Energy aware routing with link disjoint backup paths

Energy aware routing has attracted a wide spread attention in the last decade. However, relatively concentrated traffic after applying energy aware strategies makes the network be vulnerable to link failure and sudden traffic bursts. It is a challenge to take enhancing network reliability into account while keeping low energy consumption. Routing each source-to-terminal request via two disjoint paths can be a trade-off. In this paper, we propose a problem called Energy Aware Routing with Link Disjoint Backup Paths (EAR-LDBP) to minimize router power consumption when backup sharing is not allowed during off-peak hours. Each request is routed via an active path. A link disjoint backup path is computed for each request and can be used to route traffic in case of single link failure on the active path. A 0-1 integer linear programming model that can guarantee the network resilience to single link failure is formulated to minimize the power of used links and nodes while putting idle ones to sleep under constraints of link utilization and a link disjoint backup path for each request. We analyze the complexity of EAR-LDBP and show its Np-hardness. We prove that there is no polynomial-time constant-factor approximation algorithm even for two requests. Four different versions of EAR-LDBP (link/node-disjoint, link cost only/link and node cost) are shown to be equivalent and polynomial-time reducible to each other. Then two algorithms are proposed to solve this problem: 2 Stage Greedy Algorithm (2SG) and Greedy Two Link Disjoint Paths Algorithm (G2DP) with an iterative deletion strategy to improve the solution. For comparison purpose, we also use the state of the art reliable energy aware routing algorithm 2DP-NF in unsplittable way. Simulation results on synthetic networks and real network of Internet2 show the effectiveness of our algorithms. Particularly, our algorithms can save more energy (up to 39%) while using less computation time as compared to 2DP-NF (less than 1% at best). (C) 2017 Elsevier B.V. All rights reserved.