Analysis of false data injection attacks in power systems: A dynamic Bayesian game-theoretic approach

Area of Science:

• Cybersecurity
• Game Theory
• Power Systems Engineering

Background:

• False data injection (FDI) attacks pose a significant threat to the stability and reliability of power system state estimators.
• Existing methods for analyzing FDI attacks often struggle with incomplete information and dynamic environments.

Purpose of the Study:

• To propose a novel dynamic Bayesian game-theoretic approach for analyzing FDI attacks in power systems under incomplete information.
• To develop a robust method for identifying at-risk load measurements and enabling efficient resource allocation for defense.

Main Methods:

• A dynamic Bayesian game-theoretic framework is employed, incorporating a bi-level optimization model to define player payoffs.
• The attacker's type belief is continuously updated using historical data and measurement relationships.
• Convergence of type belief and Bayesian Nash equilibrium is mathematically proven, ensuring stability and reliability via the law of large numbers and central limit theorem.

Main Results:

• The proposed approach demonstrates high accuracy, achieving an average success rate of 98% in identifying at-risk load measurements.
• The dynamic Bayesian game-theoretic approach provides a computationally efficient solution with time complexity O(n_m*n_s*n_l) and space complexity O(1).

Conclusions:

• The dynamic Bayesian game-theoretic approach offers a powerful and efficient tool for detecting and mitigating FDI attacks in power systems.
• This method enables defenders to strategically allocate resources, thereby improving the overall security and resilience of power grids.