Fan Yang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Fan Yang has a total of 25 co-authors. Here are the top co-authors by number of shared publications:
- Hao Zhou
Beijing University of Posts and Telecommunications
10 shared publications - Chenglin Bai
Liaocheng University
9 shared publications - Yingjie Jiang
Beijing University of Posts and Telecommunications
7 shared publications - Du Tang
China Academy of Information and Communications Technology
6 shared publications - Hengying Xu
School of Physics Science and Information Engineering, Liaocheng University, Liaocheng, China
6 shared publications - Yaojun Qiao
Beijing University of Posts and Telecommunications
6 shared publications
Latest Publications
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 44 (21) , 9528-9535
Constellation Density Matrix and Lightweight Multi-Task Pruning Model-Based Joint Optical Performance Monitoring in PCS-EON Systems
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 44 (21) , 9528-9535
Constellation Density Matrix and Lightweight Multi-Task Pruning Model-Based Joint Optical Performance Monitoring in PCS-EON Systems
- 8 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. PP (07338724) , 1-9
Modeling and Mitigation of Power Degradation Induced by Hard Decision Errors in Power Profile Estimation
The use of pre-forward error correction (pre-FEC) hard-decision (HD) data is preferred for the practical implementation of the emerging optical performance monitoring technique, power profile estimation (PPE). However, pre-FEC HD data introduces power degradation that increases with the pre-FEC bit error rate (BER). To analyze this effect and mitigate the resulting degradation, we first derive analytical models for HD-induced power degradation in both correlation method (CM)-based and least-squares (LS)-based PPE. With the aid of the Gaussian noise (GN) model and numerical simulations, we show that HD data has a negligible impact on the spatial resolution function (SRF) but introduces significant errors in the accumulated nonlinear vectors. Based on the derived analytical models, we further propose a power degradation mitigation scheme by adjusting the weights of the accumulated nonlinear noise intensity distribution. The effectiveness of the proposed scheme is validated through both 130-GBaud simulations and DP-16QAM 64-GBaud experiments. The results demonstrate that the proposed method achieves estimation accuracy comparable to the ideal post-FEC case for both CM-based and LS-based PPE, with the root mean square error (RMSE) reduced to below 0.5 dB for LS-based PPE, whereas using HD data alone leads to severe degradation. These results support the use of HD data in practical PPE implementations and facilitate its deployment in real-time optical network monitoring.
- 8 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. PP (07338724) , 1-9
Modeling and Mitigation of Power Degradation Induced by Hard Decision Errors in Power Profile Estimation
The use of pre-forward error correction (pre-FEC) hard-decision (HD) data is preferred for the practical implementation of the emerging optical performance monitoring technique, power profile estimation (PPE). However, pre-FEC HD data introduces power degradation that increases with the pre-FEC bit error rate (BER). To analyze this effect and mitigate the resulting degradation, we first derive analytical models for HD-induced power degradation in both correlation method (CM)-based and least-squares (LS)-based PPE. With the aid of the Gaussian noise (GN) model and numerical simulations, we show that HD data has a negligible impact on the spatial resolution function (SRF) but introduces significant errors in the accumulated nonlinear vectors. Based on the derived analytical models, we further propose a power degradation mitigation scheme by adjusting the weights of the accumulated nonlinear noise intensity distribution. The effectiveness of the proposed scheme is validated through both 130-GBaud simulations and DP-16QAM 64-GBaud experiments. The results demonstrate that the proposed method achieves estimation accuracy comparable to the ideal post-FEC case for both CM-based and LS-based PPE, with the root mean square error (RMSE) reduced to below 0.5 dB for LS-based PPE, whereas using HD data alone leads to severe degradation. These results support the use of HD data in practical PPE implementations and facilitate its deployment in real-time optical network monitoring.
