A method for monitoring communication fault location in a communication network

Abstract

The application provides a communication network communication fault positioning monitoring method, which comprises the following steps: obtaining real-time running data of each communication module and preprocessing, constructing a dynamic network topology model; automatically adjusting detection parameters according to real-time running data, implementing dynamic hardware self-checking; optimizing dialing process according to self-checking data and real-time running data, detecting performance changes or potential faults; constructing a prediction model, updating fault features and the prediction model; using reinforcement learning to optimize fault response strategies; locating the fault source according to the prediction result of the prediction model and triggering response measures; automatically optimizing network configuration according to the processing result of the response measures, while providing a user-friendly interface and displaying network status and fault processing progress in real time. The application can not only improve the accuracy and speed of fault detection, but also realize early prevention and rapid response to potential faults, thereby significantly optimizing the stability and user experience of the communication network.

Description

Technical Field

[0001] This invention belongs to the field of communication networks, and in particular relates to a method for locating and monitoring communication network faults. Background Technology

[0002] In the operation of modern communication networks, ensuring network stability and reliability is crucial. Communication failures not only affect service quality but can also lead to economic losses and a decline in user trust. Traditional fault detection and localization methods mainly rely on hardware self-testing and rule-based fault detection systems. While these systems can perform fault detection and preliminary localization to a certain extent, they typically face several significant drawbacks: First, these systems often rely on predefined fault modes, making it difficult to adapt to dynamic changes in the network environment and new fault types; second, hardware self-testing usually only detects faults that have already occurred, while failing to adequately monitor impending faults and potential abnormal states, limiting the possibility of fault prevention and early intervention; finally, existing fault localization methods are often reactive, meaning that fault searching only begins after a user reports a service quality problem, resulting in a delay in fault response time.

[0003] Furthermore, a key limitation of existing technologies is the lack of an effective mechanism for processing and analyzing large-scale data, which is particularly important in modern communication networks where network complexity and data volume are constantly increasing. For example, traditional methods struggle to extract useful information from real-time data for rapid fault response.

[0004] Therefore, there is a need for an advanced system that can analyze large amounts of data in real time, predict potential faults, and quickly locate the source of faults in order to overcome these shortcomings of existing technologies and improve the reliability and maintenance efficiency of communication networks. Summary of the Invention

[0005] The purpose of this invention is to propose a communication network fault location and monitoring method that utilizes hardware self-testing technology, dialing process closed-loop feedback technology, and fault prediction technology to overcome the shortcomings of existing technologies and improve the reliability and maintenance efficiency of communication networks.

[0006] To achieve the above objectives, the present invention provides a method for locating and monitoring communication network faults, the method comprising:

[0007] The system acquires and preprocesses real-time operational data for each communication module, constructs a dynamic network topology model, and monitors the real-time operational data of each communication module. The real-time operational data includes load, error rate, and response time.

[0008] The detection parameters are automatically adjusted based on real-time operating data to perform dynamic hardware self-tests and obtain self-test data; wherein, the hardware parameters include temperature T, voltage V, current I, and signal strength S;

[0009] The dialing process is optimized based on self-test data and real-time operational data, and the network configuration is automatically adjusted in response to real-time detected performance changes or potential faults through a closed-loop feedback mechanism; the detection algorithm is as follows:

[0010]

[0011] Where, δ t Represents the fault indication function, x t Threshold represents a hardware parameter. new Indicates a dynamic threshold; if x t Exceeding Threshold new If so, it is marked as a potential fault point and a diagnostic procedure is triggered;

[0012] A predictive model is constructed based on self-test data and changes in the dynamic network topology model, and the fault characteristics and predictive model are updated based on real-time data monitoring; wherein, the objective function of the predictive model is expressed as follows:

[0013]

[0014] subjectto y i (w T φ(x i )+b)≥1-ξi,ξi≥0

[0015] Where w represents the weight vector, b represents the bias, and φ(x) i ) represents the function that maps input data to a high-dimensional space, C represents the regularization parameter, and ξ represents the regularization parameter. i Represents slack variables;

[0016] Reinforcement learning is used to predict potential faults and their impacts based on a prediction model, thereby optimizing fault response strategies; wherein, the loss function L(θ) of the reinforcement learning is defined as:

[0017]

[0018] Where D represents the training dataset collected from network operation data;

[0019] Locate the source of the fault based on the prediction results of the predictive model and trigger response measures;

[0020] The system automatically optimizes network configuration based on the results of response measures, while providing a user-friendly interface that displays network status and fault handling progress in real time. The parameter adjustment formula for automatically optimizing network configuration based on the results of response measures is expressed as follows:

[0021] Param new =Param old +κ·ΔMetric (18)

[0022] Among them, Param old This indicates the network parameters that need to be adjusted, ΔMetric is the deviation of the corresponding performance metric, κ is the adjustment sensitivity coefficient, and Param... new This indicates the updated network parameters.

