A control method and device of a router

Abstract

The application provides a router control method and device. First, based on network operation data of the router, the priority of each influence factor is determined. Then, according to the priority from high to low, the index data corresponding to the influence factor of the corresponding priority is used to predict the network state of the router within a preset time period. Until the router includes at least one initial abnormal node is predicted, the initial bandwidth expansion strategy of the router is configured based on the initial abnormal node. If it is determined that any associated abnormal node corresponding to the initial abnormal node is included, the initial bandwidth expansion strategy is updated, and the target bandwidth expansion strategy obtained by the update is used to control the bandwidth expansion of the router. Through the establishment of intelligent hierarchical prediction and strategy optimization mechanism, the application realizes accurate positioning and forward disposition of network anomalies, overcomes the hysteresis of traditional manual intervention, solves the important application lag problem during the network peak period, and improves the network transmission efficiency and user experience.

Description

Technical Field

[0001] This invention relates to the field of network communication, and in particular to a control method and apparatus for a router. Background Technology

[0002] With the acceleration of digitalization, the number of smart devices in enterprise and home networks has surged, placing higher performance demands on routers, the core of wireless networks. For example, users require the network to remain stable during peak hours to ensure a smooth experience for critical services such as video conferencing, online collaboration, and ultra-high-definition streaming media.

[0003] However, existing performance optimization methods typically involve users manually adjusting router parameters such as channel and transmit power based on the router's default settings or some basic network knowledge. However, this optimization method, which relies on manual intervention, is inherently lagging. Therefore, during peak network hours, important applications are prone to lag, which severely reduces network transmission efficiency and user experience. Summary of the Invention

[0004] This application provides a router control method and apparatus to solve the problem in the prior art where routers rely on manual intervention and static configuration, resulting in lagging network optimization and low transmission efficiency.

[0005] Firstly, this application provides a router control method, including:

[0006] Obtain network operation data of the router, the network operation data including: indicator data corresponding to at least one influencing factor related to the router's network status;

[0007] Based on the network operation data, the priority of each of the at least one influencing factor is determined;

[0008] In descending order of priority, the corresponding index data of the influencing factors of the corresponding priority are used to predict the network status of the router within a preset time period until it is predicted that the router includes at least one initial abnormal node. Then, the initial bandwidth expansion strategy of the router is configured according to the at least one initial abnormal node.

[0009] If it is determined that the router also includes any associated abnormal node corresponding to the initial abnormal node, then the initial bandwidth expansion strategy is updated based on each determined associated abnormal node, and the updated target bandwidth expansion strategy is used to control the router to expand bandwidth.

[0010] In one possible implementation, determining the priority of each of the at least one influencing factor based on the network operation data includes:

[0011] For each influencing factor, the corresponding indicator data is normalized; based on the normalized indicator data, the corresponding evaluation value is calculated.

[0012] Based on the obtained evaluation values, the priority of each of the at least one influencing factor is determined by the following method:

[0013] Based on the preset threshold range to which each evaluation value belongs, the priority of the influencing factors corresponding to the evaluation value is determined; or

[0014] Clustering analysis is performed on each of the evaluation values ​​using a clustering algorithm, and the priority of each of the at least one influencing factor is determined based on the clustering results.

[0015] In one possible implementation, the priority includes high priority, medium priority, and low priority, and determining the priority of the influencing factors corresponding to each evaluation value based on a preset threshold range to which each evaluation value belongs includes:

[0016] For each evaluation value, perform the following operations:

[0017] If the evaluation value falls within the first preset threshold range, the influencing factor corresponding to the evaluation value is taken as the dominant factor, and the priority of the influencing factor is determined as the high priority.

[0018] If the evaluation value falls within the second preset threshold range, the influencing factor corresponding to the evaluation value is taken as a secondary factor, and the priority of the influencing factor is determined as the medium priority.

[0019] If the evaluation value falls within the third preset threshold range, the influencing factor corresponding to the evaluation value is treated as a regular factor, and the priority of the influencing factor is determined to be the low priority.

[0020] In one possible implementation, the priorities include high priority, medium priority, and low priority. The step of performing cluster analysis on each of the evaluation values ​​using a clustering algorithm, and determining the priority of each of the at least one influencing factor based on the clustering results, includes:

[0021] Cluster analysis was performed on the evaluation values ​​of all influencing factors using a clustering algorithm to obtain three cluster centers and the cluster group to which each influencing factor belongs;

[0022] Compare the numerical values ​​of the three cluster centers;

[0023] The influencing factors in the cluster group corresponding to the cluster center with the largest value are taken as the dominant factors, and the priority of the influencing factors is determined as the high priority.

[0024] The influencing factors in the cluster group corresponding to the cluster center with the smallest value are taken as regular factors, and the influencing factors are determined.

[0025] The priority is the low priority;

[0026] The influencing factors in the remaining clustering groups are taken as secondary factors, and the priority of the influencing factors is determined as the medium priority.

[0027] In one possible implementation, the network status of the router within a preset time period is predicted by the following method:

[0028] The corresponding priority indicator data are input into a pre-trained behavior prediction model to output the probability distribution of the router's abnormal state within a preset time period. Based on the probability distribution of the abnormal state, the network state of the router is predicted, or...

[0029] The corresponding priority indicator data is matched with the historical anomaly database, and based on the generated matching results, the network status of the router within a preset time period is predicted.

