Central processing unit for fault tolerant computing in Columbus

One of the subsystems common to all three elements in the COLUMBUS Space Station programme is the Data Management Subsystem, DMS. The DMS is based on distributed computers connected through a Local Area Network, LAN. In general the DMS is required to withstand one internal failure and still work in a full operational mode. This means that all equipment will at least be duplicated. The computers have to have a very high error self detection coverage since no external equipment is available for this purpose. A failing computer will shut down itself after an error has been detected. The other computers on the LAN are then responsible for the reconfiguration. This paper will in general describe the fault tolerance concept proposed for the COLUMBUS DMS and in particular detail the method to achieve the high error self detection coverage in the computers.