[0023] Furthermore, the method also includes: encrypting and controlling access to all inbound and outbound data;

[0024] Establish a mechanism for continuous improvement and technological adaptation, and regularly evaluate model performance.

[0025] Furthermore, in the construction of the dynamic network topology model, the graph structure can be adjusted in real time when a new communication module is added or an old module is removed. ΔV and ΔE are used to represent changes in the graph structure, where ΔV represents newly added nodes and edges, and ΔE represents deleted nodes and edges. Then, using node features and edge connection information, a graph neural network model is trained to predict possible abnormal states and potential fault points in the network. The update formula is as follows:

[0026]

[0027] in, Let σ represent the representation of node v at layer l, and let σ represent the nonlinear activation function. c represents the neighboring nodes of v. vu W represents the normalization constant. (l) and B (l) This represents the weights and biases of the hierarchy.

[0028] Furthermore, when implementing dynamic hardware self-tests, the threshold range for normal operating conditions is set to T. norm V norm I norm S norm Exceeding these thresholds indicates a potential fault, thus requiring continuous monitoring and recording of hardware parameters, and then calculating the real-time average value μ of each parameter. t and standard deviation σ t The threshold is automatically and dynamically adjusted based on network status and performance data, as shown below:

[0029] Thresholdnew =μ t +k·σ t

[0030] Here, k represents a coefficient adjusted based on historical data and network performance to control the sensitivity of the threshold.

[0031] Furthermore, when any performance index exceeds its dynamic threshold, a feedback control mechanism is automatically triggered. The network parameter adjustment relies on a closed-loop control strategy, the update formula of which is shown below:

[0032] Param new =Param old ·(1+β·(Metric deviation))

[0033] Among them, Param new and Param old These represent the network parameters before and after adjustment, respectively. β represents the adjustment sensitivity coefficient, and Metricdeviation represents the degree to which the performance index deviates from the threshold.

[0034] Furthermore, the specific steps of updating fault characteristics and prediction models based on real-time data monitoring include:

[0035] The SVM model is updated based on newly collected fault data using an online learning algorithm. The update strategy employs incremental learning, fine-tuning the model parameters based on new fault instances, as shown below:

[0036]

[0037] Where η represents the learning rate, y i Indicates the actual fault label. x represents the model's predicted label. i This represents the fault feature vector.

[0038] Furthermore, the process of predicting potential failures and impacts based on a predictive model, using an environmental model to make multi-step predictions of future states, and selecting the action sequence that maximizes the cumulative future reward is represented as follows:

[0039]

[0040] Where H represents the prediction range; γ represents the discount factor, which represents the current value of the future reward.

[0041] Furthermore, the process of locating the fault source based on the prediction results of the prediction model, using the shortest path calculation, is expressed as follows:

[0042]

[0043] Where P represents the set of all paths from the fault detection point to other nodes.

[0044] Furthermore, the automatic optimization of network configuration based on the processing results of response measures, and the periodic updating of baseline values ​​to reflect the latest network status and performance, are described below:

[0045] Baseline new =α·Performance current +(1-α)·Baseline old

[0046] Here, α is a smoothing factor that controls the weights of historical and recent data.

[0047] Furthermore, the mechanism for continuous improvement and technological adaptation is established to periodically evaluate model performance. This periodic performance evaluation involves analyzing key performance indicators to calculate the average value Mean(X) and volatility Var(X) for each indicator, as follows:

[0048]

[0049] Where X represents the set of indicators, x i This represents a single measurement value, and N represents the total number of measurements.

[0050] The beneficial technical effects of the present invention are at least as follows:

[0051] (1) This invention provides a communication network fault location and monitoring method. Through improved hardware self-testing technology, it achieves comprehensive monitoring of the status of communication modules, including but not limited to hardware performance, signal quality, and data transmission errors. At the same time, the dialing process closed-loop feedback technology ensures that every step from the user equipment to the network center is monitored and recorded, and the network configuration and performance are optimized in real time through the closed-loop feedback mechanism.

