Distributed Cyber-physical systems (DCPS) represent a new field in automatic control and recently they are increasingly used in life critical system, where the probability of tragic failure has to be kept below very low levels. In other side, fault tolerance has been used to face failures and achieve best performance of systems. Thus, it becomes critical to develop models that are tolerant toward the failure of system. This will enable guarantee of DCPS that will continue to run even through failure status. In this paper we take self-stabilizing as an example of fault tolerance in distributed cyber physical systems, where self-stabilization provides non-masking approach to fault tolerance. Then the Event-B approach is presented as formal method for system-level modelling and analysis. It is proposed using the Rodin modelling tool for Event-B that integrates modelling and proving.