Power-efficient integrated routing of sublambda connection requests with traffic splitting in IP over WDM networks

Energy efficiency in communication networks has drawn much attention from engineers and scientists in industry and academia worldwide due to the ever-rising energy consumption of the Internet and the corresponding environmental impacts. In this paper, we focus on the problem of power-efficient provisioning of connection requests considering traffic splitting in IP over WDM networks. We adopt an integrated routing model of optical networks such that resources at the IP layer and the optical layer are managed together, thereby leading to higher flexibility and better resource utilization. We divide the power consumption of an IP over WDM network into five components, and each consumes a different amount of power. We analyze both affine and convex power profiles for these components. We deal with both static and dynamic traffic models. For static traffic model, we formulate integer programming models to minimize the total power consumption in the network. They are Integer Linear Programming (ILP) for affine power profile and Integer Quadratic Programming (IQP) for convex power profile, respectively. For dynamic traffic model, we propose power-efficient routing algorithms which are based on an auxiliary graph that assigns weights for links according to the power consumption, thereby capturing the power consumption flow of a path. Comparative studies show that traffic-splitting-enabled algorithms outperform their non-traffic-splitting counterparts with respect to power consumption and blocking probability.