Applications of genetic algorithms in bgp-based interdomain traffic engineering

Abstract

With the rapid development of the Internet, widely deployed new services such as high definition videos and voice over IP (VoIP) require higher performance guarantees. Traffic engineering can improve end to end service quality, help large autonomous systems (AS) operators improve network resource utilization and meet the challenge. At the interdomain level, traffic engineering is more challenging. Most network operators still rely on changing routing policies and BGP attributes manually. In this paper, we discuss the existing systematic techniques in BGP-based interdomain traffic engineering and propose an improved algorithm based on a multi-objective genetic algorithm. Our algorithm is scalable and efficient. We apply our solution to a provincial network of China Mobile and discuss the influence of different parameters on the performance and validity of the algorithm.