Abstract In recent years, the technology to narrow the gap between the achievable information rate and the Shannon capacity limits attracts increasing attention. In this paper, a probabilistic shaping coding method based on multi-repeat mapping is proposed. By applying block coding to the information sequences with our method, we transform the modulated constellation points from a uniform distribution to the desired distribution. Rate adaption and controllable distribution are achieved by adjusting the block length and the size of different mapping sets. The proposed algorithm has better coding efficiency, low storage and computational complexity, so is more suitable for data center optical networks. An experiment that demonstrates probabilistically shaped data transmission is successfully conducted over 10 km standard single-mode fiber (SSMF) with 16 quadrature amplitude modulation – discrete multi-tone (16-QAM-DMT). We evaluate our method for a set of code rates, i.e., 16 QAM with code rate 6/8, 7/8 and 9/12. The experimental results show that the improved signal to noise ratios (SNR) can be up to 1.1 dB at a bit error rate (BER) of 10-3 after using the proposed method for probabilistic shaping.

中文翻译：

---

一种应用于数据中心光网络的基于多重复映射的概率整形编码方法

摘要 近年来，缩小可实现信息速率与香农容量限制之间差距的技术越来越受到关注。本文提出了一种基于多重复映射的概率整形编码方法。通过使用我们的方法对信息序列应用块编码，我们将调制星座点从均匀分布转换为所需分布。通过调整块长度和不同映射集的大小来实现速率自适应和可控分布。该算法具有较好的编码效率、较低的存储和计算复杂度，更适用于数据中心光网络。一项证明概率整形数据传输的实验在 10 公里标准单模光纤 (SSMF) 上成功进行，采用 16 正交幅度调制 – 离散多音 (16-QAM-DMT)。我们针对一组码率评估我们的方法，即 16 QAM 码率 6/8、7/8 和 9/12。实验结果表明，使用所提出的概率整形方法后，在误码率 (BER) 为 10-3 的情况下，改进的信噪比 (SNR) 可高达 1.1 dB。