Secure dimensionality reduction fusion estimation against eavesdroppers in cyber-physical systems

Area of Science:

• Cyber-Physical Systems
• Information Security
• Signal Processing

Background:

• Cyber-physical systems (CPS) face security challenges due to broadcasting wireless communication.
• Eavesdroppers can collude to intercept data, compromising system privacy.
• Limited bandwidth in CPS networks restricts data transmission and security measures.

Purpose of the Study:

• To develop a robust distributed dimensionality reduction fusion estimation strategy for CPS.
• To enhance data privacy and system security against eavesdropping in limited bandwidth environments.
• To propose a method for degrading eavesdroppers' fusion estimation accuracy while maintaining system functionality.

Main Methods:

• Developed an artificial noise (AN) insertion method based on physical processes and local estimation error covariance (EEC) matrices.
• Derived criteria for selection probabilities and successful decoding probabilities to ensure AN effectiveness.
• Established a sufficient condition for AN power to guarantee data confidentiality.

Main Results:

• The proposed AN insertion method degrades the fusion EEC of colluding eavesdroppers.
• Effectiveness of the AN strategy is guaranteed by specific selection and successful decoding probabilities.
• Derived AN power ensures system confidentiality against eavesdropping.

Conclusions:

• The artificial noise insertion strategy effectively enhances data privacy in distributed estimation for CPS.
• The method provides a practical solution for securing cyber-physical systems with limited bandwidth.
• Simulation results validate the proposed approach for improving system security and confidentiality.