Optical fiber communication system with intelligent dynamic bandwidth allocation

Abstract

The application relates to the technical field of optical fiber communication, in particular to an optical fiber communication system with intelligent dynamic bandwidth allocation, which is operated based on a front-end architecture and a back-end architecture and comprises a multi-dimensional perception module, a dynamic nonlinear modeling module, a resource optimization decision module, a control execution module, a spectrum fragment integration module, a dark spectrum perception module, a power-bandwidth joint adaptation module and a real-time synchronization module; the multi-dimensional perception module is used for space division-spectrum coupling detection and quantum noise monitoring.The optical fiber communication system performs resource dynamic optimization, optical straight-through control, cross-layer interference suppression and ultra-precision synchronization in a multi-module cooperative and systematic manner, ensures hard slice time slot alignment, has shorter end-to-end time delay, meets various industrial internet and other low-time-delay application requirements, and solves and optimizes the technical problems of multi-core optical fibers in the field of optical communication in the prior art.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of optical fiber communication technology, in particular to an optical fiber communication system with intelligent dynamic bandwidth allocation. BACKGROUND

[0002] Multi-core optical fiber is an optical fiber structure integrating multiple independent cores in a single cladding, which improves transmission capacity through space division multiplexing. According to the coupling strength of the cores, it can be divided into weakly coupled, strongly coupled and heterogeneous multi-core optical fibers. Among them, seven-core optical fiber has a hexagonal close-packed structure, which performs outstandingly in capacity, space efficiency and compatibility, and is widely used in the field of communication. Its core advantage lies in its high space utilization and capacity expansion capability. Compared with single-core optical fiber, it can provide 7 times the theoretical capacity under the same cladding size while maintaining compatibility with existing optical fiber infrastructure. In addition, its symmetrical core structure reduces manufacturing complexity and effectively suppresses nonlinear crosstalk through intelligent power management, making it have significant competitiveness in backbone networks, data center interconnection and other scenarios.

[0003] However, in general, in the actual application of the field of optical fiber communication, seven-core optical fiber still faces the problem of inter-core crosstalk management. The static resource allocation mode of the existing system is difficult to adapt to dynamic business needs, leading to serious spectrum fragmentation, and the Raman scattering effect is intensified when quantum and classical channels are co-transmitted. These problems limit its application in industrial internet and other low-latency scenarios.

[0004] Based on this, the present application provides an optical fiber communication system with intelligent dynamic bandwidth allocation to solve the above-mentioned technical problems. SUMMARY

[0005] The purpose of the present application is to provide an optical fiber communication system with intelligent dynamic bandwidth allocation to solve the problems raised in the background art.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0007] The optical fiber communication system with intelligent dynamic bandwidth allocation is based on front-end architecture and back-end architecture, and includes a multi-dimensional perception module, a dynamic nonlinear modeling module, a resource optimization decision module, a control execution module, a spectrum fragmentation integration module, a dark spectrum perception module, a power-bandwidth joint adaptation module, and a real-time synchronization module.

[0008] The multi-dimensional perception module is used for space-spectrum coupling detection and quantum noise monitoring.

[0009] The dynamic nonlinear modeling module is used to generate a dynamic nonlinear interference fingerprint library and cross-layer interference prediction.

[0010] The resource optimization decision module is used to generate a bandwidth allocation scheme and a game theory collaborative allocation.

[0011] The control execution module is configured to provide a topological photonic switch and power-time slot coding;

[0012] The spectrum fragment integration module is configured to holographic diffraction reconstruction and optimized alignment spectrum position;

[0013] The dark spectrum sensing module is configured to predict unavailable spectrum regions and compressed sensing detection;

[0014] The power-bandwidth joint adaptation module is configured to dynamic PSD game and XPM compensation coding;

[0015] The real-time synchronization module is configured to photonic neural synchronization and quantum clock calibration.

[0016] Preferably, the multi-dimensional sensing module further comprises a space-spectrum coupling detection unit and a quantum noise monitoring unit;

[0017] The space-spectrum coupling detection unit detects the coupling state of the space-spectrum mode through a photonic crystal fiber grating sensor, for real-time capture of inter-core crosstalk and spectrum occupancy;

[0018] The quantum noise monitoring unit measures the quantum noise power spectral density based on a squeezed state optical interferometer, for identifying noise interference thresholds in ultra-low power scenarios.