- 1 May 2026
- Published by Optica Publishing Group in Optics Letters
- Vol. 51 (9) , 2716-2719
Vibration detection based on DSP frame pilot symbols using integrated coherent receivers
We propose a vibration detection scheme that directly utilizes integrated coherent receivers and leverages pilot symbols (PS) consistent with implementation agreements (IAs)-defined digital signal processing (DSP) frame. By applying a low-complexity sliding filter and an averaging operation to the phase tracked by the widely and commercially used PS-based carrier phase recovery algorithm, the external-cavity lasers (ECLs)-induced phase noise is effectively suppressed. Simulations demonstrated the accurate retrieval of vibrations across the frequency range of 10–40 kHz, with a linearity of R 2 =0.9998 and a sensitivity of 0.04 rad. Experimental validation in a 60-GBaud dual-polarization 16-quadrature amplitude modulation (DP-16QAM) system yielded a sensing signal-to-noise ratio (SSNR) of 23.9 dB for a ∼5.5-rad vibration at 60 kHz, achieving a sensitivity of 0.09 rad. This scheme offers a robust solution for integrated sensing in future intelligent optical networks.
- 1 May 2026
- Published by Optica Publishing Group in Optics Letters
- Vol. 51 (9) , 2716-2719
Vibration detection based on DSP frame pilot symbols using integrated coherent receivers
We propose a vibration detection scheme that directly utilizes integrated coherent receivers and leverages pilot symbols (PS) consistent with implementation agreements (IAs)-defined digital signal processing (DSP) frame. By applying a low-complexity sliding filter and an averaging operation to the phase tracked by the widely and commercially used PS-based carrier phase recovery algorithm, the external-cavity lasers (ECLs)-induced phase noise is effectively suppressed. Simulations demonstrated the accurate retrieval of vibrations across the frequency range of 10–40 kHz, with a linearity of R 2 =0.9998 and a sensitivity of 0.04 rad. Experimental validation in a 60-GBaud dual-polarization 16-quadrature amplitude modulation (DP-16QAM) system yielded a sensing signal-to-noise ratio (SSNR) of 23.9 dB for a ∼5.5-rad vibration at 60 kHz, achieving a sensitivity of 0.09 rad. This scheme offers a robust solution for integrated sensing in future intelligent optical networks.
- 17 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 26 (11) , 16864-16871
LADNet: A Lightweight Axial-Depthwise Network for Phase Unwrapping and Fading Suppression in ϕ -OTDR
In this paper, we propose a lightweight axial-depthwise network (LADNet) to address interference fading and large-dynamic-range phase wrapping in phase-sensitive optical time-domain reflectometry (ϕ-OTDR). The lightweight design leverages axis-specific patterns and local spatial correlation in ϕ-OTDR waterfall data, and captures these features using axial depthwise convolutions. We design a U-Net architecture that integrates multi-branch axial depthwise convolutions and attention mechanisms to capture the anisotropic spatiotemporal correlations of ϕ-OTDR signals. To construct paired data for training and evaluation, we generate noisy wrapped-phase samples with fading and corresponding unwrapped-phase references using a physics-driven ϕ-OTDR simulator.On simulated test data (strain range 0–500 nε), LADNet achieves a strain MAE of 1.3 nε with only 1.70 M parameters. Experiments on a 5.6-km ϕ-OTDR system further verify robust phase recovery under large-amplitude vibration, yielding an SNR of 32.2 dB and a strong linear relationship between the recovered phase and the PZT driving voltage (R2 = 0.9974). Overall, the proposed method provides an effective solution for phase demodulation in ϕ-OTDR under large amplitude vibration conditions.
- 17 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 26 (11) , 16864-16871
LADNet: A Lightweight Axial-Depthwise Network for Phase Unwrapping and Fading Suppression in ϕ -OTDR
In this paper, we propose a lightweight axial-depthwise network (LADNet) to address interference fading and large-dynamic-range phase wrapping in phase-sensitive optical time-domain reflectometry (ϕ-OTDR). The lightweight design leverages axis-specific patterns and local spatial correlation in ϕ-OTDR waterfall data, and captures these features using axial depthwise convolutions. We design a U-Net architecture that integrates multi-branch axial depthwise convolutions and attention mechanisms to capture the anisotropic spatiotemporal correlations of ϕ-OTDR signals. To construct paired data for training and evaluation, we generate noisy wrapped-phase samples with fading and corresponding unwrapped-phase references using a physics-driven ϕ-OTDR simulator.On simulated test data (strain range 0–500 nε), LADNet achieves a strain MAE of 1.3 nε with only 1.70 M parameters. Experiments on a 5.6-km ϕ-OTDR system further verify robust phase recovery under large-amplitude vibration, yielding an SNR of 32.2 dB and a strong linear relationship between the recovered phase and the PZT driving voltage (R2 = 0.9974). Overall, the proposed method provides an effective solution for phase demodulation in ϕ-OTDR under large amplitude vibration conditions.