[0030] In one possible implementation, predicting the network state of the router based on the abnormal state probability distribution includes:

[0031] Based on the probability distribution of abnormal states at k consecutive time points within the preset time period, after identifying the target time point where the probability of an anomaly occurrence is greater than or equal to the anomaly threshold, it is determined that the router includes at least one initial abnormal node, where k is a positive integer greater than or equal to 1.

[0032] The step of predicting the network status of the router within a preset time period based on the generated matching results includes:

[0033] If the matching result is successful, it is determined that the router includes at least one initial abnormal node, wherein the historical abnormal database includes the correspondence between historical indicator data and abnormal states.

[0034] In one possible implementation, determining that the router further includes an associated abnormal node corresponding to any of the initial abnormal nodes includes:

[0035] Based on historical network operation data, the correlation strength value between the initial abnormal node and other network nodes is calculated using a causal discovery algorithm.

[0036] Other network nodes whose association strength value is greater than or equal to the preset association strength threshold are identified as the associated abnormal nodes corresponding to the initial abnormal node.

[0037] The other network nodes include network devices that are in the same network as the router and have an interactive relationship with it.

[0038] In one possible implementation, the method further includes:

[0039] If the router does not include any associated abnormal node corresponding to the initial abnormal node, then the initial bandwidth expansion strategy is used to control the router to expand its bandwidth.

[0040] In one possible implementation, after the router performs bandwidth expansion, the method further includes:

[0041] Based on the collected network status data, network performance parameters, and bandwidth expansion strategies of the router, the configuration rules of the bandwidth expansion strategy are optimized through reinforcement learning.

[0042] The network status data includes the evaluation value of at least one influencing factor in the network operation data, and the network performance parameters include at least one of network transmission rate, network transmission latency, and packet loss rate.

[0043] Secondly, this application provides a router control device, comprising:

[0044] The acquisition module is used to acquire network operation data of the router, the network operation data including: indicator data corresponding to at least one influencing factor related to the router's network status;

[0045] The processing module is configured to determine the priority of each of the at least one influencing factor based on the network operation data; predict the network status of the router within a preset time period by sequentially using the index data corresponding to the influencing factors of the corresponding priority in descending order of priority, until it is predicted that the router includes at least one initial abnormal node; configure the initial bandwidth expansion strategy of the router according to the at least one initial abnormal node; if it is determined that the router also includes any associated abnormal node corresponding to the initial abnormal node, update the initial bandwidth expansion strategy based on the determined associated abnormal nodes.

[0046] The control module is used to control the router to expand bandwidth using the updated target bandwidth expansion strategy.

[0047] The beneficial effects of this application are as follows:

[0048] This application provides a router control method and apparatus. First, based on the router's network operation data, the priority of at least one influencing factor is determined. Then, according to the priority from high to low, the corresponding indicator data of the influencing factors are used sequentially to predict the router's network status within a preset time period until it is predicted that the router includes at least one initial abnormal node. Then, an initial bandwidth expansion strategy for the router is configured based on the initial abnormal node. If it is determined that any associated abnormal node corresponding to the initial abnormal node is also included, the initial bandwidth expansion strategy is updated, and the updated target bandwidth expansion strategy is used to control the router to expand its bandwidth. This method, by establishing an intelligent hierarchical prediction and strategy optimization mechanism, achieves a shift from passive response to proactive prevention. Compared to traditional optimization methods that rely on manual intervention, this method can quickly identify and handle network anomalies, effectively solving the problem of important applications lagging during peak network periods due to delays in manual operation. Simultaneously, through proactive protection of associated abnormal nodes, a comprehensive network quality assurance system is established, thereby improving network transmission efficiency and enhancing user experience. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 A flowchart illustrating a router control method provided in an embodiment of this application;

[0051] Figure 2 This is a schematic diagram of the structure of a router control device provided in an embodiment of this application;

[0052] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0054] It should be noted that the terms "first," "second," etc., used in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application.

[0055] With the deepening of digitalization, the scale of networked devices in various organizations has experienced explosive growth, increasing 10 to 100 times compared to ten years ago. Although network operation and maintenance models have gradually evolved from manual operation and maintenance to tool-based and platform-based approaches, truly intelligent operation and maintenance systems have not yet been implemented. In today's society, which heavily relies on wireless networks, routers, as core network access devices, directly determine the quality of users' network experience based on their performance. With the rapid popularization of smart terminal devices and the increasing complexity of network application scenarios, users are placing increasingly stringent requirements on wireless routers in terms of signal coverage, transmission speed, and connection stability.

[0056] However, existing performance optimization methods mainly rely on traditional network parameter settings and basic frequency band adjustments. The common practice is for users to manually adjust parameters such as channels and transmit power based on the router's default settings or limited network knowledge. This manual intervention-based optimization method has significant lag, leading to transmission delays in important applications during peak network usage periods. This severely reduces network transmission efficiency and user experience, making it difficult for wireless routers to meet the diverse needs of modern network applications.

[0057] To address the aforementioned problems, this application provides a router control method and apparatus. For ease of understanding, the router control method and apparatus provided in this application will be described in detail below with reference to the accompanying drawings.

[0058] like Figure 1 The image shows a router control method provided in an embodiment of this application. The method includes:

[0059] S101: Obtain the router's network operation data, which includes: indicator data corresponding to at least one influencing factor related to the router's network status;

[0060] S102: Based on network operation data, determine the priority of at least one influencing factor;

[0061] S103: In order of priority from high to low, the corresponding index data of the influencing factors of the corresponding priority are used to predict the network status of the router within a preset time period until it is predicted that the router includes at least one initial abnormal node. Then, the initial bandwidth expansion strategy of the router is configured according to the at least one initial abnormal node.

