Abstract
Fault tolerance (FT) is a critical aspect of industry, where systems are susceptible to disturbance and faults. Traditional FT models, based on the centralization of information to handle fault episodes, no longer meet the current industrial models based on Cyber-physical Systems (CPS). Self-healing is a promising approach for FT in CPS, consisting of the individual competence of each component in detect, diagnose and recover from failures. With this in mind, this paper discusses the engineering of self-healing fault-tolerance in industrial CPS, analyzing the maturation process of this paradigm from the local model through collaboration models and later to self-organization features. The paper also discusses the main research challenges that self-healing FT faces during this process.