The non-inclusive g -extra diagnosability of the exchanged hypercube under the PMC and MM* models

Abstract

Reliability evaluation of interconnection networks is essential for large-scale parallel systems. Traditional diagnostic strategies assume an inclusion relationship among fault sets, although the probability of such inclusion decreases rapidly as network size increases, resulting in conservative diagnosability estimates. To address this limitation, non-inclusive diagnosability eliminates the fault-set inclusion constraint, while the non-inclusive g-extra strategy further incorporates g-extra connectivity. The present study analyzes the exchanged hypercube (EHs,t) to establish its non-inclusive g-extra diagnosability under the PMC and MM* models. By defining non-inclusive fault sets and exploiting (2g +1)-extra cut properties, we establish the diagnosability under both models. For 3≤s≤t and 2g+2≤s, the non-inclusive g-extra diagnosability of EHs,t under both the PMC and MM* models is tNgEHs,t=2g+2s−g+1+g. These findings improve the accuracy of reliability evaluation for exchanged hypercubes in practical fault scenarios.