Portable WiFi access device monitoring and early warning system based on big data analysis

Abstract

The invention discloses a portable WiFi access device monitoring and early warning system based on big data analysis, and relates to the technical field of computers. Comprising an intelligent sensing and dynamic sampling module, a self-adaptive edge calculation module, a data preprocessing and correction module, a deep learning model construction module, an anomaly detection module, a multi-channel early warning module and an equipment parameter optimization module. According to the invention, the sampling frequency and the adaptive edge computing framework are dynamically adjusted through the intelligent sensing chip, so that the problem of energy consumption and performance balance is solved, and the data acquisition and processing efficiency is improved; user behavior modeling and anomaly detection are carried out by using a deep learning model and an isolated forest algorithm, so that the accuracy and the real-time performance of early warning are remarkably improved; meanwhile, an equipment parameter optimization knowledge base is established at the cloud end, equipment parameters are dynamically adjusted in combination with a remote control interface, the operation state of the equipment is optimized, and the intelligent management level and the operation efficiency of the equipment are improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and particularly to a monitoring and early warning system for portable Wi-Fi access devices based on big data analysis. Background Art

[0002] An intelligent sensing chip is embedded in the portable Wi-Fi device to collect multi-dimensional data such as the connection status, traffic usage, signal strength, and surrounding network environment of the device in real time; in practical applications, the intelligent sensing chip faces the contradiction between data acquisition accuracy and device power consumption when collecting multi-dimensional data. To ensure data accuracy, the chip needs to sample frequently, which will significantly increase power consumption and shorten the device's battery life; at the same time, when edge computing technology processes a large amount of data, it is limited by local computing resources and may not be able to process large-scale data efficiently, resulting in data processing delays and affecting the real-time monitoring effect; in addition, cloud deep learning algorithms rely on a large amount of historical data, but the quality and integrity of the data directly affect the accuracy of the model; if the historical data is missing or biased, the prediction effect of the model and the accuracy of anomaly detection will be greatly reduced. Finally, in the implementation process of the multi-channel early warning mechanism, there is a contradiction between the timeliness of information transmission and the diversity of user receiving channels. Different users may prefer different receiving methods. How to ensure that the early warning information can be transmitted to users in a timely and accurate manner is a technical problem that needs to be solved. Summary of the Invention

[0003] The purpose of the present invention is to provide a monitoring and early warning system for portable Wi-Fi access devices based on big data analysis, which solves technical problems such as the balance between energy consumption and performance, the improvement of early warning accuracy, and the optimization of device operation efficiency, and realizes efficient, accurate, and intelligent monitoring and early warning of portable Wi-Fi devices.

[0004] The purpose of the present invention can be achieved through the following technical solutions:

[0005] The present application provides a monitoring and early warning system for portable Wi-Fi access devices based on big data analysis, including:

[0006] An intelligent sensing and dynamic sampling module, which is responsible for the collection of device data and the dynamic adjustment of the sampling frequency. By embedding an intelligent sensing chip, it collects the data of the portable Wi-Fi device in real time, and dynamically adjusts the sampling frequency according to the preset data accuracy requirements and the device power status;

[0007] An adaptive edge computing module, which is responsible for local data processing and the dynamic allocation of computing resources. It adopts an adaptive edge computing framework and dynamically adjusts the computing resource allocation according to the real-time data volume; reducing the dependence on the cloud;

[0008] Data preprocessing and calibration module, which obtains historical data from the cloud big data platform, preprocesses the data, uses the interpolation filling method for missing data, and corrects deviation data to obtain a complete and accurate historical data set;

[0009] Deep learning model construction module, which constructs a deep learning model based on the preprocessed historical data. First, it uses a convolutional neural network to extract user behavior features to obtain a user behavior model; then it introduces a long short-term memory network for time series analysis and iteratively updates the model with real-time data to finally obtain an accurate user behavior prediction model;

[0010] Anomaly detection module, which uses the isolation forest algorithm to detect anomalies in real-time data based on the user behavior prediction model. When abnormal data is detected, it marks it as a potential security threat and records relevant features;

[0011] Multi-channel warning module, which is responsible for notifying users of the detected abnormal information in the form of a warning. It obtains the user's receiving channel information from the user preference database, matches the optimal transmission method for different types of warning information, generates a multi-channel warning plan, and adopts different push strategies according to the urgency of the warning information;

[0012] Device parameter optimization module, which establishes a device parameter optimization knowledge base in the cloud, automatically matches the optimal parameter adjustment plan for the detected abnormal situation, and optimizes the parameters of the portable WiFi device through the remote control interface.

