Adaptive Modulation And Coding For Large-Capacity And Flexible Short-to-Medium Reach Optical Networks

The development of Internet applications represented by high-definition video,cloud computing,cloud storage,virtual/augmented reality,etc.has promoted the continuous and rapid growth of network traffic.With the gradual popularization of the fifth generation of mobile communication and the Internet of Things,massive terminals enter the Internet,resulting in increasingly diverse link requirements,and most of the traffic will be close to the user end.Continuously growing and diversified traffic is putting tremendous pressure on cost-sensitive short-to-medium-distance(<80 km)fiber-optic communication networks represented by data center networks(DCN)and content delivery networks(CDN).Cost-effective,flexible and high-capacity transmission technologies help to cope with these challenges and provide a technical solution for the next generation of fiber optic communication systems.In this thesis,aiming at the common low-cost direct-detection systems in short-to-medium-range optical fiber communications,from the perspective of modulation and coding,a variety of multicarrier-based adaptive loading algorithms were proposed to achieve high-performance,flexible and large-capacity transmission.The research contents and results of this thesis are as follows:1)A relatively complete physical layer theoretical model for direct detection systems was constructed.To avoid the spectral selective fading caused by chromatic dispersion in intensity modulation direct detection systems,the principles and limitations of generating single sideband signals with MZM modulators and IQ modulators were investigated.On this basis,we adopted Kramers-Kronig receiver and chromatic dispersion compensation technology.The principles and implementation methods of two multicarrier modulation formats,discrete multi-tone(DMT)and staggered multi-tone(SMT)or filter-bank multicarrier / offset quadrature amplitude modulation(FBMC/OQAM),were focused on to lay the foundation for the research of high-performance adaptive loading of modulation and coding.2)The construction principle of forward error correction(FEC)coding was studied.In order to avoid complicated encoding and decoding operations,appropriate parameters were selected for hard-decision FEC and soft-decision FEC to predict the bit error rate(BER)after decoding.The theoretical error correction capability of the hard-decision Reed-Solomon code and its shortened code were calculated to provide a theoretical basis for the proposed three-dimensional(modulation format,FEC and power)adaptive loading(3D-AL)algorithm.Under the condition that the maximum modulation order is limited,the data rate of 3D-AL algorithm is about 6% larger than that of the traditional bit and power loading(BPL)algorithm,Levin-Campello(LC)algorithm.The 3D-AL algorithm allocates initial parameters based on a lookup table method,making its iterations only 27.6% ? 38%of that of LC algorithm.3)The principle of the probabilistic amplitude shaping(PAS)architecture and constant component distribution matcher(CCDM)were studied which were used to generate probabilistically shaped quadrature amplitude modulation(PS-QAM).The actual spectral efficiency of PS-QAM was derived considering CCDM and off-the-shelf FEC codes.The experimental results show that the general mutual information(GMI)widely used in the early days of the invention of PS-QAM(2015 ? 2018)is not suitable for calculating the actual data rate.The normalized general mutual information(NGMI)is adopted to accurately predict the BER after soft-decision FEC decoding.Under the premise of error-free transmission ensured by NGMI,the PS-QAMs that achieve maximum spectral efficiency were obtained for different signal-to-noise ratios(SNRs)in an additive Gaussian white noise(AWGN)channel,providing the conditions for the adaptive loading based on PS-QAM,namely entropy loading(EL).4)The principle of noise equalization of multicarrier precoding technology was obtained through theoretical derivation.Based on this principle and the water-filling algorithm,the channel capacity of the precoding system and its relationship with the channel capacity of the normal system were obtained for the first time.The theory proves that precoding reduces the channel capacity,but substantially simplifies the adaptive loading process and reduces the peak-to-average power ratio(PAPR)of multicarrier signals.Reducing PAPR effectively improves system power utilization efficiency.From the perspective of low complexity,two adaptive loading algorithms,uniform time-domain hybrid QAM(TDHQ)loading and uniform entropy loading(UEL),for single precoding set(SPS)systems were proposed.The former is the first scheme using TDHQ in a multicarrier system,and the latter is almost the simplest EL scheme.Simulation and experimental results show that both of the transmission rate and PAPR performance of UEL algorithm are better than those of the LC algorithm.According to the optical back-to-back experimental results,the receiver sensitivity is improved by about 1 d B when the net data rate is 70?85 Gb/s.The data rate of TDHQ algorithm is not as high as that of the LC algorithm,but the complexity and PAPR performance of TDHQ algorithm are better than those of the LC algorithm.The reduced PAPR value can be up to 2.56 d B.5)Based on theoretical analysis,the fact that the multiple precoding sets(MPS)system has a lower channel capacity loss than the SPS system was found.A reasonable configuration of the MPS distribution achieves little capacity loss caused by precoding.Combining the MPS system with EL by loading PS-QAM per precoding set,three adaptive loading algorithms,the equally partitioned precoding(EL-EPP)algorithm,the optimally partitioned precoding(EL-OPP)algorithm,and the K-means assisted partitioned precoding(EL-KPP)algorithm,were proposed.Simulation and experimental results show that the transmission rate of these EL algorithms is in proportion to the number of precoding sets(NPS).And when the NPS is greater than 3,the transmission rate tends to be saturated.Thus,3?4 PS-QAMs are enough for these schemes,which is much less than those required by the conventional EL(CEL)algorithm that performs loading per subcarrier.In addition,each precoding set contains multiple subcarriers,which guarantees enough symbols for each PS-QAM to eliminate the rate loss of CCDM,while CEL can hardly guarantee this condition with a normal data frame.The experimental results of the DMT system based on single sideband modulation and direct detection show that the EL-EPP and EL-OPP algorithms improve the receiver sensitivity by 3.2 d B and 4.4 d B respectively compared with the LC algorithm when the net data rate is 60 Gb/s after the transmission of 80 km standard single mode fiber(SSMF).The experimental results of the SMT system based on single sideband modulation and Kramers-Kronig detection show that the EL-KPP algorithm improves the optical signal-to-noise ratio sensitivity by 1.5 d B compared to the LC algorithm when the net rate is 90 Gb/s after 125 km optical fiber transmission in the experiments.Besides,the EL-KPP algorithm supports the net data rate of 100 Gb/s after125 km SSMF transmission with the 3 d B end-to-end system bandwidth of about 11 GHz,which is close to the theoretical data rate achieved by CEL algorithm without considering the loss of CCDM rate.