Research On Quantization Scheme Of Imdd Based Optical OQAM-OFDM Transmission System

In recent years,the emergence of new and developing technologies such as machine learning,social networking and cloud computing causes the rapid growth of user data traffic and bandwidth demands.To satisfy the requirements of high-bandwidth and large-capacity,short-reach transmission systems have gradually become one of the core reasearch of optical fiber transmission networks.The technology of short-reach transmission mainly includes direct dection based on IMDD(Intensity Modulated Direct Detection)and coherent detection based on IQ modulation.IMDD technology has the characteristics of low cost and simple implementation,and advanced optical multi-carrier modulation technology is a promissing solution to improve the performance of IMDD system.Offset Quadrature Amplitude Modulation-Orthogonal Frequency Division Multiplexing(OQAM-OFDM)based on IMDD becomes a preferred scheme in optical multi-carrier transmission systems owing to the low cost,high spectral efficiency and asynchronous transmission.This thesis focuses on the quantization technology of IMDD based optical OQAM-OFDM transmission systems.In order to effectively solve the system performance degradation caused by the peak to average power ratio(PAPR)and the requirement of high precision of digital-toanalog/analog-to-digital convertors in the OQAM-OFDM system,the research of scalar quantization and vector quantization is carried out and the main work and innovation points are as follows:1.Scalar quantization includes uniform quantization and non-uniform quantizaion.Combining clipping and quantization,a uniform quantization scheme is proposed.And the effects of uniform quantization on system performance is studied.The limitation of uniform quantization is explained,three non-uniform quantization schemes are proposed,and they are nonuniform quantization scheme based on Gaussian distribution,nonparametric histogram estimation(NPHE)and back propagation neural network(BPNN).Compared with uniform quantization scheme,they can achieve the signal to quantization noise ratio(SQNR)improvements of3.33-d B,3.88-d B and 3.87-d B,respectively.2.In order to further reduce the quantization noise,the two vector quantization schemes are proposed,namely vector quantization based on Kmeans clustering and fuzzy C-means clustering(FCM).Compared with scalar quantization,vector quantization has two advantages: Firstly,the system overhead can be reduce due to the lack of linear programming.Secondly,the system performance is improved due to the correlations between in-pharse signals and quadrature signals.The feasibility of vector quantization schemes is experimentally verified in the IMDD-based optical OQAM-OFDM transmission systems,the bit error rate(BER)thresold of forward error correction(FEC)at 10G/s and low received optical power is achieved.In the case of same quantization bit resolution,vector quantization can obtain at least 1d B SQNR improvement compared to optimal nonuniform scalar quantization schemes.