We experimentally demonstrate the highest, to the best of our knowledge, reported net rate in a SiP IQ modulator. At 100 Gbaud 32QAM (quadrature amplitude modulation), and assuming 20% FEC (forward error correction) overhead, we achieved a dual polarization net rate of 833 Gb/s. This record was achieved by adapting digital signal processing to the challenging pattern dependent distortion encountered in the nonlinear and bandwidth limited regime. First the Mach Zehnder modulator (MZM) operating point (trading off modulation efficiency and 3 dB bandwidth) and linear compensation (electrical and optical) are jointly optimized. Next, the key application of nonlinear pre- and post-compensation are explored. We show that nonlinear processing at the transmitter, in our case an iterative learning control (ILC) method, is essential as post-processing alone could not achieve reliable communications at 100 Gbaud. Nonlinear post-compensation algorithms pushed the performance under the FEC threshold with the introduction of structured intersymbol interference in post processing and a simple one-step maximum likelihood sequence detector. We provide detailed descriptions of our methodology and results.

中文翻译：

---

全硅 IQ 调制器的系统优化：实现 100 Gbaud 双极化 32QAM

据我们所知，我们通过实验证明了 SiP IQ 调制器中报告的最高净速率。在 100 Gbaud 32QAM（正交幅度调制）下，假设 20% 的 FEC（前向纠错）开销，我们实现了 833 Gb/s 的双极化净速率。这一记录是通过使数字信号处理适应非线性和带宽受限机制中遇到的具有挑战性的模式相关失真来实现的。首先，马赫曾德调制器 (MZM) 工作点（折衷调制效率和 3 dB 带宽）和线性补偿（电和光）被联合优化。接下来，探讨非线性预补偿和后补偿的关键应用。我们展示了发射机的非线性处理，在我们的例子中是迭代学习控制 (ILC) 方法，是必不可少的，因为仅靠后处理无法实现 100 Gbaud 的可靠通信。通过在后处理中引入结构化符号间干扰和简单的一步最大似然序列检测器，非线性后补偿算法将性能提升到 FEC 阈值以下。我们提供了我们的方法和结果的详细描述。