- 12 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 44 (10) , 3858-3867
Training Sequence and Amplitude Modulation Pilot Tone Aided Nonlinear Interference Monitoring for Coherent Optical Communication Systems
A nonlinear interference (NLI) monitoring method using a training sequence (TS) and amplitude modulation pilot tone (AMPT) is proposed for coherent optical transmission systems. To estimate the total noise, including NLI noise and amplifier spontaneous emission (ASE) noise, the TS is used to obtain prior information and subtract it from the received signal. Then, to separate the NLI noise, including self-channel interference (SCI) and cross-channel interference (XCI) noise, from the total noise, the AMPT induces power oscillations in the NLI noise so that the NLI and ASE powers can be extracted using different frequency characteristics. Finally, the NLI-induced optical SNR (OSNR) penalty can be estimated using these powers. Verified by a 5-channel 64-GBaud 1500-km simulation system, the proposed method can accurately estimate the SNR of total, SCI, XCI, and ASE noise, achieving NLI-induced OSNR penalty estimation errors of less than 0.39 dB. We also conducted an experiment in a 64-GBaud 500-km coherent transmission system with dual-polarization 16 quadrature amplitude modulation (DP-16QAM). The OSNR penalty estimated by the proposed method achieves a 0.41-dB monitoring error with 50-$\mu$s time-window data. The proposed method offers a better solution for achieving low-cost, real-time, and high-accuracy fiber-optical performance monitoring in coherent transmission systems.
- 12 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 44 (10) , 3858-3867
Training Sequence and Amplitude Modulation Pilot Tone Aided Nonlinear Interference Monitoring for Coherent Optical Communication Systems
A nonlinear interference (NLI) monitoring method using a training sequence (TS) and amplitude modulation pilot tone (AMPT) is proposed for coherent optical transmission systems. To estimate the total noise, including NLI noise and amplifier spontaneous emission (ASE) noise, the TS is used to obtain prior information and subtract it from the received signal. Then, to separate the NLI noise, including self-channel interference (SCI) and cross-channel interference (XCI) noise, from the total noise, the AMPT induces power oscillations in the NLI noise so that the NLI and ASE powers can be extracted using different frequency characteristics. Finally, the NLI-induced optical SNR (OSNR) penalty can be estimated using these powers. Verified by a 5-channel 64-GBaud 1500-km simulation system, the proposed method can accurately estimate the SNR of total, SCI, XCI, and ASE noise, achieving NLI-induced OSNR penalty estimation errors of less than 0.39 dB. We also conducted an experiment in a 64-GBaud 500-km coherent transmission system with dual-polarization 16 quadrature amplitude modulation (DP-16QAM). The OSNR penalty estimated by the proposed method achieves a 0.41-dB monitoring error with 50-$\mu$s time-window data. The proposed method offers a better solution for achieving low-cost, real-time, and high-accuracy fiber-optical performance monitoring in coherent transmission systems.
