Research On Optical Modulation Format Recognition Technology

In order to adapt to the rapidly developing information age,coherent optical communication technology has become more and more complicated.With the advent of re-configurable transmitters,coherent receivers are moving toward adaptive direction.Therefore,research on optical modulation format recognition technology has received extensive attention.At present,the most widely used algorithms in modulation format recognition are neural network and k-means,the neural network belongs to the supervised learning algorithms,but the training cost is high;and the k-means belongs to unsupervised learming algorithms,there is no training cost,but it is necessary to determine the k values in advance.On this premise,this paper introduces the unsupervised learming algorithms Chinese Restaurant Process(CRP)and Iterative Self-organizing Data Analysis Techniques Algorithm(ISODATA)for modulation format recognition.Both algorithms not only reduce training cost,but also become more flexible in choosing k values.The main research work is as follows:1.The structure of coherent optical communication system and the general flow of optical modulation format recognition are described.BPSK,QPSK,8PSK,8QAM and 16QAM coherent optical communication systems with bit rate of 112 Gbit/s are built by using OptiSystem and MATLAB.2.Optical modulation format recognition technology based on CRP algorithm under Stokes space representation is studied.The relationship between the recognition rate of modulation format and OSNR,number of symbols,linewidth of light source,transmission distance and launch power is simulated and analyzed,and the feasibility of the modulation format recognition technology is verified.3.The recognition technology of optical modulation format based on ISODATA algorithm under least square fitting plane representation is studied The recognition rates of different modulation formats under different OSNR,symbol number,linewidth of light source,transmission distance and launch power are simulated and analyzed,and the feasibility of the proposed optical modulation format recognition technology is verified.Due to the addition of least squares fitting plane,the overall recognition effect is higher than that of CRP algorithm.4.The performance of CRP algorithm,ISODATA algorithm and traditional k-means algorithm are compared and studied.The results show that CRP algorithm and ISODATA algorithm are better than k-means algorithm.ISODATA algorithm is better than CRP algorithm for PSK signal recognition,and CRP algorithm is better than ISODATA algorithm for QAM signal recognition.