Font Size: a A A

High-efficiency And Intelligent Signal Processing Techniques In Short-reach Coherent Receivers

Posted on:2022-06-28Degree:MasterType:Thesis
Country:ChinaCandidate:Y L ZhaoFull Text:PDF
GTID:2518306332468194Subject:Information and Communication Engineering
Abstract/Summary:PDF Full Text Request
The development of coherent optical communication is a significant factor for the capacity growth of optical communication system in the past 20 years.Nowadays,coherent optical communication has dominated the long-haul transmission.As the flow of internet grows exponentially,coherent techniques also gain more attention in short-reach systems in recent years.For short-reach systems,flexibility and cost-effectiveness are the most important fators.To improve the flexibility,optical performance monitoring(OPM)is one important function.The cost of ADC and DSP are major components of the system cost.Currently,OPM techniques mainly adopt complicated deep neural networks,which are difficult to deploy on hardware.In addition,the high-resolution ADC and complex equalization algorithms in DSP are expensive.To solve these problems,this paper studies three aspects,including optical signal processing,optical performance monitoring and adaptive equalizer.The main contributions of this paper are as follows:1.We study an optical-electronic binarized coherent receiver through simulation,and achieve 50-Gb/s single polarization transmission of QPSK signals,100-Gb/s dual-polarization QPSK signals.This receiver preprocesses optical signal through specially designed optical signal processing unit.It has potential to save system cost by reducing required resolution of ADCs.2.We simulate and experimentally verifies a high-efficient and intelligent optical performance monitoring(OPM)module based on binarized neural network.It achieves 100%accuracy of modulation format identification and higher than 95%accuracy of OSNR estimation.Moreover,it has 3 times lower memory consumption and twice as fast as float valued neural network-based OPM module.Our OPM module is promising for the real-time channel state estimation of intelligent and flexible optical network.3.We simulate and experimentally demonstrate an adaptive equalizer(AEQ)based on polarization demultiplexing in Stokes space.In a 16/64-QAM system with 5km transmission,it achieves similar BER performance compared with conventional method,while reducing the computational complexity by 45%.In addition,proposed AEQ will not suffer from failure of polarization demultiplexing and has higher tolerance of chromatic dispersion,which make it possible to transmit at least 20 km without performance loss.Hence,it is suitable to deploy in next-generation short reach systems.
Keywords/Search Tags:Coherent Optical Communication, Digital Signal Processing, Neural Networks, Stokes space
PDF Full Text Request
Related items