Multicasting for All-Optical Multifiber Networks

All-optical wavelength-routed WDM WANs can support the high bandwidth and the long session duration requirement of the application scenarios like interactive distance learning or online diagnosis of patients simultaneously in different hospitals. However, multifiber and limited sparse light splitting and wavelength conversion capabilities of switches result in a difficult optimization problem. We attack this problem using a layered graph model. The problem is defined as k edge-disjoint degree-constrained Steiner Tree problem for routing and fiber and wavelength assignment of k multicasts. A mixed integer linear programming formulation for the problem is given and a solution using CPLEX is provided. However, the complexity of the problem grows quickly with respect to the number of edges in the layered graph, which depends on the number of nodes, fibers, wavelengths, and multicast sessions. Hence, we propose two heuristics (LAMA and C-FWA) to compare with CPLEX, existing work and unicasting. Extensive computational experiments show that LAMA’s performance is very close to CPLEX and it is significantly better than existing work and C-FWA for nearly all metrics. Although, C-FWA is inferior to LAMA it is also using fiber and wavelength conversion resources more efficiently than its other alternatives. c © 2006 Optical Society of America