Decision Sciences Journal
Volume 31, Number 2
Spring 2000
Telecommunications Network Design—Comparison of Alternative
Approaches
G. Premkumar
College of Business, Iowa State University, Ames, IA 50011, email:
prem@iastate.edu
Chao-Hsien Chu
School of Information Sciences and Technology, Pennsylvania State
University, University Park, PA 16802, email: chu@ist.psu.edu
Hsinghua Chou
Sprint Corporation, 3rd Floor, 10880 College Blvd., Overland
Park, KS 66210
ABSTRACT. The design and development of the network
infrastructure to support mission-critical applications has become
a critical and complex activity. This study explores the use
of genetic algorithms (GA) for network design in the context
of degree-constrained minimal spanning tree (DCMST) problem;
compares for small networks the performance of GA with a mathematical
model that provides optimal solutions; and for larger networks,
compares GA’s performance with two heuristic methods—edge
exchange and primal algorithm. Two performance measures, solution
quality and computation time, are used for evaluation. The algorithms
are evaluated on a wide variety of network sizes with both static
and dynamic degree constraints on the network nodes. The results
indicate that GA provides optimal solutions for small networks.
For larger networks it provides better solution quality compared
to edge exchange and primal method, but is worse than the two
methods in computation time.
Subject Areas: Degree-constrained Minimum Spanning
Tree, Genetic Algorithms, Network Design, Network Modelling,
and Telecommunications. |