Optical fiber communication system capable of intelligently and dynamically distributing bandwidth

Abstract

The invention relates to the technical field of optical fiber communication, in particular to an optical fiber communication system for intelligently and dynamically distributing bandwidth, which runs based on a front-end framework and a rear-end framework. Comprising a multi-dimensional sensing module, a dynamic nonlinear modeling module, a resource optimization decision module, a control execution module, a spectrum fragment integration module, a dark spectrum sensing module, a power-bandwidth joint adaptation module and a real-time synchronization module, and the multi-dimensional sensing module is used for space division-spectrum coupling detection and quantum noise monitoring. According to the optical fiber communication system, resource dynamic optimization, optical straight-through control, cross-layer interference suppression and ultra-precision synchronization are systematically carried out through multi-module cooperation, hard slice time slot alignment is ensured, end-to-end time delay is shorter, the low-time-delay application requirements of various industrial internets and the like are met, and the application range is wide. The technical problem of the multi-core optical fiber in the optical communication field in the prior art is solved and optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of optical fiber communication, and specifically to an optical fiber communication system with intelligent dynamic bandwidth allocation. Background Art

[0002] A multi-core optical fiber is an optical fiber structure that integrates multiple independent cores within a single cladding. By means of spatial division multiplexing, the transmission capacity is enhanced. According to the core coupling strength, it can be divided into weakly coupled type, strongly coupled type, and heterogeneous multi-core optical fibers. Among them, the seven-core optical fiber, due to its hexagonal close-packed structure, shows outstanding performance in terms of capacity, space efficiency, and compatibility, and is widely used in the communication field. Its core advantages lie in its high space utilization rate and capacity expansion ability. Compared with single-core optical fibers, it can provide 7 times the theoretical capacity under the same cladding size while maintaining compatibility with existing optical fiber infrastructure. In addition, its symmetrically arranged core structure reduces the manufacturing complexity and effectively suppresses nonlinear crosstalk through intelligent power management, making it highly competitive in scenarios such as backbone networks and data center interconnections.

[0003] However, generally, in the actual application of the optical fiber communication field, the seven-core optical fiber still faces the problem of inter-core crosstalk management. The static resource allocation mode of existing systems is difficult to adapt to dynamic service requirements, resulting in serious spectrum fragmentation, and the Raman scattering effect is aggravated when quantum and classical channels are co-transmitted. These problems limit its application in low-latency scenarios such as industrial Internet.

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

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

[0006] To achieve the above purpose, the present invention provides the following technical solutions: An optical fiber communication system with intelligent dynamic bandwidth allocation, the system operates based on a front-end architecture and a 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; The multi-dimensional perception module is used for space-division-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 optical switch and power-time slot coding; The spectrum fragmentation integration module is used for holographic diffraction reconstruction and optimizing the alignment of the spectrum positions; The dark spectrum sensing module is used for predicting unavailable spectrum regions and compressive 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 photonic neural synchronization and quantum clock calibration.

[0007] Preferably, the multi-dimensional sensing module further includes a space-division-spectrum coupling detection unit and a quantum noise monitoring unit; The space-division-spectrum coupling detection unit detects the coupling state of the space-division mode and the spectrum through a photonic crystal fiber grating sensor, and is used for real-time capturing of crosstalk between cores and spectrum occupancy rate; 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.

[0008] Preferably, the dynamic non-linear modeling module further includes a non-linear channel fingerprint unit and a cross-layer interference prediction unit; The non-linear channel fingerprint unit learns the correlation characteristics between the fiber non-linear effect and the transmission power through a graph neural network, as shown in Equation (1): (1); where, is the feature vector of core node v at the l-th layer, is the non-linear coupling weight between core v and neighbor core u, is the weight matrix of the neural network at the l-th layer, which is used for extracting non-linear interference characteristics, is the ReLU activation function, which maps the linear transformation into a non-linear relationship and is used for generating a dynamic non-linear interference fingerprint library; The cross-layer interference prediction unit simulates the cross-layer interference between the quantum channel and the classical channel based on a generative adversarial network, as shown in Equation (2): (2); where, G is the generator, which simulates the interference scenario between the quantum channel and the classical channel, D is the discriminator, which judges whether the input data is the real measurement value (x) or the interference data (G(z)) synthesized by the generator, z is the noise vector, and is input into the generator to generate an interference distribution for predicting the conflict probability of the reserved spectrum window.

