Lossless Distributed Multirate Fusion Estimation Under Differential Privacy

This paper is concerned with the eavesdropping problem for distributed multi-rate fusion estimation systems, where smart sensors have different sampling rates. To prevent information leakage, differential privacy strategies are introduced into the field of multi-rate fusion systems. In this study, we define the period of steady-state fusion weights and prove that its size is unaffected by predictive compensation. To guarantee the differential privacy, two output perturbation mechanisms are proposed, where the injected noise is constructed through null space matrices and Gaussian mechanisms. Additionally, two local perturbation mechanisms are proposed to mitigate eavesdropping risks during local estimate transmission to the fusion center. Finally, a target tracking system is constructed to validate the effectiveness of the proposed privacy protection methods.