[0019] Preferably, the dynamic nonlinear modeling module further comprises a nonlinear channel fingerprint unit and a cross-layer interference prediction unit;

[0020] The nonlinear channel fingerprint unit learns the correlation features of fiber nonlinear effects and transmission power through a graph neural network, as shown in equation (1): (1); wherein, is the feature vector of core node v in the lth layer, is the nonlinear coupling weight of core v and neighbor core u, is the weight matrix of the lth layer neural network, used to extract nonlinear interference features, is the ReLU activation function, which maps linear transformation to nonlinear relationship, used to generate a dynamic nonlinear interference fingerprint library;

[0021] The cross-layer interference prediction unit simulates the cross-layer interference of quantum channels and classical channels based on a generative adversarial network, as shown in equation (2):

[0022] (2); in the formula, G is a generator, simulating the interference scene of quantum channel and classical channel, D is a discriminator, judging whether the input data is a real measurement value (x) or an interference data (G(z)) synthesized by the generator, z is a noise vector, inputting the generator to generate an interference distribution for pre-judging the conflict probability of the reserved spectrum window.

[0023] Preferably, the resource optimization decision module further comprises a multi-objective quantum annealing unit and a game theory collaborative allocation unit;

[0024] The multi-objective quantum annealing unit solves the spectrum-space-power multi-objective optimization problem by quantum annealing algorithm, as shown in formula (3): (3); in the formula, is the resource coupling constraint of fiber core i and j, is the spectrum efficiency weight of fiber core i, , is the spin operator of a quantum bit, encoding the allocation state of fiber core i, used to generate a Pareto optimal bandwidth allocation scheme;

[0025] The game theory collaborative allocation unit coordinates the distributed competition of multi-core resources based on a Nash equilibrium game model, as shown in formula (4): (4); in the formula, is the bandwidth allocated to fiber core i, is the signal-to-noise ratio of fiber core i, is the nonlinear interference coefficient of fiber core j to i, is the transmit power of fiber core i, used to balance the nonlinear interference and spectrum efficiency.

[0026] Preferably, the control execution module further comprises a topological photonic switch unit and a power-time slot coding unit;

[0027] The topological photonic switch unit realizes lossless and fast switching of the space division mode through a topological boundary state optical switch, for dynamically allocating the core resources in the multicore fiber;

[0028] The power-time slot coding unit dynamically adjusts the transmit power and time slot mapping relationship based on a nonlinear pre-distortion encoder, for suppressing nonlinear distortion under high power.

[0029] Preferably, the spectrum fragment integration module further comprises a holographic diffraction reconstruction unit and a Bayesian optimization alignment unit;

[0030] The holographic diffraction reconstruction unit coherently synthesizes the fragmented spectrum through an optical holographic diffraction chip, as shown in formula (5): (5); in the formula, is the input light field, is the transfer function of the holographic chip for spectrum synthesis, For Fourier transform, the spatial domain light field is converted into the frequency domain, which is used to recombine the discontinuous frequency spectrum blocks into continuous bandwidths;

[0031] The Bayesian optimization alignment unit dynamically adjusts the spectral offset based on the Bayesian optimization algorithm, which is used to align the fragmented spectrum positions among multiple cores.

[0032] Preferably, the dark spectrum sensing module further includes a federated learning mapping unit and a compressed sensing detection unit;

[0033] The federated learning mapping unit trains a dark spectrum distribution model across domains through a federated learning framework, which is used to predict the unavailable spectrum area at the intersection of the metropolitan area network and the backbone network;

[0034] The compressed sensing detection unit sparsely samples and reconstructs the dark spectrum based on compressed sensing technology, which is used for low-overhead detection of blind area spectrum availability.

[0035] Preferably, the power-bandwidth joint adaptation module further includes a dynamic PSD game unit and an XPM compensation coding unit;

[0036] The dynamic PSD game unit adjusts the transmission power and channel spacing through a dynamic power spectral density game algorithm, as shown in equation (6): (6); where, is the transmission power of core i at the kth iteration, is the target signal-to-noise ratio, is the noise power spectral density, which is used to balance the nonlinear interference and bandwidth utilization;

[0037] The XPM compensation coding unit pre-compensates the cross-phase modulation interference based on a Turbo-type nonlinear equalizer, which is used to compress the channel protection interval.