[0052] (2) By employing advanced data analysis and machine learning techniques, the algorithm predicts potential faults and abnormal states by analyzing historical fault data and real-time network behavior. It can identify impending fault modes and automatically trigger preventative measures.

[0053] (3) By employing advanced data analysis and machine learning techniques, the algorithm predicts potential faults and abnormal states by analyzing historical fault data and real-time network behavior. The algorithm can identify impending fault modes and automatically trigger preventative measures.

[0054] (4) By combining improved hardware self-testing technology, closed-loop feedback, and advanced fault prediction algorithms, this invention can not only improve the accuracy and speed of fault detection, but also achieve early prevention and rapid response to potential faults, thereby significantly optimizing the stability of communication networks and user experience. Attached Figure Description

[0055] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0056] Figure 1 This is a flowchart of a communication network fault location and monitoring method according to the present invention. Detailed Implementation

[0057] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0058] like Figure 1 As shown in the figure, an embodiment of the present invention provides a method for locating and monitoring communication network faults, the method comprising the following steps 1-8:

[0059] Step 1: Obtain and preprocess the real-time operating data of each communication module to construct a dynamic network topology model and monitor the real-time operating data of each communication module in real time; wherein, the real-time operating data includes load, error rate and response time.

[0060] In a preferred embodiment, a graph G = (V, E) is defined, where V is a set of nodes, each node representing a communication module; and E is a set of edges, representing connections between modules. Node characteristics. This includes parameters such as the hardware status of the communication module and signal strength, which are collected in real time by sensors.

[0061] Then, real-time operational data for each communication module is collected, including but not limited to load, error rate, and response time. The data is then normalized to make it suitable for input into the GNN model. Let... Where μ is the mean and σ is the standard deviation.

[0062] Furthermore, a dynamic graph update algorithm is implemented to adjust the graph structure in real time when a new communication module is added or an old module is removed. ΔV and ΔE are used to represent changes in the graph structure, i.e., the addition and deletion of nodes and edges. Utilizing node features and edge connectivity information, a GNN model is trained to predict possible abnormal states and potential failure points in the network. The model updates the node representation using the following formula:

[0063]

[0064] in, Let v be the representation of node v in layer l, and σ be the nonlinear activation function. c is a neighbor node of v. vu It is the normalization constant, W (l) and B (l) It refers to the weights and biases of the hierarchy.

[0065] Understandably, the system monitors the node status output by the GNN in real time, marks abnormal signals, and triggers alarms. Based on feedback, it adjusts network parameters or directly sends repair commands to the maintenance team to quickly respond to network failures.

[0066] Step 2: Automatically adjust the detection parameters based on real-time operating data, perform dynamic hardware self-test, and obtain self-test data; wherein, the hardware parameters include temperature T, voltage V, current I, and signal strength S.

[0067] In a preferred embodiment, the hardware parameters of each communication module include temperature T, voltage V, current I, and signal strength S. These parameters reflect the operating status and potential anomalies of the hardware. A threshold range for normal operating conditions is set: T norm V norm I norm S norm Exceeding these thresholds indicates a potential fault. Continue monitoring and recording hardware parameters, and calculate the real-time mean μ for each parameter. t and standard deviation σ t Dynamically adjust the failure time value based on the latest network status and performance data:

[0068] Threshold new =μ t +k·σ t (2)

[0069] Here, k is a coefficient adjusted based on historical data and network performance to control the sensitivity of the threshold.

[0070] In addition, for each monitoring cycle, the current hardware parameter x is calculated. t The difference between ThresholdNew and the dynamic inter-value. If xt Exceeding Threshold new If so, it is marked as a potential fault point, and further diagnostic procedures are triggered:

[0071]

[0072] Where δ t It is a fault indication function.

[0073] Understandably, the value of k is adjusted based on fault detection results and network feedback to optimize the accuracy and response speed of fault detection. A feedback control loop is used to optimize the value of k based on the goal of minimizing false positives and false negatives.

[0074] k new =k old ·(1+α·(FP rate-Target FP)) (4)

[0075] Where α is the learning rate, FPrate is the current false positive rate, and TargetFP is the target false positive rate.

