基于叠加训练（ST）信道估计的相干光正交频分复用系统研究

In a conventional CO-OFDM system, a large portion of the bandwidth is consumed by channel and carrier phase estimation as well as anti-ISI cyclic prefixes, which significantly decreases its spectra efficiency. To address this problem, a 3-year research project on channel estimation using superimposed training (ST) in coherent optical OFDM (CO-OFDM) systems is proposed. One major advantage of using ST based channel estimation is that no extra bandwidth is needed, since the periodic ST sequences are arithmetically added to the OFDM signals, while valuable bandwidth or time slot is required for conventional pilot-aided CO-OFDM systems using preambles or inserted subcarriers. Therefore, higher spectra efficiency is expected for ST based CO-OFDM systems. In this proposal, emphasis will be put on its enabling technologies as well as its experimental demonstrations and optimization approaches. Firstly, a joint channel and carrier phase estimation scheme using ST and inserted subcarriers is proposed, with a decision feed-back algorithm to enhance its estimated channel accuracy. Secondly, an improved ST algorithm, namely data-dependent ST (DDST) algorithm, is introduced in CO-OFDM systems for both channel and carrier phase estimation to further increase the spectra efficiency. Thirdly, the abovementioned algorithms will be investigated and evaluated in a polarization division multiplexed (PDM) CO-OFDM system. A 400 Gb/s (2-pol * 5 super-channels * 10 GBaud/s) PDM-CO-OFDM transmission over 1000 km of single mode fiber utilizing 16-QAM modulation and super-channel technology will be experimentally demonstrated.

传统的相干光正交频分复用（CO-OFDM）系统用在信道和载波相位估计、抗码间干扰上的带宽开销巨大，降低了频谱效率。为解决这一问题，本项目将叠加训练序列（ST）信道估计引入CO-OFDM系统中，研究其在CO-OFDM系统的应用和优化。ST算法将训练序列算术叠加在信号上,因而不增加系统的带宽开销，提高了频谱效率。我们首先将研究基于ST和注入导频的信道和载波相位联合估计方案，并研究通过迭代反馈消除数据信号对训练序列的干扰，从而增加信道估计精度的算法；之后，我们将研究CO-OFDM和偏振复用（PDM）CO-OFDM系统的数据相关叠加训练（DDST）算法，研究其在信道和载波相位估计的应用，从而进一步提高频谱效率；我们将利用现有的实验平台，完成以上算法的实验验证，并实现一个速率为400 Gb/s（2偏振×5超通道×10 GBaud/s，16-QAM）、传输距离1000公里的PDM-CO-OFDM系统。

传统的相干光正交频分复用（CO-OFDM）系统用在信道和载波相位估计、抗码间干扰上的带宽开销巨大，降低了频谱效率。为解决这一问题，本项目将叠加训练序列（ST） 信道估计引入 CO-OFDM 系统中，研究其在 CO-OFDM 系统的应用和优化。ST 算法将训练序列算术叠加在信号上,因而不增加系统的带宽开销，提高了频谱效率。本项目主要研究了光OFDM系统的设计和优化、长距离OFDM-PON系统的色散补偿以及基于叠加训练序列的光OFDM系统信道估计等。重要研究结果包括：研究了多载波的相干光OFDM系统、单载波相干光传输系统以及强度调制/直接探测（IM/DD）的OFDM系统中的叠加训练序列信道估计和系统验证；研究了相干光OFDM系统在调制器消光比受限情况下对信号幅度、硬裁剪系数、DAC分辨率等参数的参数设计和优化；研究了IM/DD OFDM系统中基于超奈奎斯特镜像混叠的色散补偿和信噪比增强方案；研究了基于Volterra序列的超高速率IM/DD OFDM系统等。主要研究结果包括：实验实现了基于叠加训练序列的CO-OFDM信号的5000公里光纤传输，相对于前置导频，其频谱效率有约10%的提升；通过实验，验证了采用超奈奎斯特频谱混叠技术，可以将10-GHz的OFDM信号在双边带调制时其传输距离扩展至80公里以上；实验实现了基于Volterra序列非线性补偿的单波长、单偏振170.49-Gb/s IM/DD OFDM信号的光传输。