[0009] Preferably, 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-division-power multi-objective optimization problem through a quantum annealing algorithm, as shown in Equation (3): (3); where, is the resource coupling constraint between cores i and j, is the spectral efficiency weight of core i, , is the spin operator of the qubit, encoding the allocated state of core i, for generating the Pareto optimal bandwidth allocation scheme; 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); where, is the bandwidth allocated to core i, is the signal-to-noise ratio of core i, is the non-linear interference coefficient of core j on i, is the transmit power of core i, for balancing non-linear interference and spectral efficiency.

[0010] Preferably, the control execution module further includes a topological optical switch unit and a power-time slot encoding unit; The topological optical switch unit realizes lossless and fast switching of the space division mode through a topological boundary state optical switch, for dynamically allocating core resources in a multi-core optical fiber; The power-time slot encoding unit dynamically adjusts the transmit power and time slot mapping relationship based on a non-linear pre-distortion encoder, for suppressing non-linear distortion at high power.

[0011] Preferably, the spectral fragmentation integration module further includes a holographic diffraction reconstruction unit and a Bayesian optimization alignment unit; The holographic diffraction reconstruction unit performs coherent synthesis on the fragmented spectrum through an optical holographic diffraction chip, as shown in Equation (5): (5); where, is the input optical field, is the transfer function of the holographic chip for spectrum synthesis, is the Fourier transform, converting the spatial domain optical field to the frequency domain, for recombining discontinuous spectrum blocks into a continuous bandwidth; The Bayesian optimization alignment unit dynamically adjusts the spectrum offset based on the Bayesian optimization algorithm, for aligning the positions of the fragmented spectra between multi-cores.

[0012] Preferably, the dark spectrum sensing module further includes a federated learning mapping unit and a compressive sensing detection unit; The federated learning mapping unit trains the dark spectrum distribution model across domains through the federated learning framework, for predicting the unavailable spectrum regions at the junction of the metropolitan area network and the backbone network; The compressive sensing detection unit performs sparse sampling and reconstruction of the dark spectrum based on compressive sensing technology, for low-overhead detection of the spectrum availability in blind areas.

[0013] Preferably, 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 transmit power and channel spacing through a dynamic power spectral density game algorithm, as shown in Equation (6): (6); where is the transmit power of the fiber core i in the k-th iteration, is the target signal-to-noise ratio, is the noise power spectral density, which is used to balance the non-linear interference and bandwidth utilization; The XPM compensation coding unit pre-compensates the cross-phase modulation interference based on a Turbo-type non-linear equalizer, which is used to compress the channel protection interval.

[0014] Preferably, the real-time synchronization module further includes a photonic neural synchronization unit and a quantum clock calibration unit; The photonic neural synchronization unit realizes sub-microsecond-level 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 the delay jitter; The quantum clock calibration unit calibrates the clock offset of distributed nodes based on the quantum entanglement clock synchronization protocol, which is used for precise time slot alignment of the hard slice bandwidth.

[0015] Application of the fiber optic communication system with intelligent dynamic bandwidth allocation in multi-core fiber optic communication at the junction of metropolitan area network and backbone network for dynamic bandwidth allocation.

[0016] Compared with the prior art, the beneficial effects of the present invention are: The fiber optic communication system of the present invention performs resource dynamic optimization, optical direct control, cross-layer interference suppression and ultra-precise synchronization in a multi-module collaborative and systematic manner. Through non-linear fingerprint-driven quantum annealing optimization, the XPM interference is reduced and the spectrum utilization rate is improved. At the same time, federal learning dark spectrum prediction and Bayesian alignment are carried out, and cross-domain data is aggregated to generate a dark spectrum probability map. After the optimal offset of the spectrum offset between fiber cores is calculated by Bayesian optimization, the spectrum of the fiber core is avoided from the dark spectrum area. Then, the topological photonic switch and the entangled clock cooperate to perform faster traffic switching between fiber cores. The time base error is lower through the quantum entanglement clock protocol synchronization, ensuring precise hard slice time slot alignment and shorter end-to-end delay, meeting the application requirements of low delay such as various industrial internets, and solving and optimizing the technical problems existing in multi-core fibers in the field of optical communication in the prior art. Brief Description of the Drawings