[0038] Preferably, the real-time synchronization module further includes a photonic neural synchronization unit and a quantum clock calibration unit;

[0039] The photonic neural synchronization unit achieves sub-microsecond time synchronization for multi-dimensional resource allocation through a photonic pulse neural network, as shown in equation (7): (7); where, is the synaptic weight of the photonic neural network, is the path transmission delay, T is the time slot period, which is used to eliminate delay jitter;

[0040] The quantum clock calibration unit calibrates the clock offset of distributed nodes based on a quantum entangled clock synchronization protocol, which is used for precise time slot alignment of hard-slicing bandwidth.

[0041] The intelligent dynamic allocation bandwidth optical fiber communication system is applied to dynamically allocate bandwidth in multi-core optical fiber communication at the intersection of the metropolitan area network and the backbone network.

[0042] Compared with the prior art, the present application has the following advantages:

[0043] The optical fiber communication system of the present application performs resource dynamic optimization, optical direct control, cross-layer interference suppression and ultra-precision synchronization in a modular and systematic manner. Through quantum annealing optimization driven by nonlinear fingerprint, XPM interference is reduced and spectral utilization is improved. At the same time, federal learning dark spectrum prediction and Bayesian alignment are performed, dark spectrum probability graph is generated by aggregating cross-domain data, and after the optimal offset of the spectral offset between cores is calculated by Bayesian optimization, the spectrum of the cores is avoided from the dark spectrum area. Then, the topology photon switch and entangled clock cooperate to switch the traffic between cores faster, the time base error is lower through quantum entangled clock protocol synchronization, the hard slice time slot is aligned, the end-to-end delay is shorter, and the application requirements of various industrial internet and other low latency applications are met. The technical problems of multi-core optical fiber in the field of optical communication in the prior art are solved and optimized. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 The present application is an intelligent dynamic bandwidth allocation optical fiber communication system topology.

[0045] Figure 2 The present application is an intelligent dynamic bandwidth allocation application flowchart. DETAILED DESCRIPTION

[0046] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0047] Embodiment 1, please refer to Figure 1 The present application proposes an intelligent dynamic bandwidth allocation optical fiber communication system, and further, the system is based on front-end architecture and back-end architecture.

[0048] It should be noted that the front-end architecture is composed of a multi-dimensional perception layer and an edge computing layer. The multi-dimensional perception layer collects real-time spectrum, space division mode, power and nonlinear interference data of the optical fiber network, and the edge computing layer deploys a lightweight algorithm to preprocess the data and reduce the core network computing load. The data interface protocol between the edge computing layer and the core network is gRPC. The back-end architecture is composed of an intelligent decision layer and a control execution layer. The intelligent decision layer generates a dynamic allocation strategy based on a multi-objective optimization model, and the control execution layer realizes physical layer mapping of the strategy through photon devices and DSP chips.

[0049] Specifically, the system comprises a multi-dimensional perception module, a dynamic nonlinear modeling module, a resource optimization decision module, a control execution module, a spectrum fragment integration module, a dark spectrum perception module, a power-bandwidth joint adaptation module, and a real-time synchronization module.

[0050] The multi-dimensional perception module is used for space-spectrum coupling detection and quantum noise monitoring, the dynamic nonlinear modeling module is used for generating a dynamic nonlinear interference fingerprint library and cross-layer interference prediction, the resource optimization decision module is used for generating a bandwidth allocation scheme and game theory collaborative allocation, the control execution module is used for providing a topological photon switch and power-time slot coding, the spectrum fragment integration module is used for holographic diffraction reconstruction and optimized alignment of spectral positions, the dark spectrum perception module is used for predicting unavailable spectrum regions and compressed sensing detection, the power-bandwidth joint adaptation module is used for dynamic PSD game and XPM compensation coding, and the real-time synchronization module is used for photon neural synchronization and quantum clock calibration.

[0051] In this embodiment, it should also be noted that the multi-dimensional perception module further comprises a space-spectrum coupling detection unit and a quantum noise monitoring unit.

[0052] Further, the space-spectrum coupling detection unit detects the coupling state of the space-spectrum mode through a photonic crystal fiber grating sensor, and is used for real-time capture of inter-core crosstalk and spectrum occupancy.

[0053] Further, the quantum noise monitoring unit measures the quantum noise power spectral density based on a squeezed state optical interferometer, and is used for identifying the noise interference threshold in an ultra-low power scenario.

[0054] In this embodiment, it should also be noted that the dynamic nonlinear modeling module further comprises a nonlinear channel fingerprint unit and a cross-layer interference prediction unit.

