A new genetic algorithm for the optimal communication spanning tree problem

Abstract

This paper proposes a new genetic algorithm to solve the Optimal Communication Spanning Tree problem. The proposed algo-rithm works on a tree chromosome without intermediate encoding and decoding, and uses crossovers and mutations which manipulate direc-tly trees, while a traditional genetic algorithm generally works on linear chromosomes. Usually, an initial population is constructed by the stan-dard uniform sampling procedure. But, our algorithm employs a simple heuristic based on Prim's algorithm to randomly generate an initial po-pulation. Experimental results on known data sets show that our genetic algorithm is simple and e_cient to get an optimal or near-optimal solu-tion to the OCST problem. © Springer-Verlag Berlin Heidelberg 2000.