Dynamic energy-aware multipath grooming

This paper investigates the use of multipath grooming to save energy in Wavelength Division Multiplexing (WDM) networks by splitting the bandwidth demand of requests among several paths and aggregating them with already established lightpaths. For that, two algorithms are proposed which employ an auxiliary graph which edge weights represent the energy consumed by all the operations related to the optical transmission in a lightpath to provide the bandwidth requested. Both algorithms select a subset of different sets of k lightpaths in which residual bandwidth surpasses minimally the demanded bandwidth with minimum energy consumption. These two algorithms represent the choice of criteria between energy consumption only or both energy and bandwidth. It is shown that the algorithm which considers both energy and bandwidth employs a lower number of lightpaths while reducing the energy consumption and blocking rate.