[0076] Through this dynamic hardware self-testing technology, the communication module can automatically adapt to changes in the network environment and operating status, adjusting fault detection thresholds and parameters in a timely manner, thereby improving the flexibility and accuracy of fault diagnosis. This technology is particularly suitable for complex communication network environments where the module's status changes frequently and where extremely high speed and accuracy in fault response are required.

[0077] Step 3: Optimize the dialing process based on self-test data and real-time operation data, and automatically adjust network configuration in response to real-time detected performance changes or potential faults through a closed-loop feedback mechanism.

[0078] In a preferred embodiment, performance metrics include latency (L), bandwidth utilization (U), and packet loss rate (P). These metrics are key parameters for evaluating network performance. The above performance metrics are continuously monitored and recorded, with each metric processed as time-series data. Data points are collected by time index and denoted as L(t), U(t), and P(t).

[0079] Furthermore, dynamic thresholds are set for each performance metric. These thresholds are automatically adjusted based on actual network usage and historical performance data. The threshold setting formula is as follows:

[0080]

[0081] Where μ x，t and σ x，tThese are the mean and standard deviation of indicator x over time t, respectively, and k x It is an adjustment coefficient used to control threshold sensitivity.

[0082] When any performance metric exceeds its dynamic threshold, the system automatically triggers a feedback control mechanism. This mechanism adjusts network parameters, such as routing, bandwidth allocation, or connection retry strategies, to optimize performance and reduce the impact of failures. Network parameter adjustments rely on a closed-loop control strategy based on the following update formula:

[0083] Param new =Param old ·(1+β·(Metric deviation)) (6)

[0084] Among them, Param new and Param old These are the network parameters before and after adjustment, respectively. β is the sensitivity adjustment coefficient, and Metricdeviation is the degree to which the performance index deviates from the threshold.

[0085] By implementing an adaptive network configuration feedback control system, the network can automatically adjust to performance degradation or potential failures, thereby maintaining service quality and improving overall network stability. This system is particularly suitable for environments with fluctuating demands and complex network conditions, effectively enhancing network operational adaptability and fault response speed.

[0086] Step 4: Build a prediction model based on self-test data and changes in the dynamic network topology model, and update the fault characteristics and prediction model based on real-time data monitoring.

[0087] In a preferred embodiment, a basic network fault feature model is established and trained using historical fault data. The training data includes various parameters collected from network devices, such as signal weakening (α), latency increase (δ), and packet loss rate (λ). The fault feature vector is defined as f = [α, δ, λ], and a support vector machine (SVM) is used to train the initial fault identification model, where the objective function of the model is:

[0088]

[0089] Where w is the weight vector, b is the bias, and φ(x) is the weight vector. i ) is a function that maps input data to a high-dimensional space, C is the regularization parameter, and ξi is the slack variable.

[0090] Furthermore, network operation data is continuously monitored, and the fault feature vector f defined above is extracted in real time. The real-time data is preprocessed, including standardization to adapt it to the model input. The standardization formula is:

[0091]

[0092] Where f is the original eigenvalue, μ f and σ f These are the historical mean and standard deviation of the feature, respectively.

[0093] Then, the SVM model is continuously updated based on newly collected fault data using an online learning algorithm. The update strategy employs incremental learning, fine-tuning the model parameters based on new fault instances.

[0094]

[0095] Where η is the learning rate, and y i These are actual fault labels. It is the model predicting the label, x i It is a fault feature vector.

[0096] Finally, the updated SVM model is used to identify potential faults in the network in real time. Once the model identifies a fault, an alert is immediately sent to the network management system, along with the cause of the fault and recommended response measures.

[0097] Step 5: Use reinforcement learning to predict potential failures and their impacts based on the prediction model, and optimize the failure response strategy.

[0098] In a preferred embodiment, the environment model is defined as follows:

[0099] Define the environment model f(s) t a t The model uses a neural network to predict when to take action (θ). t The subsequent network state s t+1 Network status includes key performance indicators such as latency L. t Bandwidth utilization U t Packet loss rate P t The state vector is represented as s t =[L t U t P t ].

[0100] Furthermore, features and labels are extracted from historical monitoring data. Features represent the current state and action, and labels represent the next state. Data preprocessing methods, such as normalization, are used to ensure the input data meets the requirements for neural network training. A prediction model is built using a multilayer perceptron (MLP). The model parameters θ are optimized by minimizing the prediction error, and the loss function is defined as:

[0101]

[0102] in It is a training dataset collected from network operation data.