[0062] S104: If it is determined that the router also includes any associated abnormal node corresponding to the initial abnormal node, then the initial bandwidth expansion policy is updated based on the determined associated abnormal nodes, and the updated target bandwidth expansion policy is used to control the router to expand the bandwidth.

[0063] This application provides a router control method. First, the priority of at least one influencing factor is determined. Then, according to the priority from high to low, the corresponding index data of the influencing factors are used to predict the network status of the router within a preset time period until it is predicted that the router includes at least one initial abnormal node. Based on the initial abnormal node, the router's initial bandwidth expansion strategy is configured. If it is determined that there are also associated abnormal nodes corresponding to any initial abnormal node, the initial bandwidth expansion strategy is updated, and the updated target bandwidth expansion strategy is used to control the router to expand bandwidth. This application realizes the transformation from passive response to proactive prevention by establishing an intelligent hierarchical prediction and strategy optimization mechanism. Compared with the traditional optimization method that relies on manual intervention, the above method can quickly identify and handle network anomalies, effectively solving the problem of network lag during peak hours caused by the lag of manual operation. At the same time, through the proactive protection of associated abnormal nodes, a comprehensive network quality assurance system is established, thereby improving network transmission efficiency and enhancing user experience.

[0064] In this embodiment of the application, the router's network operation data includes indicator data corresponding to at least one influencing factor closely related to the router's network status, specifically including but not limited to the following:

[0065] Data transmission metrics: such as total data volume transmitted, real-time uplink and downlink rates, and data packet transmission success rate;

[0066] Connectivity metrics: such as the number of devices connected simultaneously, the distribution of device types, and device signal strength;

[0067] Bandwidth utilization metrics: such as bandwidth utilization rate of each frequency band, channel utilization rate, and bandwidth allocation efficiency;

[0068] Network traffic metrics: such as traffic time-series distribution, peak traffic characteristics, and abnormal traffic patterns;

[0069] Network performance metrics include network latency, packet loss rate, and signal interference intensity.

[0070] The aforementioned data is continuously collected through various means such as router logs, network protocol analysis, and device status monitoring to form network operation data. After being timestamped and standardized in format, the collected data is stored in a dedicated database, providing a data foundation for subsequent analysis of influencing factors and priority determination.

[0071] In one embodiment, in step S102, the priority of at least one influencing factor is determined by the following method:

[0072] For each influencing factor, the corresponding indicator data is normalized and converted into standard values ​​in the range of [0, 1]. Based on the processed indicator data, the corresponding evaluation value of each influencing factor is calculated.

[0073] For example, collect network operation data of the router during a preset working time period, including indicator data corresponding to the total amount of data transmission, indicator data corresponding to the number of connected devices, indicator data corresponding to the bandwidth utilization rate of each frequency band, indicator data corresponding to network traffic, and indicator data corresponding to network busyness.

[0074] Using a linear normalization method, the evaluation value corresponding to each influencing factor in the above network operation data is calculated:

[0075] F1 represents the evaluation value of the total amount of data transmitted;

[0076] F2 represents the evaluation value of the number of connected devices;

[0077] F3 represents the evaluation value of bandwidth utilization for each frequency band;

[0078] Where F4 represents the evaluation value of network traffic;

[0079] F5 represents the assessment value of network busyness;

[0080] It should be noted that the normalization method used for the corresponding index data of influencing factors in this application is not specifically limited. In addition to the linear normalization method described in the embodiments, various data standardization methods such as Z-score standardization, decimal scaling normalization, and logarithmic function transformation can also be used. Those skilled in the art can select appropriate normalization methods according to the actual data characteristics and application scenarios, and these transformation methods are all within the protection scope of this application.

[0081] Based on the obtained evaluation values, the priority of at least one influencing factor is determined using the following method, where the priority is categorized as high, medium, and low:

[0082] The first method:

[0083] Based on the preset threshold range to which each evaluation value belongs, the priority of the influencing factors corresponding to the evaluation value is determined;

[0084] Specifically, for each evaluation value, the following operations are performed:

[0085] If the evaluation value falls within the first preset threshold range, the influencing factor corresponding to the evaluation value will be taken as the dominant factor, and the priority of the influencing factor will be determined as high priority.

[0086] If the evaluation value falls within the second preset threshold range, the influencing factor corresponding to the evaluation value will be taken as a secondary factor, and the priority of the influencing factor will be determined as medium priority.

[0087] If the evaluation value falls within the third preset threshold range, the influencing factors corresponding to the evaluation value will be treated as regular factors, and the priority of the influencing factors will be determined as low priority.

[0088] The numerical ranges of the first preset threshold range, the second preset threshold range, and the third preset threshold range decrease sequentially.

[0089] For example, if the evaluation value F1 for the total data transmission is determined to be 0.9, the evaluation value F2 for the number of connected devices is 0.3, the evaluation value F3 for the bandwidth utilization rate of each frequency band is 0.95, the evaluation value F4 for network traffic is 0.35, the evaluation value F5 for network congestion is 0.65, and the first preset threshold range is (0.75, 1], the second preset threshold range is (0.5, 0.75], and the third preset threshold range is (0.0, 0.5];

[0090] The evaluation values ​​falling within the first preset threshold range include the total data transmission volume evaluation value F1 and the bandwidth utilization rate of each frequency band evaluation value F3. The evaluation values ​​falling within the second preset threshold range include the network congestion evaluation value F5. The evaluation values ​​falling within the third preset threshold range include the number of connected devices evaluation value F2 and the network traffic evaluation value F4. That is, the total data transmission volume and the bandwidth utilization rate of each frequency band are the dominant factors with high priority, the network congestion is the secondary factor with medium priority, and the number of connected devices and the network traffic are the conventional factors with low priority.