[0055] Further, the nonlinear channel fingerprint unit learns the correlation features of fiber nonlinear effects and transmission power through a graph neural network, as shown in equation (1): (1); wherein, is a feature vector of a core node v at the lth layer, is a nonlinear coupling weight of the core v and a neighbor core u, is a weight matrix of the lth layer neural network, used for extracting nonlinear interference features, is a ReLU activation function, which maps a linear transformation to a nonlinear relationship, and is used for generating a dynamic nonlinear interference fingerprint library;

[0056] It should also be noted that the input features are defined as follows:

[0057] The feature vector of the core node v contains the following physical parameters (dimension 4): ;

[0058] wherein, it also needs to be explained that:

[0059] : the launch power of fiber core v (unit: dBm);

[0060] : the spectral occupancy of fiber core v (unit: %);

[0061] : the average crosstalk coefficient between fiber core v and the nearest neighbor fiber core (unit: dB);

[0062] : the real-time signal-to-noise ratio of fiber core v (unit: dB);

[0063] wherein, it also needs to be explained that the neighbor relationship is defined as:

[0064] The neighbor set of fiber core v is defined as the fiber core that is physically adjacent;

[0065] In a seven-core optical fiber, the neighbors of the center core 3 are cores 2, 4, 5, and 6 (hexagonal arrangement structure);

[0066] wherein, it also needs to be explained that the calculation of the nonlinear coupling weight

[0067] Based on the calibration experiment of the nonlinear effect of the optical fiber (XPM / FWM), the weight formula is:

[0068] ;

[0069] wherein, it also needs to be explained that:

[0070] : the spectral interval between fiber core v and u (unit: GHz);

[0071] : the nonlinear coefficient of the optical fiber (typical value: );

[0072] wherein, it also needs to be explained that the GNN training and verification:

[0073] Dataset construction: collect the crosstalk data of the multi-core optical fiber under different powers {P}_{v}\in [5,15]\mathrm{d}\mathrm{B}\mathrm{m} and spectral intervals {\Delta \lambda}_{vu}\in [25,100]\mathrm{G}\mathrm{H}\mathrm{z} , a total of 10,000 groups of samples;​

[0074] The loss function is: ;

[0075] : L2 regularization coefficient;

[0076] Output dynamic nonlinear interference fingerprint library, XPM interference intensity of core 3 and 5 ;

[0077] Further, the cross-layer interference prediction unit simulates the cross-layer interference of the quantum channel and the classical channel based on the generative adversarial network, as shown in equation (2):

[0078] (2); In the formula, G is a generator, simulating the interference scenario of the quantum channel and the classical channel, D is a discriminator, judging whether the input data is a real measurement value (x) or an interference data synthesized by the generator (G(z)), z is a noise vector, input into the generator to generate an interference distribution for predicting the conflict probability of the reserved spectrum window;

[0079] It should be further pointed out that the real data x is collected:

[0080] Experimental setup: core 3 transmits classical service (1545nm, 10Gbps), core 5 transmits quantum channel (1550nm, single-photon level signal);

[0081] It should be further pointed out that the measurement parameters are:

[0082] Raman scattering power : Use an optical spectrum analyzer (OSA) to measure the noise power at the quantum channel wavelength (unit: dBm);

[0083] Quantum error rate : The quantum receiving end counts the error rate change (unit: %);

[0084] Spectrum offset of classical channel : Use a coherent receiver to measure the center frequency offset (unit: GHz);

[0085] Dataset: x=[{P}_{\mathrm{R}\mathrm{a}\mathrm{m}\mathrm{a}\mathrm{n}},\Delta \mathrm{B}\mathrm{E}\mathrm{R},\Delta f] , a total of 5,000 samples;

[0086] It should be further pointed out that the definition of the noise vector z is:

[0087] Dimension 4, distributed as follows:

[0088] \mathrm{z}=[\Delta P,\Delta \lambda,{N}_{\mathrm{p}\mathrm{h}\mathrm{o}\mathrm{t}\mathrm{o}\mathrm{n}},\Delta T] ;

[0089] : Transmit power fluctuation;

[0090] : Spectral interval random offset;

[0091] : Quantum signal photon number;

[0092] : Environmental temperature disturbance;

[0093] It should be noted that the generator G and the discriminator D are designed:

[0094] Generator G:

[0095] Input noise vector z (4 dimensions);

[0096] Output synthetic interference data ;

[0097] Network structure: 3-layer fully connected neural network (hidden layer dimension 64, activation function ReLU);

[0098] Discriminator D:

[0099] Input real data x or generated data G(z) (3 dimensions);

[0100] Output discriminant probability D(\mathrm{x})\in [0,1] ;

[0101] Network structure: 3-layer fully connected neural network (hidden layer dimension 64, activation function LeakyReLU);

[0102] It should be noted that the conflict probability mapping:

[0103] The discriminator output D(x) is mapped to the conflict probability: ;

[0104] Slope factor k = 10: calibrated by experiment (when the bit error rate change sensitivity is the highest).

