Research On Nonlinear Equalization Technology In Multi-core Fiber Transmission System

With the rise of new services such as artificial intelligence,highdefinition video and virtual reality,the network traffic has increased sharply,which poses a challenge to the transmission capacity requirements of traditional optical fiber communication systems.The main physical multiplexing dimensions of optical fiber are divided into five categories:time,frequency,polarization,amplitude/phase and space,At present,space division multiplexing(SDM),the spatial dimension in optical fiber communication system,has become the key technology to break through the capacity limitation of optical fiber communication system.It mainly realizes mode multiplexing and core number expansion in a single optical fiber.Among them,multi-core fiber may bend or deform in the actual transmission process,resulting in crosstalk between different cores and affecting the communication quality.In addition,fiber loss,dispersion and nonlinear effects in multi-core fiber will affect the transmission of optical signal.Among these damages,inter core crosstalk can be solved by MIMO(multi input multi output),and optical fiber loss can be compensated by optical fiber amplifier.However,the nonlinear effect in optical fiber will accumulate with the extension of transmission distance.Therefore,the damage mechanism and compensation algorithm of nonlinearity are the research focus of multi-core optical fiber transmission system.This paper focuses on the research based on the nonlinear effect in multi-core optical fiber.Firstly,the link damage in multi-core and optical fiber is introduced.Then,the multi-core optical fiber transmission system is built according to the transmission theory,and the machine learning algorithm and artificial neural network are introduced at the receiver to compensate the nonlinear damage and balance the signal.The main work of this paper includes the following aspects:(1)The nonlinear joint channel transmission model of multi-core fiber is studied.Aiming at the problem that the theoretical modeling of nonlinear transmission in multi-core optical fiber transmission system is not accurate enough,this paper jointly models the nonlinear effect in multi-core optical fiber and the nonlinear transmission of inter core crosstalk,and establishes a seven core optical fiber channel model with a total bit rate of 1.12tbit/s and a transmission distance of 1200km.(2)A multi-core fiber channel equalization algorithm based on Knearest neighbor method is proposed.By introducing KNN algorithm into the damage equalization of multi-core optical fiber channel,the proposed scheme effectively reduces the damage caused by nonlinear secondary stress factors in multi-core optical fiber channel;According to the characteristics of KNN algorithm for multi-core fiber channel equalization,an improved AKNN algorithm is proposed.Compared with the traditional algorithm,the tolerance of optical signal-to-noise ratio is increased by 1.1 dB and the average transmission distance is extended by 900km.(3)A multi-core optical fiber equalization algorithm based on artificial neural network is proposed.By combining the artificial neural algorithm with multi-core optical fiber communication system,the optical signal-to-noise ratio tolerance of the proposed algorithm is reduced by about 1dB and the average transmission distance is extended by 400km compared with AKNN algorithm.