Abstract:
Self-diagnosing of modular computing systems (CS) is investigated on the system level at the presence of multiple faults and with the use of unreliable tests. Dependence of self-diagnosing efficiency on properties of the unreliable tests is studied by the method of imitation-statistical simulation. We compare the self-diagnosing efficiency for CS using the unreliable tests corresponding to the known PMC-model and the unreliable tests suggested by the authors, and state conditions under which the use of the last tests is more effective.