- 12 March 2026
- Published by Optica Publishing Group in Journal of Optical Communications and Networking
- Vol. 18 (4) , B60
Least-squares longitudinal power profile estimation based on pilot symbols and training sequences in DSP frames
Longitudinal power profile estimation (PPE) is an essential forward-sensing technique for optical network monitoring, enabling early detection of abnormal loss and supporting the reliable operation of coherent transmission systems. The conventional least-squares PPE (LS-PPE) method either relies on post-FEC reference symbols, which introduce considerable digital signal processing (DSP) overhead, or employs hard-decision (HD) symbols, which inevitably suffer from a systematic power offset that increases with the pre-FEC bit-error rate (BER). In this paper, we present a comprehensive study of a new, to our knowledge, LS-PPE framework that leverages the pilot symbols (PSs) and training sequence (TS) inherently defined in DSP super-frames. By selectively constructing perturbation matrices from known PS and TS positions while discarding error-prone HD data, the proposed scheme eliminates systematic offset without relying on post-FEC information. Extensive simulations with 130 GBaud signals across multiple modulation formats and transmission distances demonstrate the robustness of the proposed method under diverse system conditions. Experimental validation on an 80 GBaud dual-polarization 16QAM transmission over 250 km standard single-mode fiber confirms estimation accuracy comparable to post-FEC benchmarks and superior robustness over HD-based LS-PPE, highlighting the potential of the PS&TS-based framework for practical deployment.
- 12 March 2026
- Published by Optica Publishing Group in Journal of Optical Communications and Networking
- Vol. 18 (4) , B60
Least-squares longitudinal power profile estimation based on pilot symbols and training sequences in DSP frames
Longitudinal power profile estimation (PPE) is an essential forward-sensing technique for optical network monitoring, enabling early detection of abnormal loss and supporting the reliable operation of coherent transmission systems. The conventional least-squares PPE (LS-PPE) method either relies on post-FEC reference symbols, which introduce considerable digital signal processing (DSP) overhead, or employs hard-decision (HD) symbols, which inevitably suffer from a systematic power offset that increases with the pre-FEC bit-error rate (BER). In this paper, we present a comprehensive study of a new, to our knowledge, LS-PPE framework that leverages the pilot symbols (PSs) and training sequence (TS) inherently defined in DSP super-frames. By selectively constructing perturbation matrices from known PS and TS positions while discarding error-prone HD data, the proposed scheme eliminates systematic offset without relying on post-FEC information. Extensive simulations with 130 GBaud signals across multiple modulation formats and transmission distances demonstrate the robustness of the proposed method under diverse system conditions. Experimental validation on an 80 GBaud dual-polarization 16QAM transmission over 250 km standard single-mode fiber confirms estimation accuracy comparable to post-FEC benchmarks and superior robustness over HD-based LS-PPE, highlighting the potential of the PS&TS-based framework for practical deployment.
- 28 January 2026
- Published by Optica Publishing Group in Optics Letters
- Vol. 51 (3) , 724-727
Pilot symbols and amplitude modulation pilot tone aided nonlinear interference monitoring for coherent systems
A nonlinear interference (NLI) monitoring method is proposed for coherent transmission systems by jointly using implementation agreement-defined pilot symbols (PS) and amplitude modulation pilot tone (AMPT). The PS is employed to extract the total noise from the received signal, while the AMPT enables separation of self-channel interference (SCI) and cross-channel interference (XCI) power from amplifier spontaneous emission (ASE) power. In a 60-GBaud dual-polarization quadrature phase-shift keying (DP-QPSK) 1500-km wavelength-division multiplexing (WDM) simulation system, the NLI signal-to-noise ratio (SNR) and NLI-induced optical SNR (OSNR) penalty were estimated with errors less than 0.26/0.07 dB for a single-channel (SCI-only) system and 0.52/0.30 dB for a 5-channel (SCI+XCI) system within a ∼160-µs time window. In a single-channel 60-GBaud DP-QPSK 500-km experimental system, the estimation errors of NLI SNR and OSNR penalty were below 0.45 dB and 0.35 dB, respectively, using ∼ 300-µs data. The proposed PS-AMPT NLI monitoring method exhibits high accuracy and strong compatibility with deployed optical networks.
