Performance of fault-tolerant diagnostics in the hypercube systems

Journal Article · Tue Aug 01 04:00:00 EDT 1989 · IEEE Trans. Comput.; (United States)

DOI:https://doi.org/10.1109/12.30870· OSTI ID:5587654

Ghafoor, A; Sole, P

In this paper, they introduce the concept of fault-tolerant self-diagnosis for distributed systems and show that there exists a performance tradeoff between the complexity of a self-diagnostic algorithm and the level of fault tolerance inherited by the algorithm. For the study, they select hypercube systems and show that designing an optimal algorithm for such systems has an equivalent coding theory formulation which belongs to the class of NP-hard problems. Subsequently, they propose an ''efficient'' diagnostic scheme for these systems and study the performance tradeoff of the proposed algorithm which is based on a combinatorial structure called Hadamard matrix. The authors make an essential use of its properties of symmetrical partitioning and covering in hypercube networks. Using known translate weight distributions, they evaluated the tradeoff between the fault tolerance and traffic complexity of the proposed diagnostic algorithm for hypercubes of small sizes. An interesting compromise is exhibited for the hypercube with an arbitrary size.

• Research Organization: Syracuse Univ., NY (USA). Dept. of Electrical and Computer Engineering; School of Computer and Information Science, Syracuse Univ., Syracuse, NY (US)
• OSTI ID: 5587654
• Journal Information: IEEE Trans. Comput.; (United States), Journal Name: IEEE Trans. Comput.; (United States) Vol. 38:8; ISSN ITCOB
• Country of Publication: United States
• Language: English

Related Subjects

99 GENERAL AND MISCELLANEOUS
990210* -- Supercomputers-- (1987-1989)
ALGORITHMS
COMPUTERS
DATA PROCESSING
DESIGN
DIGITAL COMPUTERS
DISTRIBUTED DATA PROCESSING
FAULT TOLERANT COMPUTERS
HYPERCUBE COMPUTERS
MATHEMATICAL LOGIC
PERFORMANCE
PROCESSING
SIZE