Table of Contents
Fetching ...

Nonlinearity Cancellation Based on Optimized First Order Perturbative Kernels

Alex Alvarado, Astrid Barreiro, Gabriele Liga

TL;DR

The paper tackles nonlinear distortions in fiber-optic channels modeled by the Manakov equation and proposes interference cancellation based on power-optimized first-order regular perturbation (FRP) kernels, specifically the NBGD kernels. A genie-aided estimator leverages these kernels to cancel the perturbative nonlinear interference $\Delta \mathbf{a}$, with the FRP model expressed as $\mathbf{y}_0 \approx \mathbf{a}_0 + \Delta \mathbf{a}_0 + \mathbf{n}_0$ and $\Delta \mathbf{a}_0 = j \frac{8}{9} \gamma E_s \sum_{(klm)\in \mathcal{M}} (\mathbf{a}_k^\dagger \mathbf{a}_l) \mathbf{a}_m S_{klm}$. Numerical results for PM-16QAM show up to $2.5$ dB improvement in effective SNR and a $0.42$ bit/2D GMI gain, with the optimum launch power shifted by $+1.5$ dB, highlighting the importance of per-launch-power kernel optimization. The results also indicate robustness relies on matching kernel optimization to the operating power, motivating future work on low-complexity receivers and broader comparisons to existing cancellation schemes. Overall, the work demonstrates the theoretical and practical potential of NBGD-based nonlinear cancellation for single-span, single-channel fiber systems.

Abstract

The potential offered by interference cancellation based on optimized regular perturbation kernels of the Manakov equation is studied. Theoretical gains of up to 2.5 dB in effective SNR are demonstrated.

Nonlinearity Cancellation Based on Optimized First Order Perturbative Kernels

TL;DR

The paper tackles nonlinear distortions in fiber-optic channels modeled by the Manakov equation and proposes interference cancellation based on power-optimized first-order regular perturbation (FRP) kernels, specifically the NBGD kernels. A genie-aided estimator leverages these kernels to cancel the perturbative nonlinear interference , with the FRP model expressed as and . Numerical results for PM-16QAM show up to dB improvement in effective SNR and a bit/2D GMI gain, with the optimum launch power shifted by dB, highlighting the importance of per-launch-power kernel optimization. The results also indicate robustness relies on matching kernel optimization to the operating power, motivating future work on low-complexity receivers and broader comparisons to existing cancellation schemes. Overall, the work demonstrates the theoretical and practical potential of NBGD-based nonlinear cancellation for single-span, single-channel fiber systems.

Abstract

The potential offered by interference cancellation based on optimized regular perturbation kernels of the Manakov equation is studied. Theoretical gains of up to 2.5 dB in effective SNR are demonstrated.

Paper Structure

This paper contains 4 sections, 2 equations, 2 figures, 1 table.

Figures (2)

  • Figure 1: System model under consideration. The genie interference estimation uses NBGD kernels $S_{klm}$ and the transmitted sequence to ideally estimate the perturbative NLI term $\Delta \mathbf{a}$, setting an upper bound on performance. The traditional DSP chain is obtained setting $\Delta \mathbf{a}=0$.
  • Figure 2: Effective SNR (left), GMI (middle), and post-FEC FER (right) for a traditional DSP chain (black lines) and the proposed (genie) interference cancellation scheme (green curves). The ASE-only case ($\gamma=0$) is shown with dotted lines and the NLI-only case (zero ASE noise) with dashed lines. Effective SNR results for mismatched interference cancellation are also shown (purple, blue, and red squares).