Dynamic wavelength allocation method based on DCI wavelength division II type equipment

Abstract

The invention provides a dynamic wavelength allocation method based on DCI wavelength division II type equipment. The method comprises the following steps: acquiring an occupation state and a transmission quality index of each wavelength channel of a current network; establishing a wavelength channel quality evaluation model, and scoring each wavelength channel according to the transmission quality index of each wavelength channel of the network; according to the bandwidth requirement, the priority and the QoS requirement of the service request, generating a traffic characteristic matrix containing the delay sensitivity and the bandwidth requirement through a traffic prediction algorithm; based on a network state and the traffic characteristic matrix, through resource availability calculation and path quality evaluation, forming an allocable wavelength resource pool and a quality score thereof; for the allocatable wavelength resource pool, generating an optimal wavelength allocation scheme by adopting an improved heuristic algorithm and comprehensively considering the resource utilization rate, the signal quality and the service priority; and according to the performance feedback of the optimal wavelength allocation result and the network state change, continuous optimization of wavelength allocation is realized through a self-adaptive adjustment mechanism. The method can effectively deal with the dynamic flow change in the DCI network, improve the resource utilization rate, ensure the signal quality and meet the QoS requirement of the service, thereby providing an efficient and reliable wavelength allocation solution for the data center interconnection network.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of optical communication resource allocation, in particular to a dynamic wavelength allocation method based on DCI wavelength division type II equipment. BACKGROUND

[0002] Data center interconnection (DCI) technology is a key infrastructure supporting modern cloud computing and big data applications. Wavelength division multiplexing (WDM) technology significantly improves the transmission capacity and efficiency between data centers by transmitting multiple different wavelength optical signals in a single optical fiber. With the explosive growth of data traffic and the diversification of business demands, higher requirements are placed on the dynamic and intelligent nature of DCI networks.

[0003] Existing wavelength allocation techniques mainly include static pre-allocation and simple dynamic allocation. Static pre-allocation configures wavelength resources in advance based on historical traffic patterns, with the advantages of simple configuration and good stability. Simple dynamic allocation allocates wavelengths based on real-time service requests, which can adapt to some degree of traffic changes.

[0004] The most relevant technology currently is a dynamic wavelength allocation algorithm based on traffic awareness. This technology monitors network traffic conditions, combines shortest path algorithms and load balancing strategies to achieve dynamic adjustment of wavelength resources. Its working principle is to first collect traffic information of each link in the network, then calculate the optimal wavelength allocation scheme according to the pre-set allocation strategy, and finally complete wavelength switching through control plane configuration instructions.

[0005] However, there are two key problems with existing technology: first, there is a lack of deep perception of the unique performance parameters of DCI wavelength division type II equipment, which cannot fully utilize the advanced functions of the equipment for fine management; second, when facing sudden traffic and rapid changes in network state, the response speed and optimization effect of the allocation algorithm are insufficient, which can easily lead to resource waste or degradation of service quality. SUMMARY

[0006] The present application aims to solve the problems of lack of deep perception of the unique performance parameters of DCI wavelength division type II equipment, inability to fully utilize the advanced functions of the equipment for fine management, and insufficient response speed and optimization effect of the allocation algorithm when facing sudden traffic and rapid changes in network state in existing technology, and provides a dynamic wavelength allocation method based on DCI wavelength division type II equipment, characterized by: Based on the port monitoring data of DCI wavelength division type II equipment, the occupancy status and transmission quality indicators of each wavelength channel in the current network are obtained through optical power detection and signal quality analysis; A wavelength channel quality evaluation model is established, and each wavelength channel is scored according to the transmission quality indicators of each wavelength channel in the network; According to the bandwidth demand, priority and QoS requirement of the service request, a traffic feature matrix containing delay sensitivity and bandwidth demand is generated through a traffic prediction algorithm; Based on the network state and the traffic feature matrix, a pool of allocable wavelength resources and its quality score are formed through resource availability calculation and path quality assessment; For the pool of allocable wavelength resources, an improved heuristic algorithm is adopted to generate an optimal wavelength allocation scheme by comprehensively considering resource utilization rate, signal quality and service priority; According to the performance feedback of the optimal wavelength allocation result and the change of network state, continuous optimization of wavelength allocation is realized through an adaptive adjustment mechanism.

[0007] The application can obtain the occupation state and transmission quality index of each wavelength channel in the current network in real time by monitoring the port monitoring data of the DCI wavelength division II type device, combining optical power detection and signal quality analysis. The wavelength channel quality assessment model is established, and each wavelength channel is scored according to the transmission quality index of each wavelength channel in the network, which can realize fine management of the wavelength channel, allocate resources according to the actual transmission quality, and improve the resource utilization efficiency. According to the bandwidth demand, priority and QoS requirement of the service request, a traffic feature matrix containing delay sensitivity and bandwidth demand is generated through a traffic prediction algorithm, which can dynamically respond to traffic changes, predict traffic demand in advance, and improve the response speed and adaptability of the system. Based on the network state and the traffic feature matrix, a pool of allocable wavelength resources and its quality score are formed through resource availability calculation and path quality assessment, which can efficiently allocate resources and ensure the rationality and optimization effect of resource allocation. For the pool of allocable wavelength resources, an improved heuristic algorithm is adopted to generate an optimal wavelength allocation scheme by comprehensively considering resource utilization rate, signal quality and service priority, which can quickly generate the optimal wavelength allocation scheme in a complex network environment and improve the overall performance of the system. According to the performance feedback of the optimal wavelength allocation result and the change of network state, continuous optimization of wavelength allocation is realized through an adaptive adjustment mechanism, which can adaptively adjust according to the real-time network state and service demand, and ensure the long-term stability and optimization effect of the system. The application solves the problems in the prior art through deep perception of device performance parameters, dynamic traffic prediction, fine resource management and adaptive adjustment mechanism, provides an efficient, dynamic and optimized wavelength allocation method, and significantly improves the application effect of DCI wavelength division II type device in data center interconnection network.

[0008] The above description is only a summary of the technical scheme of the present disclosure, in order to more clearly understand the technical means of the present disclosure, the content of the specification can be implemented, and in order for the above and other purposes, features and advantages of the present disclosure to be more obvious and easy to understand, the following preferred embodiments are described in detail below, and the accompanying drawings are described as follows. BRIEF DESCRIPTION OF DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings needed to be used in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description only represent some of the embodiments of the present disclosure, and all other drawings obtained by those of ordinary skill in the art based on the drawings without creative labor should also fall within the scope of the present disclosure.

[0010] Figure 1 The flow chart of the dynamic wavelength allocation method based on the DCI wavelength division type II device provided by the embodiments of the present disclosure.

[0011] Figure 2 The flow chart of the method for generating a traffic feature matrix containing latency sensitivity and bandwidth demand provided by the embodiments of the present disclosure.

[0012] Figure 3 The flow chart of the method for forming an allocable wavelength resource pool and its quality score provided by the embodiments of the present disclosure.

[0013] Figure 4 The flow chart of the method for generating an optimal wavelength allocation scheme using an improved heuristic algorithm provided by the embodiments of the present disclosure.

[0014] Figure 5 The flow chart of the method for realizing continuous optimization of wavelength allocation according to performance feedback and network state changes of the allocation result provided by the embodiments of the present disclosure.

[0015] Figure 6 The structural schematic diagram of a computer device provided by the embodiments of the present disclosure. DETAILED DESCRIPTION

[0016] The embodiments of the present disclosure will be described in detail below with reference to the drawings.

[0017] It should be clear that the embodiments of the present disclosure are described below through specific, concrete examples, and those of ordinary skill in the art can easily understand other advantages and effects of the present disclosure from the disclosure. Obviously, the described embodiments are only some of the embodiments of the present disclosure, not all. The present disclosure can also be implemented or applied by other different specific embodiments, and the details in the specification can be modified or changed based on different views and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict. Based on the embodiments in the present disclosure, all other embodiments obtained by those of ordinary skill in the art without creative labor should also fall within the scope of the present disclosure.

