Advanced Resource Allocation Strategies for MCF-based SDM-EONs: Crosstalk Aware and Machine Learning Assisted Algorithms.

In space division multiplexed elastic optical networks (SDM-EONs), parallel transmission of lightpaths is enabled using multicore fibers (MCFs) in the network. However, the intercore crosstalk (XT) between parallel transmissions degrades the quality of service and reduces the utilization of available capacity. This impairment results in a tradeoff between spectrum utilization and XT accumulation. In this paper, we discuss various approaches to solve the routing, modulation, core, and spectrum assignment (RMCSA) problem while balancing the tradeoff, namely, Tridental Resource Assignment algorithm (TRA), and Spectrum Wastage Avoidance-based Resource Allocation (SWARM) algorithm. We also propose offline optimizations such as machine learning (ML)-aided threshold optimization, integer linear programming-based priority path selection (PPS) for routing, and customized weights in the tridental coefficient (TC) to improve the performance of TRA. The ML-aided optimizer and PPS improve the performance of "any" RMCSA algorithm. The customized weights in TC and intelligent resource selection strategy improve TRA even further. Extensive simulation experiments show significant reductions in bandwidth blocking probability, by several orders of magnitude for a variety of scenarios.