[0017] Figure 1Topological diagram of the fiber optic communication system for intelligent dynamic bandwidth allocation according to the present invention; Figure 2 Application flow chart of the intelligent dynamic bandwidth allocation according to the present invention. Specific embodiments

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0019] Embodiment 1, please refer to Figure 1 , the present invention proposes a fiber optic communication system for intelligent dynamic bandwidth allocation. Further, the system operates based on a front-end architecture and a back-end architecture; Among them, 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 is used to collect the spectrum, spatial division mode, power, and non-linear interference data of the fiber optic network in real time. The edge computing layer is used to deploy lightweight algorithms to preprocess the data and reduce the computing load of the core network. Among them, 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-making layer and a control execution layer. The intelligent decision-making layer is used to generate a dynamic allocation strategy based on a multi-objective optimization model, and the control execution layer is used to realize the physical layer mapping of the strategy through photon devices and DSP chips; Specifically, the system includes a multi-dimensional perception module, a dynamic non-linear 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; The multi-dimensional perception module is used for spatial division-spectrum coupling detection and quantum noise monitoring. The dynamic non-linear modeling module is used to generate a dynamic non-linear interference fingerprint library and cross-layer interference prediction. The resource optimization decision module is used to generate a bandwidth allocation scheme and game theory collaborative allocation. The control execution module is used to provide a topological photon switch and power-time slot coding. The spectrum fragmentation integration module is used for holographic diffraction reconstruction and optimized alignment of spectrum positions. The dark spectrum perception module is used to predict unavailable spectrum regions and compressive 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. [[ID=2l]]

[0020] In this embodiment, it should also be noted that the multi-dimensional perception module further includes a spatial division-spectrum coupling detection unit and a quantum noise monitoring unit; Furthermore, the spatial-division and spectrum-coupling detection unit detects the coupling state of spatial-division modes and spectrum through a photonic crystal fiber grating sensor, and is used to capture the crosstalk between cores and the spectrum occupancy rate in real time; Furthermore, the quantum noise monitoring unit measures the quantum noise power spectrum density based on a squeezed-state optical interferometer, and is used to identify the noise interference threshold in an ultra-low power scenario.

