A multi-band dynamic optical network transmission optimization method and system

Abstract

The application provides a multi-band dynamic optical network transmission optimization method and system, the steps of the method comprising: determining whether a fiber node is a compensation point based on the power spectrum of the fiber node in an unoptimized optical network; calculating a target gain curve corresponding to the compensation point based on the power spectrum corresponding to the compensation point; inputting the target gain curve into a pre-trained machine learning model, constructing an initial pump power vector based on the output value of the machine learning model; determining the final applied pump power by iteratively calculating the initial pump power vector through multiple iteration rounds, in each iteration round, calculating a circulating pump power vector based on the current iteration pump power vector, calculating an environmental difference based on the circulating pump power vector and the initial pump power vector, calculating a new iteration pump power vector of the next round based on the environmental difference, and determining whether the new iteration pump power vector of the next round corresponds to the final applied pump power.

Description

Technical Field

[0001] This invention relates to the field of optical amplifier technology, and in particular to a method and system for optimizing multi-band dynamic optical network transmission. Background Technology

[0002] With the rapid development of new internet technologies and the explosive growth of terminals and applications, traffic has been growing exponentially. As the underlying network for transmission, optical networks urgently need new expansion methods to cope with the massive increase in traffic. Multiband transmission is considered a promising alternative for recent optical network upgrades. This approach utilizes existing spectrum beyond the traditional C-band, including the O, E, S, and L bands. Given that widely deployed ITU-T G.652D optical fibers exhibit low attenuation in these bands, the need to deploy additional optical fibers can be effectively avoided in multiband systems. Furthermore, the various devices required in these bands are relatively mature. Currently, multiband wavelength division multiplexing (WDM) systems have proven to be an effective method for achieving greater optical network capacity.

[0003] However, expanding the bands leads to more significant nonlinear effects, especially stimulated Raman scattering between channels, which causes non-uniform effects on signals in different bands. This directly results in complex changes in the power spectrum of the transmitted signal, specifically the transfer of channel power from high-frequency channels to low-frequency channels. This phenomenon causes the signal quality of high-frequency channels to be significantly lower than that of low-frequency channels, and in severe cases, it can lead to the inability to transmit signals at high frequencies. This problem reduces the effectiveness of multi-band expansion. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a method and system for optimizing multi-band dynamic optical network transmission, in order to eliminate or improve one or more defects existing in the prior art.

[0005] One aspect of the present invention provides a method for optimizing transmission in a multi-band dynamic optical network, the method comprising the following steps:

[0006] The power spectrum of each fiber node in the unoptimized optical network is obtained. The power spectrum records the power values ​​of the fiber node's location at different frequencies. Based on the maximum and minimum power values ​​in the power spectrum, it is determined whether the location of the fiber node is a compensation point.

[0007] The target gain curve corresponding to the compensation point is calculated based on the power spectrum corresponding to the compensation point. The target gain curve includes multiple gain values ​​corresponding to different frequencies.

[0008] The target gain curve is input into a pre-trained machine learning model, and an initial pump power vector is constructed based on the output value of the machine learning model.

[0009] The initial pump power vector is iteratively calculated through multiple iterations to determine the final applied pump power. In each iteration, the cyclic pump power vector is calculated based on the current iterative pump power vector. The environmental difference is calculated based on the cyclic pump power vector and the initial pump power vector. The iterative pump power vector for the new iteration is calculated based on the environmental difference, and it is determined whether the iterative pump power vector for the new iteration corresponds to the final applied pump power.

[0010] In the above scheme, due to the transfer of channel power from the high-frequency channel to the low-frequency channel in the actual application of Raman amplifiers, the signal quality of the high-frequency channel is significantly lower than that of the low-frequency channel, and in severe cases, the high-frequency signal cannot be transmitted. This problem reduces the effectiveness of multi-band expansion. In the step of confirming the compensation point, this scheme determines the compensation point based on the maximum and minimum power values ​​in the power spectrum. The compensation point ensures that the compensation point is set at the position of the maximum and minimum power values ​​in the power spectrum, thus ensuring the effectiveness of multi-band expansion. Furthermore, this scheme calculates the environmental difference based on the cyclic pump power vector and the initial pump power vector in each round through multiple iterations, so that the cyclic pump power vector is close to the initial pump power vector corresponding to the target gain curve, ensuring that the final pump power effect matches the expected target.

