Towards Creation of Automated Prediction Systems for Trust and Dependability Evaluation

We advance the ability to design reliable Cyber-Physical Systems of Systems (CPSoSs) by integrating artificial intelligence to the engineering methods of these systems. The current practice relies heavily on independent validation of software and hardware components, with only limited evaluation during engineering integration activities. Furthermore, our changing landscape of real-time adaptive systems allows software components to be dynamically included or re-distributed within a Cyber-Physical System (CPS), with mostly unknown implications on the overall systems integrity, reliability and security. This paper introduces an approach consisting of scientific and engineering processes that enable development of concepts for automated prediction systems for evaluating the dependability and trust of CPSoS. This significantly moves the security and reliability design process ahead by opening the doors for far more relevant design strategies and the opportunity to develop protocols, methods, and tools aimed at dealing with a wide variety of platforms with poorly calibrated reliability characteristics.