Testing of fault-tolerant hardware through partial control of inputs

The problem of testing fault-tolerant redundant digital systems is investigated. To test redundant systems through normal voter outputs, independent control of the output of each replicated unit is required. In the past it was assumed that independent control of the output of a replicated unit requires independent control of all of its inputs. The authors show that partial control of inputs is actually required. The critical input set problem, which is the problem of finding a set of inputs that need to be independently controlled, is formulated. Solutions are offered for different testing strategies, including exhaustive testing and deterministic testing, and for different levels of circuit description