An Algorithm Providing Fault - Tolerance for Layered Distributed Systems

This paper presents a new approach for the fault-tolerance in layered distributed systems. We develop the dynamic regeneration method in which faulty software compone nts are dynamically regenerated. In contrast to other techniques which duplicated critical components, this method does not increase the complexity of the system and tolerates an unlimited numbe r of failures. To apply this technique, we developp a method for the design of reliable systems. We tran sform an initial software architecture in a model which encloses homogeneous elements. Our study relies on t he OSI standard model defined within the ISO organization. So, from an initial layered distributed architecture, we exhibit a homogeneous communication chain. The building and the maintenance of this chain in an unreliable environment is achieved by an algorithm using dynamic regeneration .