[0018] It is to be understood that the embodiments described hereinbelow within the scope of the appended claims. It will be apparent to one of ordinary skill in the art that aspects described herein can be implemented in a wide variety of forms, and that any specific structure and / or function described herein is merely illustrative. An aspect described herein can be implemented alone or in combination with any other aspect(s). For example, an apparatus can be implemented using any number of the aspects set forth herein. Additionally, an apparatus can be implemented using other structure and / or functionality.

[0019] It is also to be understood that the diagrams provided in the following embodiments are only schematic and that the dimensions of certain elements in the figures can be exaggerated relative to other elements for illustrative purposes and are presented only to explain certain aspects of the present disclosure. The same reference numerals in different drawings represent the same or similar elements.

[0020] Furthermore, in the following description, numerous specific details are set forth in order to provide a thorough understanding of the examples. However, it will be apparent to one of ordinary skill in the art that the aspects described herein can be practiced without these specific details.

[0021] The detailed description is described with reference to the accompanying drawings Figure 1 The present application provides a swallowing electric stimulation timing generation method based on multi-modal BCI and bionic timing electric stimulation, comprising the following steps: Step S1, based on the port monitoring data of the DCI wave division II type device, through optical power detection and signal quality analysis, the occupation state and transmission quality index of each wavelength channel of the current network are obtained.

[0022] In specific implementation, the occupation state parameters of each wavelength channel of the current network include whether the wavelength channel is occupied by service, the current transmission service type, the occupied bandwidth size, the service duration, etc. The occupation state parameters can be obtained by querying the wavelength management table and service configuration database of the DCI wave division II type device. By querying the wavelength management table built-in the DCI wave division II type device, the configuration state and service binding information of each wavelength channel are read in real time; then the service configuration database of the device is accessed to obtain the detailed parameters such as the current transmission service type, the occupied bandwidth size, the service establishment time and the expected duration; at the same time, the actual use of the wavelength channel is verified combining the optical power monitoring data and the signal detection result, to ensure the accuracy of the state information; finally, the state change events are regularly synchronized through the device management interface, including the operation records such as new service establishment, service release and service change.

[0023] In terms of obtaining transmission quality indicators, firstly, through the SNMP protocol or device-specific API interface, the key parameters of the DCI wave division type II device, such as optical power, bit error rate, dispersion compensation state, etc. are collected in real time. The collection frequency can be set to 10 times per second to ensure rapid response to network state changes. At the same time, a multi-level data collection architecture can be established, including three levels of monitoring points at the port level, device level and network level, covering network state information comprehensively. Then the collected raw data is filtered and normalized, and the Kalman filter algorithm is used to eliminate noise interference, and the Z-score standardization method is used to unify the data format. A data quality inspection mechanism is established to mark and correct abnormal data, ensuring the accuracy and reliability of the monitoring data.

[0024] Step S2, a wavelength channel quality evaluation model is established, and each wavelength channel is scored according to the transmission quality indicators of each wavelength channel in the network.

[0025] The wavelength channel quality evaluation model adopts a weighted scoring mechanism, and the specific mathematical formula is: Where Q is the wavelength channel quality score (0-100 point system), is the normalized transmission quality indicator, which is calculated according to the transmission quality indicator of the wavelength channel; is the normalized transmission distance influence factor, is the normalized device aging degree, is the normalized environmental factor score, which considers factors such as temperature stability, humidity control and electromagnetic interference. , . , is the device running time (months), is the device design life, is the coefficient.

[0026] In specific implementation, first, collect the monitoring data of each wavelength channel in the past 30 days as training samples, use 70% data for model training and 30% data for verification test. The wavelength channel quality evaluation model is trained by support vector machine (SVM) regression algorithm, and the input data includes 6-dimensional feature vector of wavelength channel signal-to-noise ratio, bit error rate, optical power, dispersion compensation state, transmission distance, device running time, etc. and corresponding historical quality label data.

[0027] Then use the SVM algorithm with radial basis function (RBF) kernel to build the regression model, the specific implementation process is as follows: first, establish the RBF kernel function Where γ is the kernel parameter, which controls the width of the radial basis function, the larger the γ value, the narrower the function, and the higher the model complexity; in wavelength channel quality evaluation, γ determines how the model balances the similarity between different features (such as signal-to-noise ratio, bit error rate, optical power, etc.), and are the representations of input feature vectors. These notations are used to describe how the kernel function calculates the similarity between two data points.

[0028] Then the SVM regression objective function is constructed: where C is the penalty parameter, which controls the tolerance of training error. C controls the fitting strictness of the model to the training data. In the wavelength channel quality assessment, C determines the balance between fitting the training data and maintaining the generalization ability of the model. The C value that best fits the wavelength channel quality assessment is found by grid search optimization. A larger C value will make the model fit the training data more strictly, but may cause overfitting; a smaller C value will make the model pay more attention to the generalization ability, but may underfit. w is the weight vector, is the slack variable.

[0029] When using the support vector machine (SVM) regression model for wavelength channel quality assessment, the key parameters that need to be optimized include the penalty parameter and the RBF kernel parameter. Specifically, the parameters are optimized by grid search method: set the search range of C as [0.1, 1, 10, 100, 1000], and the search range of γ as [0.001, 0.01, 0.1, 1, 10], use 5-fold cross-validation to evaluate the performance of each parameter combination, and select the parameter combination with the smallest mean square error as the optimal solution.

[0030] In the model training phase, the 6-dimensional feature vector (signal-to-noise ratio, bit error rate, optical power, dispersion compensation state, transmission distance, and equipment running time) is used as input, and the corresponding quality score is used as output label. The quality score is the target value of model training, which is used to supervise the learning process of the model. By using the Lagrange multiplier method, the constrained optimization problem is converted into an unconstrained optimization problem. Introduce the Lagrange multiplier and to construct the Lagrange function. The quadratic programming problem of the SVM regression model optimal weight vector w and bias term b is solved by the sequential minimal optimization (SMO) algorithm, and the support vector and the corresponding Lagrange multiplier are obtained. When the model is predicted, for a new feature vector x, calculate where is the Lagrange multiplier, b is the bias term, and the output is the quality score in the 0-100 point system. is the signal-to-noise ratio, bit error rate, optical power, dispersion compensation state, transmission distance, and equipment running time parameter; is the kernel function, which is used to calculate the similarity between the support vector and the new feature vector ​To train the wavelength channel quality evaluation model. Through these mechanisms, the SVM regression model can predict new data points based on key points (support vectors) in the training data, thereby achieving wavelength channel quality evaluation.

[0031] The model can automatically learn the influence weight of each parameter on the quality score based on historical data. The model parameters are retrained every 7 days using the latest data, and the network environment changes are adapted through incremental learning mechanisms. When the distribution difference between new data and the original training set exceeds the set threshold, the model update is automatically triggered. The output is the real-time quality score (0-100 scale) of each wavelength channel, where 90-100 is excellent, 80-89 is good, 70-79 is general, and less than 70 is attention needed. For example, in a certain DCI data center interconnection scenario, the current monitoring of the 20th channel (wavelength 1550.12nm) of the C band shows a signal-to-noise ratio of 25dB, a bit error rate of , an optical power of -5dBm, a normal dispersion compensation state, a transmission distance of 80km, and a device running time of 18 months. After SVM model calculation, the quality score is 92 points, belonging to the excellent level. While the 45th channel of the L band has a transmission distance of 150km and a serious device aging, although the signal-to-noise ratio is 22dB, the comprehensive score is only 68 points, which needs to be focused on and considered for path optimization or device maintenance.

[0032] A real-time alarm mechanism is established. When the wavelength channel quality score is lower than the preset threshold (such as 70 points), the system automatically generates an alarm information and pushes it to the network administrator. At the same time, hierarchical storage of state data is realized, with high-frequency data saved for 7 days, medium-frequency data saved for 30 days, and low-frequency summary data saved for a long time. Finally, the processed state data is stored in a distributed real-time database, and a time series database architecture is used to support high-concurrency read and write operations. An event-driven state change notification mechanism is established. When a significant change in wavelength channel state is detected (such as a change in quality score of more than 10 points), an event notification is triggered immediately, and update information is pushed to the wavelength allocation decision module through a message queue, ensuring that the decision algorithm can perform optimization calculation based on the latest network state.