[0103] Furthermore, the reward function R(s) t a t Defined based on the improvement or deterioration of network performance, it is calculated as follows:

[0104] R(s t a t )=-(αΔL t +βΔU t +γΔP t (11)

[0105] Where, ΔL t ΔU t ΔP t These are the changes in post-action latency, bandwidth utilization, and packet loss rate, respectively.

[0106] In each decision cycle, an environmental model is used to make multi-step predictions of future states, and the action sequence that maximizes the cumulative reward in the future is selected. Dynamic programming or other optimization algorithms are used to solve the following optimization problem:

[0107]

[0108] Where H is the prediction range and γ is the discount factor, representing the current value of future rewards.

[0109] In addition, the system continuously collects the latest performance data (such as latency, bandwidth utilization, packet loss rate, etc.) from network devices and performs necessary preprocessing on this data, such as scaling and noise reduction, to ensure the quality and consistency of the input data.

[0110] Online learning technology is used to predict model f(s) t a t Incremental updates are performed on θ. This involves fine-tuning the neural network parameters θ using newly collected data to adapt to changes in network behavior. Update methods can employ stochastic gradient descent (SGD) or the Adam optimizer, and the update formula is:

[0111]

[0112] Where η is the learning rate. It is the gradient of the loss function with respect to the parameters.

[0113] Furthermore, the updated model f is used, combined with the current network state s. tThe optimal sequence of actions is recalculated using a Model Predictive Control (MPC) strategy. This may involve simulating multiple future time steps to predict and select actions that maximize long-term rewards. The calculated optimal strategy is then applied to network management, such as automatically adjusting routing algorithms, reconfiguring network bandwidth, or modifying load balancing settings to address predicted network conditions and potential failures.

[0114] Furthermore, key performance indicators (KPIs), such as network response time, fault recovery time, and service interruption time, are collected periodically to evaluate overall network performance and the effectiveness of fault management strategies. The effectiveness of the strategies is assessed by comparing changes in performance indicators before and after implementation. Statistical analysis methods, such as t-tests or ANOVA, are used to determine whether the performance improvements are significant.

[0115] Based on performance evaluation results and feedback obtained from network operations, adjust the parameters of the prediction model and decision-making strategy. For example, if the prediction model is found to perform poorly under certain specific conditions, the model structure can be adjusted or the dataset can be retrained. Establish a continuous improvement mechanism to ensure that the system adapts to network changes and new operating conditions over time. This may involve introducing new data sources, updating data preprocessing procedures, or introducing advanced machine learning algorithms.

[0116] Step 6: Locate the source of the fault based on the prediction results of the prediction model and trigger response measures.

[0117] In a preferred embodiment, a threshold-based detection method is used to identify anomalous behavior. For network parameters such as latency (L... t ), bandwidth utilization (U t ) and packet loss rate (P t ), Defined value L max U max P max When the observed parameters exceed these thresholds, fault detection is triggered.

[0118]

[0119] The fault source localization employs graph-based analysis methods, such as shortest path algorithms and network flow analysis, to determine the fault origin. A network topology graph G(V, E) is constructed, where the set of nodes V represents network devices, and the set of edges E represents connections. Each edge e has a weight w. e This represents the probability of connection failure, determined by both historical failure data and real-time performance data. Response measures are formulated based on the failure type and location. This includes route reconfiguration, bandwidth reallocation, or activating a backup system. A decision tree is used to determine the most appropriate response; each branch of the decision tree represents a failure handling option, with conditions based on the severity and scope of the failure.

[0120] Specifically, response measures are developed based on the type and location of the fault. This includes route reconfiguration, bandwidth reallocation, or activating backup systems. A decision tree is used to determine the most appropriate response measure, with each branch of the decision tree representing a fault handling option based on the severity and scope of the fault.

[0121] In addition, implement the selected fault response strategy and monitor its effectiveness. If the fault is not handled correctly, iterative adjustments are made, and alternative fault response strategies are adopted. Post-implementation network performance data, such as improved latency, bandwidth utilization, and packet loss rate, are collected to verify the effectiveness of the fault handling.

[0122] Step 7: Automatically optimize network configuration based on fault handling results to improve overall system performance and stability. The system automatically optimizes network configuration based on the handling results of response measures, while providing a user-friendly interface that displays network status and fault handling progress in real time.

