A Flexible Scalable Solution for All-Optical Multifiber Multicasting: SLAM

We previously proposed a mixed-integer linear-programming formulation, solved by CPLEX, which is a state-of-the-art optimization tool, and two heuristic solutions, layered all-optical multicasting algorithm (LAMA) and conservative fiber and wavelength assignment (C-FWA), for all-optical multicasting in wavelength-routed multifiber wavelength-division multiplexed networks with sparse wavelength conversion and light-splitting restrictions. However, LAMA was not suitable for large and dynamic multicasting problems. In this paper, we evaluate and propose scalable layered all-optical multicasting (SLAM), which is a scalable version of LAMA, for any size static or dynamic multicasting problems. We demonstrate that SLAM performs very close to the optimal (the lower bound/CPLEX) and LAMA, and significantly better than the existing work and C-FWA in terms of nearly all metrics, including the session- and group- blocking probabilities (SBPs and GBPs), since it does not separate routing and fiber-wavelength assignment steps as compared to the other candidates like the existing work and C-FWA. Additionally, transmitter-optimized SLAM, which is a version of SLAM that is specifically tuned to minimize the number of transmitters used, spends three times less extra transmitters than SLAM, without a performance loss in other metrics (one extra transmitter for 12 multicast sessions). Therefore, the adjustable parameters of SLAM make it very flexible to balance different objectives.