[0033] Step S3, according to the bandwidth demand, priority and QoS requirement of the service request, generate a traffic feature matrix containing delay sensitivity and bandwidth demand through traffic prediction algorithm. For specific implementation steps, see Appendix Figure 2 .

[0034] Step S310, collect service request parameters through the service management interface, including source and destination node ID, bandwidth requirement value, service priority (1-5 level), maximum tolerable delay, packet loss rate requirement, and other QoS indicators. For example, a certain video conference service request from Beijing data center (node ID: BJ-DC01) to Shanghai data center (node ID: SH-DC03) with a bandwidth requirement of 10 Gbps, service priority of level 4 (high priority), maximum tolerable delay of 20 ms, and packet loss rate requirement of less than 10-6, and the service type is identified as real-time audio and video transmission.

[0035] Step S320, use the ARIMA time series model to combine the historical 30-day traffic data for short-term traffic prediction to predict the traffic trend in the next 1 hour. The specific implementation process is as follows: first, pre-process the historical traffic data, including data cleaning, outlier detection, and missing value filling, and use the moving average method to smooth the data noise; then perform stationarity test to determine the stationarity of the sequence through ADF test, and if it is not stationary, perform difference processing until the sequence is stationary; then determine the ARIMA model parameters , where p is the autoregressive order, d is the difference order, and q is the moving average order. The optimal parameter combination is selected through AIC information criterion and BIC criterion, and the value range of p and q is generally 0-5, and d is usually 1-2; then use the maximum likelihood estimation method to train the model parameters to establish the prediction equation ; is the traffic value at the current time, is the autoregressive coefficient, is the error term. is the moving average coefficient. Finally, perform rolling prediction, update the model every 15 minutes and predict the traffic value every 15 minutes in the next 1 hour. For example, for the traffic between Beijing and Shanghai data centers, based on the traffic data sampled every 15 minutes in the past 30 days (a total of 2880 data points), the ARIMA(2,1,1) model is determined after stationarity test, i.e. 2-order autoregression, 1-time difference, and 1-order moving average. Under the condition that the trained model predicts the current traffic to be 85 Gbps, the traffic in the next 1 hour will be increased to 90 Gbps, 95 Gbps, 98 Gbps, and 96 Gbps, respectively, showing a trend of first rising and then stabilizing, with a prediction accuracy of 92%, providing a reliable basis for early reservation and dynamic adjustment of wavelength resources.

[0036] ​Step S330, based on the short-term traffic variation trend prediction, then build a multi-dimensional feature vector. First, establish the feature quantization rules, bandwidth demand by dividing the system maximum support bandwidth (such as 400Gbps) for normalization, the absolute value into 0-1 interval relative value. Delay sensitivity using segmented mapping, real-time business (delay requirement <10ms) encoding 0.9, quasi real-time business (delay requirement 10-50ms) encoding 0.6, non real-time business (delay requirement >50ms) encoding 0.3. Priority from 1-5 level using linear transformation formula Mapping to 0.2-1.0 interval, ensure the minimum priority still have basic weight. Then through the vector construction algorithm, the quantified features of each service are assembled into a standardized vector in a fixed order, using an array structure [bandwidth normalized value, delay sensitivity, priority normalized value, predicted growth rate, service type code] for storage. For example, a certain financial transaction business from Shanghai data center to Shenzhen data center, bandwidth demand 80Gbps, delay requirement <5ms, priority level 5, predicted traffic growth 20%, belongs to real-time business type, then build the feature vector for [80 / 400, 0.9, 0.2+(5-1)×0.2, 0.20, 1] = [0.2, 0.9, 1.0, 0.20, 1], the vector directly as the subsequent intelligent allocation algorithm standardization input.

[0037] Finally, an N x M dimensional traffic feature matrix is generated, where N is the number of service connections, and M is the feature dimension (including bandwidth demand, latency sensitivity, priority, predicted growth rate, service type, etc. 5 dimensions), which is used as the standardized input of the wavelength allocation decision algorithm. The specific implementation process is as follows: first, establish a feature matrix data structure, stored in the form of a two-dimensional array, with row index corresponding to service connection ID and column index corresponding to feature dimension; then, standardize the data in each dimension, normalize the bandwidth demand by dividing by the maximum supported bandwidth of the system (such as 400Gbps), map the latency sensitivity to 0.9, 0.6, and 0.3 according to <10ms, 10-50ms, and >50ms, respectively, linearly map the priority from 1-5 levels to the 0.2-1.0 interval, divide the predicted growth rate by 100 in percentage form, and use one-hot encoding for service type (real-time = 1, 0, 0, quasi-real-time = 0, 1, 0, non-real-time = 0, 0, 1); then, use a matrix operation library (such as NumPy) to build a standardized matrix, with each row representing a complete feature vector of a service connection; finally, establish a matrix index and query interface to support fast retrieval of corresponding feature vectors by service ID. For example, in a certain DCI network, there are 3 services to be allocated: service 1 is a video conference from Beijing to Shanghai (100Gbps bandwidth, latency <20ms, priority 4, predicted growth 10%, real-time service), service 2 is a data backup from Shenzhen to Guangzhou (50Gbps bandwidth, latency <100ms, priority 2, predicted growth 5%, non-real-time service), and service 3 is an online transaction from Shanghai to Hangzhou (200Gbps bandwidth, latency <10ms, priority 5, predicted growth 15%, real-time service). The generated 3 x 5 dimensional feature matrix is: [[0.25, 0.9, 0.8, 0.10, 1], [0.125, 0.3, 0.4, 0.05, 0], [0.5, 0.9, 1.0, 0.15, 1]], where the first row [0.25, 0.9, 0.8, 0.10, 1] represents the standardized feature vector of service 1. This matrix is directly input into the intelligent wavelength allocation decision algorithm for optimization calculation.

[0038] Step S4, based on network status and traffic features, through resource availability calculation and path quality evaluation, form a pool of allocable wavelength resources and their quality scores, see the specific implementation in the attached Figure 3 .

[0039] Step S410, according to the real-time network topology and the occupation state of each wavelength channel of the current network, an initial candidate resource set is constructed. In a specific implementation, the occupation state of each wavelength channel of the current network is obtained by querying, the real-time occupation state of all 96 wavelength channels in the C band and the L band is obtained, the link state information obtained by the network topology discovery protocol (such as OSPF-TE) is combined, the end-to-end reachable idle wavelength channel combination is identified, and considering the wavelength continuity constraint and the device port availability, an initial candidate set containing the source node, the destination node, the available wavelength list and the relay path is formed. Through efficient topology discovery and resource identification, this step reduces unnecessary calculation and search, and improves the response speed of the system Step S420, the quality of each candidate wavelength and path combination is evaluated, and the wavelength and path combination quality score is calculated by comprehensively considering factors such as transmission distance, signal-to-noise ratio, bit error rate and device performance. The evaluation process adopts a hierarchical calculation mode. Firstly, the optical signal transmission quality is calculated at the physical layer, including optical power loss (0.2 dB attenuation per kilometer), dispersion accumulation effect and nonlinear effect influence; then the bit error rate and jitter index are evaluated at the link layer, and the net bit error rate after forward error correction is calculated by real-time monitoring data; then the end-to-end delay (optical fiber transmission delay + device processing delay) and reliability index are comprehensively evaluated at the path layer, and finally the path quality score in the 0-100 point system is obtained through weighted average algorithm. Through multi-dimensional and hierarchical calculation, the accuracy and comprehensiveness of the quality evaluation are ensured.