[0011] In some embodiments of the present invention, in the step of calculating the cyclic pump power vector based on the current iterative pump power vector, the corresponding actual gain value is determined based on the current iterative pump power vector, and the cyclic pump power vector is determined based on the actual gain value.

[0012] In some embodiments of the present invention, the step of calculating the iterative pump power vector for a new round based on the environmental difference includes:

[0013] The tuning parameters are calculated based on the initial pump power vector and the current iterative pump power vector;

[0014] Calculate the new iteration pump power vector based on the current iterative pump power vector, tuning parameters, and environmental differences.

[0015] In some embodiments of the present invention, in the step of calculating tuning parameters based on the initial pump power vector and the current iterative pump power vector, the quotient of the initial pump power vector and the current iterative pump power vector is calculated as the tuning parameters.

[0016] In some embodiments of the present invention, in the step of calculating the iterative pump power vector for a new round based on the current iterative pump power vector, tuning parameters, and environmental differences, the iterative pump power vector for the new round is calculated based on the following formula:

[0017] P_new = P_last - λ * P_dif;

[0018] Where P_new represents the iterative pump power vector of the new round, P_last represents the current iterative pump power vector, P_dif represents the environmental difference, and λ represents the tuning parameters.

[0019] In some embodiments of the present invention, in the step of determining whether the iterative pump power vector of a new round corresponds to the final applied pump power, the iterative pump power vector of the new round is compared with the target gain curve to determine whether the iterative pump power vector of the new round corresponds to the final applied pump power.

[0020] In some embodiments of the present invention, in the step of comparing the iterative pump power vector of a new round with the target gain curve to determine whether the iterative pump power vector of the new round corresponds to the final applied pump power, the root mean square error and the maximum absolute error are calculated based on the gain value corresponding to each power point in the iterative pump power vector of the new round and the gain value corresponding to each power point in the target gain curve. The root mean square error and the maximum absolute error are compared with the corresponding root mean square error threshold and the maximum absolute error threshold, respectively, to determine whether the iterative pump power vector of the new round corresponds to the final applied pump power.

[0021] In some embodiments of the present invention, in the step of comparing the root mean square error and the maximum absolute error with the corresponding root mean square error threshold and maximum absolute error threshold, respectively, to determine whether the iterative pump power vector of the new round corresponds to the final applied pump power, if the root mean square error is less than the root mean square error threshold and the maximum absolute error is less than the maximum absolute error threshold, then the pump power of each dimension in the iterative pump power vector of the new round is determined to be the final applied pump power.

[0022] In some embodiments of the present invention, in the step of determining whether the position corresponding to the fiber optic node is a compensation point based on the maximum and minimum power values ​​in the power spectrum, the difference between the maximum and minimum power values ​​in the power spectrum is calculated. If the difference between the maximum and minimum power values ​​in the power spectrum is greater than a preset difference threshold, then the position corresponding to the fiber optic node is determined to be a compensation point.

[0023] A second aspect of the present invention also provides a multi-band dynamic optical network transmission optimization system, the system comprising a computer device including a processor and a memory, the memory storing computer instructions, the processor executing the computer instructions stored in the memory, and the system implementing the steps of the method described above when the computer instructions are executed by the processor.

[0024] A third aspect of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the aforementioned multi-band dynamic optical network transmission optimization method.

[0025] Additional advantages, objects, and features of the invention will be set forth in part in the description which follows, and will also become apparent in part to those skilled in the art upon studying the text, or may be learned by practice of the invention. The objects and other advantages of the invention will become apparent from the description and the accompanying drawings.

[0026] Those skilled in the art will understand that the objectives and advantages achievable with the present invention are not limited to those specifically described above, and that the above and other objectives achievable with the present invention will become clearer from the following detailed description. Attached Figure Description

[0027] The accompanying drawings, which are provided to further illustrate the invention and form part of this application, are not intended to limit the scope of the invention.

[0028] Figure 1 This is a schematic diagram of one implementation method of the multi-band dynamic optical network transmission optimization method of this scheme;

[0029] Figure 2 This is a schematic diagram of another implementation of the multi-band dynamic optical network transmission optimization method of this scheme;

[0030] Figure 3 This is a flowchart illustrating the overall workflow of this solution.

[0031] Figure 4 This is a flowchart illustrating the single-compensation-point pump configuration and iterative optimization process of this scheme.

[0032] Figure 5 This diagram illustrates the effect of optimizing signal transmission quality in this solution. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0034] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.

