Novel Node-Arc Model and Multiiteration Heuristics for Static Routing and Spectrum Assignment in Elastic Optical Networks

Abstract

We consider an elastic optical network and study the static routing and spectrum assignment (RSA) problem aiming to minimize the maximum number of frequency slots required to accommodate all lightpath demands. We introduce a novel node-arc integer linear programming (ILP) model, which jointly decides optimal routes and assigned spectra for lightpaths between all source-destination pairs. To reduce the total number of variables, and therefore lower the computational complexity, our new node-arc model extends previous work by representing the spectrum assigned to each lightpath by two variables (i.e., the starting and ending boundary frequency slot indexes). To achieve scalability, we also develop an efficient spectrum-window-based greedy heuristic algorithm, and further propose three multiiteration-based algorithms that consider the effect of demand-serving sequences. Analytical and numerical results show that comparing with an existing approach, our solution to the RSA problem based on the new node-arc ILP model achieves a significant computational complexity improvement. Further numerical results demonstrate that the proposed multiiteration algorithms obtain solutions closer to optima than existing algorithms.