A Dynamic traffic-aware energy-efficient algorithm based on sleep-scheduling for autonomous systems

Abstract

Over recent years, energy conservation of Internet infrastructure has attracted a great deal of attention to reduce the economical cost and environmental pollution produced by networking sectors. Due to over-provisioning of networking resources for peak traffic periods, an important proportion of these resources may remain under-utilized in low traffic periods. In this paper, we propose an online green algorithm which switches off some extra nodes and links in order to conserve energy while guaranteeing the network performance. To achieve this, nodes with less occurrence in the shortest path tree and their connected links are selected to be switched-off if their utilization is not more than the average network utilization. Due to the dynamic nature of network traffic, our solution is equipped with a dynamic engine which re-switches on some nodes and links in higher traffic loads. The proposed approach does not require any centralized controller for switching off devices and uses a link-state protocol like OSPF to share required information with minimal overhead. Simulation results obtained using a real network topology show that our proposal switches off up to 28% of network nodes and up to 74% of the maximum number of switchable links with low effect in network performance.