[0021] In this embodiment, it should also be noted that the dynamic non-linear modeling module further includes a non-linear channel fingerprint unit and a cross-layer interference prediction unit; Furthermore, the non-linear channel fingerprint unit learns the correlation characteristics between the fiber non-linear effect and the transmission power through a graph neural network, as shown in Equation (1): (1); where is the feature vector of core node v at the l-th layer, is the non-linear coupling weight between core v and neighboring core u, is the weight matrix of the neural network at the l-th layer, which is used to extract non-linear interference characteristics, is the ReLU activation function, which maps the linear transformation into a non-linear relationship and is used to generate a dynamic non-linear interference fingerprint library; Among them, it should also be noted that the input feature definition: The feature vector of core node v includes the following physical parameters (dimension is 4): ; Among them, it should also be noted that: : The transmission power of core v (unit: dBm); : The spectrum occupancy rate of core v (unit: %); [[ID=�6]] : The average crosstalk coefficient between core v and its nearest neighbor core (unit: dB); : The real-time signal-to-noise ratio of core v (unit: dB); Among them, it should also be noted that the neighbor relationship is defined: The neighbor set of core v is defined as the physically adjacent cores; In a seven-core optical fiber, the neighbors of the central core 3 are cores 2, 4, 5, and 6 (hexagonal arrangement structure); Among them, it should also be noted that the calculation of the non-linear coupling weight : Based on the calibration experiment of the fiber non-linear effect (XPM / FWM), the weight formula is: ; Among them, it should also be noted that: : The spectral interval between the cores v and u (unit: GHz); : The fiber nonlinear coefficient (typical value: ); Among them, it should also be noted that for the GNN training and verification: Dataset construction: Collect the crosstalk data of multi-core optical fiber at 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; The loss function is: ; : The L2 regularization coefficient; Output the dynamic nonlinear interference fingerprint library, the XPM interference intensity between cores 3 and 5 ; Furthermore, the cross-layer interference prediction unit simulates the cross-layer interference between the quantum channel and the classical channel based on the generative adversarial network, as shown in Equation (2): (2); In the formula, G is the generator, simulating the interference scenario between the quantum channel and the classical channel, D is the discriminator, judging whether the input data is the real measurement value (x) or the interference data synthesized by the generator (G(z)), z is the noise vector, input into the generator to generate the interference distribution for predicting the conflict probability of the reserved spectrum window; Among them, it should also be noted that for the acquisition of the real data x: Experimental setup: Core 3 transmits classical services (at 1545 nm, 10 Gbps), and core 5 transmits quantum channels (at 1550 nm, single-photon level signal); Among them, it should also be noted that for the measurement parameters: Raman scattering power : Measure the noise power (unit: dBm) at the quantum channel wavelength using an optical spectrum analyzer (OSA); Quantum bit error rate : The change in the quantum bit error rate statistically measured at the quantum receiver (unit: %); Classical channel spectral shift : Measure the central frequency shift (unit: GHz) using a coherent receiver; Dataset: x = [{P}_{\mathrm{R}\mathrm{a}\mathrm{m}\mathrm{a}\mathrm{n}},\Delta \mathrm{B}\mathrm{E}\mathrm{R},\Delta f] , with a total of 5,000 groups of samples; Among them, it should also be noted that the definition of the noise vector z: The dimension is 4, and the distribution is as follows: \mathrm{z}=[\Delta P,\Delta \lambda,{N}_{\mathrm{p}\mathrm{h}\mathrm{o}\mathrm{t}\mathrm{o}\mathrm{n}},\Delta T] ; : Fluctuation of the emission power; : Random offset of the spectral interval; : Number of photons of the quantum signal; : Environmental temperature perturbation; Among them, it should also be noted that the design of the generator G and the discriminator D: Generator G: Input the noise vector z (4 - dimensional); Output the synthetic interference data ; Network structure: 3 - layer fully - connected neural network (the dimension of the hidden layer is 64, and the activation function is ReLU); Discriminator D: Input the real data x or the generated data G(z) (3 - dimensional); Output the discrimination probability D(\mathrm{x})\in [0,1] ; Network structure: 3 - layer fully - connected neural network (the dimension of the hidden layer is 64, and the activation function is LeakyReLU); Among them, it should also be noted that the conflict probability mapping: Map the discriminator output D(x) to the conflict probability: ; The slope factor k = 10: Calibrated through experiments (set when the sensitivity of the bit error rate change is the highest).

[0022] In this embodiment, it should also be noted that the resource optimization decision - making module further includes a multi - objective quantum annealing unit and a game - theory collaborative allocation unit; Furthermore, the multi-objective quantum annealing unit solves the spectrum-space-division-power multi-objective optimization problem through the quantum annealing algorithm, as shown in Equation (3): (3); where is the resource coupling constraint between cores i and j, is the spectrum efficiency weight of core i, , is the spin operator of the qubit, encoding the allocation state of core i, and is used to generate the Pareto optimal bandwidth allocation scheme; Furthermore, 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); where is the bandwidth allocated to core i, is the signal-to-noise ratio of core i, is the non-linear interference coefficient of core j on i, is the transmit power of core i, which is used to balance non-linear interference and spectrum efficiency.

[0023] In this embodiment, it should also be noted that the control execution module further includes a topological optical switch unit and a power-time slot coding unit; Furthermore, the topological optical switch unit realizes the lossless and fast switching of the space-division mode through the topological boundary state optical switch, and is used for dynamically allocating the core resources in the multi-core optical fiber; Furthermore, the power-time slot coding unit dynamically adjusts the relationship between the transmit power and the time slot mapping based on the non-linear pre-distortion encoder, and is used to suppress the non-linear distortion at high power.

[0024] In this embodiment, it should also be noted that the spectrum fragmentation integration module further includes a holographic diffraction reconstruction unit and a Bayesian optimization alignment unit; Furthermore, the holographic diffraction reconstruction unit performs coherent synthesis on the fragmented spectrum through the optical holographic diffraction chip, as shown in Equation (5): (5); where is the input optical field, is the transfer function of the holographic chip for spectrum synthesis, is the Fourier transform, which converts the spatial domain optical field into the frequency domain and is used to recombine the discontinuous spectrum blocks into a continuous bandwidth; Furthermore, the Bayesian optimization alignment unit dynamically adjusts the spectrum offset based on the Bayesian optimization algorithm, and is used to align the positions of the fragmented spectra between multi-cores.

