| Abstract: | A new diagnosis method, called Double-Syndrome diagnostic, is proposed, which can identify faulty nodes by comparing 2 different syndromes. For the same system, the average number of faulty nodes identified correctly by the Double-Syndrome diagnostic is much greater than the t-diagnosability and the(t_1/t_1)-diagnosability of the system. Furthermore, in order to identify the remaining faulty nodes in the system, two strategies of fault diagnostic are proposed, one is called(k, t)-fault diagnosable strategy, another is called(k, t/t)-fault diagnosable strategy. Besides, the conditional(k, t)-diagnosable((k, t/t)-diagnosable) system is introduced. Furthermore, the conditional diagnosabilities are proved for some regular(k, t)-diagnosable and(k, t/t)-diagnosable networks such as n-dimensional hypercube network and n-dimensional star network. And then, for a system, its(k,t)-conditional diagnosability and its(k, t/t)-conditional diagnosability are identical, and in the worst case, they are equal to their traditional conditional diagnosability. |
