Study On The Key Technologies Of Large-Capacity Low-Cost Metro Optical Networks

As emerging large-bandwidth low-latency applications grow fast,more and more data centers are built around the citys in which users live by internet service providers,cloud service providers and network operators.Thus,most of the global internet traffic will be borne in metro optical networks(MONs).When the capacity demand of MONs is growing fast,the growth rate of income for providers and operators will fall behind the one of expense owing to the policy of "Facilitating Faster and More Affordable Internet Connection".Therefore,how to realize a low-cost and large-capacity MON will be a critical problem in the study of MONs.Large-capacity low-cost MONs depend on low-cost high-speed transmission technology and intelligent management&control technology.The link impairment mitigation of low-cost direct-detection(DD)transmission systems is a main challenge in short-reach metro data center interconnect(DCI)networks.For metro core networks with longer reach,they face a challenge of intelligent management&control in low-margin elastic optical networks.Aimed at the challenges in large-capacity and low-cost MONs,metro DD transmission technologies and MON soft failure diagnosis technologies are studied in this dissertation.The main study content and innovation are as following:1.Study on nonlineariry compensation technologies in metro DD transmission systemsa)Chromatic dispersion and signal to signal beating inference(SSBI)are primary impairments that limit the capacity-distance product in DD transmission systems.A novel single sideband(SSB)four-ary pulse amplitude modulation(PAM4)transmission scheme is designed,which is based on a low-cost dual-drive Mach-Zehnder modulator(DD-MZM)and Kramers-Kronig(KK)detection.112-Gbps 80-km SSB-PAM4 DD transmission experiments show that this scheme can effectively mitigate chromatic dispersion and reduce the computational complexity of the equalizer by 87%,which will reduce the hardware cost and computational complexity in metro DCI DD transmission technologies further.b)The transmission performance of advanced modulation formats is severely limited by the saturation effect of electric amplifiers,modulation nonlinearity and fiber nonlinearity.An I/Q Volterra filter(VF)is proposed to mitigate nonlinear distortions in single-polarization 16-ary quadrature amplitude modulation(16QAM)DD systems.In addition,a sparse I/Q VF is realized by l1 regularization and re-training.In 112-Gbps single-polarization 16QAM DD transmission experiments,the maximal transmission distance is extended from 480 km to 960 km with the help of I/Q VF.The capacity distance product reaches the international advanced level.At 960 km,the computational complexity is reduced by 58%for the sparse I/Q VF when the BER stays the same.Compared with a conventional real-valued I/Q linear filter,the proposed I/Q VF can improve the tolerance to system nonlinearities significantly.Thus,it can extend the application range of DD transmission technologies and reduce the cost of MONs further.2.Study on quantization noise compensation technologies for low-resolution digital to analog converters(DACs)in metro DD transmission systemsLow-resolution DACs is beneficial for reducing the cost,size and power consumption of optical transceivers,but they also face the challenge of large quantization noise.This dissertation investigates the applications of clipping,digital resolution enhancer(DRE)and error-feedback noise shaping(EFNS)in metro DD transmission systems by detailed simulation and experiments.The extensive results show that both DRE and EFNS can make the performance of a 4-bit DAC close to the one of an 8-bit DAC.Compared with DRE,EFNS has lower computational complexity and processing latency.In addition,no channel response is required for EFNS.Therefore,EFNS is more suitable for low-latency low-power-consumption metro DD transmission systems.This work can give an important guide for the design of low-resolution high-speed DD transmission schemes and thus reduce the hardware cost,size and power consumption of transmitters.3.Study on soft failure diagnosis technologies in metro low-margin elastic optical networksLow-margin elastic optical network is an inevitable trend in metro core networks.Soft failure diagnosis is a key technology to guarantee the availability and reliability of low-margin elastic optical neworks.In this dissertation,a spectrum area-based feature extraction method is proposed for digital spectra for the first time.This method considers the influence of common soft failures on the symmetry,power and signal to noise ratio of the signal spectra.Thus,this method can effectively reduce the input features of the model when the diagnosis performance is guaranteed.A low-complexity dual-stage soft failure detection and identification framework is designed based on the spectrum area-based feature extraction method.In the dual-stage soft failure detection(SFD)scheme,the low-frequency characteristic of soft failure occurrence is considered to reduce the monitoring and processing overhead.In an experimental system including three soft failures,dual-stage SFD structure reduce the number of digital spectrum calculation by 61.6%,and realize a false positive rate(FPR)of 0.42%and a false negative rate(FNR)of 1.47%.In addition,an accuracy of 99.55%is achieved for support vector machine-based soft failure identication.The performance of digital spectrum-based soft failure diagnosis technology reaches the international advanced level.This technology can significantly reduce the hardware monitoring and processing cost when the diagnosis performance is guaranteed,which is beneficial to realize low-cost intelligent management&control of MONs.