Design And Optimization Of Passive Optical Network System Based On Sparse Code Division Multiple Access

As the level of global informatization continues to increase,many industrial powers have put forward strategies to transform traditional economies into digital economies.Under the guidance of the CPC Central Committee and the State Council’s "Network Power" strategy,the optical access network has been considered as a part of the national "new infrastructure" category and the "14th Five-Year Plan".As the foundation supporting the comprehensive digital transformation of society,its importance and enabling effect increasingly prominent.However,with the increase of network communication users and communication equipment,the optical access network gradually faces the problem of insufficient bandwidth.To further increase network bandwidth,researchers have proposed various multiplexing schemes to improve spectral efficiency.The multiplexing dimensions mainly includes time,frequency,polarization state,and space.However,in these multiplexing schemes,signals must be guaranteed to be orthogonal.Limited by the orthogonality,these multiplexing technologies cannot support the simultaneous sharing of the same orthogonal resources among different users,which restricts the access bandwidth of users to a certain extent.As a non-orthogonal multiple access technology,sparse code division multiple access(SCMA)technology has the advantages of large capacity and high spectral efficiency,and is regarded as one of the alternative technologies for 5G.This thesis proposes to introduce SCMA into passive optical network(PON),and focuses on the improvement of receiver performance and the design of anti-interference codebook in SCMA-PON.The main research contents are as follows:1.The SCMA-PON scheme is proposed.For the problem that the original structure of each user codebook is destroyed due to the new users,an access scheme based on user groups is designed to effectively reuse the codebook,and the feasibility of the scheme is verified based on the VPI simulation software.In addition,the effects of the transmission distance,the number of iterations of the MPA algorithm used by the receiver for decoding,and the order of scheduling on the performance of SCMA-PON are also explored.Simulations show that the convergence performance of the MPA algorithm used for decoding at the receiving end is robust to chromatic dispersion,and as the modulation order increases,the SCMA-PON system is more sensitive to chromatic dispersion.2.For the error propagation problem of the traditional multi-user detection scheme LSMPA,this thesis proposes an end-to-end graph neural network decoding scheme to replace the traditional equalization and multi-user detection,which can effectively avoid error propagation and improve reception.machine sensitivity.The simulation results show that the graph neural network demodulation scheme proposed in this thesis improves the optical power of the traditional LS-MPA receiver by 1.5d B@b2b,1.5d B@20km,2d B@60km,and the scheme is not effective for the non-conventional effects introduced by increasing the transmit power.Linear damage also has strong compensation ability.3.In SCMA,the codebook used by the user will greatly affect the transmission performance.Aiming at the high complexity of the current codebook design optimization criterion,this thesis proposes a low-complexity pairwise error rate optimization criterion,which uses genetic algorithm to jointly optimize the phase and amplitude information.And a new label mapping rule is proposed.Since the optimization criterion adopted in this thesis has low computational complexity,it is more suitable for large-scale codebook design.The simulation results show that the codebook design in this thesis has better bit error performance in the case of low OSNR.