Interdomain Traffic Control over Multiple Links Based on Genetic Algorithm

摘要

Network operators must have control over the flow of traffic into, out of, and across their networks. Usually, changes of interdomain traffic affect intradomain traffic in other domains. We propose an approach to control interdomain traffic over multiple links based on prefixes. Transit Autonomous System (AS) measure the flow of traffic and cooperate with other ASes in balancing interdomain traffic. Using the data of interdomain traffic and intradomain traffic measured in transit networks, the approach evaluates the cost of transit networks. We present a genetic algorithm to specify a link for prefix in AS neighbors with the objective of minimizing costs and configuration changes. In order to decrease the complex of calculation, we analysis the distribution of traffic and propose an approach to select popular prefixes. A example verify the availability and effectiveness of the algorithm.