Privacy-Preserving Fusion for Multi-Sensor Systems Under Multiple Packet Dropouts

TL;DR

This paper introduces a privacy-preserving fusion estimation method for multi-sensor wireless networks that effectively handles packet dropouts and eavesdropping, ensuring secure and accurate state estimation.

Contribution

It proposes a distributed encoding-based privacy mechanism integrated with a control-theoretic fusion framework, addressing privacy and robustness in sensor networks under packet loss and attacks.

Findings

The method maintains bounded estimation error for legitimate users.

Eavesdropper's estimation error diverges, ensuring data confidentiality.

Simulation confirms effectiveness in a three-tank system.

Abstract

Wireless sensor networks (WSNs) are critical components in modern cyber-physical systems, enabling efficient data collection and fusion through spatially distributed sensors. However, the inherent risks of eavesdropping and packet dropouts in such networks pose significant challenges to secure state estimation. In this paper, we address the privacy-preserving fusion estimation (PPFE) problem for multi-sensor systems under multiple packet dropouts and eavesdropping attacks. To mitigate these issues, we propose a distributed encoding-based privacy-preserving mechanism (PPM) within a control-theoretic framework, ensuring data privacy during transmission while maintaining the performance of legitimate state estimation. A centralized fusion filter is developed, accounting for the coupling effects of packet dropouts and the encoding-based PPM. Boundedness conditions for the legitimate user's estimation error covariance are derived via a modified algebraic Riccati equation. Additionally, by demonstrating the divergence of the eavesdropper's mean estimation error, the proposed PPFE algorithm's data confidentiality is rigorously analyzed. Simulation results for an Internet-based three-tank system validate the effectiveness of the proposed approach, highlighting its potential to enhance privacy without compromising estimation accuracy.