Evolutionary Algorithms (EAs) have been used todevelop methods for Traffic Engineering (TE) over IP-basednetworks in the last few years, being used to reach the bestset of link weights in the configuration of intra-domain routingprotocols, such as OSPF. In this work, the multiobjective natureof a class of optimization problems provided by TE with Qualityof Service constraints is identified. Multiobjective EAs (MOEAs)are developed to tackle these tasks and their results are comparedto previous approaches using single objective EAs. The effectof distinct genetic representations within the MOEAs is alsoexplored. The results show that the MOEAs provide more flexiblesolutions for network management, but are in some cases unableto reach the level of quality obtained by single objective EAs.Furthermore, a freely available software application is describedthat allows the use of the mentioned optimization algorithms bynetwork administrators, in an user-friendly way by providingadequate user interfaces for the main TE tasks.
展开▼
