A method of ensuring fault tolerance in redundant computer control systems

A method is worked out for ensuring fault tolerance at the symptoms of non- byzantine and byzantine faults in redundant computer control systems of a d efinite class. The consistency of results of the local processing of fault symptoms for all channels is provided by the original structure of checkpoi nts and also by the algorithm for ensuring fault tolerance with the use of a system log and hardware support-a localization unit. The method under dis cussion insignificantly reduces the capacity of a system, makes it possible to counteract byzantine faults at a number of channels in the system equal to three, and totally precludes their masking (disguise).