Work

Analytical Statistical Modeling of PDL-Induced SNR Margins for Reliable and Smarter Optical Networks

Year: 2026
Type: preprint
Abstract: Efficient optical network planning and operation require sufficient signal-to-noise ratio (SNR) margins to meet reliability targets without unnecessarily limiting lightpath provisioning. Polarization-dependent loss (PDL) introduces SNR fluctuations that must be accounted for in this trade-off. We develop analytical models that translate characterized device PDL values and distributed link noise into margins for a target outage probability. A truncated Central Limit Theorem based approximation is complemented by two moment-matched log-skew-normal models adapted to characterized frequency-port-dependent per-device PDL. Their accuracy is assessed against a previously derived exact solution, Monte Carlo simulations and laboratory measurements. We then extend the study to 36,544 lightpath-channel cases in the Japanese Fukui and Wakayama metro network models. A realistic characterization of node port-frequency combinations represents PDL, while GNPy accounts for amplified spontaneous emission and nonlinear interference. Both moment-matching models remain conservative relative to the numerical exact model in all analyzed network cases. The recursive third-moment model reduces the mean margin by 2.818 dB in Fukui and 3.551 dB in Wakayama compared with an accumulated PDL benchmark based on industry guidelines. The Gaussian approximation instead underestimates some margins in longer cascades. Beyond predictive accuracy, the analytical models support lightpath-specific PDL margins that can be recomputed after network reconfiguration, provided that port-frequency device characterization and distributed-noise estimates are available in the operational network.
Language: English
Open Access status: green
OpenAlex ID: W7234401754
Cites: 40
Cited by:
Related to: 10