Hybrid fault diagnosability of the exchanged crossed cube under the HPMC model

The rapid growth of data-intensive and AI-driven applications has heightened the reliability and diagnosability requirements of high-performance computing systems, in which hybrid faults involving both nodes and links occur frequently. Although the exchanged crossed cube ECQs,t is a cost-effective interconnection network with strong topological properties, its hybrid fault diagnosability under contemporary hybrid models remains insufficiently studied. Recently, the concept of h-edge g-good-neighbor conditional diagnosability has been proposed to significantly enhance the accuracy of hybrid fault diagnosis. Motivated by this gap, this paper investigates the h-edge g-good-neighbor conditional diagnosability of ECQs,t under the HPMC model, and proves that tghECQs,t=2gs−g+2−h−1, under the constraints 2≤s≤t and 0≤h≤g≤s. Simulation experiments verify the correctness of the proposed theoretical results. This study not only establishes the exact hybrid diagnosability of ECQs,t but also demonstrates its superior capability in supporting hybrid fault diagnosis for large-scale parallel systems.