[0123] In a preferred embodiment, key metrics for network performance evaluation are selected, including latency (L). t ), bandwidth utilization (U t ), and packet loss rate (P t These metrics will be used to assess the current state of the network and optimization needs. A baseline threshold L is defined for each performance metric. baseline U baseline P baseline These baselines are set based on the network's design specifications and historical performance data. The baseline values ​​are updated periodically to reflect the network's latest status and performance, using the following update method:

[0124] Baseline new =α·Performance current +(1-α)·Baseline old (16)

[0125] Here, α is a smoothing factor that controls the weights of historical and recent data.

[0126] Specifically, for each performance cycle, the deviation of the current performance metric from the baseline is calculated:

[0127] ΔL=L t -L baseline ΔU=U t -U baseline ΔP=P t -P baseline (17)

[0128] Use deviation values ​​to evaluate network performance and determine the need for optimization.

[0129] Specifically, feedback control logic is designed based on performance deviations to automatically adjust network parameters to optimize performance. For example, if ΔL exceeds a preset tolerance range, bandwidth is automatically increased or routing strategies are changed. The parameter adjustment formula can be expressed as:

[0130] Param new =Param old +κ·ΔMetric (18)

[0131] Where Param represents the network parameters that need to be adjusted, ΔMetric is the deviation of the corresponding performance metric, and K is the adjustment sensitivity coefficient.

[0132] Finally, after implementing parameter adjustments, continue to monitor performance changes to evaluate the effectiveness of the adjustments. If the adjustments do not achieve the expected results, continue adjusting the k value or consider different adjustment strategies to find the optimal method for performance improvement.

[0133] Specifically, it also provides a user-friendly interface that displays network status and troubleshooting progress in real time.

[0134] Further details are as follows:

[0135] Set up a real-time data acquisition system to continuously monitor key network performance indicators, such as latency (L). t ), bandwidth utilization (U t Packet loss rate (P) t This data is automatically collected through network devices and sensors and transmitted to a central monitoring system via the network. Monitoring thresholds are defined for each performance metric, such as L. max U max P max When the indicator exceeds the threshold, the system automatically triggers an alarm and records the event.

[0136] Specifically, develop a graphical user interface (GUI) that can display real-time data, historical trends, and alert status. The interface should include various view options, such as dashboards, charts, and map views, to suit the needs of different users.

[0137] The user interface includes features such as real-time data update display, historical data analysis, fault reporting, and user-customized alarm settings. Users can directly adjust monitoring thresholds and configure network management tasks through the interface.

[0138] Furthermore, the charts and reports used to drive the user interface ensure the accuracy and timeliness of the information displayed.

[0139] The performance metrics displayed are calculated using formulas, for example:

[0140]

[0141] ActualMetric is the actual measured value, while MinimumMetric and MaximumMetric are the historical minimum and maximum values ​​of the indicator, respectively.

[0142] The interface allows users to provide feedback on network performance issues and user experience; this feedback is directly used to improve system functionality and performance. An interactive help system is implemented, providing operation guides and troubleshooting support to help users effectively manage the network and resolve problems.

[0143] Step 8: Encrypt and control access to all inbound and outbound data.

[0144] In a preferred embodiment, all inbound and outbound data is encrypted using the TLS protocol to ensure data security during transmission. The latest version of the TLS protocol, such as TLS 1.3, and a strong cipher suite, such as ECDHE-RSA-AES256-GCM-SHA384, are employed to prevent man-in-the-middle attacks and data leakage.

[0145] In addition, all sensitive data stored on servers and backup storage devices is encrypted using AES-256. Encryption keys are strictly managed using a Key Management Service (KMS), with keys rotated periodically, and access to the keys is strictly restricted to authorized personnel.

[0146] Furthermore, access permissions are assigned based on users' job roles, such as network administrators, operations and maintenance personnel, and auditors. Each role has clearly defined access permissions for data, ensuring that only authorized users can access sensitive information. MFA is implemented for all system users, combining passwords with biometric technologies (such as fingerprint or facial recognition) or hardware-based authentication tokens to enhance authentication security.

[0147] Then, SHA-256 hash values ​​are generated for all transmitted and stored data to ensure data integrity. The hash values ​​are verified during data backup and recovery; any hash mismatch will trigger a security alert. A comprehensive audit logging system is implemented to record all access and modification operations to sensitive data, including timestamps, user IDs, executed operations, and the objects of those operations. The audit logs themselves must also be encrypted and backed up regularly, and the audit policy and record retention period should comply with relevant laws and regulations.

