Energy Aware Traffic Engineering for Intra-Domain Networks Using Computational Intelligence Techniques

Because of various ecological, environmental, and economic issues, energy efficient networking has been a subject of interest in recent years. In a typical backbone network all the routers and their ports are always active and consume energy. Average link utilization in internet service providers is about 30-40%. Energy aware traffic engineering aims to change routing algorithms so that low utilized links are deactivated and their load is distributed over other routes. As a consequence, by turning off these links and their respective devices and ports, network energy consumption is significantly decreased. In this paper we propose four algorithms for energy aware traffic engineering in intra-domain networks. Sequential Link Elimination (SLE) removes links based on their role in maximum network utilization. As a heuristic method, Extended Minimum Spanning Tree (EMST) uses minimum spanning trees to eliminate redundant links and nodes. Energy Aware DAMOTE (EAD) is another heuristic method that turns off links with low utilization. The forth approach is based on genetic algorithms that randomly searches for feasible network architectures in a potentially huge solution space. Evaluation results on Abilene network with real traffic matrix indicate that about 35% saving can be obtained by turning off underutilized links and routers on off-peak hours with respect to QOS. Furthermore, experiments with GA confirm that a subset of links and core nodes with respect to QOS can be switched off when traffic is in its off-peak periods and energy can be saved about 40%.