Modeling and Detecting False Data Injection Attacks against Railway Traction Power Systems

Subhash Lakshminarayana, Zhan-Teng Teo, Rui Tan, David K. Y. Yau

Introduction

Modern urban railway networks are sophisticated Cyber-Physical Systems (CPS) that rely on integrated communication and control for efficient operation. A critical component of this infrastructure is the Traction Power System (TPS), which delivers electrical energy to trains via third rails or overhead lines. While traditionally protected by "air-gapping," the increasing connectivity of these systems has exposed them to cyber-threats, specifically False Data Injection (FDI) attacks. In these scenarios, an attacker compromises local sensors to feed malicious data into the control loops, potentially leading to massive energy waste or catastrophic electrical failures.

The Traction Power System Model

In a typical DC traction system, power is supplied by substations and consumed or generated by trains as they move along the track. The electrical topology is highly dynamic because the "nodes" (trains) are constantly changing position, which in turn changes the resistance between components.

Electrical topology of a typical DC Traction Power System with two substations and two trains.

As shown in the figure above, trains interact with the grid in two primary modes: tractioning (consuming power to accelerate) and regenerating (injecting power back into the grid during braking). To maintain system stability, trains use local control logic to adjust their power demand based on the measured line voltage ($V_i$).

Control characteristics of trains: (a) tractioning power reduction at low voltages and (b) regenerative power reduction at high voltages.

This control logic, illustrated in the graphs above, creates a vulnerability. By injecting false voltage readings, an attacker can trick a train into unnecessarily reducing its regenerative braking (wasting energy) or drawing excessive power, which could trip circuit breakers and cause sudden, dangerous braking incidents.

Global Attack Detection (GAD) Framework

To counter these threats, the authors propose a multi-layered Global Attack Detection (GAD) system. The framework operates at a central monitoring station and leverages the unique physics of moving trains to identify even "stealthy" attacks that might bypass traditional security measures.

  1. Bad Data Detection (BDD): This initial layer uses state estimation to check if reported sensor values are physically consistent with the global electrical model. While effective against crude attacks, sophisticated attackers can craft "stealthy" vectors that satisfy these basic equations.
  2. Position Integrity Verification (PIV): Because the electrical model depends entirely on the train's location, the system cross-checks multiple sources (GPS, wheel sensors, track-side beacons) to ensure the attacker hasn't spoofed the train's position.
  3. Secondary Attack Detection (SAD): This is the core innovation. The authors prove a "Discrete Solution Property": if the train's position is known and fixed, the electrical equations of the TPS have only a finite number of discrete solutions. A smart attacker trying to remain stealthy must choose one of these specific points. SAD monitors the "distance" between the reported state and the previous known-good state; if the jump is inconsistent with the physics of the previous time-step, an alarm is raised.

Key Results and Insights

The proposed system was evaluated using a MATLAB simulation of a realistic 10km track with four trains and four substations. The results demonstrate the severe impact of undefended FDI attacks and the effectiveness of the GAD framework:

  • Impact of Attacks: Without detection, FDI attacks increased total system energy consumption by 28.3% and successfully breached safety-critical voltage limits.
  • Detection Performance: The novel SAD algorithm achieved a 96% detection probability for sophisticated attacks that had already bypassed the initial BDD layer.
  • Operational Reliability: By using an adaptive sliding window (GAD-W), the system maintained an extremely low False Positive rate of 0.0009, ensuring that legitimate fluctuations in power demand do not trigger false alarms.

Conclusion

By exploiting the physical constraints of moving trains—specifically the relationship between position and electrical state—the GAD framework provides a robust defense against cyber-physical sabotage in railway networks. While the system currently focuses on DC urban metros, the underlying principle of using discrete physical solutions for anomaly detection offers a promising path for securing other dynamic infrastructure systems.

Summary updated

Abstract

Modern urban railways extensively use computerized sensing and control technologies to achieve safe, reliable, and well-timed operations. However, the use of these technologies may provide a convenient leverage to cyber-attackers who have bypassed the air gaps and aim at causing safety incidents and service disruptions. In this article, we study False Data Injection (FDI) attacks against railway Traction Power Systems (TPSes). Specifically, we analyze two types of FDI attacks on the train-borne voltage, current, and position sensor measurements—which we call efficiency attack and safety attack—that (i) maximize the system’s total power consumption and (ii) mislead trains’ local voltages to exceed given safety-critical thresholds, respectively. To counteract, we develop a Global Attack Detection (GAD) system that serializes a bad data detector and a novel secondary attack detector designed based on unique TPS characteristics. With intact position data of trains, our detection system can effectively detect FDI attacks on trains’ voltage and current measurements even if the attacker has full and accurate knowledge of the TPS, attack detection, and real-time system state. In particular, the GAD system features an adaptive mechanism that ensures low false-positive and negative rates in detecting the attacks under noisy system measurements. Extensive simulations driven by realistic running profiles of trains verify that a TPS setup is vulnerable to FDI attacks, but these attacks can be detected effectively by the proposed GAD while ensuring a low false-positive rate.
BibTeX
@article{lakshminarayana2018modeling,
  author = {Subhash Lakshminarayana and Zhan-Teng Teo and Rui Tan and David K. Y. Yau},
  title = {{Modeling and Detecting False Data Injection Attacks against Railway Traction Power Systems}},
  journal = {ACM Trans. Cyber Phys. Syst},
  year = {2018},
  doi = {10.1145/3226030},
}
Figure