 
Summary: Exact algorithm for delayconstrained capacitated
minimum spanning tree network
Y.J. Lee and M. Atiquzzaman
Abstract: The delayconstrained capacitated minimum spanning tree (DCCMST) problem of
finding several broadcast trees from a source node is discussed. While the traditional CMST
problem deals with only the traffic capacity constraint served by a port of the source node, and
delayconstrained minimum spanning tree (DCMST) considers only the maximum endend
delay constraint, the DCCMST problem deals with both the mean network delay and traffic
capacity constraints. The DCCMST problem consists of finding a set of minimum cost spanning
trees to link endnodes to a source node satisfying the traffic requirements at endnodes and the
required mean delay of the network. In the DCCMST problem, the objective function is to mini
mise the total link cost. A dynamic programmingbased threephase algorithm that solves the
DCCMST problem is proposed. In the first phase, the algorithm generates feasible solutions to
satisfy the traffic capacity constraint. It finds the CMSTs in the second phase, and allocates the
optimal link capacities to satisfy the mean delay constraint in the third phase. Performance evalu
ation shows that the proposed algorithm has good efficiency for any network with less than 30
nodes and light traffic. The proposed algorithm can be applied to any network regardless of its con
figuration, and used for the topological design of local networks and for efficient routing algorithms
capable of constructing least cost broadcast trees.
1 Introduction