[0105] It should be noted that the resource optimization decision module further comprises a multi-objective quantum annealing unit and a game theory collaborative allocation unit in the embodiment.

[0106] Further, the multi-objective quantum annealing unit solves the spectrum-space-power multi-objective optimization problem by quantum annealing algorithm, as shown in equation (3): (3); wherein, is the resource coupling constraint of the fiber core i and j, is the spectrum efficiency weight of the fiber core i, , is the spin operator of the quantum bit, which encodes the allocation state of the fiber core i, and is used to generate the Pareto optimal bandwidth allocation scheme;

[0107] Further, the game theory collaborative allocation unit coordinates the distributed competition of multi-core resources based on the Nash equilibrium game model, as shown in equation (4): (4); wherein, is the bandwidth allocated to the fiber core i, is the signal-to-noise ratio of the fiber core i, is the nonlinear interference coefficient of the fiber core j to i, is the transmission power of the fiber core i, which is used to balance the nonlinear interference and spectrum efficiency.

[0108] It should be noted that the control execution module further comprises a topological photon switch unit and a power-time slot coding unit in the embodiment.

[0109] Further, the topological photon switch unit realizes lossless and fast switching of the space division mode through the topological boundary state optical switch, which is used to dynamically allocate the core resources in the multicore fiber;

[0110] Further, the power-time slot coding unit dynamically adjusts the transmission power and time slot mapping relationship based on the nonlinear pre-distortion encoder, which is used to suppress nonlinear distortion under high power.

[0111] It should be noted that the spectrum fragment integration module further comprises a holographic diffraction reconstruction unit and a Bayesian optimization alignment unit in the embodiment.

[0112] Further, the holographic diffraction reconstruction unit coherently synthesizes the fragmented spectrum through the optical holographic diffraction chip, as shown in equation (5): (5); wherein, is the input light field, is the transfer function of the holographic chip for spectrum synthesis, is the Fourier transform, which converts the spatial light field into the frequency domain, and is used to recombine the discontinuous spectrum blocks into continuous bandwidth;

[0113] Further, the Bayesian optimization alignment unit dynamically adjusts the spectral offset based on a Bayesian optimization algorithm, for aligning the fragmented spectral positions among the multiple cores.

[0114] In this embodiment, it should be further explained that the dark spectrum sensing module further includes a federated learning mapping unit and a compressed sensing detection unit.

[0115] Further, the federated learning mapping unit trains the dark spectrum distribution model across domains through a federated learning framework, for predicting the unavailable spectrum region at the interconnection of the metropolitan area network and the backbone network.

[0116] Further, the compressed sensing detection unit performs sparse sampling and reconstruction on the dark spectrum based on compressed sensing technology, for low-overhead detection of the availability of the blind area spectrum.

[0117] In this embodiment, it should be further explained that the power-bandwidth joint adaptation module further includes a dynamic PSD game unit and an XPM compensation coding unit.

[0118] Further, the dynamic PSD game unit adjusts the transmission power and channel spacing through a dynamic power spectral density game algorithm, as shown in equation (6): (6); in the equation, is the transmission power of core i at the kth iteration, is the target signal-to-noise ratio, is the noise power spectral density, used to balance the nonlinear interference and bandwidth utilization;

[0119] Further, the XPM compensation coding unit pre-compensates the cross-phase modulation interference based on a Turbo-type nonlinear equalizer, for compressing the channel protection interval.

[0120] In this embodiment, it should be further explained that the real-time synchronization module further includes a photonic neural synchronization unit and a quantum clock calibration unit.

[0121] Further, the photonic neural synchronization unit realizes sub-microsecond time synchronization of multi-dimensional resource allocation through a photonic pulse neural network, as shown in equation (7): (7); in the equation, is the synaptic weight of the photonic neural network, is the path transmission delay, T is the time slot period, used to eliminate delay jitter;

[0122] Further, the quantum clock calibration unit calibrates the clock offset of the distributed nodes based on a quantum entangled clock synchronization protocol, for precise time slot alignment of the hard-sliced bandwidth.