[0040] Step S430, resource conflict detection and constraint verification are performed, and candidate resources that cause wavelength conflict with existing services or do not meet the QoS requirements are excluded. The conflict detection includes three levels of wavelength conflict detection (occupation conflict of the same wavelength on the same optical fiber), port conflict detection (device port multiplexing restriction) and bandwidth conflict detection (insufficient link residual bandwidth), and the QoS constraint verification includes delay constraint verification (whether the total path delay meets the service requirement), bandwidth constraint verification (whether the path can provide bandwidth to meet the demand) and reliability constraint verification (whether the path availability meets the SLA requirement), and the effective candidate resources that meet all constraint conditions are screened out through Boolean logic judgment. This step ensures that the generated candidate resources meet all constraint conditions, and improves the feasibility of resource allocation.

[0041] Step S440, a multi-dimensional resource scoring model is established, and the wavelength channel quality, path quality, load balancing degree and historical performance are weighted and combined.

[0042] The scoring model adopts the analytic hierarchy process to determine the weight distribution: the wavelength channel quality weight obtained in step S2 is 35%, the path quality weight of the candidate wavelength-path combination obtained in step S420 is 30%, the load balancing degree weight is 20% (obtained based on the current path load rate and the overall network load distribution), and the historical performance weight is 15% (based on the average performance in the past 30 days). The dimensional indicators are standardized, and the wavelength channel quality, path quality, load balancing degree, and historical performance are converted into dimensionless values in the 0-1 interval; then the weighted sum method is used to calculate the comprehensive score S = 0.35 x Qchannel+ 0.30 x Rpath+ 0.20 x Lbalance+ 0.15 x Phistory; at the same time, a dynamic adjustment factor a is introduced, when the network congestion degree exceeds 80%, the load balancing weight is increased to 30%, when the failure rate rises, the path reliability weight is increased to 40%, the weight distribution is updated every 5 minutes by real-time monitoring of network state changes, to ensure that the scoring model can adapt to the dynamic changes of the network environment. This step considers multiple key indicators and dynamically adjusts the weights according to the network state, ensuring the comprehensiveness and adaptability of the scoring model.

[0043] In step S450, a standardized resource pool containing available wavelength resources and their comprehensive quality scores is formed to provide optimized input for subsequent intelligent allocation decisions. The resource pool adopts a three-layer B+ tree index structure: the first layer establishes a hash index according to the source-destination node pair, the second layer establishes a range index according to the bandwidth level (10G, 40G, 100G, 400G), and the third layer establishes an ordered index according to the quality score. Each resource entry is stored in a structured data format, containing 12 complete attribute fields such as wavelength ID (32-bit identifier), path information (node sequence and link ID list), quality score (0-100 floating point number), available bandwidth (Gbps unit), estimated latency (microsecond level precision), reliability level (five-level classification), etc. This step ensures the efficiency and read operation consistency of the resource pool through an efficient index structure and real-time update mechanism.

[0044] The real-time updating mechanism is implemented through an event-driven architecture. When a network state change is detected, an incremental updating thread is triggered. The write-once-copy technology is used to ensure the consistency of read operations, and the update delay is controlled within 100 milliseconds. For example, when a network state changes (e.g., the occupancy state of a wavelength channel changes, the link state changes, etc.), the system generates an event. These events trigger an incremental updating thread, which is responsible for updating the relevant information in the resource pool. To ensure the consistency of read operations, the write-once-copy technology is used. This means that when updating the resource pool, the original data is not directly modified, but a copy is created for modification. After the modification is completed, the original data is replaced. Through optimized design, the update delay is controlled within 100 milliseconds, ensuring the real-time performance of the system. The standardized API interface is designed using the RESTful architecture, providing the GET / wavelength-pool / search interface to support multi-dimensional queries, including combined retrieval according to source and destination nodes, bandwidth range, quality threshold, and latency limit conditions, returning a resource list in JSON format and arranging it in descending order according to the comprehensive score.

[0045] Step S5: An improved heuristic algorithm is used to generate an optimal wavelength allocation scheme by considering resource utilization, signal quality, and service priority. In specific implementation, first, a multi-objective optimization function is designed, maximizing resource utilization, optimizing signal quality, and prioritizing high-priority services as optimization objectives. Then, an improved genetic algorithm or particle swarm optimization algorithm is used to find the Pareto optimal solution set through iterative search. Next, a service priority constraint and QoS guarantee mechanism are introduced to ensure that the resource needs of critical services are prioritized. Finally, through multiple rounds of optimization iteration and solution set screening, an optimal wavelength allocation scheme is generated that meets all constraint conditions, including the specific wavelength and transmission path corresponding to each service. For specific implementation steps, see Appendix 2. Figure 4 .

[0046] Step S510, a multi-objective optimization function is constructed, resource utilization, signal quality and service priority are taken as optimization objectives, and a weighted summation method is used to establish the objective function F=0.3*U+0.4*Q+0.3*P, wherein U is a resource utilization index, Q is a signal quality index, and P is a priority satisfaction index. The specific implementation process is as follows: first, the resource utilization index U is calculated, the formula U=∑(allocated bandwidth / total available bandwidth)*100% is used, and the network overall utilization is obtained by counting the bandwidth occupation of all wavelength channels; then, the signal quality index Q is calculated, the weighted average method Q=∑(service bandwidth*channel quality score) / ∑service bandwidth is used based on the wavelength channel quality score obtained in step 1, and the signal quality of high-bandwidth services is ensured to be given priority; then, the priority satisfaction index P is calculated, the resource allocation satisfaction of each priority service is counted through the traffic characteristic matrix counted in step S2, wherein the weight of the 5-level service is 1.0, the weight of the 4-level service is 0.8, the weight of the 3-level service is 0.6, the weight of the 2-level service is 0.4, and the weight of the 1-level service is 0.2; finally, the three indexes are normalized and then substituted into the objective function for calculation.

[0047] Step S520, a wavelength allocation solving framework based on hybrid intelligent optimization is designed, the multi-objective optimization problem is converted into a solvable mathematical model, and the optimal wavelength and path allocation scheme is found through intelligent algorithm. The role of this step is to achieve the best balance of resource utilization, signal quality and service priority under the condition of meeting all constraints. The specific implementation process is as follows: first, a problem coding scheme is constructed, a three-layer coding structure is used to represent the allocation scheme: the first layer is a service-wavelength mapping vector B=[b1,b2,...,bn], wherein bi represents the wavelength ID allocated to service i; the second layer is a service-path mapping vector R=[r1,r2,...,rn], wherein ri represents the path number selected by service i; the third layer is a quality weight vector W=[w1,w2,...,wn], wherein wi represents the sensitivity weight of service i to signal quality. Then a hybrid optimization algorithm is designed, the coded wavelength allocation scheme{ B , R , W} is taken as input, the global search ability of particle swarm optimization (PSO) and the local fine search of simulated annealing (SA) are combined to find the optimal wavelength and path allocation scheme based on the multi-objective optimization function: the population size of 30 particles is used in the PSO stage, each particle contains a position vector (current solution) and a velocity vector (search direction), the particles are guided to move by individual optimum and global optimum, and after 200 iterations, a candidate solution set is obtained; in the SA stage, the optimal solution of PSO is taken as the initial solution, the initial temperature T0=50 is set, the cooling coefficient α=0.9 is set, 50 neighborhood searches are performed at each temperature, new solutions are generated by random disturbance and accepted according to the Metropolis criterion, and finally the local optimum is converged to obtain the optimized wavelength allocation scheme A dynamic fitness evaluation mechanism can be established. Based on the basic objective function F = 0.3 * U + 0.4 * Q + 0.3 * P, a network state awareness factor is introduced to realize the fitness evaluation value after adjusting the weight. When the overall network load rate exceeds 80%, the resource utilization rate weight is dynamically adjusted to 0.4; when the signal quality is generally detected to be declining, the signal quality weight is increased to 0.5; when high-priority services are queued, the priority weight is increased to 0.4, so that the algorithm can adjust the optimization target according to the real-time network state. Then, a constraint processing strategy is realized, and a feasibility repair mechanism is used to process constraint violations: for wavelength conflict, the conflict detection matrix is used to quickly identify and reassign free wavelengths to conflicting services; for capacity overrun, a service degradation strategy is used to appropriately reduce the bandwidth demand of non-critical services; for QoS violation, an alternative path is found for delay-sensitive services that meet the requirements through path reselection. Finally, convergence judgment and solution output are performed, and multiple convergence conditions are set: the improvement amplitude of the objective function is less than 0.01% for 20 consecutive iterations, or the maximum number of iterations is reached 500 times, or all constraints meet the degree of 98% or more, the algorithm is terminated, and the optimal wavelength allocation scheme that meets the conditions is output For example, when processing the allocation requirements of 5 services in a certain DCI network, the algorithm converges at the 180th iteration, and the final scheme realizes a resource utilization rate of 92.5%, an average signal quality of 89.3, and a priority satisfaction degree of 96.7%. Compared with the traditional heuristic algorithm, the comprehensive performance is improved by 15%, providing a scientific and reliable decision basis for subsequent configuration execution. Combined with the global search ability of particle swarm optimization (PSO) and the local fine search ability of simulated annealing (SA), the optimal wavelength and path allocation scheme can be effectively found.