[0091] The second method:

[0092] Clustering algorithms are used to perform cluster analysis on each evaluation value, and the priority of at least one influencing factor is determined based on the clustering results.

[0093] Specifically, cluster analysis is performed on the evaluation values ​​of all influencing factors using a clustering algorithm to obtain three cluster centers and the cluster group to which each influencing factor belongs;

[0094] Compare the numerical values ​​of the three cluster centers;

[0095] The influencing factors in the cluster group corresponding to the cluster center with the largest value are taken as the dominant factors, and the priority of the influencing factors is determined to be high priority;

[0096] The influencing factors in the cluster group corresponding to the cluster center with the smallest value are taken as regular factors, and the priority of the influencing factors is determined to be low priority.

[0097] The influencing factors in the remaining cluster groups are taken as secondary factors, and the priority of the influencing factors is determined as medium priority.

[0098] For example, as described in the example above, the evaluation value F1 for the total data transmission is 0.9, the evaluation value F2 for the number of connected devices is 0.3, the evaluation value F3 for the bandwidth utilization of each frequency band is 0.95, the evaluation value F4 for network traffic is 0.35, the evaluation value F5 for network congestion is 0.65, and the K-means++ algorithm is used for cluster analysis.

[0099] Finally, the analysis results show that the first cluster group includes the evaluation value F1 of the total data transmission volume and the evaluation value F3 of the bandwidth utilization rate of each frequency band, with a cluster center value of 0.925; the second cluster group includes the evaluation value F2 of the number of connected devices and the evaluation value F4 of the network traffic, with a cluster center value of 0.325; and the third cluster group includes the evaluation value F5 of the network busyness, with a cluster center value of 0.65.

[0100] The cluster center value of the first cluster group is greater than that of the second cluster group, which is greater than that of the third cluster group. The influencing factors in the first cluster group are taken as the dominant factors, namely, the total amount of data transmission and the bandwidth utilization rate of each frequency band are both dominant factors, with a high priority. The influencing factors in the second cluster group are taken as secondary factors, namely, the number of connected devices and network traffic are both secondary factors, with a medium priority. The influencing factors in the third cluster group are taken as normal factors, namely, the network busyness is a normal factor, with a low priority.

[0101] In one embodiment, in step S103, the index data corresponding to the influencing factors of the corresponding priorities are used sequentially in descending order of priority, and the network status of the router within a preset time period is predicted by the following method:

[0102] The first method:

[0103] The corresponding priority index data is input into the pre-trained behavior prediction model to output the abnormal state probability distribution of the router within a preset time period. Based on the abnormal state probability distribution, the network state of the router is predicted.

[0104] The pre-trained behavior prediction model in this application is the NextK-BehaviorsPrediction (NBP) model. This model outputs the probability distribution of abnormal states of the router at k consecutive time points within a preset time period, where k is a positive integer greater than or equal to 1.

[0105] The NBP model is trained using the cross-entropy loss function, which optimizes the model parameters by minimizing the difference between the predicted probability and the true probability, as shown in the following formula:

[0106] Formula 1

[0107] Wherein, δ2 represents the training dataset generated based on historical network operation data of the router. To define the true labels in the training dataset, representing the probability that the k-th subsequent action is action i, The probability of the k-th subsequent action predicted by the model is action i, K represents the number of predicted actions, and p represents the number of negative samples, which is used to enhance the robustness of the model.

[0108] After the NBP model is trained, the current priority indicator data is organized into a model input sequence according to time series characteristics. The NBP model is used to process and analyze the input sequence to obtain the probability values ​​of the router's abnormal state at the next k time points, thus forming an abnormal state probability distribution.

[0109] Based on a preset anomaly probability threshold, determine whether the probability of an anomaly occurring is greater than or equal to the target time point of the preset anomaly probability threshold;

[0110] When at least one target time point is identified, it is determined that the router contains at least one initial abnormal node.

[0111] The abnormal state probability distribution reflects the likelihood of network anomalies occurring at various points in the future. The system accurately identifies abnormal states by comparing probability thresholds. When the initial abnormal node cannot be predicted based on the current priority data, the system automatically switches to the next priority influencing factor data and continues the prediction process until the initial abnormal node is successfully predicted or all priority data has been traversed.

[0112] For example, the bandwidth utilization rate of each frequency band is a primary factor with high priority. The network operation data corresponding to the bandwidth utilization rate of each frequency band in the preset time period is input into the NBP model, as shown in Table 1. At time t0, the bandwidth utilization rate of each frequency band of the router is 65%; at time t1, it is 68%; at time t2, it is 72%; at time t3, it is 85%; at time t4, it is 88%; at time t5, it is 92%; at time t6, it is 95%; and at time t7, it is 96%.

[0113] Table 1

[0114]

[0115] The Z-score normalization method is used to standardize the above index data. The processed feature vector sequence is then passed through the Long Short-Term Memory (LSTM) encoder of the NBP model to extract deep temporal features.

