Digital backpropagation (DBP) is an electronic scheme for compensating nonlinear distortions in fiber transmission systems. Due to the nonlinearity-induced spectral broadening, the data must be oversampled to avoid aliasing, which increases the complexity and power consumption of the scheme. In this work, we show that aliasing can alternatively be prevented by distributed antialiasing filters, at a lower complexity. We proposed a new modified split-step Fourier method (SSFM) with easy-to-implement low-pass filters (LPFs) in the linear steps to avoid aliasing due to spectral broadening. Both the forward fiber propagation and a transmitter-side DBP are simulated using the modified SSFM. High-order modulation formats such as 256-ary quadrature-amplitude-modulation (256-QAM) and 1024-QAM transmissions at 28 Gbaud and 64 Gbaud over 1000 km fiber are considered, and our results show that the complexity of the DBP can be reduced by up to 50%. The optimal bandwidth of the LPFs is studied for both forward propagation and the DBP.

中文翻译：

---

抗锯齿发送端数字反向传播

数字反向传播 (DBP) 是一种用于补偿光纤传输系统中非线性失真的电子方案。由于非线性引起的频谱展宽，必须对数据进行过采样以避免混叠，这增加了方案的复杂性和功耗。在这项工作中，我们表明可以通过分布式抗混叠滤波器以较低的复杂度来防止混叠。我们提出了一种新的改进的分步傅立叶方法 (SSFM)，在线性步骤中具有易于实施的低通滤波器 (LPF)，以避免由于光谱展宽而造成混叠。前向光纤传播和发射机侧 DBP 均使用改进的 SSFM 进行模拟。考虑了高阶调制格式，例如 256 进制正交幅度调制 (256-QAM) 和 1024-QAM 在 1000 公里光纤上以 28 Gbaud 和 64 Gbaud 传输，我们的结果表明 DBP 的复杂性可以是减少高达 50%。研究了前向传播和 DBP 的 LPF 的最佳带宽。