Author_Institution :
Dept. of Comput. Software, Aizu Univ., Fukushima, Japan
Abstract :
We give efficient algorithms for node disjoint path problems in incomplete star graphs which are defined in this paper to reduce the large gaps in the size of systems based on star graph topologies. Four disjoint path paradigms in incomplete star graphs are discussed: (1) disjoint paths between a pair of nodes s and t, (2) disjoint paths from a node s to a set T of nodes, (3) disjoint paths from a set S of nodes to a set T of nodes, and (4) disjoint paths between node pairs (si ,ti). We give algorithms which can find the maximum number of disjoint paths for these paradigms in optimal time. For an n-dimensional incomplete star graph Gn,m, the length of the disjoint paths constructed by our algorithms is at most d(Gn,m)+c, where d(Gn,m) is the diameter of G and c is a small constant. This paper also shows that the k-wide-diameter d n-2W(Gm,n), k-Rabin-diameter dn-2 R(Gm,n), k-set-diameter dn-2S(Gm,n), and k-pair-diameter dn-2P(Gm,n) of Gn,m are at d(Gn,m)+c
Keywords :
multiprocessor interconnection networks; incomplete star networks; n-dimensional incomplete star graph; node disjoint paths algorithms; Concurrent computing; Distributed computing; Fault tolerant systems; Hypercubes; Intelligent networks; Multiprocessing systems; Multiprocessor interconnection networks; Network topology; Routing; Software algorithms;