Reliability, failure detection and prevention in cyber-physical systems (CPSs) with agents

The complexity of cyber-physical systems (CPSs) opens up several challenges regarding the assurance of high levels of reliability. Therefore, we need to develop a new data quality analysis and new CPS components/subsystems for the detection, isolation, recovery, and prevention of failures and intrusion events. Furthermore, we have to develop new reliability assessment models and metrics in order to properly evaluate CPSs. Agent technologies, through their characteristics such as autonomy, social ability, reactivity, and proactivity, have the potential to address these challenges and requirements. We propose a multi-agent based solution, namely ReliaCPS, for detecting and preventing the failures of the system components of CPSs. We also use a case study of a CPS for monitoring ambient parameters to evaluate the performance of the proposed approach. Our results indicate improvements in terms of reliability metrics (32.69% for the mean time between failures and mean time to failure, 50% for the mean time to repair and 0.94% for reliability function over time).