[0035] Existing technologies, while extending bands, lead to more significant nonlinear effects, particularly stimulated Raman scattering (SRS) between channels. This SRS causes non-uniform effects on signals across different bands, directly resulting in complex changes in the power spectrum of transmitted signals. Specifically, channel power shifts from high-frequency channels to low-frequency channels. This phenomenon leads to significantly lower signal quality in high-frequency channels compared to low-frequency channels, and in severe cases, it can prevent signal transmission at high frequencies. This problem reduces the effectiveness of multi-band capacity expansion. Furthermore, when networks are highly dynamic, the co-transmission of signals with varying degrees of damage in the channels poses greater challenges to repeater equipment. Specifically, the changes in the signal power spectrum become more complex, requiring repeater equipment with more flexible characteristics to cope with different scenarios.

[0036] Existing solutions primarily employ optical switches and optical attenuators for power equalization. This involves attenuating the power of all signals to their minimum power level, flattening the power spectrum, and then re-amplifying them to ensure normal signal transmission. This process is repeated during subsequent transmissions to guarantee signal reception at the transmission endpoint. However, this approach leads to a significant degradation in signal transmission quality, thereby shortening the transmission distance and making it highly unfavorable for long-distance signal transmission.

[0037] Raman amplifiers, with their wide spectral range amplification capability, potential for generating arbitrary gain spectra, and low noise figure, are considered excellent repeater amplification devices for multi-band capacity expansion schemes. The introduction of machine learning methods has further enabled rapid tuning of Raman amplifiers; by constructing an inverse mapping model from collected data, the corresponding pump parameters can be predicted based on the desired target parameters. For the problems in multi-band dynamic optical networks, designing Raman amplifier gain spectra according to specific situations and using multi-stage Raman amplifiers to handle ultra-long-distance transmission scenarios is a promising solution.

[0038] However, current research on machine learning-based Raman amplifier gain spectrum generation schemes relies on specific scenarios, and performance evaluations only focus on the same or similar scenarios. Due to the sensitivity of Raman amplifiers to changes in application scenarios—meaning that changes directly lead to a decrease in gain spectrum generation accuracy, specifically a significant difference between the actual and expected gain spectrum—the introduction of network dynamics further increases the uncertainty of Raman amplifier deployment scenarios. Using existing machine learning-based Raman amplifiers alone is insufficient for transmission optimization in multi-band dynamic optical networks. Addressing this issue by re-collecting data for new model training in new scenarios or constructing a general model across multiple scenarios is unreasonable, costly, and inefficient, and cannot guarantee efficient signal transmission optimization in multi-band dynamic optical networks.

[0039] In response to the above situation, there is an urgent need for a transmission optimization method based on Raman amplifiers that can be deployed in multi-band dynamic optical networks. This method should be able to ensure the accurate generation of Raman amplifier gain spectrum in various scenarios, thereby avoiding the degradation of optimization effect due to Raman amplifier errors. Moreover, compared with existing methods, this method should be able to effectively improve signal transmission quality and ensure effective optimization of transmission in multi-band dynamic optical networks.

[0040] like Figure 1 and 3 As shown, this invention proposes a multi-band dynamic optical network transmission optimization method, the steps of which include:

[0041] Step S100: Obtain the power spectrum of each fiber node in the unoptimized optical network. The power spectrum records the power values ​​of the fiber node's location at different frequencies. Determine whether the location of the fiber node is a compensation point based on the maximum and minimum power values ​​in the power spectrum.

[0042] In practice, an optical network is composed of multiple optical fibers, and the connection points between the optical fibers are called optical fiber nodes. The power spectrum of an optical fiber node is measured by a power measurement device.

[0043] In the specific implementation process, in the step of determining whether the location corresponding to the fiber node is a compensation point based on the maximum and minimum power values ​​in the power spectrum, the difference between the maximum and minimum power values ​​in the power spectrum is calculated, and the location corresponding to the fiber node is determined as a compensation point based on the difference between the maximum and minimum power values ​​in the power spectrum.

[0044] In the actual implementation process, a Raman amplifier needs to be set at the compensation point.

[0045] Step S200: Calculate the target gain curve corresponding to the compensation point based on the power spectrum corresponding to the compensation point. The target gain curve includes multiple gain values ​​corresponding to different frequencies.

