Jamming attack against remote state estimation over multiple wireless channels: A reinforcement learning based game...

Area of Science:

• Cyber-Physical Systems
• Game Theory
• Wireless Communication Security

Background:

• Wireless remote estimation systems (RESs) are vulnerable to channel jamming attacks.
• Smart attackers can adapt to sensor countermeasures, creating a dynamic game scenario.
• Understanding this sensor-attacker interplay is crucial for robust system design.

Purpose of the Study:

• To investigate the strategic interactions between sensors and attackers in wireless RESs.
• To model both closed-loop and open-loop scenarios based on attacker knowledge.
• To develop effective countermeasures against channel jamming attacks.

Main Methods:

• Designed a potential game model to represent sensor-attacker dynamics.
• Considered scenarios where sensor actions are known (closed-loop) or unknown (open-loop) to the attacker.
• Employed reinforcement learning algorithms to find game solutions.

Main Results:

• The proposed game model effectively captures the strategic interplay.
• Reinforcement learning methods successfully yielded solutions for both scenarios.
• Simulation results validated the effectiveness of the developed approach.

Conclusions:

• The study provides a robust framework for analyzing security in wireless RESs.
• Reinforcement learning offers a viable solution for mitigating jamming attacks.
• The findings contribute to enhancing the resilience of cyber-physical systems.