[0148] Furthermore, data that needs to be analyzed or shared should be de-identified using techniques such as dynamic data masking and tag exchange to protect personally identifiable information. Ensure the de-identification process is conducted under the supervision of authorized personnel and is reversible if necessary. Implement privacy impact assessments and regularly review the use of de-identified data and privacy compliance to ensure all operations do not infringe on user privacy.

[0149] Finally, develop specific data breach response procedures, including immediately isolating affected systems, assessing the scope and nature of the breach, notifying all relevant parties, and taking measures to limit further damage. Conduct a thorough investigation of the data breach to determine the cause, assess the impact, and implement remedial measures based on the nature and severity of the incident. Adjust security policies and processes to prevent similar incidents from recurring in the future.

[0150] Step 9: Establish a mechanism for continuous improvement and technology adaptation, and regularly evaluate model performance.

[0151] In a preferred embodiment, periodic (e.g., quarterly) system performance evaluations are performed by analyzing key performance indicators (KPIs) such as network latency (L). t ), bandwidth usage (U t ) and packet loss rate (P t To evaluate system performance, the average and volatility of each indicator are calculated:

[0152]

[0153] Where X represents the set of indicators, x i It is a single measurement value, and N is the total number of measurements.

[0154] Furthermore, statistical analysis methods, such as deviation analysis and trend analysis, are used on the collected data to determine whether the system performance meets the predetermined business objectives and technical requirements. New optimization objectives are set based on changes in business needs and technological developments, such as reducing the target value of average latency ΔL, increasing bandwidth efficiency ΔU, or improving packet processing efficiency ΔP.

[0155] Finally, identify system weaknesses and changes in user requirements through user feedback and automatically collected performance data. Organize regular retrospective meetings to discuss feedback and data analysis results. Adjust system parameters or optimize algorithms based on the analysis results; for example, adjust network monitoring thresholds or update fault detection logic to improve accuracy and response speed. Develop a detailed system upgrade plan based on the evaluation and feedback adjustments, including hardware updates, software upgrades, and configuration optimizations. Implement iterative upgrades, ensuring all improvements are tested and verified in a controlled environment before full deployment. Use version control and deployment management tools to ensure a smooth upgrade process and data consistency.

[0156] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0157] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of apparatuses or units may be electrical, mechanical, or other forms.

[0158] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0159] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for locating and monitoring communication faults in a communication network, characterized in that, The method includes: The system acquires and preprocesses real-time operational data for each communication module, constructs a dynamic network topology model, and monitors the real-time operational data of each communication module. The real-time operational data includes load, error rate, and response time. The detection parameters are automatically adjusted based on real-time operating data to perform dynamic hardware self-tests and obtain self-test data; wherein, the hardware parameters include temperature T, voltage V, current I, and signal strength S; The dialing process is optimized based on self-test data and real-time operational data, and the network configuration is automatically adjusted in response to real-time detected performance changes or potential faults through a closed-loop feedback mechanism; the detection algorithm is as follows: Where, δ t Represents the fault indication function, x t Threshold represents a hardware parameter. new Indicates a dynamic threshold; if x t Exceeding Threshold new If so, it is marked as a potential fault point and a diagnostic procedure is triggered; A predictive model is constructed based on self-test data and changes in the dynamic network topology model, and the fault characteristics and predictive model are updated based on real-time data monitoring; wherein, the objective function of the predictive model is expressed as follows: subject to y i (w T φ(x i )+b)≥1-ξ i ,ξ i ≥0 Where w represents the weight vector, b represents the bias, and φ(x) i ) represents the function that maps input data to a high-dimensional space, C represents the regularization parameter, and ξ represents the regularization parameter. i Let y represent a slack variable. i Indicates the actual fault label; Reinforcement learning is used to predict potential faults and their impacts based on a prediction model, thereby optimizing fault response strategies; wherein, the loss function L(θ) of the reinforcement learning is defined as: Where D represents the training dataset collected from network operation data, s t Indicates the current network state, a t θ represents the action, θ represents the model parameters, and s t+1 Indicates the next network state, f(s) t ,a t ;θ) represents an environmental model that uses a neural network to predict the outcome of taking a t After s t+1 ; Locate the source of the fault based on the prediction results of the predictive model and trigger response measures; The system automatically optimizes network configuration based on the results of response measures, while providing a user-friendly interface that displays network status and fault handling progress in real time. The parameter adjustment formula for automatically optimizing network configuration based on the results of response measures is expressed as follows: Param new =Param old +κ·ΔMetric Among them, Param old This indicates the network parameters that need to be adjusted, ΔMetric is the deviation of the corresponding performance metric, κ is the adjustment sensitivity coefficient, and Param... new This indicates the updated network parameters.