[0025] In this embodiment, it should also be noted that the dark spectrum sensing module further includes a federated learning mapping unit and a compressive sensing detection unit; Further, the federated learning mapping unit trains the dark spectrum distribution model across domains through the federated learning framework, which is used to predict the unavailable spectrum area at the junction of the metropolitan area network and the backbone network; Further, the compressive sensing detection unit performs sparse sampling and reconstruction on the dark spectrum based on compressive sensing technology, which is used for low-overhead detection of the spectrum availability in the blind area.

[0026] In this embodiment, it should also be noted that the power-bandwidth joint adaptation module further includes a dynamic PSD game unit and an XPM compensation coding unit; Further, the dynamic PSD game unit adjusts the transmit power and channel spacing through the dynamic power spectral density game algorithm, as shown in Equation (6): (6); where, is the transmit power of the fiber core i in the k-th iteration, is the target signal-to-noise ratio, is the noise power spectral density, which is used to balance the nonlinear interference and the bandwidth utilization rate; Further, the XPM compensation coding unit performs pre-compensation on the cross-phase modulation interference based on the Turbo-type nonlinear equalizer, which is used to compress the channel protection interval.

[0027] In this embodiment, it should also be noted that the real-time synchronization module further includes a photonic neural synchronization unit and a quantum clock calibration unit; Further, the photonic neural synchronization unit realizes sub-microsecond time synchronization for multi-dimensional resource allocation through the 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 the delay jitter; Further, the quantum clock calibration unit calibrates the clock offset of the distributed nodes based on the quantum entanglement clock synchronization protocol, which is used for accurate time slot alignment of the hard slice bandwidth.

[0028] Embodiment 2, please refer to Figure 2 , in practical applications, the fiber optic communication system with intelligent dynamic bandwidth allocation of the present invention is applied to the dynamic bandwidth allocation of multi-core fiber optic communication at the junction of the metropolitan area network and the backbone network. Specifically, it includes the following steps: (1) Multi-dimensional data acquisition and preprocessing: The spectral occupancy rate and mode coupling coefficient of cores 1 to 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 optical interferometer. Sparse sampling is performed on the dark spectrum at the junction of the metropolitan area network backbone network. For core 3 in the multi-core optical fiber (7 cores), a hard slice bandwidth with a 1 ms time delay requirement is allocated for the industrial Internet service. Core 5 carries ordinary Internet traffic. The steps of space-division - spectrum coupling detection, quantum noise monitoring, and compressive sensing detection are as follows: (1.1) Space-division - spectrum coupling detection: The photonic crystal fiber grating sensor collects the real-time spectral occupancy rate of core 3 and the crosstalk data with 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. The spectrum - space-division coupling matrix: {\mathrm{C}}_{3,5}=[0.92,-15] ; (1.2) Quantum noise monitoring: The squeezed state optical interferometer measures the noise floor of core 3 under single-photon level signals, the quantum noise power spectral density , the quantum noise threshold , triggering the low-power optimization mode; (1.3) Compressive sensing detection: A 10% sampling rate is performed at the junction of the metropolitan area network and the backbone network. Dark spectrum is detected at 1570 - 1572 nm in core 4, and the bit error rate , the dark spectrum marking vector D = [0,0,0,1,0,0,0], then core 4 is marked as the dark spectrum; (2) Nonlinear modeling and cross-layer interference prediction: (2.1) Nonlinear channel fingerprint: The transmission power of core 3 , the spectral interval of core 5 ; Execute the following algorithm: ; The predicted value of the XPM interference intensity (normalized); (2.2) Cross-layer interference prediction: The power distribution of the 1550 nm quantum channel and the 1545 nm classical channel of core 3. The generative adversarial network simulates the change in the quantum bit error rate caused by Raman scattering, and the quantum - classical channel conflict probability ; (3) Multi-objective optimization dynamic decision-making: (3.1) Multi-objective quantum annealing: Spectrum requirements of core 3 (1540~1550nm), , the dark spectrum marker D, construct the Hamiltonian: ; The Pareto optimal solution allocates 1540~1545nm to core 3, avoiding 1546~1550nm of core 5, power , guard interval ; (3.2) Game theory for collaborative allocation: The current power of core 5 , spectrum occupancy rate, 85%, perform Nash equilibrium solution: ; Therefore, the equilibrium strategy is: core 5 shifts the spectrum to the right to 1546.5~1551.5nm, and the power drops to 8.8dBm; (4) Physical layer dynamic execution feedback: (4.1) Holographic diffraction reconstruction: The fragmented spectrum of core 3 (1540~1541.5nm, 1542~1545nm), perform optical diffraction field reconstruction: ; Reorganized into a continuous spectrum of 1540~1545nm, insertion loss <0.2dB; (4.2) Dynamic PSD game: The initial PSD of core 3 ; The PSD of core 5 after adjustment , the power iteration formula is: ; Converge to , the guard interval is compressed to 1.5GHz; (4.3) Photonic neural synchronization: The hard slice time slot requirements of core 3, (1ms period, 50ns jitter), the algorithm expression of optical pulse synchronization is: ; The time delay jitter drops to 5ns, the time slot alignment error <0.1%; (5) Dark spectrum avoidance and cross-domain collaboration: (5.1) Federated learning mapping: The metropolitan area network reports the dark spectrum interval of core 4, 1570~1572nm, and the federated aggregation updates the global dark spectrum probability model , and core 4 is prohibited from allocating new services in 1570~1572nm; (5.2) Bayesian optimization alignment: The available spectrum of core 4 is 1565~1570nm, 1572~1575nm, and the Bayesian optimization objective function is: ; Optimal offset Shift the spectrum of core 4 to the right to 1566.2 - 1571.2 nm and 1573.2 - 1576.2 nm; Through the above steps, mainly through the optimization of quantum annealing driven by non-linear fingerprints, the XPM interference is reduced to 0.08, and the spectrum utilization rate is increased by 15%. At the same time, federated learning dark spectrum prediction and Bayesian alignment are also carried out to aggregate cross-domain data and generate a dark spectrum probability map. After the optimal offset of the spectrum offset between cores is calculated by Bayesian optimization, the spectrum of the core is avoided from the dark spectrum area, and then the topological photonic switch and the entangled clock cooperate to switch the traffic between cores within 1 and synchronize the time base error to 0.1 ns through the quantum entanglement clock protocol to ensure the alignment of hard slice time slots, and the end-to-end delay is stable at 1.05 , meeting the requirements of various industrial Internet and solving and optimizing the technical problems of multi-core optical fibers in the field of optical communication in the prior art.

