- Title
- Privacy-preserving control of multiagent systems
- Creator
- Fang, Wentuo
- Relation
- University of Newcastle Research Higher Degree Thesis
- Resource Type
- thesis
- Date
- 2022
- Description
- Research Doctorate - Doctor of Philosophy (PhD)
- Description
- This thesis is aimed at developing secure and privacy-preserving consensus algorithms for networked systems. Given the technical challenges arising from the decentralised design of such algorithms, existing findings are limited to systems with simple first-order dynamics. Similar to other control problems, bridging the gap between first-order and higher-order dynamics requires more advanced technical developments. This thesis presents a Paillier encryption--based average consensus algorithm for a networked system with second-order dynamics and randomness added to the network weights. The conditions for privacy preservation, especially those that depend on consensus rate, are thoroughly studied, with theoretical analysis and numerical verification. For networked systems under directed topology, a new privacy-preserving algorithm is developed to achieve confidential consensus. As a prerequisite to applying the algorithm, a structural consensus problem and solvability conditions for an explicitly constructed controller are presented. The controller is then implemented with encryption to achieve consensus while preventing the leakage of individual information to external eavesdroppers and/or malicious internal neighbours. When applying algorithms to a practical multi-agent system, limitations in bandwidth and computing power necessitate quantisation. Further, because each piece of quantised data has a boundary, signals in the system, including control input, are constrained, which may result in the system becoming unstable. Therefore, the non-uniform, non-convex input--constrained flocking control problem of continuous-time multi-agent systems is investigated. A distributed flocking control algorithm is proposed for each agent using local information from neighbouring agents subject to non-uniform and non-convex control constraints. Based on a constraint scaling factor and a specific coordinate transformation, a Lyapunov function is constructed to address the nonlinearities caused by input constraints. It is demonstrated that all agents from an explicitly specified region of initial states and the same velocity eventually converge to a common point, while each agent's inputs remain within their own constrained non-convex set. A simulation example is given to demonstrate the effectiveness of the theoretical results.
- Subject
- multi-agent system; privacy-preserving; input constraints; Paillier encryption; flocking; consensus; network security
- Identifier
- http://hdl.handle.net/1959.13/1512443
- Identifier
- uon:56617
- Rights
- Copyright 2022 Wentuo Fang
- Language
- eng
- Full Text
- Hits: 29
- Visitors: 43
- Downloads: 15
| Thumbnail | File | Description | Size | Format | |||
|---|---|---|---|---|---|---|---|
| View Details Download | ATTACHMENT01 | Thesis | 1 MB | Adobe Acrobat PDF | View Details Download | ||
| View Details Download | ATTACHMENT02 | Abstract | 243 KB | Adobe Acrobat PDF | View Details Download |