[0116] It was identified that the bandwidth utilization rate showed a significant upward trend starting from time t3;

[0117] The accelerating growth rate indicates that abnormal risks are accumulating;

[0118] It was identified that the bandwidth utilization rate at time t7 (96%) was close to the historical extreme level;

[0119] The LSTM encoder generates a context vector based on the processed feature vector sequence and deep temporal features. Then, based on the context vector and the attention mechanism, it generates the hidden state at the current time step. Finally, through a fully connected layer and a sigmoid activation function, the hidden state is transformed into anomaly probability values.

[0120] If the model outputs the probability distribution of the abnormal state of the router at three consecutive time points within a preset time period, then the final probability distribution of the abnormal state at the next three time points is as follows: the probability of the abnormal state at predicted time point t8 is 0.85, the probability of the abnormal state at predicted time point t9 is 0.92, and the probability of the abnormal state at predicted time point t10 is 0.96.

[0121] If the preset anomaly probability threshold is 0.80, the anomaly probability of each predicted time point is compared with the threshold. It is then determined that the predicted time points t8, t9, and t10 are all greater than the anomaly probability threshold. The predicted time points t8, t9, and t10 are all taken as target time points. It is determined that the router contains at least one initial abnormal node, which is the bandwidth overload anomaly that occurs in the future time period [t8, t10].

[0122] Based on this determination, the initial bandwidth expansion strategy configuration process is initiated to pre-allocate bandwidth resources to the identified abnormal nodes, thereby completing the deployment of protective measures before the actual anomaly occurs and realizing the transformation from a passive response to a proactive prevention operation and maintenance mode.

[0123] The second method:

[0124] The corresponding priority indicator data is matched with the historical anomaly database. Based on the generated matching results, the network status of the router within a preset time period is predicted. The historical anomaly database includes the correspondence between historical indicator data and anomaly status.

[0125] If the similarity between the indicator data and one or more historical indicator data in the historical anomaly database is greater than or equal to the preset similarity threshold, the matching result is determined to be a successful match, indicating that the router contains at least one initial abnormal node, which means that the router has entered a state that is highly similar to the early stage of the historical abnormal event, and its network status is highly likely to become abnormal within the preset time period.

[0126] If the similarity between the indicator data and all historical indicator data in the historical anomaly database is less than the preset similarity threshold, the matching result is determined to be a matching failure, indicating that the router's network status is normal within the preset time period.

[0127] The method described above, by identifying initial abnormal nodes highly similar to historical anomaly patterns, can issue early warnings before faults have a substantial impact on services. This transforms the operation and maintenance mode from passive response to proactive intervention, effectively shortening the mean time to repair (MTBL). Furthermore, matching based on a validated historical anomaly database effectively filters out transient noise interference, reducing false alarm rates. In addition, compared to training complex prediction models, the matching process has low computational overhead and low response latency, making it suitable for real-time or near-real-time prediction in resource-constrained routers or large-scale network management systems.

[0128] In this embodiment of the application, after predicting that the router includes at least one initial abnormal node, the initial bandwidth expansion strategy of the router is configured according to the initial abnormal node, specifically including the following steps:

[0129] First, identify one or more service flows involved in the initial abnormal node and obtain the preset priority of each service flow. Then, analyze the bandwidth requirements of each service flow, which are determined comprehensively based on historical transmission modes, real-time service types, and service quality requirements. Next, calculate the resource allocation weight of each service based on service priority and bandwidth requirements, with higher priority services receiving larger resource allocation weights. Finally, allocate corresponding bandwidth resources to services of different priorities according to the resource allocation weights and configure the initial bandwidth expansion strategy.

[0130] In practice, the weight of business resource allocation is calculated using the following formula:

[0131] Formula 2

[0132] in, The resource allocation weight represents the i-th business. This represents the bandwidth requirement of the i-th service. The priority coefficient representing the i-th service (the higher the priority, the smaller the coefficient value).

[0133] By employing the above methods, we can ensure that high-priority services (such as video conferencing and real-time gaming) receive sufficient bandwidth guarantees during network anomalies, while reducing resource allocation for low-priority services (such as file downloads and software updates). This approach guarantees the service quality of critical services, effectively improves user experience, and provides a basic configuration scheme for subsequent strategy optimization.

[0134] In one embodiment, in step S104, it is determined whether the router includes an associated abnormal node corresponding to any initial abnormal node by the following method:

[0135] A causal discovery algorithm is used to calculate the correlation strength between the initial anomalous node and other network nodes. These other network nodes include network devices that are in the same network as the router and have interactive relationships, such as switches, gateways, wireless access points, secondary routers, and network firewalls.

[0136] In this embodiment, the causal discovery algorithm uses the Hawkes process model to calculate the association strength value between nodes using the following formula:

[0137] Formula 3

[0138] in, Characterizes the occurrence rate of basic events, reflecting the inherent frequency of anomalies in the network environment. The excitation coefficient characterizes the intensity of the excitation effect of historical anomalous events on the current moment. Characterizing the decay coefficient, which controls the rate at which the influence of historical anomalous events decays. The time of occurrence of the i-th historical anomalous event is represented by t, where t represents the current observation time. It represents the intensity of the abnormal event at the current moment.

[0139] It should be noted that the basic event occurrence rate Incentive coefficient attenuation coefficient and the time of occurrence of the i-th historical anomaly It is determined based on historical network operation data. Specifically, it is obtained by statistically analyzing the abnormal event sequences in the historical network operation data and fitting the data using the maximum likelihood estimation method. , , The parameter values ​​reflect the spatiotemporal distribution characteristics of historical anomalous events in the network environment.

