Design and Verification of Trusted Collective Adaptive Systems

Abstract

Collective adaptive systems (CAS) often adopt cooperative operating strategies to run distributed decision-making mechanisms. Sometimes, their effectiveness massively relies on the collaborative nature of individuals’ behavior. Stimulating cooperation while preventing selfish and malicious behaviors is the main objective of trust and reputation models. These models are largely used in distributed, peer-to-peer environments and, therefore, represent an ideal framework for improving the robustness, as well as security, of CAS. In this article, we propose a formal framework for modeling and verifying trusted CAS. From the modeling perspective, mobility, adaptiveness, and trust-based interaction represent the main ingredients used to define a flexible and easy-to-use paradigm. Concerning analysis, formal automated techniques based on equivalence and model checking support the prediction of the CAS behavior and the verification of the underlying trust and reputation models, with the specific aim of estimating robustness with respect to the typical attacks conducted against webs of trust.