Orthogonal Genetic Algorithms with Application to Multimedia Multicast Routing

Abstract

Some major steps of a genetic algorithm can be considered to be “experiments”. For example, a crossover operator samples the genes from the parents to produce some potential offspring, and this operation can be considered to be a sampling experiment. Based on this observation, we recently propose to integrate experimental design methods into the genetic algorithms [1–4], so that the resulting algorithms can be more powerful and statistically sound. In this chapter, we summarize our recent results on this research direction [1,2]. In particular, we introduce the basic concept of experimental design methods and describe a well-known experimental design method called orthogonal design. Then we describe how to apply the orthogonal design to design a new type of genetic algorithms called orthogonal genetic algorithms. In additionally, we describe an orthogonal genetic algorithm for an optimization problem in communication networks called multimedia multicast routing. We present numerical results to demonstrate the effectiveness of orthogonal genetic algorithms.