A Novel Approach to Comparison-Based Diagnosis for Hypercube-Like Multiprocessor Systems

Abstract

Interconnection networks have been an active re- search area for parallel and distributed computer sys- tems. The diagnosability has played an important role in the reliability of interconnection networks. In this paper, we present a novel idea on system diag- nosis called local diagnosability. There is a strong relationship between the local diagnosability and the traditional global one. For this local sense, we focus more on a single processor and require only identify- ing the status of this particular processor correctly. We propose a sufficient condition to determine the local diagnosability of a given processor. Moreover, we prove that the diagnosability of an n-dimensional hypercube-like network HLn is n for n ≥ 5 in this local sense, and show that the local diagnosability of each node in an n-dimensional hypercube-like net- work equals to its degree with up to n − 2 faulty edges.