[0140] In obtaining the correlation strength value Then, it is compared with a preset association strength threshold, which is determined by statistical analysis of verified anomaly propagation cases in historical network operation data, and can effectively distinguish between real associations and random co-occurrences. Network devices with association strength values ​​greater than or equal to this threshold are identified as associated anomaly nodes corresponding to the initial anomaly node, thereby establishing an accurate anomaly propagation path map and providing a reliable basis for subsequent bandwidth expansion strategy optimization.

[0141] In this embodiment of the application, after determining that the router also includes associated abnormal nodes corresponding to the initial abnormal node, the initial bandwidth expansion strategy is updated based on each determined associated abnormal node.

[0142] Specifically, the initial bandwidth expansion strategy is dynamically adjusted based on the number and type of associated abnormal nodes and their topological relationship with the initial abnormal node to form a target bandwidth expansion strategy. For example, the scale of bandwidth expansion is adjusted based on the number of associated abnormal nodes, or the priority of bandwidth resource allocation is determined based on the type of associated abnormal nodes, or the timing of bandwidth expansion is determined based on the topological relationship between associated abnormal nodes and the initial abnormal node.

[0143] After obtaining the target bandwidth expansion strategy, the router is controlled to expand the bandwidth using the target bandwidth expansion strategy. In the specific execution process, while allocating expansion resources to the initial abnormal node, spare resources are pre-allocated to associated abnormal nodes or they are placed in a high-priority monitoring queue to achieve rapid response when associated abnormal nodes trigger abnormal alarms.

[0144] The above methods ensure timely handling of initial abnormal nodes and proactive protection of associated abnormal nodes, thus constructing a comprehensive network anomaly handling mechanism.

[0145] In another embodiment, if the router does not include any associated abnormal node corresponding to any initial abnormal node, then the initial bandwidth expansion strategy is used to control the router to expand its bandwidth.

[0146] In one embodiment, after the router expands its bandwidth, it enters the policy optimization phase. Based on the collected network status data, network performance parameters, and the bandwidth expansion policy adopted by the router, the configuration rules of the bandwidth expansion policy are optimized through a reinforcement learning mechanism.

[0147] In this embodiment, the learning mechanism adopts the Q-learning algorithm, taking network state data as state input, wherein the network state data includes the evaluation value of at least one influencing factor in the network operation data; taking bandwidth expansion strategy as executable action; and taking the improvement effect of network performance parameters as reward signal, wherein the network performance parameters include at least one of network transmission rate, network transmission latency, and packet loss rate.

[0148] Through continuous iterative learning, the system can autonomously explore and accumulate experience in implementing various bandwidth expansion strategies under different network conditions, thereby gradually optimizing strategy configuration rules. When encountering similar network conditions subsequently, it can automatically select the bandwidth expansion strategy with the best expected effect, achieving continuous improvement in network resource allocation efficiency.

[0149] Based on the same inventive concept, this application also provides a router control device. The implementation principle of the router control device is similar to that of the router control method. The specific implementation of the router control device can be found in the aforementioned router control method embodiments, and repeated details will not be described again.

[0150] like Figure 2 The diagram shown is a schematic representation of a router control device according to an embodiment of this application, comprising:

[0151] The acquisition module 21 is used to acquire the network operation data of the router, the network operation data including: index data corresponding to at least one influencing factor related to the network status of the router;

[0152] Processing module 22 is configured to determine the priority of each of the at least one influencing factor based on the network operation data; predict the network status of the router within a preset time period by sequentially using the index data corresponding to the influencing factors of the corresponding priority in descending order of priority, until it is predicted that the router includes at least one initial abnormal node; configure the initial bandwidth expansion strategy of the router according to the at least one initial abnormal node; if it is determined that the router also includes any associated abnormal node corresponding to the initial abnormal node, update the initial bandwidth expansion strategy based on the determined associated abnormal nodes.

[0153] Control module 23 is used to control the router to expand bandwidth using the updated target bandwidth expansion strategy.

[0154] In one embodiment, the processing module 22 is specifically used for:

[0155] For each influencing factor, the corresponding indicator data is normalized; based on the normalized indicator data, the corresponding evaluation value is calculated.

[0156] Based on the obtained evaluation values, the priority of each of the at least one influencing factor is determined by the following method:

[0157] Based on the preset threshold range to which each evaluation value belongs, the priority of the influencing factors corresponding to the evaluation value is determined; or

[0158] Clustering analysis is performed on each of the evaluation values ​​using a clustering algorithm, and the priority of each of the at least one influencing factor is determined based on the clustering results.

[0159] In one embodiment, the priority includes high priority, medium priority, and low priority, and the processing module 22 is specifically used for:

[0160] For each evaluation value, perform the following operations:

[0161] If the evaluation value falls within the first preset threshold range, the influencing factor corresponding to the evaluation value is taken as the dominant factor, and the priority of the influencing factor is determined as the high priority.

[0162] If the evaluation value falls within the second preset threshold range, the influencing factor corresponding to the evaluation value is taken as a secondary factor, and the priority of the influencing factor is determined as the medium priority.

[0163] If the evaluation value falls within the third preset threshold range, the influencing factor corresponding to the evaluation value is treated as a regular factor, and the priority of the influencing factor is determined to be the low priority.

