Advanced Institute of Convergence Information Technology
摘要:
The performance of parallel and distributed computer systems depends on the effectiveness of the underlying interconnection network. Embedding is ofgreat importance in the applications ofparallel and distributed computer systems. This paper presents embedding multiple networks into a faulty Incrementally Extensible Hypercube, including linear array, ring, mesh, torus, and double-rooted complete binary tree. To obtain the replaceable node ofthe faulty node, 2-expansion is permitted such that up to (n+ 1) faults can be tolerated with dilation 3, congestion I and load 1. It is demonstrated that O(n) faults can be tolerated. Furthermore, the presented embedding methods are optimized mainly for balancing the processor loads, while minimizing load as far as possible. Based on these results, this methodology of embedding enables extremely high-speed parallel and distributed computation. Moreover, the technology can be applied in grid computing and cloud computing.
