More on the Efficiency of Interval Routing.pdfVIP

More on the Efficiency of Interval Routing Savio S.H. Tse and Francis C.M. Lau Department of Computer Science The University of Hong Kongfsshtse,fcmlaug@cs.hku.hk March 1997 Abstract Interval routing is a space-efficient routing method for computer networks. The method is said to be optimal if it can generate optimal routing paths for all messages. A path is optimal if it is a shortest path between the two nodes in- volved. A seminal result (The Computer Journal, 34 (1991), 475–476) in the area, however, has pointed out that “the interval routing algorithm cannot be op- timal in networks with arbitrary topology.” The statement is correct but the lower bound that was derived is not. We give the counterproof in this paper, and the corrected bound. 1 Introduction Interval routing [5, 7] is a space-efficient routing method for general topologies. It keeps a table of sizeO(d log n) at every node, where d is the degree of the node, andn is the number of nodes. Interval routing has attracted a fair amount of attention in recent years partly because it has been adopted as the routing method in a com- mercial routing chip [3]. The space concern aside, an interval routing scheme (IRS), in order to be practical, must try to achieve reasonable performance which is mea- sured in terms of the lengths of the routing paths generated by the scheme. An IRS is optimal if all the routed paths are shortest paths.Correspondence: Dr F.C.M.Lau, Department ofComputer Science, TheUniversity ofHongKong, Hong Kong / Email: fcmlau@cs.hku.hk / Fax: (+852) 2559 8447 1 The graphwe consider is connected, of whichE is the set of edges, and V the set of the nodes. Every edge inE is bidirectional. There are n nodes in V . To implement interval routing, each node is labeled with a unique integer, called node number, from the cyclicly ordered set L = f0; : : : ; n 1g.1 In the following, we identify a node by its node number. In the simplest kind of IRS, the one-label interval routing scheme (1-IRS),