[0046] Step S300: Input the target gain curve into a pre-trained machine learning model, and construct an initial pump power vector based on the output value of the machine learning model;

[0047] In some embodiments of the present invention, the machine learning model is a feedforward neural network model. During the training of the machine learning model, the power curves of the Raman amplifier pump output power are collected when the pump is not turned on and when the pump is turned on, respectively, by adjusting the pump output power in the existing scenario. The switching gain curve corresponding to the current power is calculated to construct the original dataset, and the machine learning model used in the optimization scheme is trained based on the original dataset.

[0048] To build the machine learning model, the original dataset was collected. The specific data collection scenario is as follows: First, in the light source section, a broadband noise light source was used to generate multi-band optical signals. This light source can achieve high flatness and high power stability, ensuring long-term normal operation. An adjustable optical attenuator was connected after the ASE light source for adjusting the optical signal power. In the transmission medium section, standard single-mode fiber was used for data collection. In the Raman amplifier deployment section, the Raman amplifier used had multiple pump wavelengths, including multiple pump input directions that were forward or backward. It is worth noting that when using backward pumping, an optical isolator needs to be deployed at the fiber input port to avoid damage to the light source equipment. In the optical signal receiving section, the signal was output to the spectrum analyzer after passing through the optical fiber, and the optical power spectrum within the signal wavelength range was measured.

[0049] In this embodiment, the method for collecting and processing data is as follows: First, the no-gain spectral line is recorded when the Raman pump is turned off, and this is set as the original state power spectrum. Then, the Raman pump is turned on, the output power of each pump source is adjusted, and the signal power spectral lines corresponding to different input pump powers are recorded, denoted as the on-gain spectral lines. The switching gain of the Raman amplifier is obtained by subtracting the on-gain spectral line from the original state power spectrum. After mapping the switching gain to the output power, several sets of original two-dimensional datasets are obtained for training and validating the machine learning model. The machine learning model used here is a feedforward neural network, where the switching gain is the model input, the model input dimension is consistent with the switching gain dimension, the pump power is the model output, the model output dimension is consistent with the power dimension, and the learning objective is to minimize the prediction error of the pump configuration parameters. The specific error evaluation parameter is the root mean square error. and maximum absolute error And the model parameters are optimized based on gradient descent.

[0050] like Figure 4 As shown, in step S400, the initial pump power vector is iteratively calculated through multiple iterations to determine the final applied pump power. In each iteration, the cyclic pump power vector is calculated based on the current iterative pump power vector. The environmental difference is calculated based on the cyclic pump power vector and the initial pump power vector. The iterative pump power vector for the new iteration is calculated based on the environmental difference, and it is determined whether the iterative pump power vector for the new iteration corresponds to the final applied pump power.

[0051] In the above scheme, due to the transfer of channel power from the high-frequency channel to the low-frequency channel in the actual application of Raman amplifiers, the signal quality of the high-frequency channel is significantly lower than that of the low-frequency channel, and in severe cases, the high-frequency signal cannot be transmitted. This problem reduces the effectiveness of multi-band expansion. In the step of confirming the compensation point, this scheme determines the compensation point based on the maximum and minimum power values ​​in the power spectrum. The compensation point ensures that the compensation point is set at the position of the maximum and minimum power values ​​in the power spectrum, thus ensuring the effectiveness of multi-band expansion. Furthermore, this scheme calculates the environmental difference based on the cyclic pump power vector and the initial pump power vector in each round through multiple iterations, so that the cyclic pump power vector is close to the initial pump power vector corresponding to the target gain curve, ensuring that the final pump power effect matches the expected target.

[0052] like Figure 2 As shown, in some embodiments of the present invention, the step of calculating the cyclic pump power vector based on the current iterative pump power vector includes step S410, which involves determining the corresponding actual gain value based on the current iterative pump power vector, and determining the cyclic pump power vector based on the actual gain value.

[0053] In the specific implementation process, in the step of determining the corresponding actual gain value based on the current iterative pump power vector, the pump power of the Raman amplifier is adjusted to each power value in the iterative pump power vector, and the corresponding gain value is collected as the actual gain value corresponding to the iterative pump power vector.

[0054] In some embodiments of the present invention, the step of calculating the environmental difference based on the circulating pump power vector and the initial pump power vector includes step S420, which calculates the difference between the values ​​of each dimension of the circulating pump power vector and the initial pump power vector to obtain the environmental difference.

