Trust Quantification for Networked Cyber-Physical Systems

Abstract

Cyber-physical systems (CPS) are highly integrated hardware–software devices that electro-mechanical components are tightly coupled with advanced computational algorithms for data collection, processing, communication, and control. Internet of Things is the emerging application of CPS. The main research challenge in designing CPS devices and systems is the quantification of complex system behaviors, such as consciousness, adaptation, and evolution. Particularly trust becomes an important element that affects system behavior in the networked society. To capture the unique human societal and systems aspects of trustworthiness quantification for CPS, in this paper, trustworthiness is measured by the perceptions of ability, benevolence, and integrity quantitatively. Ability measures one’s sensing and reasoning capability and influence to others. Benevolence captures the genuineness of intention and the extent of reciprocity in information exchange. Integrity provides the confidence about system dependability and predictability. A generic probabilistic graph model is developed to represent CPS functionality at mesoscale and demonstrate the perception-based quantification of ability and benevolence. Trust-based CPS network design and optimization are also demonstrated with the metrics of ability and benevolence.