基于深度学习的数据中心光通信色散估计与管理

瞿国庆; 于树科

光学技术, 2018, 44 (5): 617, 网络出版: 2018-10-08

基于深度学习的数据中心光通信色散估计与管理

Estimation and management of fiber dispersion in the data center based on the deep learning

论文大纲

瞿国庆 ^1,2,*于树科 ¹

作者单位

¹ 江苏商贸职业学院物联网技术研究所, 江苏南通 226011

² 南通智大信息技术有限公司, 江苏南通 226010

人工神经网络深度学习数据中心光纤色散光纤通信 artificial neural network deep learning data center fiber dispersion fiber communication

摘要

最大似然估计能够准确地估计光通信的色散, 但其对高速符号传输的计算复杂度较高, 云计算数据中心的光网络是一种高速光通信的情况, 为了降低数据中心中光通信色散估计的计算成本, 设计了一种基于深度学习的数据中心光通信均衡器。基于人工神经网络的均衡器分为两个阶段, 第一阶段采用光信道的脉冲响应数据对人工神经网络进行训练, 对人工神经网络的模型参数进行优化, 建立人工神经网络的非线性响应模型; 第二阶段采用训练的人工神经网络均衡器对光信道的传输数据进行处理, 实现对光信道色散的估计与补偿。按照数据中心的光网络方案进行了仿真实验, 结果显示, 基于人工神经网络的均衡器提高了光通信的光信噪比, 并且延长了光通信的传输距离。

Abstract

Fiber dispersion can be estimated by maximum likelihood sequence estimation accurately, but it suffers from the high computational complexity in the condition of high-speed transmission symbol rate, the optical networks of data centers of cloud computing usually work with high-speed transmission symbol rate, in order to reduce the computational cost of estimation of fiber dispersion in data center, an efficient equalizer of the optical communication of data center based on deep learning is designed. The equalizer based on artificial neural network(ANN) consists of two phases, in the first phase, the impulse response datasets of the optical channel are adopted to train ANN and optimize the parameters of ANN model, and the non-linear response model of ANN is constructed; in the second phase, the trained ANN equalizer is used to process the transmitted data of optical channels, and the fiber dispersion is estimated and compensated. Simulation experimental results according to the optical network scenario of data center show that, the equalizer based on the ANN model improves the optical signal to noise ratio, and extend the transmission distance of optical communication.

参考文献

[1] 李耀芳, 吴斌, 肖杰, 等. 基于骨干网的数据中心网络服务规划设计［J］. 计算机工程与科学,2017,39(7):1281-1287.

Li Yaofang, Wu Bin, Xiao Jie, et al. Data center network service planning based on backbone networks［J］. Computer Engineering and Science,2017,39(7):1281-1287.

[2] 张虎, 陈健. 高速相干光通信系统中残留色散均衡器研究［J］. 光子学报,2014,43(9):906004-0906004.

Zhang Hu, Chen Jian. Investigation of residual chromatic eispersion equalizer in high-speed optical coherent communication systems［J］. Acta Photonica Sinica,2014,43(9):906004-0906004.

[3] Tuna C. Comparison of BER performances of MMSE equalizer and viterbi algorithm for the MLSE solution in 60 GHz communication channel［C］// International Conference on Electrical and Electronics Engineering. IEEE,2016:1212-1215.

[4] 臧大伟, 曹政, 王展, 等. 基于AWGR的OCS/EPS数据中心光电混合网络［J］. 计算机学报,2016,39(9):1868-1882.

Zang Dawei, Cao Zheng, Wang Zhan, et al. AWGR-based OCS/EPS hybrid datacenter network［J］. Chinese Journal of Computers,2016,39(9):1868-1882.

[5] Sudarsan V. Optical materials: fundamentals and applications［J］. Functional Materials,2012:285-322.

[6] Krehlik P. Directly modulated lasers in negative dispersion fiber links［J］. Opto-Electronics Review,2007,15(2):71-77.

[7] Bakr M. Nonlinear optimization in electrical engineering with applications in MATLAB［M］. Institution of Engineering and Technology,2013.

[8] Cunningham D G, White I H, Ingham J D, et al. Performance and power dissipation comparisons between 28 Gb/s NRZ, PAM, CAP and optical OFDM systems for data communication applications［J］. Journal of Lightwave Technology,2012,30(20):3273-3280.

[9] 陈功, 贺林, 刘登宝. 基于系数可调FFE的10Gb/s发送端的设计［J］. 微电子学,2016,46(3):356-359.

Chen Gong, He Lin, Liu Dengbao. Design of a 10 Gb/s transmitter based on adjustable coefficient FFE［J］. Microelectronics,2016,46(3):356-359.

[10] 付强, 肖沈阳, 郑思远, 等. 基于信道估计的判决反馈均衡器在时变水声信道中的应用研究［J］. 南京大学学报(自然科学),2015(s1):55-59.

Fu Qiang, Xiao Shenyang, Zheng Siyuan, et al. Implementation of channel-estimation-based adaptive decision feedback equalization in time-varying underwater acoustic channels［J］. Journal of Nanjing University (Natural Sciences),2015(s1):55-59.

[11] 黄泽. 紫外非视距通信的MLSE均衡器［J］. 计算机工程,2012,38(20):72-75.

Huang Ze. MLSE equalizer of ultraviolet non-line-of-sight communication［J］. Computer Engineering,2012,38(20):72-75.

[12] Forestieri E, Prati G, Colavolpe G, et al. Maximum-likelihood sequence detection with closed-form metrics in OOK optical systems impaired by GVD and PMD［J］. Journal of Lightwave Technology,2006,24(8):3073-3087.

瞿国庆, 于树科. 基于深度学习的数据中心光通信色散估计与管理[J]. 光学技术, 2018, 44(5): 617. QU Guoqing, YU Shuke. Estimation and management of fiber dispersion in the data center based on the deep learning[J]. Optical Technique, 2018, 44(5): 617.

基于深度学习的数据中心光通信色散估计与管理

关于本站 Cookie 的使用提示

全站搜索

基于深度学习的数据中心光通信色散估计与管理

相关论文

相关资讯

关于本站 Cookie 的使用提示

全站搜索