A new clustering algorithm for the topological design of multilevel data networks.

Abstract

This thesis presents IMULEC, a new clustering algorithm for the interactive design of multilevel data networks. The algorithm is capable of handling different types of concentrators and provides the line-layout of the data network, under various given constraints, such as: line capacities, maximum traffic per concentrator, maximum number of input ports to a concentrator etc. A comparison is made, for the single concentrator level case, with the algorithm NEWCLUST [3]. The results show that IMULEC reduces both network design cost and total execution times (with special savings in the clustering process).