Description

Computer systems are becoming more and more interesting for ultra-high dependable applications as for example flight control systems, where they replace and supplement traditional hydraulic or mechanical mechanisms. Such computer systems inherently require a distributed solution in order that the corruption of a part of the system does not cause the system to fail as a whole. The spatial distribution, however, implies the implementation of a communication infrastructure, such that the participants in the system are able to exchange information. For economic reasons the communication infrastructure is often realized as a shared resource and dedicated communication protocols have to be implemented to coordinate its usage. Such a communication protocol can be based on the synchronization of local clocks of the participants. This thesis addresses the design of fault-tolerant algorithms for initial synchronizing the local clocks of the participants in the distributed system as well as for recovery procedures to re-gain synchronization once it is lost.