Abstract
The emerging dynamic architectures of autonomous digital ecosystems raise new challenges in the process of assuring trust and safety. In particular, the admission of software smart agents into autonomous dynamic ecosystems will become a significant future topic. In this work we propose the concept of predictive simulation, which elevates from the concept of virtual Hardware-in-the-Loop (vHiL) testbed, to support rapid runtime evaluation of software smart agents in autonomous digital ecosystems. Based on this testbed, we introduce a novel strategy for building trust in software components that enter an ecosystem as black boxes without executing their behavior which can be potentially malicious, but by executing corresponding digital twins which are abstract models fed with real-time data.