[0123] Embodiment 2, please refer to Figure 2In practical applications, the application of the intelligent dynamic bandwidth allocation optical fiber communication system of the application to the multi-core optical fiber communication dynamic bandwidth allocation at the junction of the metropolitan area network and the backbone network specifically includes the following steps:

[0124] (1) Multi-dimensional data acquisition and preprocessing: the spectral occupancy and mode coupling coefficient of cores 1-7 in the multi-core optical fiber are collected by the photonic crystal fiber grating sensor, the quantum noise power spectral density of core 2 in the L band (1565-1625 nm) is measured by the squeezed state light interferometer, the dark spectrum at the junction of the metropolitan area network and the backbone network is sparsely sampled, core 3 in the multi-core optical fiber (7 cores) is allocated a hard slice bandwidth with a 1 ms latency requirement for industrial internet business, core 5 carries ordinary internet traffic, and the steps of space-spectrum coupling detection, quantum noise monitoring and compressed sensing detection are;

[0125] (1.1) Space-spectrum coupling detection:

[0126] The photonic crystal fiber grating sensor collects the real-time spectral occupancy of core 3 and the crosstalk data of adjacent core 5, the occupancy rate of core 3 in the C band (1540-1545 nm) is 92%, and the crosstalk coefficient with core 5 is -15 dB, and the spectrum-space coupling matrix is: {\mathrm{C}}_{3,5}=[0.92,-15] ;

[0127] (1.2) Quantum noise monitoring:

[0128] The noise floor of core 3 under a single photon level signal is measured by the squeezed state light interferometer, the quantum noise power spectral density , the quantum noise threshold , and the low power optimization mode is triggered;

[0129] (1.3) Compressed sensing detection:

[0130] The junction of the metropolitan area network and the backbone network is sampled at a rate of 10%, and it is detected that core 4 has a dark spectrum at 1570-1572 nm, the bit error rate , the dark spectrum marker vector D=[0,0,0,1,0,0,0], and then core 4 is marked as a dark spectrum;

[0131] (2) Nonlinear modeling and cross-layer interference prediction:

[0132] (2.1) Nonlinear channel fingerprint:

[0133] The transmission power of core 3 , and the spectral interval of core 5 ;

[0134] The following algorithm is executed: ;

[0135] XPM interference strength prediction value (normalized);

[0136] (2.2) Cross-layer interference prediction:

[0137] 1550nm quantum channel and core 3 1545nm classical channel power distribution, generate quantum error rate changes caused by stimulated Raman scattering, quantum-classical channel conflict probability ;

[0138] (3) Multi-objective optimization dynamic decision:

[0139] (3.1) Multi-objective quantum annealing:

[0140] Core 3 spectrum requirement (1540~1550nm), , dark spectrum mark D, construct Hamiltonian: ;

[0141] Pareto optimal solution for core 3 to allocate 1540~1545nm, avoid core 5 1546~1550nm, power , guard interval ;

[0142] (3.2) Game theory collaborative allocation:

[0143] Core 5 current power , spectrum occupancy, 85%, Nash equilibrium solution: ; Therefore, the equilibrium strategy is: core 5 right spectrum to 1546.5~1551.5nm, power to 8.8dBm;

[0144] (4) Physical layer dynamic execution feedback:

[0145] (4.1) Holographic diffraction reconstruction:

[0146] Fragmented spectrum of core 3 (1540~1541.5nm, 1542~1545nm), optical diffraction field reconstruction: ;

[0147] Recombined into continuous spectrum 1540~1545nm, insertion loss <0.2dB;

[0148] (4.2) Dynamic PSD game:

[0149] Initial PSD of core 3 ; Adjusted PSD of core 5 , power iteration formula: ;

[0150] Converge to , the guard interval is compressed to 1.5GHz;

[0151] (4.3) Photon neural synchronization:

[0152] Hard slice time slot requirement of core 3, (1ms period, 50ns jitter), the algorithm expression of optical pulse synchronization is: ;

[0153] The delay jitter is reduced to 5ns, and the time slot alignment error is less than 0.1%; (5) Dark spectrum avoidance and cross-domain cooperation:

[0154] (5.1) Federal learning mapping:

