Abstract

Nonlinear distortions in passive optical network (PON) systems degrade signal transmission performance. Joint linear and nonlinear pre-equalizers based on the Volterra series, memory polynomials and artificial neural networks (ANNs) are utilized to mitigate linear and nonlinear impairments. Centralized pre-equalization at the transmitter side of the optical line terminal (OLT) keeps the receiver side digital signal processing (DSP) at the optical network unit (ONU) simple. In this study, performance assessments of these three different pre-equalization schemes in intensity modulation and direct detection (IM/DD)-based 25-GBaud PON systems were evaluated. The joint linear and nonlinear pre-equalization schemes were based on the weight coefficients and the model obtained by channel estimation on the receiver side. We studied and compared the performance of three pre-equalization methods under various implementation conditions, such as received optical power (ROP), number of equalizer taps, and semiconductor optical amplifier (SOA) current. Besides, we studied the equalization performance under different modulation orders, which were applied for data rate optimization to support different power budgets, being the potential solutions for future PON systems with various fiber-reach and split-ratio distributions. As a proof-of-concept, the equalization performance of the 25-Gbaud signals modulated with non-return-to-zero (NRZ), four-level pulse amplitude modulation (PAM-4) and PAM-8 formats, corresponding to 25-, 50-, and 75-Gbps data rates, respectively, were tested. Both numerical simulations and experimental demonstrations were conducted to assess the performance of different pre-equalization schemes. With ANN pre-equalization, more than 4dB improvement in power budget was achieved for 25-GBaud PAM4 and PAM8 signals after 20 km fiber transmission. ANN-based pre-equalizer performed better than ANN-based post-equalizer even when the received signals were in low SNR. The power budgets of the 25-Gbaud NRZ, PAM4 and PAM8 signals after 20 km fiber transmission were 36.2 dB, 29.5 dB and 18 dB, respectively.

SOA pattern effect mitigation by neural network based pre-equalizer for 50G PON

Lei Xue, Lilin Yi, Rui Lin, Luyao Huang, and Jiajia Chen
Opt. Express 29(16) 24714-24722 (2021)

Low-complexity Volterra-inspired neural network equalizer in 100-G band-limited IMDD PON system

Luyao Huang, Wenqing Jiang, Yongxing Xu, Weisheng Hu, and Lilin Yi
Opt. Lett. 47(21) 5692-5695 (2022)

100 Gbit/s VSB-PAM-n IM/DD transmission system based on 10 GHz DML with optical filtering and joint nonlinear equalization

Jiao Zhang, Jianjun Yu, Xinying Li, Yiran Wei, Kaihui Wang, Li Zhao, Wen Zhou, Miao Kong, Xiaolong Pan, Bo Liu, and Xiangjun Xin
Opt. Express 27(5) 6098-6105 (2019)

Low-complexity clustered polynomial nonlinear filter-based electrical dispersion pre-compensation scheme in IM/DD transmission systems

Xiaoqian Huang, Fei Xie, Du Tang, Shuangyue Liu, and Yaojun Qiao
Opt. Express 31(20) 32529-32542 (2023)

Real-time validation of downstream 50G/25G and 50G/100G flexible rate PON based on Miller encoding, NRZ, and PAM4 modulation

Doutje van Veen and Vincent Houtsma
J. Opt. Commun. Netw. 15(8) C147-C154 (2023)