[0029] In the description of this specification, the descriptions referring to terms such as "one embodiment", "example", "specific example", etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0030] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the art in the relevant technical field can understand and utilize the present invention well. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. An optical fiber communication system with intelligent dynamic bandwidth allocation, characterized in that, It includes a multi-dimensional perception module, a dynamic non-linear modeling module, a resource optimization decision-making 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; The multi-dimensional perception module is used for space-division-spectrum coupling detection and quantum noise monitoring; The dynamic non-linear modeling module is used for generating a dynamic non-linear interference fingerprint library and cross-layer interference prediction; The resource optimization decision-making module is used for generating a bandwidth allocation scheme and game theory collaborative allocation; The control execution module is used for providing a topological photonic switch and power-time slot coding; The spectrum fragmentation integration module is used for holographic diffraction reconstruction and optimizing the alignment of spectrum positions; The dark spectrum perception module is used for predicting unavailable spectrum regions and compressive 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 photonic neural synchronization and quantum clock calibration.

2. The fiber optic communication system for intelligent dynamic bandwidth allocation according to claim 1, characterized in that, The multi-dimensional perception module further includes a space-division-spectrum coupling detection unit and a quantum noise monitoring unit; The space-division-spectrum coupling detection unit detects the coupling state of the space-division mode and the spectrum through a photonic crystal fiber grating sensor, and is used for real-time capturing of crosstalk between cores and spectrum occupancy rate; 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.