[0155] The metropolitan area network reports the dark spectrum interval of core 4 as 1570-1572nm, and the federal aggregation updates the global dark spectrum probability model , and prohibits core 4 from allocating new services at 1570-1572nm;

[0156] (5.2) Bayesian optimization alignment:

[0157] The available spectrum of core 4 is 1565-1570nm and 1572-1575nm, and the Bayesian optimization objective function is: ;

[0158] The optimal offset moves the spectrum of core 4 to 1566.2-1571.2nm and 1573.2-1576.2nm; through the above steps, mainly through quantum annealing optimization driven by nonlinear fingerprint, XPM interference is reduced to 0.08, spectrum utilization is improved by 15%, and at the same time, federal learning dark spectrum prediction and Bayesian alignment are carried out, cross-domain data is aggregated, and a dark spectrum probability graph is generated. After the optimal offset of the spectrum offset between cores calculated by the Bayesian optimization, the spectrum of the core avoids the dark spectrum region, and then the topology photon switch and entangled clock cooperate to switch the traffic between cores within 1 , the time base error is synchronized to 0.1ns through the quantum entangled clock protocol, the hard slice time slot is aligned, and the end-to-end delay is stabilized at 1.05 , which meets the various industrial internet demands and solves and optimizes the technical problems of multi-core optical fiber in the field of optical communication in the prior art.

[0159] In the description of the specification, reference to "one embodiment", "an example", "a specific example" or the like means that a particular feature, structure, material or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the application. The appearances of the phrases "in one embodiment", "an example", "a specific example" or the like in various places in the specification are not necessarily referring to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0160] The preferred embodiments of the application disclosed above are only to help explain the application. The preferred embodiments do not describe all the details of the application and limit the application to the specific embodiments described. Obviously, many modifications and variations can be made in light of the contents of the specification. The specification selects and specifically describes these embodiments in order to better explain the principles and practical application of the application, so that those skilled in the art can well understand and utilize the application. The application is limited only by the claims and their full scope and equivalents.

Claims

1. An optical fiber communication system with intelligent dynamic bandwidth allocation, characterized in that, The multi-dimensional perception module, the dynamic nonlinear modeling module, the resource optimization decision module, the control execution module, the spectrum fragmentation integration module, the dark spectrum perception module, the power-bandwidth joint adaptation module, and the real-time synchronization module are included. The multi-dimensional perception module is used for space-spectrum coupling detection and quantum noise monitoring. The dynamic nonlinear modeling module is used for generating a dynamic nonlinear interference fingerprint library and cross-layer interference prediction. The resource optimization decision module is used for generating a bandwidth allocation scheme and game theory collaborative allocation. The control execution module is used for providing a topological photon switch and power-time slot coding. The spectrum fragmentation integration module is used for holographic diffraction reconstruction and optimized alignment spectrum position. The dark spectrum perception module is used for predicting unavailable spectrum regions and compressed sensing detection. The power-bandwidth joint adaptation module is used for dynamic PSD game and XPM compensation coding. The real-time synchronization module is used for photon neural synchronization and quantum clock calibration.

2. The intelligent dynamically allocated bandwidth optical fiber communication system of claim 1, wherein, The multi-dimensional perception module further includes a space-spectrum coupling detection unit and a quantum noise monitoring unit. The space-spectrum coupling detection unit detects the coupling state of the space-spectrum mode through a photonic crystal fiber grating sensor, and is used for real-time capture of inter-core crosstalk and spectrum occupancy. The quantum noise monitoring unit measures quantum noise power spectral density based on a squeezed state optical interferometer, and is used for identifying noise interference thresholds in ultra-low power scenarios.

3. The intelligent dynamically allocated bandwidth optical fiber communication system of claim 2, wherein, The dynamic nonlinear modeling module further includes a nonlinear channel fingerprint unit and a cross-layer interference prediction unit. The nonlinear channel fingerprint unit learns the correlation characteristics of fiber nonlinear effects and transmission power through a graph neural network, as shown in equation (1): （1）； In the formula, is the core node In the first Eigen vector of the layer, is the core Nonlinear coupling weight with the adjacent core , the nonlinear interference feature extraction layer is the weight matrix of the first Nonlinear interference feature extraction layer ReLU activation function, mapping linear transformation to nonlinear relationship, used to generate dynamic nonlinear interference fingerprint library The cross-layer interference prediction unit simulates cross-layer interference between quantum channels and classical channels based on a generative adversarial network, as shown in equation (2): （2）； In the formula, is a generator, simulating the interference scenario of quantum channels and classical channels, is a discriminator, judging whether the input data is a real measurement value or interference data synthesized by the generator , is a noise vector, input to the generator to generate an interference distribution for predicting the conflict probability of the reserved spectrum window.

