Privacy-aware and resource-saving network control

Abstract

The broadcasting nature of the fast developing wireless networking technologies has aroused concerns about data privacy since neighboring illegitimate users are able to overhear confidential information contained in the transmitted data. At the same time, due to the limited resources of the remote sides, how to use fewer resources without sacrificing too much system performance has also attracted increasing interest. In this thesis, we focus on privacy-aware and resource-saving network control problems in wireless networks. Specifically, we investigate an optimal encryption scheduling under an operation constraint, introduce an event-triggered mechanism into cognitive radio sensor networks (CRSNs), and propose a joint sensor and actuator placement to minimize an infinite linear-quadratic Gaussian (LQG) cost. For privacy-aware network control, we investigate the optimal encryption scheduling for remote state estimation under an operation constraint. As the information about eavesdroppers is unknown to the estimator, we introduce the concept of eavesdropper-invariant schedules and derive associated structural results. In addition, we propose a practical algorithm that compares a finite number of points to obtain an ε-optimal encryption schedule. For resource-saving network control, we introduce a stochastic event-triggered mechanism into CRSNs. To achieve a better trade-off between the estimation performance and communication consumption, we propose both open-loop and closed-loop schedulers. The parameter design problems in both schedules are efficiently solved by convex programming. We also consider the joint sensor and actuator (SaA) placement to minimize the infinite-horizon LQG cost for a discrete dynamic noisy system. This problem is first reformulated as a joint bilinear problem by relaxing the Boolean constraints. After deriving a compact search region that the optimal solution of the relaxed problem belongs to, we introduce a branch and bound (B&B) algorithm to obtain the global optimal solution of the relaxed problem. We then generate a suboptimal solution to the original problem from the relaxed one and further analyze the optimality gap. Finally, we illustrate the results with numerical examples for the above problems. At the end of this thesis, we also provide some possible directions for future work.

Date of Award	2021
Original language	English
Awarding Institution	The Hong Kong University of Science and Technology
Supervisor	Ling SHI (Supervisor)