2. The communication network communication fault location and monitoring method according to claim 1, characterized in that, The method further includes: encrypting and controlling access to all inbound and outbound data; Establish a mechanism for continuous improvement and technological adaptation, and regularly evaluate model performance.

3. The communication network fault location and monitoring method according to claim 1, characterized in that, In the constructed dynamic network topology model, the graph structure can be adjusted in real time when a new communication module is added or an old module is removed. ΔV and ΔE are used to represent changes in the graph structure, where ΔV represents newly added nodes and edges, and ΔE represents deleted nodes and edges. Then, using node features and edge connection information, a graph neural network model is trained to predict abnormal states and potential fault points in the network. The update formula is as follows: in, Let σ represent the representation of node v at layer l, and let σ represent the nonlinear activation function. c represents the neighboring nodes of v. vu W represents the normalization constant. (l) and B (l) This represents the weights and biases of the hierarchy.

4. The communication network fault location and monitoring method according to claim 1, characterized in that, When performing dynamic hardware self-tests, the threshold range for normal operating conditions is set to T. norm V norm ,I norm ,S norm Exceeding these thresholds indicates a fault, thus continuously monitoring and recording hardware parameters, and then calculating the real-time average value μ of each parameter. t and standard deviation σ t The threshold is automatically and dynamically adjusted based on network status and performance data, as shown below: Threshold new =μ t +k·σ t Here, k represents a coefficient adjusted based on historical data and network performance to control the sensitivity of the threshold.

5. The communication network communication fault location and monitoring method according to claim 4, characterized in that, When any performance metric exceeds its dynamic threshold, a feedback control mechanism is automatically triggered. The network parameter adjustment depends on a closed-loop control strategy, the update formula of which is shown below: Param new =Param old ·(1+β·(Metric deviation)) Among them, Param new and Param old These represent the network parameters before and after adjustment, respectively. β represents the adjustment sensitivity coefficient, and Metricdeviation represents the degree to which the performance index deviates from the threshold.

6. The communication network communication fault location and monitoring method according to claim 1, characterized in that, The specific steps for updating fault characteristics and prediction models based on real-time data monitoring include: The SVM model is updated based on newly collected fault data using an online learning algorithm. The update strategy employs incremental learning, fine-tuning the model parameters based on new fault instances, as shown below: Where η represents the learning rate, y i Indicates the actual fault label. x represents the model's predicted label. i This represents the fault feature vector.

7. The communication network communication fault location and monitoring method according to claim 1, characterized in that, The process of predicting potential failures and impacts using a predictive model, performing multi-step predictions of future states using an environmental model, and selecting the action sequence that maximizes the cumulative future reward is represented as follows: Where H represents the prediction range, a t:t+H Let R(s) represent the action sequence, γ represent the discount factor, and represent the current value of future rewards. k ,a k ) represents the reward function.

8. The communication network fault location and monitoring method according to claim 1, characterized in that, The process of locating the fault source based on the prediction results of the prediction model, using the shortest path calculation, is as follows: Where P represents the set of all paths from the fault detection point to other nodes, w e Indicates the weight.

9. The communication network fault location and monitoring method according to claim 1, characterized in that, The automatic optimization of network configuration based on the processing results of response measures, and the periodic updating of baseline values ​​to reflect the latest network status and performance, are described below: Baseline new =α·Performance current +(1-α)·Baseline old Among them, Baseline new The latest data represents the baseline value. α is a smoothing factor that controls the weighting of historical and recent data. Performance current Represents performance data, Baseline old Historical data representing baseline values.

10. A method for locating and monitoring communication network faults according to claim 2, characterized in that, The mechanism for continuous improvement and technological adaptation is established to periodically evaluate model performance. This periodic performance evaluation involves analyzing key performance indicators to calculate the average value Mean(X) and volatility Var(X) for each indicator, as follows: Where X represents the set of indicators, x i This represents a single measurement value, and N represents the total number of measurements.

Images