[0164] In one embodiment, the priority includes high priority, medium priority, and low priority, and the processing module 22 is specifically used for:

[0165] Cluster analysis was performed on the evaluation values ​​of all influencing factors using a clustering algorithm to obtain three cluster centers and the cluster group to which each influencing factor belongs;

[0166] Compare the numerical values ​​of the three cluster centers;

[0167] The influencing factors in the cluster group corresponding to the cluster center with the largest value are taken as the dominant factors, and the priority of the influencing factors is determined as the high priority.

[0168] The influencing factors in the cluster group corresponding to the cluster center with the smallest value are taken as regular factors, and the priority of the influencing factors is determined as the low priority.

[0169] The influencing factors in the remaining clustering groups are taken as secondary factors, and the priority of the influencing factors is determined as the medium priority.

[0170] In one embodiment, the processing module 22 predicts the network status of the router within a preset time period using the following method:

[0171] The corresponding priority indicator data are input into a pre-trained behavior prediction model to output the probability distribution of the router's abnormal state within a preset time period. Based on the probability distribution of the abnormal state, the network state of the router is predicted, or...

[0172] The corresponding priority indicator data is matched with the historical anomaly database, and based on the generated matching results, the network status of the router within a preset time period is predicted.

[0173] In one embodiment, the processing module 22 is specifically used for:

[0174] Based on the probability distribution of abnormal states at k consecutive time points within the preset time period, after identifying the target time point where the probability of an anomaly occurrence is greater than or equal to the anomaly threshold, it is determined that the router includes at least one initial abnormal node, where k is a positive integer greater than or equal to 1.

[0175] The step of predicting the network status of the router within a preset time period based on the generated matching results includes:

[0176] If the matching result is successful, it is determined that the router includes at least one initial abnormal node, wherein the historical abnormal database includes the correspondence between historical indicator data and abnormal states.

[0177] In one embodiment, the processing module 22 is specifically used for:

[0178] Based on historical network operation data, the correlation strength value between the initial abnormal node and other network nodes is calculated using a causal discovery algorithm.

[0179] Other network nodes whose association strength value is greater than or equal to the preset association strength threshold are identified as the associated abnormal nodes corresponding to the initial abnormal node.

[0180] The other network nodes include network devices that are in the same network as the router and have an interactive relationship with it.

[0181] In one embodiment, the control module 23 is further configured to:

[0182] If the router does not include any associated abnormal node corresponding to the initial abnormal node, then the initial bandwidth expansion strategy is used to control the router to expand its bandwidth.

[0183] In one embodiment, the processing module 22 is further configured to:

[0184] Based on the collected network status data, network performance parameters, and bandwidth expansion strategies of the router, the configuration rules of the bandwidth expansion strategy are optimized through reinforcement learning.

[0185] The network status data includes the evaluation value of at least one influencing factor in the network operation data, and the network performance parameters include at least one of network transmission rate, network transmission latency, and packet loss rate.

[0186] Based on the same inventive concept, the present application also provides an electronic device. The implementation principle of the electronic device is similar to that of the aforementioned router control method. The specific implementation method of the electronic device can be found in the aforementioned router control method embodiment, and the repeated parts will not be described again.

[0187] like Figure 3 The diagram shown is a structural schematic of an electronic device provided in an embodiment of this application, comprising:

[0188] Memory 31 is used to store program instructions;

[0189] The processor 32 is used to call program instructions stored in the memory and execute the steps included in the aforementioned router control method according to the obtained program instructions.

[0190] This application provides a router control method and apparatus. First, based on the router's network operation data, the priority of at least one influencing factor is determined. Then, according to the priority from high to low, the corresponding indicator data of the influencing factors are used sequentially to predict the router's network status within a preset time period until it is predicted that the router includes at least one initial abnormal node. Then, an initial bandwidth expansion strategy for the router is configured based on the initial abnormal node. If it is determined that any associated abnormal node corresponding to the initial abnormal node is also included, the initial bandwidth expansion strategy is updated, and the updated target bandwidth expansion strategy is used to control the router to expand its bandwidth. This method, by establishing an intelligent hierarchical prediction and strategy optimization mechanism, achieves a shift from passive response to proactive prevention. Compared to traditional optimization methods that rely on manual intervention, this method can quickly identify and handle network anomalies, effectively solving the problem of important applications lagging during peak network periods due to delays in manual operation. Simultaneously, through proactive protection of associated abnormal nodes, a comprehensive network quality assurance system is established, thereby improving network transmission efficiency and enhancing user experience.

[0191] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0192] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0193] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1The function specified in one or more boxes.

[0194] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0195] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the equivalent technology of this application, this application also intends to include such modifications and variations.

Claims

1. A router control method, characterized in that, The method includes: Obtain network operation data of the router, the network operation data including: indicator data corresponding to at least one influencing factor related to the router's network status; Based on the network operation data, the priority of each of the at least one influencing factor is determined; In descending order of priority, the corresponding index data of the influencing factors of the corresponding priority are used to predict the network status of the router within a preset time period until it is predicted that the router includes at least one initial abnormal node. Then, the initial bandwidth expansion strategy of the router is configured according to the at least one initial abnormal node. If it is determined that the router also includes any associated abnormal node corresponding to the initial abnormal node, then the initial bandwidth expansion strategy is updated based on each determined associated abnormal node, and the updated target bandwidth expansion strategy is used to control the router to expand bandwidth. The associated abnormal nodes are determined through the following steps: Based on historical network operation data, the correlation strength value between the initial abnormal node and other network nodes is calculated using a causal discovery algorithm. Other network nodes whose association strength value is greater than or equal to the preset association strength threshold are identified as the associated abnormal nodes corresponding to the initial abnormal node. The other network nodes include network devices that are in the same network as the router and have an interactive relationship with it.

