Detail

Verifying fault-tolerance requirements in architecture models

When designing and creating critical system a special attention must be given to dependability requirements. Such requirements do not change what the system does, but introduce constraints on how it should do it. Dependability requirements can be split into four categories: fault avoidance, fault removal, fault detection and fault tolerance. The goal of this work is not to introduce fault tolerance in an architecture, but to check if a given architecture satisfies the wanted fault tolerance requirements. In order to do so we create two views of the system: an architectural view, where a model of the architecture of the system is defined using a component-based architecture modeling language, and a dependability view, in which dependability strategies are used to decompose the system in a way that verifies the wanted fault tolerance requirements. To ensure that the requirements verified in the dependability view are valid in the architecture view, we introduce the concept of zone that binds the reliability view to the architecture view. By verifying that each zone satisfies a set of constraints given by the dependability strategies, we ensure that the architecture satisfies the fault tolerance requirements.