Performance Evaluation of Fragmentation Metrics in Elastic Optical Networks: A Comprehensive Analysis and Classification Framework
Year: 2026
Type: article
Abstract: Spectrum fragmentation in Elastic Optical Networks (EONs) results in serious resource wastage.Despite numerous works and efforts, the community still lacks both a systematized classification as well as an inter-class evaluation framework.This gap was addressed with a proposed taxonomy concerning both SMF and Multi-Core Fibers (MCF) driven by a multi-faceted utility score.A dynamic benchmark for 21 SMF-compatible candidates screened, 18 quantitatively evaluated using a dynamic RMSA simulation, on three different network topologies.Main results reveal that the learned representative metric is the most accurate blocking predictor, but by a small margin (mean per-run AUC 0.714 versus 0.701 for the best representative analytical metric), while pooled evaluation protocols inflate every metric's AUC to at least 0.97 through load leakage.Conversely, the analytical metrics with the highest predictive accuracy yield significantly higher blocking when used for routing (up to 4.66 percentage points at 150 Erlangs).Three deployment zones are derived accordingly.
Authors: Omar Y. Shaaban, Ahmed M. Saheb
Institution: University of Baghdad
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W7204665688