2. The method according to claim 1, characterized in that, Determining the priority of each of the at least one influencing factor based on the network operation data includes: For each influencing factor, the corresponding indicator data is normalized; based on the normalized indicator data, the corresponding evaluation value is calculated. Based on the obtained evaluation values, the priority of each of the at least one influencing factor is determined by the following method: Based on the preset threshold range to which each evaluation value belongs, the priority of the influencing factors corresponding to the evaluation value is determined; or Clustering analysis is performed on each of the evaluation values ​​using a clustering algorithm, and the priority of each of the at least one influencing factor is determined based on the clustering results.

3. The method according to claim 2, characterized in that, The priorities include high priority, medium priority, and low priority. Determining the priority of the influencing factors corresponding to each evaluation value based on a preset threshold range includes: For each evaluation value, perform the following operations: If the evaluation value falls within the first preset threshold range, the influencing factor corresponding to the evaluation value is taken as the dominant factor, and the priority of the influencing factor is determined as the high priority. If the evaluation value falls within the second preset threshold range, the influencing factor corresponding to the evaluation value is taken as a secondary factor, and the priority of the influencing factor is determined as the medium priority. If the evaluation value falls within the third preset threshold range, the influencing factor corresponding to the evaluation value is treated as a regular factor, and the priority of the influencing factor is determined to be the low priority.

4. The method according to claim 2, characterized in that, The priorities include high priority, medium priority, and low priority. The step of performing cluster analysis on each of the evaluation values ​​using a clustering algorithm, and determining the priority of each of the at least one influencing factor based on the clustering results, includes: Cluster analysis was performed on the evaluation values ​​of all influencing factors using a clustering algorithm to obtain three cluster centers and the cluster group to which each influencing factor belongs; Compare the numerical values ​​of the three cluster centers; The influencing factors in the cluster group corresponding to the cluster center with the largest value are taken as the dominant factors, and the priority of the influencing factors is determined as the high priority. The influencing factors in the cluster group corresponding to the cluster center with the smallest value are taken as regular factors, and the priority of the influencing factors is determined as the low priority. The influencing factors in the remaining clustering groups are taken as secondary factors, and the priority of the influencing factors is determined as the medium priority.

5. The method according to any one of claims 1 to 4, characterized in that, The network status of the router within a preset time period is predicted using the following method: The corresponding priority indicator data are input into a pre-trained behavior prediction model to output the probability distribution of the router's abnormal state within a preset time period. Based on the probability distribution of the abnormal state, the network state of the router is predicted, or... The corresponding priority indicator data is matched with the historical anomaly database, and based on the generated matching results, the network status of the router within a preset time period is predicted.

6. The method according to claim 5, characterized in that, The step of predicting the network state of the router based on the abnormal state probability distribution includes: Based on the probability distribution of abnormal states at k consecutive time points within the preset time period, after identifying the target time point where the probability of an anomaly occurrence is greater than or equal to the anomaly threshold, it is determined that the router includes at least one initial abnormal node, where k is a positive integer greater than or equal to 1. The step of predicting the network status of the router within a preset time period based on the generated matching results includes: If the matching result is successful, it is determined that the router includes at least one initial abnormal node, wherein the historical abnormal database includes the correspondence between historical indicator data and abnormal states.

7. The method according to any one of claims 1 to 4, characterized in that, The method further includes: If the router does not include any associated abnormal node corresponding to the initial abnormal node, then the initial bandwidth expansion strategy is used to control the router to expand its bandwidth.

8. The method according to any one of claims 1 to 4, characterized in that, After the router performs bandwidth expansion, the method further includes: Based on the collected network status data, network performance parameters, and bandwidth expansion strategies of the router, the configuration rules of the bandwidth expansion strategy are optimized through reinforcement learning. The network status data includes the evaluation value of at least one influencing factor in the network operation data, and the network performance parameters include at least one of network transmission rate, network transmission latency, and packet loss rate.

9. A control device for a router, characterized in that, include: The acquisition module is used to acquire network operation data of the router, the network operation data including: indicator data corresponding to at least one influencing factor related to the router's network status; The processing module is configured to determine the priority of each of the at least one influencing factor based on the network operation data; predict the network status of the router within a preset time period by sequentially using the index data corresponding to the influencing factors of the corresponding priority in descending order of priority, until it is predicted that the router includes at least one initial abnormal node; configure the initial bandwidth expansion strategy of the router according to the at least one initial abnormal node; if it is determined that the router also includes any associated abnormal node corresponding to the initial abnormal node, update the initial bandwidth expansion strategy based on the determined associated abnormal nodes. The control module is used to control the router to perform bandwidth expansion using the updated target bandwidth expansion strategy; The processing module determines the associated abnormal nodes through the following steps: Based on historical network operation data, the correlation strength value between the initial abnormal node and other network nodes is calculated using a causal discovery algorithm. Other network nodes whose association strength value is greater than or equal to the preset association strength threshold are identified as the associated abnormal nodes corresponding to the initial abnormal node. The other network nodes include network devices that are in the same network as the router and have an interactive relationship with it.

Images