[0048] Through the dynamic fitness evaluation mechanism, the weights of the optimization target are dynamically adjusted according to the network state, improving the adaptability and flexibility of the algorithm. Through the constraint processing strategy, it is ensured that the generated wavelength allocation scheme meets all constraint conditions, including wavelength conflict, bandwidth capacity, and QoS delay constraints.

[0049] Step S530, implement a constraint processing mechanism, process wavelength conflict constraints, bandwidth capacity constraints, and QoS delay constraints through a penalty function method, apply a penalty coefficient to solutions that violate constraints, and simultaneously use a repair operator to locally adjust infeasible solutions so that the optimal wavelength allocation scheme that does not meet the constraint conditions meets the constraint conditions. Through the constraint processing mechanism, it is ensured that the optimal wavelength allocation scheme generated in step S520 meets all constraint conditions, including wavelength conflict, bandwidth capacity, and QoS delay constraints, to obtain a repaired wavelength allocation scheme .

[0050] The specific implementation process is as follows: first, a constraint detection function is established to detect the optimal wavelength allocation scheme whether the wavelength conflict, bandwidth capacity and QoS latency constraints are violated. The wavelength conflict constraint is checked by constructing a wavelength occupation matrix detection, where i represents the fiber link and j represents the wavelength channel, if the same wavelength is occupied by multiple services on the same link >1, a conflict penalty is triggered; the bandwidth capacity constraint is checked by calculating the ratio of the total demand bandwidth to the available bandwidth of each link, and a capacity overrun penalty is applied when the ratio exceeds 1; the QoS latency constraint is checked by accumulating the transmission latency and device processing latency of all links on the path, and a latency penalty is applied in proportion to the excess part compared to the service latency requirement.

[0051] Penalty function formula is designed where α, β, γ are the penalty coefficients of conflict, capacity and latency respectively (value range 10-100), original fitness value, i.e. the fitness value without considering constraint violation; is the conflict degree, is the capacity overrun degree, is the latency overrun degree. In this embodiment, the fitness value is the objective function F=0.3×U+0.4×Q+0.3×P. Then the repair operator mechanism is implemented, when an infeasible solution is detected, the wavelength reallocation repair operator is used first, which searches for a free wavelength on the same path for the conflicting service to replace; if there is no available wavelength, the path reselection repair operator is used, which searches for an alternative path for the service that satisfies the constraints; if it still cannot be repaired, the service degradation repair operator is used, which reduces the bandwidth demand or relaxes the latency requirement of non-critical services. After the above steps, the repaired wavelength allocation scheme is obtained. For example, in a certain DCI network, there are 3 services to be allocated: service A (Beijing to Shanghai, λ1 wavelength, 100Gbps), service B (Shanghai to Shenzhen, λ1 wavelength, 80Gbps), and service C (Beijing to Shenzhen, path through Beijing-Shanghai-Shenzhen, λ2 wavelength, 150Gbps), if service A and service B both select λ1 wavelength on the Shanghai-Shenzhen link in the initial allocation scheme, a wavelength conflict is generated, the conflict detection function calculates =1 (one conflict), at the same time the total latency of the path selected by service C is 45ms but its QoS requirement is 30ms, a latency overrun is generated, =(45-30) / 30=0.5, if the original fitness =85 points, the penalty coefficients α=50 and γ=30, then the fitness after penalty = 20, which is marked as a poor solution; after the repair operator is activated, the wavelength conflict is eliminated by replacing the λ3 wavelength on the Shanghai-Shenzhen link for service B, and then the path for service C is recalculated, and a direct path from Beijing to Shenzhen with a delay of only 25 ms and the λ2 wavelength is found to be available, so the fitness of the repaired solution returns to 82, successfully satisfying all the constraints.

[0052] By the penalty function method and the repair operator, the wavelength allocation scheme is ensured to satisfy all the constraints, including wavelength conflict, bandwidth capacity, and QoS delay constraint. By using the repair operators such as wavelength reallocation, path reselection, and service degradation, the infeasible solution can be effectively repaired, and the quality of the solution is improved.

[0053] Step S540, by simulating the natural selection and genetic variation mechanism, a high-quality global optimal solution is found from the repaired wavelength allocation scheme within a limited number of iterations to obtain the final optimized wavelength allocation scheme . The iterative optimization process is executed, the maximum number of iterations is set to 500 generations, and the process is terminated in advance when the optimal solution is not improved for 50 consecutive generations or the preset convergence precision is reached, and each generation of the evolution process includes selection, crossover, mutation, and environmental selection four operations, and the tournament selection strategy is used to maintain the population diversity. The specific implementation process is as follows: first, the repaired wavelength allocation scheme selection operation, the tournament selection strategy is used to select the parent individuals from the current population, 3 individuals are randomly selected for competition each time, and the one with the highest fitness wins as the parent, and the process is repeated until a parent set equal to the population size is selected, which ensures the high selection probability of excellent individuals and maintains the population diversity; then the crossover operation is performed, the selected parent individuals are paired, and the simulated binary crossover (SBX) is performed with a crossover probability of 0.8; for real number coded chromosomes, two offspring individuals are generated by the SBX operator, and a distribution index is introduced in the crossover process to control the distribution range of the offspring, ensuring that the offspring are generated near the parents while having certain exploration ability; then the mutation operation is performed, the offspring individuals generated by the crossover are subjected to polynomial mutation with a mutation probability of 0.1, and the mutation strength is controlled by the distribution index The control prevents the algorithm from converging to a local optimum too early by adding small random perturbations on the gene sites to increase the population diversity; finally, the environmental selection operation is performed to combine the parent population and the offspring population to form a temporary population with a size of twice, sort them from high to low according to the fitness, select the first 100 individuals to form the next generation population, and apply the elite reservation strategy to ensure that the optimal solution of the current generation is not lost; after each generation evolution is completed, the population average fitness and the optimal fitness are calculated, and when the optimal solution improvement amplitude is less than 0.01% for 50 consecutive generations, it is determined to converge and terminate the iteration in advance. The individual with the highest fitness is selected from the final population as the optimal solution, a detailed allocation scheme is generated, and the final optimized wavelength allocation scheme is obtained For example, in a certain DCI network wavelength allocation optimization, the fitness of the initial population of 100 individuals is distributed between 60 and 75 points, and after tournament selection, the probability of selecting individuals with a fitness of 72 points or more reaches 80%; after the first generation evolution, the optimal fitness is improved to 78.5 points, and the average fitness is improved from 67.2 points to 69.8 points; after 50 generations of evolution, the optimal fitness reaches 92.3 points, and the improvement amplitude of 10 consecutive generations is less than 0.1 point; when the evolution continues to the 120th generation, the optimal solution of the last 50 generations remains at 94.7 points without significant improvement, the algorithm is automatically terminated and the optimal allocation scheme is output: service A is allocated to wavelength λ1 and transmitted through path 1, service B is allocated to wavelength λ3 and transmitted through path 2, and service C is allocated to wavelength λ2 and transmitted through path 1, which realizes a resource utilization rate of 96.8%, an average signal quality of 91.5 points, and a priority satisfaction degree of 98.2%, which are significantly improved compared with the initial random allocation scheme in various indicators.

