods like port scanning and dynamic injection to identify vulnerabilities before malicious actors can exploit them.

What Problem Does This Paper Address?

For decades, industrial systems—the "gears and valves" of our infrastructure—were protected by physical isolation. Today, the rise of the Industrial Internet and 5G has bridged the gap between operational technology (OT) and traditional IT. While this enables "smart" manufacturing and better efficiency, it opens a massive attack surface.

The authors highlight that a breach in an ICS, such as a city’s heating system, isn't just a data leak; it’s a physical safety hazard with the potential for "serious social harm." The challenge lies in the fact that these systems are incredibly diverse and complex. Unlike a fleet of identical Windows laptops, every industrial plant has a unique, customized configuration of sensors, controllers, and legacy hardware, making standardized security difficult to implement.

Methods

The paper summarizes three primary technical approaches to penetration testing within an industrial context:

  1. Port Scanning: This is the reconnaissance phase. Testers use tools like nmap to scan TCP and UDP ports to identify active services and potential entry points. The authors specifically discuss how firewalls in these environments handle packets (e.g., returning ICMP "unreachable" messages), which helps testers map the network's defensive perimeter.
  2. Dynamic External Injection Testing: This involves running the system's code in a controlled, real-world environment to observe its behavior under stress. The goal is to feed the system both "correct" and "erroneous" data to test its robustness and fault tolerance.
  3. Intrusion Trojan Detection: The paper describes a forensic method for dealing with "web page hanging horses"—malicious code implanted in web interfaces used to manage industrial systems. They highlight a specific technique using Referer field recursive analysis and Wireshark to trace the communication data, restore the malicious files, and locate the original source of the infection.

Datasets

Not discussed in the paper. The paper functions primarily as a technical review and methodology overview rather than an experimental study using a specific benchmark dataset. However, the authors emphasize that for dynamic injection testing, obtaining "real external input data" is critical, and when unavailable, high-fidelity simulated data must be used.

Key Insights and Ideas

  • The "Non-Interference" Rule: The authors draw a sharp line between a hacker's attack and a penetration test. A successful pen-test must identify vulnerabilities without affecting normal business operations or damaging the internal structure of the system. In an industrial setting, a system crash could mean a factory line stopping or a boiler exploding.
  • Individualized Customization: Because industrial systems are built to specific production requirements and process standards, security cannot be a "plug-and-play" solution. Every penetration test must be customized to the specific logic and hardware of the target plant.
  • Multi-Domain Threats: The paper notes that data in ICS isn't just moving through Ethernet cables; it travels via electromagnetic waves and other spatial transmissions. This "multi-domain" reality makes the intrusion channels more abundant and the testing more difficult.

Key Results

As a review paper, the "results" are qualitative rather than quantitative. The authors conclude that:

  • Penetration testing is the most effective mechanism for proving that a network's defenses are operating as intended.
  • The integration of deep learning is the inevitable next step for ICS security. As the scale of industrial data grows, manual detection becomes impossible, and AI will be required to identify subtle patterns of intrusion.
  • The use of recursive analysis on the Referer field is a proven way to accurately locate the storage location of malicious code in web-based management consoles.

Limitations and Future Work

The primary limitation identified is the sheer complexity and heterogeneity of ICS. Because no two systems are alike, creating a universal automated testing tool is currently out of reach.

For future work, the authors point toward:

  • Deep Learning Integration: Developing algorithms that can handle the massive data scales of modern industrial networks.
  • Multi-Algorithm Fusion: Combining different detection ideas to improve the comprehensive performance of security tests.
  • Addressing New Channels: Researching how to better secure non-traditional data transmission forms, such as electromagnetic spatial information, which remain vulnerable to interference and interception.

Summary updated

Abstract

The relationship between industrial control system and Internet is becoming closer and closer, and its network security has attracted much attention. Penetration testing is an active network intrusion detection technology, which plays an indispensable role in protecting the security of the system. This paper mainly introduces the principle of penetration testing, summarizes the current cutting-edge penetration testing technology, and looks forward to its development.
BibTeX
@inproceedings{jin2022research,
  author = {Kai Jin and Zhanji Niu and Jieping Liu and Jinxue Bai and Lei Zhang},
  title = {{Research on network security technology of industrial control system}},
  booktitle = {MATEC},
  year = {2022},
  doi = {10.1051/matecconf/202235503067},
}
Figure