4. The intelligent dynamically allocated bandwidth optical fiber communication system of claim 3, wherein, The resource optimization decision module further includes a multi-objective quantum annealing unit and a game theory collaborative allocation unit. The multi-objective quantum annealing unit solves the spectrum-space-power multi-objective optimization problem through a quantum annealing algorithm, as shown in equation (3): （3）； wherein is a core with a resource coupling constraint, is a core a spectral efficiency weight, is a spin operator of a qubit, encoding an allocation state of a core for generating a Pareto optimal bandwidth allocation scheme; The game theory collaborative allocation unit coordinates distributed competition of multi-core resources based on a Nash equilibrium game model, as shown in equation (4): （4）； In the formula, For fiber core allocated bandwidth For fiber core signal-to-noise ratio, For fiber core right The nonlinear interference coefficient, For fiber core The transmit power is used to balance nonlinear interference and spectral efficiency.

5. The intelligent dynamically allocated bandwidth optical fiber communication system of claim 4, wherein, The control execution module further includes a topological photon switch unit and a power-time slot coding unit. The topological photon switch unit realizes lossless and fast switching of space-spectrum modes through a topological boundary state optical switch, and is used for dynamically allocating core resources in a multi-core fiber. The power-time slot coding unit dynamically adjusts the transmission power and time slot mapping relationship based on a nonlinear pre-distortion encoder, and is used for suppressing nonlinear distortion under high power.

6. The intelligent dynamically allocated bandwidth optical fiber communication system of claim 5, wherein, The spectrum fragmentation integration module further includes a holographic diffraction reconstruction unit and a Bayesian optimization alignment unit. The holographic diffraction reconstruction unit coherently synthesizes fragmented spectrum through an optical holographic diffraction chip, as shown in equation (5): （5）； wherein is the input light field, is the transfer function of the holographic chip for spectral synthesis, is the Fourier transform converting the spatial light field into the frequency domain for recombining the non-continuous spectral blocks into a continuous bandwidth; The Bayesian optimization alignment unit dynamically adjusts the spectrum offset based on a Bayesian optimization algorithm, and is used for aligning fragmented spectrum positions among multiple cores.

7. The intelligent dynamically allocated bandwidth optical fiber communication system of claim 6, wherein, The dark spectrum perception module further includes a federated learning mapping unit and a compressed sensing detection unit. The federal learning mapping unit trains the dark spectrum distribution model across domains through a federal learning framework for predicting unavailable spectrum regions at the metropolitan backbone network interface; The compressed sensing detection unit sparsely samples and reconstructs the dark spectrum based on compressed sensing technology for low-overhead detection of blind area spectrum availability.

8. The intelligent dynamically allocated bandwidth optical fiber communication system of claim 7, wherein, The power-bandwidth joint adaptation module further includes a dynamic PSD game unit and an XPM compensation coding unit; The dynamic PSD game unit adjusts the transmission power and channel spacing through a dynamic power spectral density game algorithm, as shown in equation (6): （6）； wherein is the fiber core In the first iteration, the transmit power, is the target signal-to-noise ratio, is the noise power spectral density, used to balance non-linear interference and bandwidth utilization; The XPM compensation coding unit pre-compensates for cross-phase modulation interference based on a Turbo-type nonlinear equalizer for compressed channel protection interval.

9. The intelligent dynamically allocated bandwidth optical fiber communication system of claim 8, wherein, The real-time synchronization module further includes a photonic neural synchronization unit and a quantum clock calibration unit; The photonic neural synchronization unit achieves sub-microsecond time synchronization for multi-dimensional resource allocation through a photonic pulse neural network, as shown in equation (7): （7）； In the formula, is a synaptic weight of a photonic neural network, is a path transmission delay, is a time slot period for eliminating delay jitter; The quantum clock calibration unit calibrates the clock offset of distributed nodes based on a quantum entangled clock synchronization protocol for precise time slot alignment of hard-sliced bandwidth.

10. Application of the intelligent dynamic bandwidth allocation optical fiber communication system of any one of claims 1-9 in multi-core optical fiber communication at the metropolitan backbone network interface for dynamically allocating bandwidth.

Images