A novel approach to fault tolerant computing [in space systems]

The realization of fault tolerant computers requires a considerable effort, both for their development and validation. In addition, the redundancy required to achieve the fault tolerance increases power consumption, mass and volume of the computers. In order to mitigate these problems, a standardized Fault Management Element (FME) has been developed, in which the complete set of fault management functions necessary to realize fault tolerant computers are provided "once and for all" in standardized and fully validated form. The fault management technology of this FME is based on the "Byzantine" fault tolerant computer for the Russian Service Module of the International Space Station, and for the logistics vehicle ATV servicing the station. Using this FME, fault tolerant computers need not be developed in the usual sense, but are basically realized by a configuration process, which, in simplified terms, comprises an integration of an FME with each of the foreseen redundant application processor boards, and their cross-strapping via the preconceived high-speed data links of the FMEs. By this novel approach the considerable development and validation effort to realize fault tolerant computers is practically eliminated. Moreover, the FME, which is currently available as printed circuit board, will be available as ASIC in the near future, such that the above mentioned power, mass and volume problems are also greatly reduced.