[0055] In the specific implementation process, in the step of calculating the difference between the values ​​of each dimension of the cyclic pump power vector and the initial pump power vector to obtain the environmental difference, the difference between the values ​​of each dimension of the cyclic pump power vector and the initial pump power vector is calculated as the corresponding vector of the environmental difference.

[0056] In some embodiments of the present invention, the step of calculating the iterative pump power vector for a new round based on the environmental difference includes:

[0057] Step S430: Calculate the tuning parameters based on the initial pump power vector and the current iterative pump power vector;

[0058] Step S440: Calculate the new round of iterative pump power vector based on the current iterative pump power vector, tuning parameters, and environmental differences.

[0059] In some embodiments of the present invention, in the step of calculating tuning parameters based on the initial pump power vector and the current iterative pump power vector, the quotient of the initial pump power vector and the current iterative pump power vector is calculated as the tuning parameters.

[0060] In some embodiments of the present invention, in the step of calculating the iterative pump power vector for a new round based on the current iterative pump power vector, tuning parameters, and environmental differences, the iterative pump power vector for the new round is calculated based on the following formula:

[0061] P_new = P_last - λ * P_dif;

[0062] Where P_new represents the iterative pump power vector of the new round, P_last represents the current iterative pump power vector, P_dif represents the environmental difference, and λ represents the tuning parameters.

[0063] The above scheme works by fine-tuning the pump configuration parameters of the previous iteration based on the differences in the scenario during each iteration. This process can gradually approach the optimal pump configuration, thereby making the actual gain curve gradually approach the target gain curve.

[0064] In the specific implementation process, in the step of calculating the target gain curve corresponding to the compensation point based on the power spectrum corresponding to the compensation point, the gain value corresponding to each frequency point in the target gain curve is calculated based on the following formula:

[0065]

[0066] in, P represents the gain value corresponding to frequency point i, N represents the total number of frequency points, and P represents the gain value corresponding to frequency point i. i G represents the power value corresponding to frequency point i in the power spectrum. 平均 This represents the average gain across all frequency points introduced by the desired Raman amplifier.

[0067] In practical implementation, since the Raman pump source has an upper limit on power output, an upper limit is set for the maximum gain value introduced by the Raman amplifier to avoid unreasonable target gain curves affecting the normal output of the Raman amplifier. At the same time, the gain of the original amplifier needs to be corrected according to the introduced average Raman gain. It is worth noting that in actual networks, wavelength switching occurs during transmission, which can lead to phenomena such as power jumps between frequency points in the power spectrum. The gain curve generated using the above formula may have unreasonable phenomena, leading to abnormal values ​​when the machine learning model predicts the Raman pump, which is not conducive to achieving sufficient transmission optimization. To address the power jump phenomenon, this embodiment uses interpolation to correct the gain value of individual frequency points to ensure that the Raman amplifier generates gain with high accuracy.

[0068] In this embodiment, after the target gain curve is designed, the initially trained machine learning model can be used to predict Raman pumping. Specifically, the gain curve is input into the machine learning model in descending order of frequency points. The model calculates and outputs the predicted initial Raman pumping configuration. This configuration is deployed to the Raman compensation point, and the amplifier originally used is adjusted according to the introduced average gain value so that the average output signal power remains unchanged after compensation. The power spectrum at the compensation point is re-acquired, and the difference between the power spectrum at the corresponding frequency point and the power spectrum before compensation is calculated and then summed with the introduced average gain value to calculate the actual gain curve introduced under the current pumping configuration.

[0069] In some embodiments of the present invention, the step of determining whether the iterative pump power vector of a new round corresponds to the final applied pump power includes step S450, comparing the iterative pump power vector of the new round with the target gain curve to determine whether the iterative pump power vector of the new round corresponds to the final applied pump power.

[0070] In some embodiments of the present invention, in the step of comparing the iterative pump power vector of a new round with the target gain curve to determine whether the iterative pump power vector of the new round corresponds to the final applied pump power, the root mean square error and the maximum absolute error are calculated based on the gain value corresponding to each power point in the iterative pump power vector of the new round and the gain value corresponding to each power point in the target gain curve. The root mean square error and the maximum absolute error are compared with the corresponding root mean square error threshold and the maximum absolute error threshold, respectively, to determine whether the iterative pump power vector of the new round corresponds to the final applied pump power.

