A Proactive Algorithm for the Mitigation of Fragmentation Losses in Elastic Links

Abstract

This paper presents simulations of a novel algorithm for spectrum assignment on an elastic link that achieves reductions of nearly 50% of the losses incurred under the first-fit algorithm with the same random sequence of requests. Instead of trying to reduce the fragmentation itself, the new algorithm aims to optimize the matching between each request and the spectral void chosen for its assignment. For this purpose, it prioritizes functional voids over dysfunctional ones, which are deferred in order to maximize their chance of recovering functionality through coalescence with neighbouring voids brought about by connection terminations. The simulations were performed for a 2-class traffic with requests for bitrates of 400 Gb/s and 1 Tb/s, with uniform and non-uniform traffic profiles, and with slot numbers optimized for short (400 km), intermediate (1600 km) and long (8000 km) distances.