A fault-tolerant multiple processor system for space instrumentation

The paper is not based on the design for any single space project but has several ESA, NASA and Soviet projects as the background for the development of a system which will automatically reconfigure itself when a failure occurs. A functional triplicated TMR (triple modular redundancy) configuration is designed which allows the system performance to degrade gracefully after a sequence of individual failures. The distributed system is maintained by use of the 'system information' which is the data which the system holds concerning its own status. The management of multiple redundant copies of this system information allows majority voting, modular fault recovery and system reconfiguration.<
>