[0071] In some embodiments of the present invention, in the step of comparing the root mean square error and the maximum absolute error with the corresponding root mean square error threshold and maximum absolute error threshold, respectively, to determine whether the iterative pump power vector of the new round corresponds to the final applied pump power, if the root mean square error is less than the root mean square error threshold and the maximum absolute error is less than the maximum absolute error threshold, then the pump power of each dimension in the iterative pump power vector of the new round is determined to be the final applied pump power.

[0072] In practice, the thresholds for root mean square error (RMSE) and maximum absolute error (Emax) can be determined based on the specific circumstances of the optical network.

[0073] Using the above scheme, if both evaluation indicators are below the threshold, the point is considered to have successfully completed the pump configuration of the Raman compensation point; if either evaluation indicator is above the threshold, iterative optimization is required based on the specific error situation.

[0074] In some embodiments of the present invention, in the step of determining whether the position corresponding to the fiber optic node is a compensation point based on the maximum and minimum power values ​​in the power spectrum, the difference between the maximum and minimum power values ​​in the power spectrum is calculated. If the difference between the maximum and minimum power values ​​in the power spectrum is greater than a preset difference threshold, then the position corresponding to the fiber optic node is determined to be a compensation point.

[0075] In the specific implementation process, the Raman amplifier pump configuration at all compensation points is completed, and the quality parameters at the completion of all service transmissions are recorded. The main evaluation index for signal transmission quality is the generalized signal-to-noise ratio (GSNR), which simultaneously considers the impact of linear noise P_ASE and nonlinear noise P_NLI on the signal. Its formula is as follows:

[0076]

[0077] P represents the actual measured signal power.

[0078] In practical implementation, due to the need for transmission optimization of multi-band dynamic optical networks, it is necessary to perceive and evaluate the actual network status under the current condition. Due to the strong nonlinear effect of stimulated Raman scattering, the most significant manifestation of the affected signal status is the signal power spectrum. In addition, it is currently difficult to measure parameters other than power during signal transmission. Based on these two points, the specific method for perceiving and evaluating the actual network status is to collect the signal power spectrum at the end of each fiber optic span. The specific evaluation method for the signal power spectrum is to calculate the degree of power spectrum tilt. The calculation method is to subtract the power of the signal at the minimum wavelength from the power of the signal at the maximum wavelength, and record this as the maximum signal power difference. When the maximum signal power difference exceeds a set threshold, this point needs to be determined as the Raman compensation point.

[0079] Using the above scheme, before determining the compensation point, the signal power spectrum is collected sequentially at the end of each optical fiber. The collected power spectrum is analyzed, and a certain threshold is usually set. When the power spectrum tilt exceeds the threshold for the first time, the point is set as the compensation point. Subsequent compensation points can be set sequentially according to the distance between the first compensation point and the source node. The position of the compensation point in this step can also be determined by the position of the traditional optical switch equalization station, that is, Raman amplifiers are directly used to replace each optical switch equalization station in turn.

[0080] Since the power spectrum at the end of the fiber always exceeds the preset threshold when Raman compensation is not performed, the compensation point cannot be determined by measuring the power spectrum. The number of segments from the starting point to the first compensation point can be calculated and recorded as the maximum number of uncompensated segments. For subsequent compensation points, the maximum number of uncompensated segments can be calculated from the first compensation point and set sequentially until the transmission end point.

[0081] This solution addresses the problem that existing multi-band power equalization schemes only consider the use of optical switches and attenuators in all scenarios, lacking consideration for the actual transmission power spectrum, which leads to significant degradation in signal transmission quality. It proposes an alternative to traditional optical switch equalization stations, providing strong technical support for multi-band capacity expansion schemes. This solution uses machine learning to adjust the gain spectrum generated by the Raman amplifier and iteratively optimizes the gain accuracy based on the difference between the deployment scenario and the original scenario, achieving effective adjustment of the signal power spectrum. Ultimately, it can significantly improve signal transmission quality while ensuring normal signal transmission.

[0082] In practice, each iteration of the optimization process includes two steps: error assessment and pump configuration update. The termination condition is the same as described in the error assessment: when the error is below a threshold, the optimization is considered complete; otherwise, pump configuration updates continue for multiple iterations. In this embodiment, the upper limit for the number of iterations is set to 7. The optimization of the compensation point is usually completed within 5 iterations. If the error assessment still fails when the iteration terminates, the initial gain curve design needs to be re-evaluated and some values ​​adjusted.