[0054] Step S550, output the optimal allocation scheme, select the individual with the highest fitness from the final optimized wavelength allocation scheme as the optimal solution, and generate a detailed allocation scheme containing the specific wavelength ID, transmission path node sequence, expected signal quality and resource occupation of each service. The specific implementation process is as follows: first, the final population of the evolutionary algorithm (i.e., the final optimized wavelength allocation scheme The process involves extracting the individual with the highest fitness, converting its chromosome encoding into an actual wavelength allocation mapping, and then using a decoding function to convert the real-number encoding into specific wavelength IDs (e.g., λ1 corresponds to 1550.12nm), path node sequences (e.g., BJ-DC01→SH-DC02→GZ-DC03), and quality parameters. A standardized allocation scheme data structure is then constructed, storing complete allocation information for each service in JSON format, including 12 key fields such as service ID, source and destination nodes, allocated wavelength, transmission path, expected latency, signal quality score, occupied bandwidth, and resource utilization. Next, a final feasibility verification of the scheme is performed, using a constraint check algorithm to confirm that all wavelength allocations are conflict-free, bandwidth requirements are met, and QoS indicators are satisfied. If problems are found, a new individual is selected from the suboptimal ones. Finally, a detailed execution instruction set is generated, including specific configuration commands for each DCI device, optical cross-connect matrix settings, and service establishment timing, providing a standardized interface for the automated configuration of the network management system.

[0055] By utilizing the natural selection and mutation mechanisms of genetic algorithms, high-quality global optima can be found within a finite number of iterations. A tournament selection strategy is employed to maintain population diversity and prevent the algorithm from prematurely converging to local optima. By setting multiple convergence conditions, the algorithm is ensured to terminate promptly upon reaching the optimal solution, thereby improving optimization efficiency.

[0056] Step S6: Based on the performance feedback of the allocation results and changes in network status, continuously optimize wavelength allocation through an adaptive adjustment mechanism. (See attached...) Figure 5 It can be seen that the implementation process includes the following specific steps: Step S610: Establish a multi-layered performance monitoring system. Deploy a distributed monitoring agent to collect key performance indicators of allocated services every 10 seconds, including end-to-end latency (target <50ms), packet loss rate (target < The system monitors transmission quality parameters such as jitter (target <1ms) and throughput utilization; it also monitors physical layer indicators such as optical signal-to-noise ratio, bit error rate, and optical power for wavelength channels; and it establishes a user satisfaction evaluation model to calculate a satisfaction score (0-100 points) based on service response time and the number of service interruptions. Monitoring data is stored in a time-series database, supporting real-time querying and historical trend analysis.

[0057] Step S620: Design an intelligent trigger condition judgment mechanism. Establish a multi-dimensional threshold system: performance thresholds include latency exceeding 120% of the SLA requirement, packet loss rate exceeding... , a wavelength channel quality score decrease of more than 15 points, a link utilization rate of more than 85%, and a path reliability decrease caused by adjacent link failure. A sliding window algorithm (window size 5 minutes) is used for trend analysis, and when the indicators deteriorate or a single mutation exceeds the emergency threshold in three consecutive monitoring periods, the re-optimization process is triggered immediately. At the same time, a machine learning prediction model is introduced to predict the performance trend in the next 15 minutes based on historical data, and active triggering is realized.

[0058] In step S630, an incremental optimization algorithm is implemented. A local search strategy is used to first identify performance substandard service connections and establish an impact analysis map to determine the adjustment range; then a constraint satisfaction algorithm is applied to search for alternative wavelengths and path combinations for problem connections while maintaining more than 90% of existing service paths unchanged; then a simulated annealing algorithm is used for local optimization, with the objective function considering adjustment cost (weight 30%), performance improvement degree (weight 50%), and resource utilization rate (weight 20%); finally, conflict detection and feasibility verification are performed to ensure that the optimization scheme does not have a negative impact on other stable services. The specific implementation process is as follows: first, performance substandard service identification is performed by scanning the real-time performance indicators of all active service connections, and services with a delay exceeding 120% of the SLA requirement, a packet loss rate exceeding 10^-5, or a throughput decrease exceeding 20% are marked as problem services. At the same time, a business dependency graph is constructed to analyze the sharing relationship of problem services and other services on wavelengths and paths, and the minimum adjustment impact range is determined. Then, a constraint satisfaction search algorithm is implemented to search for candidate wavelength-path combinations that meet the bandwidth and QoS requirements for each problem service in the current network topology. A backtracking search method is used to gradually build a feasible solution space, and a pruning strategy is applied to exclude obviously infeasible combinations to ensure search efficiency. Next, a simulated annealing optimization engine is deployed with an initial temperature T0=100, a cooling coefficient α=0.95, and a termination temperature Tf=0.1. At each temperature, 100 neighborhood searches are performed to generate neighborhood solutions by randomly perturbing the current solution, and the objective function value is calculated , accept or reject the new solution according to the Metropolis criterion, and gradually converge to the local optimum. Finally, global conflict detection is performed, and the wavelength occupation matrix of the new and old schemes is compared to check whether there is a wavelength conflict, bandwidth overrun or path loop, and the network connectivity and path reachability are verified by graph theory algorithm to ensure the global feasibility of the optimization scheme. For example, in a certain DCI network, there are 5 service connections running normally, among which service A (Beijing to Shanghai, wavelength λ1, path BJ→TJ→SH, current delay 35 ms, SLA requirement ≤25 ms) and service B (Tianjin to Shanghai, wavelength λ1, path TJ→SH, current delay normal 18 ms) share the λ1 wavelength resource of TJ→SH link. When it is detected that the delay of service A is out of limit, the incremental optimization algorithm starts: first, identify service A as the problem service, build the influence graph to find that it shares the TJ→SH link with service B; then search for an alternative scheme for service A, find that the λ3 wavelength is available on the path BJ→SH (direct connection), and the expected delay is 20 ms, which meets the SLA requirement; then evaluate the adjustment scheme by the simulated annealing algorithm, calculate the adjustment cost as 15 points (need to reconfigure the optical cross-connect), the performance improvement degree as 85 points (the delay is reduced from 35 ms to 20 ms), and the resource utilization rate as 70 points (release the λ1 resource on TJ→SH), and the comprehensive objective function value F = 0.3 × 15 + 0.5 × 85 + 0.2 × 70 = 61 points, which is better than the current scheme of 40 points; finally, perform conflict detection to confirm that the λ3 wavelength is not occupied by other services on the BJ→SH path, and generate the optimization scheme after verification: adjust service A from the BJ→TJ→SH path of λ1 wavelength to the BJ→SH direct connection path of λ3 wavelength, which not only solves the delay problem of service A, but also releases the competition pressure of TJ→SH link for service B, and realizes the improvement of the overall network performance.

[0059] Step S640, perform the gradual switching strategy. A three-stage switching process is developed: in the preparation stage, the target service is pre-established on the new wavelength path, end-to-end connectivity test and performance verification are performed, and it is ensured that the new path meets the QoS requirements; in the switching stage, the lossless switching mode of building first and then tearing down is adopted, the service traffic is smoothly migrated from the old path to the new path through the protection switching function of the device, and the switching time is controlled within 50 ms; in the cleaning stage, the old path resources are released after the new path runs stably for 5 minutes, and the network state database and resource management table are updated. The entire process is guaranteed to be atomic through the transaction mechanism, such as automatically rolling back to the original state in case of switching failure, to ensure service continuity and network stability.

