Research On Equalization Techniques In Faster-than-Nyquist Passive Optical Networks

Due to the wide spreading of bandwidth-hungry applications such as cloud service,internet of things and virtual reality,demand on the capacity of optical communication systems is growing continuously.PON beyond 10 Gb/s per wavelength has become a current research hotspot.Since low cost and power consumption are the most important factors for access networks,25-Gb/s PON with low-cost 10G-class optics becomes a more promising scheme.However,there is severe inter-symbol interference in such bandwidth-limited systems,which will further decrease the optical power budget and chromatic dispersion tolerance.Therefore,it is inevitable to deal with these challenges in the cost-effective 25-Gb/s PON.Researches show that faster-than-Nyquist(FTN)signaling has great advantages in bandwidth-limited,high-capacity transmission systems.If appropriate digital signal processing technologies are adopted at the receiver,FTN signal could increase the symbol rate by 25%without degradation of bit error rate performance.Faced with the rate demands for future access networks,feasible and low-complexity equalization techniques in FTN-PON are thoroughly investigated in this dissertation.The specific research contents and results are as follows:(1)Low-Complexity MLSD Based on NRZ.Due to the problems of current equalization methods,such as the high complexity of maximum likelihood sequence estimation(MLSE)and the limited equalization performance of feed-forward equalization(FFE),low-complexity maximum likelihood sequence detection(MLSD)based on non-return-to-zero(NRZ)signal is proposed,which is a cascade structure of FFE,post-filter and MLSD.FFE is used to equalize the received signal;the post-filter is applied to eliminate in-band noise enhanced by FFE and shape the channel response to a known finite impulse response pattern;MLSD is finally used to remove the remaining inter-symbol interference by channel information provided by post-filter,which further improves the transmission performance.(2)Optimization Method of Low-Complexity MLSD.This method is an optimization method of low-complexity MLSD based on NRZ,which is mainly aimed at the optimization of post-filter.Post-filter provides the equivalent channel information for MLSD,so the computational complexity of MLSD increases exponentially with the increase of channel memory length.When channel memory length is set as 2,the post-filter can be designed to have a fixed partial response,tap coefficients of which can be optimized by a parameter a.The channel coefficients obtained by this method are more matched with the actual channel.Compared with low-complexity MLSD based on NRZ,the transmission performance of this method is improved and the complexity of it is lower.(3)Equalization Based on EDB Receiver.In bandwidth-limited systems,the received signal is approximately converted to electrical duo-binary(EDB)signal due to the severe inter-symbol interference.Therefore,the received signal can be treated as EDB,then the signal can be equalized by a simple 3-level FFE,and finally,the signal can be decoded into NRZ signal.In this paper,a 28-Gb/s FTN-PON simulation and experimental system is built to compare the above three equalization methods.Simulation and experimental results show that the equalization method based on EDB receiver can not only guarantee good transmission performance but also possess a relatively low computational complexity.In summary,equalization techniques in FTN-PON systems are thoroughly studied in this paper.Low-complexity MLSD based on NRZ,its optimization method and equalization based on EDB reveiver are proposed,which equalization perf-ormance are verified through simulation and experiment.The research results are of great significance to the realization and commercialization of 25-Gb/s PON in the future.