[0083] In this embodiment, after the pump configuration is successfully iteratively optimized by the Raman compensation points, the same process as described above can be performed sequentially along the optical signal propagation direction according to the expected estimated Raman compensation points. After completing compensation for all pre-estimated Raman compensation points, the signal transmission quality at the signal transmission endpoint is recorded. It is worth noting that since optical signal transmission in a multi-band dynamic optical network is not limited to a single transmission endpoint, we need to record the optical signals at each endpoint separately. In this embodiment, to fully evaluate the impact of linear and nonlinear interference on the signal, the generalized signal-to-noise ratio (GSNR) is used as the evaluation metric for signal quality.

[0084] In this embodiment, to illustrate the optimization effect of the multi-band dynamic optical network transmission optimization method based on multi-stage Raman amplifiers, the baseline scheme is the existing method of jointly constructing an equalization station using optical switches and optical attenuators. Specifically, the relay equipment adjustment method in the topology involves removing the Raman pumps at all compensation points and deploying an equalization station at each Raman compensation point. To fully demonstrate the comparison, the amplifier's input gain is adjusted to ensure that the average signal power value remains constant when the relay equipment outputs. The GSNR at the signal transmission endpoint is recorded similarly to the method described above. Furthermore, in this embodiment, the network topology is modified to simultaneously use both the Raman compensation scheme and the equalization station scheme, and the GSNR at the signal transmission endpoint is recorded similarly to the method described above.

[0085] In this embodiment, the GSNR curves of the two schemes are illustrated graphically according to frequency points, and the signal transmission quality optimization effect diagram is shown below. Figure 5 As shown, the baseline scheme uses a three-stage optical switch equalizer; the two-stage Raman amplifier uses Raman compensation for the first two compensation points under the same service conditions, and the equalizer scheme for the last compensation point; the three-stage Raman amplifier uses Raman compensation for all compensation points. The GSNR curves are subtracted according to frequency points to quantify the significant optimization effect of the optimization scheme. In this embodiment, the average GSNR is used for specific characterization. The two-stage Raman amplifier can improve the average GSNR of the optical signal by approximately 1.6 dB, while the three-stage Raman amplifier can improve the average GSNR of the optical signal by approximately 2.2 dB.

[0086] The beneficial effects of this plan include:

[0087] 1. Adaptive optimization schemes are adopted for multi-band dynamic optical networks based on actual transmission conditions. This avoids the limitation of traditional optical switch attenuation mode equalization schemes that apply the same treatment to all situations, and overcomes the dependence on optical switch attenuation schemes. This avoids the problem of transmission quality degradation caused by the introduction of optical switches and optical attenuators.

[0088] 2. This method can adapt to dynamic network environments and effectively handle dynamic changes in the network. It efficiently utilizes the parameter characteristics of the signal transmission process during signal processing, and the optimized scheme uses a Raman amplifier with better noise performance, which can significantly improve transmission quality and extend transmission distance.

[0089] 3. In the process of regulating the Raman amplifier, the limitations of traditional machine learning-based regulation schemes, which are subject to application scenarios, are overcome. The system can adaptively and iteratively optimize the pump parameter configuration according to changes in the scenario, ensuring a high-efficiency and high-precision Raman amplifier configuration. This greatly ensures the effectiveness of the optimization scheme in improving signal quality in various scenarios.

[0090] This invention also provides a multi-band dynamic optical network transmission optimization system. The system includes a computer device, which includes a processor and a memory. The memory stores computer instructions, and the processor executes the computer instructions stored in the memory. When the computer instructions are executed by the processor, the system implements the steps of the method described above.

[0091] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the aforementioned multi-band dynamic optical network transmission optimization method. The computer-readable storage medium can be a tangible storage medium, such as random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, floppy disks, hard disks, removable storage disks, CD-ROMs, or any other form of storage medium known in the art.

[0092] Those skilled in the art will understand that the exemplary components, systems, and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Whether implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention. When implemented in hardware, it can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this invention are programs or code segments used to perform the desired tasks. The programs or code segments can be stored in a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave.

[0093] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.

[0094] In this invention, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with or in place of features of other embodiments.

