Consensus-Based Power System State Estimation Algorithm Under Collaborative Attack

Zhijian Cheng, Guanjun Chen, Xiaomeng Li, Hongru RenSensors 2024 · DOI

TL;DR

This paper introduces a novel distributed state estimation algorithm for power systems that can withstand collaborative cyberattacks, specifically Denial of Service (DoS) and False Data Injection (FDI) attacks. By integrating these attack models into a consensus-based framework, the algorithm aims to maintain accurate state estimation even when faced with sophisticated, coordinated threats.

What Problem Does This Paper Address?

Modern power systems are increasingly complex, relying on vast networks of sensors and communication systems to monitor their operational state. This interconnectedness, while enabling efficient control, also makes them vulnerable to cyberattacks. Specifically, attackers can disrupt operations by either overwhelming communication channels (DoS attacks) or injecting false data to mislead the system (FDI attacks). These attacks can compromise the accuracy of state estimation, which is crucial for grid stability and reliable power delivery. Existing research often focuses on single attack types, leaving a gap in understanding and defending against collaborative attacks where multiple attack vectors are used simultaneously. This paper addresses this critical gap by developing a robust distributed state estimation algorithm capable of handling these combined threats.

Methods

The core of this paper is a consensus-based distributed state estimation algorithm. Here's a breakdown of the key methodological components:

  • Hybrid Measurement Model: The algorithm accounts for measurements from both traditional Remote Terminal Units (RTUs) and modern Phasor Measurement Units (PMUs), reflecting the reality of mixed-technology power grids.
  • Collaborative Attack Modeling: The authors construct specific mathematical models for:
  • DoS Attacks: Targeting the innovation signal transmitted from sensors to estimators.
  • FDI Attacks: Targeting the estimation information exchanged between neighboring estimators. These models are integrated to simulate realistic collaborative attack scenarios.

  • Consensus-Based Distributed Estimator: Building upon existing consensus theory, the algorithm allows local estimators to iteratively refine their state estimates by exchanging information with their neighbors. This distributed approach avoids the computational burden of centralized methods.

  • Suboptimal Filter Design: To reduce computational complexity, a suboptimal filter gain is derived by making certain approximations. This is crucial for scalability in large power systems.
  • Lyapunov Stability Analysis: The paper rigorously proves the stability of the proposed estimator using Lyapunov stability theory. This provides a theoretical guarantee that the algorithm will converge to accurate estimates under specified conditions.
  • Consensus Gain Matrix Design: A method is provided to design the consensus gain matrix, which is critical for ensuring stable and effective information exchange between estimators.

Datasets

The paper evaluates its proposed algorithm using a simulated IEEE benchmark 14-bus power system. This is a standard, well-established benchmark in power system research, allowing for reproducible comparisons. The authors mention that the measurement configuration is based on existing literature ([45]).

Key Insights and Ideas

  • Collaborative Attacks are a Real Threat: The paper highlights that attackers can combine different attack strategies (DoS and FDI) to achieve more significant disruption than single-attack methods.
  • Consensus is Key to Robustness: The distributed consensus mechanism is shown to be effective in maintaining estimation accuracy even under these complex attack scenarios. By averaging and agreeing on estimates, the system can filter out malicious information.
  • Suboptimality for Scalability: The trade-off between optimal performance and computational feasibility is addressed by proposing a suboptimal estimator. This practical approach makes the algorithm more applicable to real-world, large-scale power grids.
  • Theoretical Guarantees Matter: The use of Lyapunov stability analysis provides a strong theoretical foundation, assuring researchers and engineers that the proposed method is reliable.

Key Results

The paper demonstrates the effectiveness of its proposed algorithm through simulations on the IEEE 14-bus system:

  • Accurate State Tracking: Figures 3 and 4 show that the estimated states (real and imaginary parts of bus voltage) closely track the actual states, even under collaborative attacks.
  • Robustness to Different Attack Scenarios: Figures 6 and 7 compare the Mean Square Error (MSE) under three scenarios: collaborative attacks, only FDI attacks, and only DoS attacks. The results indicate that the MSE does not change significantly across these scenarios, demonstrating the algorithm's robustness to both individual and combined attacks.
  • Impact of DoS Probability: Figures 8 and 9 show that as the probability of DoS attacks increases (lower $\alpha_1$), the estimation performance degrades, confirming that collaborative attacks do indeed impact performance, but the algorithm still maintains a reasonable level of accuracy.
  • Comparison with Existing Methods: Figures 10 and 11 compare the proposed algorithm (DCEA) with a decentralized approach (DEA) from a previous study ([24]). The DCEA consistently outperforms DEA, particularly highlighting the benefit of the consensus term in improving estimation performance.

Limitations and Future Work

  • Suboptimal Nature: While the suboptimal approach reduces computational cost, it might lead to slightly less accurate estimates compared to a hypothetical optimal solution. The paper notes that the introduced error is "reasonable" and stability is maintained.
  • Linear System Assumption: The current model is based on linear discrete-time systems. Real-world power systems can exhibit nonlinear dynamics.
  • Fixed Communication Topology: The analysis assumes a fixed communication graph. Dynamic or changing network topologies could pose further challenges.

The authors suggest several promising directions for future research:

  • Extending to Nonlinear Systems: Adapting the algorithm to handle nonlinear power system dynamics.
  • Extending to Nonlinear Power Systems: A broader goal of applying these techniques to more complex nonlinear power system models.
  • Redundancy for Sparse Attacks: Investigating system observability redundancy to defend against sparse sensor attacks.

Summary updated

Abstract

Due to its vulnerability to a variety of cyber attacks, research on cyber security for power systems has become especially crucial. In order to maintain the safe and stable operation of power systems, it is worthwhile to gain insight into the complex characteristics and behaviors of cyber attacks from the attacker's perspective. The consensus-based distributed state estimation problem is investigated for power systems subject to collaborative attacks. In order to describe such attack behaviors, the denial of service (DoS) attack model for hybrid remote terminal unit (RTU) and phasor measurement unit (PMU) measurements, and the false data injection (FDI) attack model for neighboring estimation information, are constructed. By integrating these two types of attack models, a different consensus-based distributed estimator is designed to accurately estimate the state of the power system under collaborative attacks. Then, through Lyapunov stability analysis theory, a sufficient condition is provided to ensure that the proposed distributed estimator is stable, and a suitable consensus gain matrix is devised. Finally, to confirm the viability and efficacy of the suggested algorithm, a simulation experiment on an IEEE benchmark 14-bus power system is carried out.
BibTeX
@inproceedings{cheng2024consensus,
  author = {Zhijian Cheng and Guanjun Chen and Xiaomeng Li and Hongru Ren},
  title = {{Consensus-Based Power System State Estimation Algorithm Under Collaborative Attack}},
  booktitle = {Sensors},
  year = {2024},
  doi = {10.3390/s24216886},
}
Figure