[0060] In the second embodiment of the present application, a multi-domain cooperative wavelength allocation scheme can be adopted. First, a cross-domain information exchange protocol is established, a standardized RESTAPI interface and a JSON data format are adopted, and real-time sharing of wavelength occupation states, link quality indicators and available resource information between different management domains is realized. The inter-domain state data is synchronized once every 30 seconds to ensure the visibility of each domain to the whole network resources. Then, a hierarchical distributed decision architecture is designed, and the first three steps of the technical scheme are independently executed in each domain. In the first step of real-time network state monitoring, each domain obtains the wavelength channel occupation state and transmission quality indicators in the domain by querying the wavelength management table and service configuration database of the DCI wavelength II type device in the domain, and shares them with other domains through the cross-domain protocol. In the second step of dynamic traffic demand analysis, each domain uses an ARIMA time series model to analyze the historical traffic data in the domain, receives the bandwidth demand, priority and QoS requirement of the cross-domain service request, and constructs a multi-dimensional matrix containing cross-domain traffic characteristics. In the third step of wavelength resource evaluation, each domain forms a domain allocable wavelength resource pool based on the network state and the whole network traffic characteristics, and reports it to the global coordinator. The global coordinator collects resource information from each domain to execute the fourth step of intelligent wavelength allocation decision. An improved heuristic algorithm is used to consider the cross-domain resource utilization, signal quality and service priority, generate a cross-domain optimal allocation scheme, and issue it to each domain for execution. Then, a cross-domain path calculation and resource reservation mechanism is realized. A segmented path calculation method is adopted, the optimal path in the source domain is calculated first, then the global coordinator calculates the cross-domain path segment based on the wavelength resource pool reported by each domain, and finally the target domain completes the end-to-end path splicing. A two-stage resource reservation protocol is established to ensure the atomic allocation of cross-domain resources. Finally, in the fifth step of dynamic adjustment and optimization, through inter-domain SLA negotiation and resource settlement mechanism, each domain cooperates to optimize the resources according to the performance feedback of the cross-domain service: when the performance of the cross-domain service is detected to be substandard, the incremental optimization algorithm of multi-domain cooperation is triggered, and the distributed simulated annealing algorithm is used to search for cross-domain alternative wavelengths and path combinations while keeping 90% of the service paths in the domain unchanged. Through the inter-domain billing and compensation mechanism, the service quality and fairness in the multi-domain cooperative scenario are ensured. The whole cross-domain cooperation process strictly follows the five-step sequence of the single-domain technical scheme. Through the distributed architecture, the execution range of each step is expanded to the cooperative operation of multiple management domains, realizing the unified optimization and dynamic management of cross-domain wavelength resources.

[0061] The above only describes the preferred embodiments of the present application, and does not limit the patent scope of the present application. Any equivalent structural transformation made according to the content of the present application specification and drawings, or direct / indirect application in other related technical fields within the inventive concept of the present application is included in the patent protection scope of the present application.

[0062] A computer device according to an embodiment of the present disclosure includes a memory and a processor. The memory is configured to store non-transitory computer readable instructions. Specifically, the memory can include one or more computer program products that can include various forms of computer readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM) and / or cache memory, etc. The non-volatile memory may, for example, include read-only memory (ROM), hard disk, flash memory, etc.

[0063] The processor can be a central processing unit (CPU) or other forms of processing units having data processing and / or instruction execution capabilities, and can control other components in the computer device to perform desired functions. In one embodiment of the present disclosure, the processor is configured to execute the computer readable instructions stored in the memory, so that the computer device performs all or part of the steps of the learning achievement prediction method based on learning behavior data mining according to the aforementioned embodiments of the present disclosure.

[0064] Those skilled in the art will understand that, in order to solve the technical problem of how to obtain a good user experience effect, the present embodiment can also include well-known structures such as a communication bus, an interface, etc., which should also be included in the protection scope of the present disclosure.

[0065] As shown in Figure 6 A structural schematic diagram of a computer device according to an embodiment of the present disclosure is shown. It shows a structural schematic diagram of a computer device suitable for implementing the computer device in the embodiments of the present disclosure. Figure 6 The computer device shown is only an example and should not bring any limitation to the functions and use range of the embodiments of the present disclosure.

[0066] As shown in Figure 6 As shown, the computer device can include a processor (such as a central processor, a graphics processor, etc.) that can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) or loaded from a storage device into a random access memory (RAM). In the RAM, various programs and data required for the operation of the computer device are also stored. The processor, the ROM, and the RAM are connected to each other through a bus. An input / output (I / O) interface is also connected to the bus.

[0067] Generally, the following devices can be connected to the I / O interface: input devices, including, for example, sensors or visual information collection devices; output devices, including, for example, display screens; storage devices, including, for example, magnetic tapes, hard disks, etc.; and communication devices. The communication devices can allow the computer device to communicate wirelessly or wired with other devices, such as edge computing devices, to exchange data. Although the computer device is shown with various devices, it should be understood that not all of the shown devices are required to be implemented or possessed. More or fewer devices can be alternatively implemented or possessed.

[0068] In particular, according to embodiments of the present disclosure, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for executing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network through the communication device, or installed from the storage device, or installed from the ROM. When the computer program is executed by the processor, all or part of the steps of the learning achievement prediction method based on learning behavior data mining of the embodiments of the present disclosure are performed.

[0069] Detailed descriptions of the present embodiments can refer to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0070] The computer-readable storage medium according to the embodiments of the present disclosure has non-transitory computer-readable instructions stored thereon. When the non-transitory computer-readable instructions are run by the processor, all or part of the steps of the learning achievement prediction method based on learning behavior data mining of the embodiments of the present disclosure described above are performed.

[0071] The computer-readable storage medium described above includes, but is not limited to, optical storage media (for example, CD-ROM and DVD), magneto-optical storage media (for example, MO), magnetic storage media (for example, magnetic tape or mobile hard disk), media with built-in rewritable non-volatile memory (for example, memory card), and media with built-in ROM (for example, ROM cartridge).

[0072] Detailed descriptions of the present embodiments can refer to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0073] The basic principles of the present disclosure are described above in combination with specific embodiments, but it should be noted that the advantages, advantages, effects, etc. mentioned in the present disclosure are only examples and are not limiting, and these advantages, advantages, effects, etc. cannot be considered as the must-have of each embodiment of the present disclosure. In addition, the above specific details are only for the purpose of example and for the purpose of understanding, and are not limiting, and the above details do not limit the present disclosure to the above specific details.

[0074] In this disclosure, relational terms such as first and second and the like can be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. The illustrative block diagrams of the devices, apparatuses, equipment, systems referred to in this disclosure are shown as block diagrams only and do not necessarily imply a connection, arrangement, configuration as shown therein. As those skilled in the art will appreciate, these devices, apparatuses, equipment, systems can be connected, arranged, configured in any manner as will serve the purpose. The words "comprise," "comprising," "include," "including," and the like can be used synonymously with the word "containing" or "containing" and are therefore to be interpreted in the same manner as the term "including," "including," "containing," or "containing." The word "or" as used in this disclosure is intended to mean at least one of the items it connects. For example, the phrase "A or B" is intended to mean "A or B or both A and B." The words "and" and "or" as used in this disclosure, can be used synonymously with the word "and / or." The word "such as" as used in this disclosure is intended to mean "such as but not limited to."

[0075] In addition, as used in this disclosure, the term "or" as used in the listing of items prefaced by "at least one of" indicates a disjunctive list, e.g., a listing of at least one of A, B, or C means A or B or C or AB or AC or BC or ABC (i.e., A and B and C). Further, the phrase "example of" as used in this disclosure is not meant to be limiting in any way.

[0076] It is also important to note that the systems and methods of the disclosure can be embodied in a variety of forms without departing from the teachings of the disclosure. Such forms include components, steps, locations, and / or a combination thereof adapted in various ways as equivalents of the forms described specifically herein.

[0077] Various changes, modifications and improvements in the herein described technologies can be made within the teachings of the technology, as defined by the appended claims, without departing from the technical teachings of the disclosure. Further, the elements of the disclosure recited in the claims are specified by way of examples only and are not meant to be limiting. Processes, machines, manufacture, compositions of matter, means, methods, or steps, presently existing or later to be developed, that perform substantially the same function as the corresponding processes, machines, manufacture, compositions of matter, means, methods, or steps described herein, could be utilized. Accordingly, the appended claims include within their scope all processes, machines, manufacture, compositions of matter, means, methods, or steps substantially as such.

