Research On Modulationg Formates Identification In Coherent Optical Communication System

With the emergence of new devices,as well as the rapid development of big data,cloud computing,demand on the bandwidth is increasing rapidly.High-speed,large-capacity optical communications networks become the mainstream in the development of next-generation communications networks.Compared to traditional 10G optical system,the current system with capacity of 100Gbit/s has been deployed in a large scale.System with capacity of 400Gbit/s will be taken into commercialization in no time.Research objectives have focused on terabit system.Because of the limitation of electronic bottleneck,the improvement of capacity mainly depends on the use of higher order modulation formats,from QPSK to 16QAM,even to 32QAM and 64QAM.Besides,telephone,television broadcasting,e-health,e-commerce,video on demand and other business will occupy large amount of market share.Service providers will offer different services according to customers'needs.To address this problem,an elastic optical network supporting multiple modulation formats and signal rate is needed in the future.Next-generation optical network presents some new features,including resource scheduling,dynamic bandwidth and other adaptive reconfigurable concept.In that situation,modulation format identification is proposed to provide information to the next compensation algorithm and guarantee the optimal decoding of signal.This paper focuses on MFI in coherent optical communication system,including extraction and recognition of signal features.Researchs are as follows:(1)Coherent optical communication system was established,which can transmit different modulation formats signals.Discussion on the performance of various recognition methods and analysis on the performance were finished,including the impact of ASE noise and polarization mode dispersion.(2)We theoretical derived the distribution of different modulation formats in Stokes space.We take advantage of the symmetry of data to recognition the modulation format with certain data.Compared with the existing algorithms.Computational complexity was reduced largely by using new clustering algorithm.Meanwhile,the range was extended to 32QAM.(3)The innovations including three aspects about the stokes space based modulation format identification:a.Two-stage method combining the feature of frequence spectrum and the feature of stokes space was utilized.Maximum of normalized instantaneous amplitude spectrum was proposed to distinguish between QAM signal and PSK signal.Based on this result further identification utilizing clustering analysis was done;b.Fuzzy C-means and hierarchical clustering were combined for cluster analysis in this method;c.Meanwhile,certain rule of selecting the initial cluster centers was applied in FCM to avoid falling into local optimum results.Simulation results show that the algorithm can effectively enhance the OSNR penalty at least 1dB.The OSNR threshold for QPSK,8PSK,16QAM,32QAM signals are individualy 17dB,22dB,28dB,30dBand the tolerance to PMD is 22ps and 14ps respectively for QPSK and 32QAM.