- 28 January 2026
- Published by Optica Publishing Group in Optics Letters
- Vol. 51 (3) , 724-727
Pilot symbols and amplitude modulation pilot tone aided nonlinear interference monitoring for coherent systems
A nonlinear interference (NLI) monitoring method is proposed for coherent transmission systems by jointly using implementation agreement-defined pilot symbols (PS) and amplitude modulation pilot tone (AMPT). The PS is employed to extract the total noise from the received signal, while the AMPT enables separation of self-channel interference (SCI) and cross-channel interference (XCI) power from amplifier spontaneous emission (ASE) power. In a 60-GBaud dual-polarization quadrature phase-shift keying (DP-QPSK) 1500-km wavelength-division multiplexing (WDM) simulation system, the NLI signal-to-noise ratio (SNR) and NLI-induced optical SNR (OSNR) penalty were estimated with errors less than 0.26/0.07 dB for a single-channel (SCI-only) system and 0.52/0.30 dB for a 5-channel (SCI+XCI) system within a ∼160-µs time window. In a single-channel 60-GBaud DP-QPSK 500-km experimental system, the estimation errors of NLI SNR and OSNR penalty were below 0.45 dB and 0.35 dB, respectively, using ∼ 300-µs data. The proposed PS-AMPT NLI monitoring method exhibits high accuracy and strong compatibility with deployed optical networks.
- 13 January 2026
- Published by Optica Publishing Group in Optics Express
- Vol. 34 (2) , 1802-1816
Low-complexity symbol-rate equalization schemes for short-reach coherent-lite optical interconnects
In this paper, we propose two low-complexity multiple-input multiple-output (MIMO) equalization schemes for short-reach coherent-lite optical interconnects: dynamic pruning MIMO (DP-MIMO) and dynamic clustering MIMO (DC-MIMO). Based on the conventional 4 × 4 real-valued (RV) MIMO architecture, both schemes aim to reduce the computational complexity of each channel filter according to real-time channel characteristics. DP-MIMO sparsifies filter coefficient matrices and dynamically adjusts the sparsity level of each channel filter based on the energy distribution across all channels, achieving adaptive tap optimization. DC-MIMO reduces redundant computations by clustering filter coefficients and further introduces an energy-adaptive mechanism that dynamically adjusts the number of clusters, balancing complexity and system performance. To further evaluate the benefits of the energy-adaptive mechanism, the static pruning MIMO (SP-MIMO) and static clustering MIMO (SC-MIMO) are also introduced as their static counterparts. The proposed schemes are experimentally validated in a C-band 80-GBaud dual-polarization 16QAM (DP-16QAM) system over 1-km, 2-km, and 5-km standard single-mode fiber (SSMF) links. Experimental results show that, compared with conventional 4 × 4 RV MIMO, DP-MIMO can achieve around 40% reduction in real-valued multiplications (RMs) per transmitted symbol within a 0.20-dB receiver optical power (ROP) penalty. Compared with SP-MIMO at a similar complexity level, DP-MIMO introduces a smaller ROP penalty. DC-MIMO achieves around 60% RMs reduction within a 0.15-dB ROP penalty compared with 4 × 4 RV MIMO, and compared with SC-MIMO at similar performance, DC-MIMO achieves lower complexity. To verify the robustness of the proposed schemes against in-phase/quadrature (IQ) skew impairments, an additional 2-ps IQ skew was introduced. Both DP-MIMO and DC-MIMO maintain stable performance under this condition, with DP-MIMO reducing around 40% of complexity within a 0.2-dB ROP penalty and DC-MIMO reducing around 60% of complexity within a 0.1-dB ROP penalty.