[0078] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein can be applied to other aspects without departing from the scope of the disclosure. Thus, the present disclosure is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0079] The foregoing description has been presented for the purposes of illustration and description. Furthermore, the description is not intended to limit the embodiments of the disclosure to the forms disclosed herein. Although the various example aspects and embodiments have been described herein with regard to particular aspects and embodiments, those skilled in the art will recognize that certain modifications, changes, substitutions, additions and sub-combinations can be made without departing from the spirit of the disclosure.

Claims

1. A dynamic wavelength allocation method based on DCI wavelength division II equipment, characterized in that: Based on port monitoring data from DCI wavelength division II equipment, the occupancy status and transmission quality indicators of each wavelength channel in the current network are obtained through optical power detection and signal quality analysis. Establish a wavelength channel quality assessment model and score each wavelength channel based on the transmission quality indicators of each wavelength channel in the network; Based on the bandwidth requirements, priority, and QoS requirements of the service requests, a traffic feature matrix containing latency sensitivity and bandwidth requirements is generated through a traffic prediction algorithm. Based on the network status and the traffic feature matrix, an allocable wavelength resource pool and its quality score are formed through resource availability calculation and path quality assessment. For the allocable wavelength resource pool, an improved heuristic algorithm is used to generate the optimal wavelength allocation scheme by comprehensively considering resource utilization, signal quality and service priority. Based on performance feedback from the optimal wavelength allocation results and changes in network state, an adaptive adjustment mechanism is used to continuously optimize wavelength allocation.

2. The method as described in claim 1, characterized in that, The establishment of the wavelength channel quality assessment model specifically includes: A wavelength channel quality assessment formula is established through a weighted scoring mechanism, which weights the transmission quality indicators, transmission distance, equipment aging degree, and environmental factors. The wavelength channel quality assessment formula is trained using a support vector machine (SVM) regression algorithm. During the model training phase, the wavelength channel's signal-to-noise ratio, bit error rate, optical power, dispersion compensation status, transmission distance, and device runtime are used as inputs, and the wavelength channel quality score is used as the output label. The quadratic programming problem of the SVM regression algorithm is solved using the Sequence Minimum Optimization (SMO) algorithm to obtain the support vectors and their corresponding Lagrange multipliers. ; Based on the Lagrange multipliers Construct the wavelength channel quality assessment model ,in Here, b is a Lagrange multiplier, and b is a bias term. For support vectors, The new feature vector, the support vector and feature vector include the signal-to-noise ratio, bit error rate, optical power, dispersion compensation status, transmission distance and device runtime of the parameter wavelength channel.

3. The method as described in claim 2, characterized in that, The step of generating a traffic feature matrix containing latency sensitivity and bandwidth requirements based on the bandwidth demand, priority, and QoS requirements of service requests, using a traffic prediction algorithm, specifically includes: Collect QoS indicator parameters for service requests. These QoS indicator parameters include source and destination node IDs, bandwidth requirements, service priority, maximum tolerable latency, and packet loss rate requirements. Short-term traffic flow trend prediction is performed using the ARIMA time series model combined with historical traffic data. A multidimensional feature vector is constructed based on the predicted short-term traffic change trend; wherein the multidimensional feature vector includes one or more of the following: bandwidth demand value, latency sensitivity, service priority, predicted growth rate, and service type code. The multiple multidimensional feature vectors are combined to form a traffic feature matrix.

4. The method as described in claim 3, characterized in that, The process of forming an allocable wavelength resource pool and its quality score based on network state and traffic characteristic matrices through resource availability calculation and path quality assessment specifically includes: Based on the real-time network topology and the current occupancy status of each wavelength channel in the network, an initial candidate resource set is constructed; Based on transmission distance, wavelength channel signal-to-noise ratio, wavelength channel bit error rate, and device performance parameters, a quality assessment is performed on each candidate wavelength and path combination in the initial candidate resource set to obtain a quality score for each candidate wavelength and path combination. For the initial set of candidate resources, perform resource conflict detection and constraint verification to exclude candidate resources that conflict with the service in terms of wavelength or do not meet QoS requirements; A multi-dimensional resource scoring model is established, which weights and combines the wavelength channel quality, the quality of each candidate wavelength and path combination, load balancing, and historical performance. Based on a multi-dimensional resource scoring model, a standardized resource pool is formed that includes available wavelength resources and a comprehensive quality score for wavelength channels.

5. The method as described in claim 4, characterized in that, For the allocatable wavelength resource pool, an improved heuristic algorithm is used to generate an optimal wavelength allocation scheme by comprehensively considering resource utilization, signal quality, and service priority. Specifically, this includes: For the available wavelength resources in the standardized resource pool, a multi-objective optimization function is constructed based on resource utilization, signal quality, and service priority. Design a wavelength allocation solution framework based on hybrid intelligent optimization, and use intelligent algorithms to find the first wavelength and path allocation scheme for the multi-objective optimization function; For the first wavelength and path allocation scheme, a second wavelength and path allocation scheme is obtained by using a constraint processing mechanism based on the penalty function method to process wavelength conflict constraints, bandwidth capacity constraints and QoS delay constraints. By simulating natural selection and genetic variation mechanisms, a high-quality third wavelength and path allocation scheme is found from the second wavelength and path allocation scheme within a limited number of iterations. Output the third wavelength and path allocation scheme, select the individual with the highest fitness from the third wavelength and path allocation scheme as the optimal solution, and generate a detailed allocation scheme that includes the specific wavelength ID, transmission path node sequence, expected signal quality and resource usage for each service.

6. The method as described in claim 5, characterized in that, The construction of the multi-objective optimization function based on resource utilization, signal quality, and service priority specifically includes: The multi-objective optimization function is: Where U is the resource utilization rate index, Q is the signal quality index, P is the priority satisfaction index, and a, b, and c are coefficients.

7. The method as described in claim 6, characterized in that, The design is based on a hybrid intelligent optimization wavelength allocation solution framework. Specifically, the multi-objective optimization function uses an intelligent algorithm to find the first wavelength and path allocation scheme, including: A three-layer coding structure is constructed to represent the allocation scheme {B,R,W}, where B is the service and wavelength mapping vector, R is the service and path mapping vector, and W is the quality weight vector; Using {B,R,W} as input, the optimal solution is obtained by combining the particle swarm optimization (PSO) algorithm and the simulated annealing (SA) algorithm to obtain the optimized wavelength allocation scheme. For the optimized wavelength allocation scheme, based on the constraint processing strategy and multiple convergence conditions, the first wavelength and path allocation scheme that meets the conditions is output.

8. The method as described in claim 7, characterized in that, The process of obtaining a second wavelength and path allocation scheme based on the first wavelength and path allocation scheme, through a constraint processing mechanism and using a penalty function method to handle wavelength conflict constraints, bandwidth capacity constraints, and QoS delay constraints, specifically includes: Establish a constraint detection function to detect whether the first wavelength and path allocation scheme violate wavelength conflict, bandwidth capacity and QoS delay constraints; When there are illegal constraints between the first wavelength and the path allocation scheme, the multi-objective optimization function is adjusted by a penalty function; The first wavelength and path allocation scheme is repaired by an operator mechanism to obtain the second wavelength and path allocation scheme.

9. The method as described in claim 8, characterized in that, The process of finding a high-quality third wavelength and path allocation scheme from the second wavelength and path allocation scheme within a limited number of iterations by simulating natural selection and genetic variation mechanisms specifically includes: The iterative optimization process is executed, with a maximum number of iterations of 500 generations. The process terminates early when the optimal solution does not improve or reaches the preset convergence accuracy after 50 consecutive generations. Each generation of evolution includes four operations: selection, crossover, mutation, and environmental selection. A tournament selection strategy is used to maintain population diversity.

10. The method as described in claim 1, characterized in that, Before obtaining the occupancy status and transmission quality indicators of each wavelength channel in the current network through optical power detection and signal quality analysis, the following may also be included: Establish a cross-domain information exchange protocol, using standardized REST API interfaces and JSON data format, to exchange wavelength occupancy status, link quality indicators, and available resource information between different management domains, ensuring the visibility of network resources to each domain.

Images