The tremendous growth in high-bandwidth applications and consequent increase in power consumption has underscored the importance of energy efficient design strategies for backbone optical networks. In this paper, we exploit the knowledge of the demand holding times to obtain more power efficient grooming. We consider time-varying, periodic traffic, where the setup and teardown times of the demands may be fixed or vary within a larger time window. The problem of joint scheduling (in time) and traffic grooming, with the goal of minimizing energy consumption for sliding demands, has not been considered previously. We first present an efficient integer linear program formulation for scheduling and allocating resources to a single subwavelength traffic demand, such that incremental energy consumption due to the new demand is minimized. Next, we extend the proposed approach and present a new heuristic that jointly performs scheduling and resource allocation for a large set of periodic demands with the goal of reducing the overall energy consumption of the network for the entire demand set. Through simulations, we demonstrate that significant energy savings can be achieved by considering the demand holding time and show that joint allocation of demands yields additional improvements.
Read full abstract