[0095] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations of the embodiments of the present invention are possible. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimizing transmission in a multi-band dynamic optical network, characterized in that, The steps of this method include: The power spectrum of each fiber node in the unoptimized optical network is obtained. The power spectrum records the power values ​​of the fiber node's location at different frequencies. Based on the maximum and minimum power values ​​in the power spectrum, it is determined whether the location of the fiber node is a compensation point. The target gain curve corresponding to the compensation point is calculated based on the power spectrum corresponding to the compensation point. The target gain curve includes multiple gain values ​​corresponding to different frequencies. The target gain curve is input into a pre-trained machine learning model, and an initial pump power vector is constructed based on the output value of the machine learning model. The initial pump power vector is iteratively calculated through multiple iterations to determine the final applied pump power. In each iteration, the cyclic pump power vector is calculated based on the current iterative pump power vector. The environmental difference is calculated based on the cyclic pump power vector and the initial pump power vector. The iterative pump power vector for the new iteration is calculated based on the environmental difference, and it is determined whether the iterative pump power vector for the new iteration corresponds to the final applied pump power.

2. The multi-band dynamic optical network transmission optimization method according to claim 1, characterized in that, In the step of calculating the cyclic pump power vector based on the current iterative pump power vector, the corresponding actual gain value is determined based on the current iterative pump power vector, and the cyclic pump power vector is determined based on the actual gain value.

3. The multi-band dynamic optical network transmission optimization method according to claim 1, characterized in that, The step of calculating the iterative pump power vector for the next round based on the environmental difference includes: The tuning parameters are calculated based on the initial pump power vector and the current iterative pump power vector; Calculate the new iteration pump power vector based on the current iterative pump power vector, tuning parameters, and environmental differences.

4. The multi-band dynamic optical network transmission optimization method according to claim 3, characterized in that, In the step of calculating tuning parameters based on the initial pump power vector and the current iterative pump power vector, the quotient of the initial pump power vector and the current iterative pump power vector is calculated as the tuning parameters.

5. The multi-band dynamic optical network transmission optimization method according to claim 3, characterized in that, In the step of calculating the iterative pump power vector for the next round based on the current iterative pump power vector, tuning parameters, and environmental differences, the iterative pump power vector for the next round is calculated based on the following formula: P_new = P_last - λ * P_dif; Where P_new represents the iterative pump power vector of the new round, P_last represents the current iterative pump power vector, P_dif represents the environmental difference, and λ represents the tuning parameters.

6. The multi-band dynamic optical network transmission optimization method according to any one of claims 1 to 5, characterized in that, In the step of determining whether the iterative pump power vector of the new round corresponds to the final applied pump power, the iterative pump power vector of the new round is compared with the target gain curve to determine whether the iterative pump power vector of the new round corresponds to the final applied pump power.

7. The multi-band dynamic optical network transmission optimization method according to claim 6, characterized in that, In the step of comparing the iterative pump power vector of the new round with the target gain curve to determine whether the iterative pump power vector of the new round corresponds to the final applied pump power, the root mean square error and the maximum absolute error are calculated based on the gain value corresponding to each power point in the iterative pump power vector of the new round and the gain value corresponding to each power point in the target gain curve. The root mean square error and the maximum absolute error are compared with the corresponding root mean square error threshold and the maximum absolute error threshold, respectively, to determine whether the iterative pump power vector of the new round corresponds to the final applied pump power.

8. The multi-band dynamic optical network transmission optimization method according to claim 1, characterized in that, In the step of calculating the target gain curve corresponding to the compensation point based on the power spectrum of the compensation point, the gain value corresponding to each frequency point in the target gain curve is calculated based on the following formula: in, P represents the gain value corresponding to frequency point i, N represents the total number of frequency points, and P represents the gain value corresponding to frequency point i. i Gi represents the power value corresponding to frequency point i in the power spectrum, and Gaverage represents the average gain across all frequency points introduced by the desired Raman amplifier.

9. The multi-band dynamic optical network transmission optimization method according to claim 1 or 8, characterized in that, In the step of determining whether the location corresponding to the fiber optic node is a compensation point based on the maximum and minimum power values ​​in the power spectrum, the difference between the maximum and minimum power values ​​in the power spectrum is calculated. If the difference between the maximum and minimum power values ​​in the power spectrum is greater than a preset difference threshold, then the location corresponding to the fiber optic node is determined to be a compensation point.

10. A multi-band dynamic optical network transmission optimization system, characterized in that, The system includes a computer device, which includes a processor and a memory. The memory stores computer instructions, and the processor executes the computer instructions stored in the memory. When the computer instructions are executed by the processor, the system implements the steps of the method as described in any one of claims 1 to 9.

Images