Proactive Grooming With Delay Optimization in Sliceable Elastic Optical Network

Typical energy-efficient dynamic grooming solutions may find a detour and longer path where delay from both the IP layer and optical layer is increased on extra nodes and links. In this paper, analysis of both delay and energy is made on typical grooming operations. An integer linear programming (ILP) model for minimization of both delay and energy is formulized and an alternative model is given with more delay consideration on the distribution over expected IP links. By considering the expected resource distribution of service requests, the available network resource can be re-priced by amortized cost over the expected distribution and a delay proactive control algorithm is designed over this distribution. A proactive delay heuristic is then proposed as an independent policy by a sorting function which is distinguished from other policies by cost functions on a virtual graph. By mining a frequent subset of IP links between node pairs, the problem is simplified without loss of statistical advantage over the resource distribution. The proposed grooming heuristic is performed on a simulation platform in terms of path delay, optical energy consumption, and electrical energy consumption. The simulation results show that the proposed grooming heuristic decreased average path delay without impacting energy performance. Different factors such as sliceability of a network resource, frequent degrees of expected resource, and grooming control periods, which affect the performance of the proposed heuristic, are compared to give thorough evaluations.