A comparative evaluation of four basic system-level diagnosis strategies for hypercubes

Sequential diagnosis for BC graphs

“…For sufficient 10g(N It) large N, it was shown in[5] that this inequality issatisfied by t = J N log log N , where 0< J � 1. Therefore,10gN the lower bound of the degree of diagnosability is O( NloglogN ) .…”

“…and Kim[5] first gave an estimate on the degree of sequential diagnosability for an n-dimensional hypercube.Khanna …”

An adaptive system-level diagnosis approach for hypercube multiprocessors

“…Although we cannot prove that the numbers of tests and test links required by our algorithm are also lower than in the t-fault pessimistic diagnosis algorithm, it is shown to be correct in the next section by probabilistic analysis. In Figures 6 and 7, the traditional diagnosis algorithm for hypercubes in 111, the adaptive safe diaposis algorithm in [SI and t 6 e one-step t-fault pessimistic diagnosis algorithm for hypercubes in [7], are chosen for comparison with our approach in terms of the three measures of diagnosis cost. Note that the adaptive safe diagnosis algorithm can have different options for the increment in the connectivity of the testing graph at each iteration [SI, which may result in different diagnosis costs.…”

“…Thus we call this approach a hierarchical adaptive diagnosis scheme. The main results for this scheme is that its diagnosis cost is not only lower than the traditional diagnosis algorithm in [ll], but it is also lower than the adaptive diagnosis method of [8] and the pessimistic diagnosis method of [7]. This paper is organized as follows.…”

“…There are two kinds of diagnosis, which are socalled precise diagnosis and pessimistic diagnosis [7]. Different from precise diagnosis [4], .pessimistic diagnosis [5] identifies the suspicious units instead of the exactly faulty units.…”