Secure Reduced-dimensional Coding Scheme for Distributed Estimation with Communication Constraints

This paper addresses the problem of secure state estimation in distributed sensor networks with communication constraints. We propose a reduced-dimensional coding scheme based on the PredVAR model, which extracts dynamics from high-dimensional measurements while enhancing communication efficiency and privacy. A distributed estimator is developed under the proposed coding framework, and the impact of dimensionality reduction on estimation performance is analyzed. To defend against adversarial inference, we explicitly model a subspace-based eavesdropper and introduce a lightweight, time-varying perturbation strategy using orthogonal transformations. Simulation results demonstrate the effectiveness of our framework in balancing estimation accuracy, communication efficiency, and resilience against eavesdropping attacks.