3. The optical fiber communication system with intelligent dynamic bandwidth allocation according to claim 2, characterized in that The dynamic non-linear modeling module further includes a non-linear channel fingerprint unit and a cross-layer interference prediction unit; The non - linear channel fingerprint unit learns the correlation features between the fiber non - linear effect and the transmit power through a graph neural network, as shown in Equation (1): (1); where is the feature vector of the core node v at the l - th layer, is the non - linear coupling weight between the core v and its neighbor core u, is the weight matrix of the l - th layer neural network, used to extract non - linear interference features, is the ReLU activation function, which maps the linear transformation into a non - linear relationship and is used to generate a dynamic non - linear interference fingerprint library; The cross-layer interference prediction unit simulates the cross-layer interference between the quantum channel and the classical channel based on a generative adversarial network, as shown in Equation (2): (2); where G is a generator that simulates the interference scenario of the quantum channel and the classical channel, D is a discriminator that determines whether the input data is a true measurement value (x) or interference data synthesized by the generator (G(z)), z is a noise vector, and the input generator is used to generate an interference distribution for predicting the conflict probability of the reserved spectrum window.

4. The fiber optic communication system for intelligent dynamic bandwidth allocation according to claim 3, wherein The resource optimization decision-making 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-division-power multi-objective optimization problem through the quantum annealing algorithm, as shown in Equation (3): (3); In the formula, is the resource coupling constraint between cores i and j, is the spectrum efficiency weight of core i, , is the spin operator of the qubit, encoding the allocation state of core i, and is used to generate the Pareto optimal bandwidth allocation scheme; 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); where The bandwidth allocated to core i The signal-to-noise ratio of core i The non-linear interference coefficient of core j on i The transmission power of core i, used to balance non-linear interference and spectral efficiency.

5. The fiber optic communication system for intelligent dynamic bandwidth allocation according to claim 4, wherein The control execution module further includes a topological photonic switch unit and a power-time slot coding unit; The topological photonic switch unit realizes lossless and fast switching of the space-division mode through a topological edge state optical switch, and is used for dynamically allocating core resources in a multi-core optical fiber; The power-time slot coding unit dynamically adjusts the relationship between the transmission power and the time slot mapping based on a non-linear pre-distortion encoder, and is used for suppressing non-linear distortion at high power.

6. The fiber optic communication system for intelligent dynamic bandwidth allocation according to 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 the fragmented spectrum through an optical holographic diffraction chip, as shown in Equation (5): (5); where is the input optical field, is the transfer function of the holographic chip for spectrum synthesis, is the Fourier transform, which converts the spatial-domain optical field into the frequency domain and is used to recombine discontinuous spectrum blocks into a continuous bandwidth; The Bayesian optimization alignment unit dynamically adjusts the spectrum offset based on the Bayesian optimization algorithm, and is used for aligning the fragmented spectrum positions between multi-cores.

7. The optical fiber communication system for intelligently and dynamically allocating bandwidth according to claim 6, characterized in that, The dark spectrum perception module further includes a federated learning mapping unit and a compressive sensing detection unit; The federated learning mapping unit cross-domain trains a dark spectrum distribution model through a federated learning framework, and is used for predicting unavailable spectrum regions at the junction of the metropolitan area network and the backbone network; The compressive sensing detection unit sparsely samples and reconstructs the dark spectrum based on compressive sensing technology, and is used for low-overhead detection of the availability of the blind area spectrum.

8. The optical fiber communication system for intelligent dynamic bandwidth allocation according to 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 interval through the dynamic power spectral density game algorithm, as shown in Equation (6): (6); where is the transmission power of the fiber core i in the k-th iteration, is the target signal-to-noise ratio, is the noise power spectral density, which is used to balance the non-linear interference and bandwidth utilization; The XPM compensation coding unit pre-compensates the cross-phase modulation interference based on a Turbo-type non-linear equalizer, and is used for compressing the channel protection interval.

9. The fiber optic communication system for intelligent dynamic bandwidth allocation according to claim 8, characterized in that The real-time synchronization module further includes a photon neural synchronization unit and a quantum clock calibration unit; The photon neural synchronization unit achieves sub-microsecond time synchronization for multi-dimensional resource allocation through a photon pulse neural network, as shown in Equation (7): (7); In the formula, is the synaptic weight of the photon neural network, is the path transmission delay, and T is the time slot period, which is used to eliminate delay jitter; The quantum clock calibration unit calibrates the clock offset of distributed nodes based on the quantum entanglement clock synchronization protocol, and is used for accurate time slot alignment of hard slice bandwidth.

10. Application of the optical fiber communication system for intelligent dynamic bandwidth allocation in multi-core optical fiber communication at the junction of a metropolitan area network and a backbone network for dynamically allocating bandwidth.

Images