- 13 January 2026
- Published by Optica Publishing Group in Optics Express
- Vol. 34 (2) , 1802-1816
Low-complexity symbol-rate equalization schemes for short-reach coherent-lite optical interconnects
In this paper, we propose two low-complexity multiple-input multiple-output (MIMO) equalization schemes for short-reach coherent-lite optical interconnects: dynamic pruning MIMO (DP-MIMO) and dynamic clustering MIMO (DC-MIMO). Based on the conventional 4 × 4 real-valued (RV) MIMO architecture, both schemes aim to reduce the computational complexity of each channel filter according to real-time channel characteristics. DP-MIMO sparsifies filter coefficient matrices and dynamically adjusts the sparsity level of each channel filter based on the energy distribution across all channels, achieving adaptive tap optimization. DC-MIMO reduces redundant computations by clustering filter coefficients and further introduces an energy-adaptive mechanism that dynamically adjusts the number of clusters, balancing complexity and system performance. To further evaluate the benefits of the energy-adaptive mechanism, the static pruning MIMO (SP-MIMO) and static clustering MIMO (SC-MIMO) are also introduced as their static counterparts. The proposed schemes are experimentally validated in a C-band 80-GBaud dual-polarization 16QAM (DP-16QAM) system over 1-km, 2-km, and 5-km standard single-mode fiber (SSMF) links. Experimental results show that, compared with conventional 4 × 4 RV MIMO, DP-MIMO can achieve around 40% reduction in real-valued multiplications (RMs) per transmitted symbol within a 0.20-dB receiver optical power (ROP) penalty. Compared with SP-MIMO at a similar complexity level, DP-MIMO introduces a smaller ROP penalty. DC-MIMO achieves around 60% RMs reduction within a 0.15-dB ROP penalty compared with 4 × 4 RV MIMO, and compared with SC-MIMO at similar performance, DC-MIMO achieves lower complexity. To verify the robustness of the proposed schemes against in-phase/quadrature (IQ) skew impairments, an additional 2-ps IQ skew was introduced. Both DP-MIMO and DC-MIMO maintain stable performance under this condition, with DP-MIMO reducing around 40% of complexity within a 0.2-dB ROP penalty and DC-MIMO reducing around 60% of complexity within a 0.1-dB ROP penalty.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 44 (15) , 1-10
A Wide-Range and Accurate Skew Estimation Method Based on Coprime Spectral-Block Training Signal in Coherent Systems
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 44 (15) , 1-10
A Wide-Range and Accurate Skew Estimation Method Based on Coprime Spectral-Block Training Signal in Coherent Systems
- 21 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 44 (4) , 1259-1269
Low-Complexity Nonlinear Interference Estimation Using WRP1-Aided Coarse Step-Size Power Profile Estimation
As an emerging optical performance monitoring (OPM) technique, power profile estimation (PPE) has gained significant research attention for its ability to reveal the longitudinal signal power evolution along optical fibers. Among its various applications, nonlinear interference (NLI) noise estimation is particularly promising. In this paper, we propose a low-complexity NLI noise estimation method based on coarse step-size PPE combined with the weighted first-order regular perturbation model (WRP1). The WRP1 model accounts for the correlation between neighboring symbols in the nonlinear operator, thereby mitigating the inaccuracies introduced by coarse step-size power estimation and enhancing overall NLI estimation performance. The proposed method is validated through numerical simulations over a 1000 km transmission system and experimental demonstrations over 500 km and 1200 km optical links. The results demonstrate that the proposed method achieves an estimation error of less than 0.5 dB in both the nonlinear signal-to-noise ratio ($\text{SNR}_{\text{NL}}$) and OSNR penalty, which is comparable to that of conventional fine step-size approaches. Meanwhile, it reduces computational complexity by over 90%, providing a highly efficient solution for practical PPE-enabled NLI estimation.
- 21 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 44 (4) , 1259-1269
Low-Complexity Nonlinear Interference Estimation Using WRP1-Aided Coarse Step-Size Power Profile Estimation
As an emerging optical performance monitoring (OPM) technique, power profile estimation (PPE) has gained significant research attention for its ability to reveal the longitudinal signal power evolution along optical fibers. Among its various applications, nonlinear interference (NLI) noise estimation is particularly promising. In this paper, we propose a low-complexity NLI noise estimation method based on coarse step-size PPE combined with the weighted first-order regular perturbation model (WRP1). The WRP1 model accounts for the correlation between neighboring symbols in the nonlinear operator, thereby mitigating the inaccuracies introduced by coarse step-size power estimation and enhancing overall NLI estimation performance. The proposed method is validated through numerical simulations over a 1000 km transmission system and experimental demonstrations over 500 km and 1200 km optical links. The results demonstrate that the proposed method achieves an estimation error of less than 0.5 dB in both the nonlinear signal-to-noise ratio ($\text{SNR}_{\text{NL}}$) and OSNR penalty, which is comparable to that of conventional fine step-size approaches. Meanwhile, it reduces computational complexity by over 90%, providing a highly efficient solution for practical PPE-enabled NLI estimation.