Accuracy Assessment of Nondispersive Optical Perturbative Models through Capacity Analysis

Area of Science:

• Optical Communications
• Information Theory
• Nonlinear Optics

Background:

• Simplified models based on perturbation theory are common for fiber-optical channels.
• These models, primarily for low-power regimes, are often applied at moderate or high powers.
• The impact of simplifying assumptions on channel capacity at higher powers remains unclear.

Purpose of the Study:

• To investigate the per-sample capacity of three continuous-time optical models under zero-dispersion conditions.
• To assess the validity and limitations of simplified fiber-optical channel models in the high-power regime.
• To compare the capacity of regular perturbative, logarithmic perturbative, and stochastic nonlinear Schrödinger (NLS) channels.

Main Methods:

• Considered single-channel data transmission using three optical models: regular perturbative, logarithmic perturbative, and stochastic NLS.
• Employed zero-dispersion fibers and a sampling receiver to derive analytically tractable discrete-time models.
• Established tight bounds for the regular perturbative channel capacity and obtained the capacity for the logarithmic perturbative channel.

Main Results:

• Derived a novel upper bound for the capacity of the zero-dispersion NLS channel.
• Demonstrated that the capacity of the three models diverges at high powers due to differing capacity pre-logs.
• Highlighted significant discrepancies in capacity predictions among the models at higher power levels.

Conclusions:

• Simplified fiber-optical channel models exhibit distinct capacity behaviors at high powers.
• Careful consideration is needed when applying simplified models in the high-power regime of optical communications.
• The choice of model significantly impacts the predicted performance and capacity of optical channels.