[0013] Furthermore, according to the preset data accuracy requirements and the device battery status, the sampling frequency is dynamically adjusted, specifically including:

[0014] The intelligent sensing chip is embedded in the portable WiFi device to collect the connection status, traffic usage, and signal strength data of the device in real-time, obtain the current battery status of the device, judge whether the battery is lower than the preset threshold. When the battery is lower than the threshold, the low-frequency sampling mode is adopted. When the battery is higher than the threshold, the high-frequency sampling mode is adopted. The data acquisition accuracy is adjusted according to the sampling frequency, and the collected data is stored and analyzed to form a device operation status report.

[0015] Furthermore, an adaptive edge computing framework is adopted to dynamically adjust the computing resource allocation according to the real-time data volume, specifically including:

[0016] The adaptive edge computing framework is used to obtain the real-time data volume, judge whether the data volume exceeds the preset local processing capacity threshold. When it exceeds the threshold, part of the data is uploaded to the cloud for distributed processing. When it does not exceed the threshold, the processing is preferentially completed at the edge node;

[0017] Dynamically adjust the computing resource allocation strategy according to the real-time data volume and the status of local computing resources. When the data volume continues to increase, increase the computing resource allocation ratio of the edge nodes. When the data volume decreases, reduce the computing resource allocation ratio of the edge nodes;

[0018] Allocate computing tasks between the edge nodes and the cloud through a preset load balancing algorithm. When the load of the edge nodes is high, migrate some computing tasks to the cloud. When the load of the cloud is high, migrate some computing tasks to the edge nodes;

[0019] Obtain the usage status of computing resources of the edge nodes and the cloud in real time through a preset resource monitoring mechanism. When the resource utilization rate of the edge nodes reaches the preset threshold, trigger the resource allocation adjustment strategy. When the resource utilization rate of the cloud reaches the preset threshold, trigger the resource allocation adjustment strategy;

[0020] Dynamically adjust the computing resource allocation ratio of the edge nodes and the cloud according to the preset resource allocation strategy. When the resource utilization rate of the edge nodes continuously exceeds the preset threshold, increase the computing resource allocation ratio of the edge nodes. When the resource utilization rate of the cloud continuously exceeds the preset threshold, increase the computing resource allocation ratio of the cloud;

[0021] Conduct resource optimization allocation between the edge nodes and the cloud through a preset resource optimization algorithm. When the resource utilization rate of the edge nodes is lower than the preset threshold, reduce the computing resource allocation ratio of the edge nodes. When the resource utilization rate of the cloud is lower than the preset threshold, reduce the computing resource allocation ratio of the cloud.

[0022] Furthermore, obtain historical data from the cloud big data platform, and preprocess the data. Use the interpolation filling method for missing data and perform correction processing on deviated data to obtain a complete and accurate historical data set, specifically including:

[0023] Obtain historical data from the cloud big data platform, identify missing values and deviated values in the data, and use the linear interpolation method to fill in the missing parts of the missing values to generate a data sequence without missing values;

[0024] Determine the abnormal range of deviated values through standard deviation and mean, use the median substitution method to correct deviated data, and merge the filled missing data and the corrected deviated data to generate a complete data set;

[0025] Perform standardization processing on the complete data set to eliminate the dimension difference, generate a standardized data set, use the principal component analysis method to reduce the dimension of the standardized data set, extract the main feature data, and use the K-means clustering algorithm to perform clustering analysis on the feature data to generate a historical data set with accurate classification.

[0026] Furthermore, based on the preprocessed historical data, a deep learning model is constructed. First, a convolutional neural network is used to extract user behavior features to obtain a user behavior model, which specifically includes:

[0027] Obtain the preprocessed historical data, input it into the convolutional neural network for feature extraction, and perform feature extraction on the input data through multiple layers of convolution and pooling operations to automatically learn the spatial and temporal correlations in the data, generating a preliminary feature dataset;

[0028] Train the convolutional neural network according to the preliminary feature dataset, continuously optimize the network parameters through the backpropagation algorithm, enable the network to better capture the key features of user behavior, and update the user behavior model according to the feature dataset to obtain the user behavior model.

[0029] Furthermore, after constructing a deep learning model based on the preprocessed historical data, first using a convolutional neural network to extract user behavior features to obtain a user behavior model, it also includes:

[0030] Obtain the time series data of user behavior, input it into the long short-term memory network for time series analysis to generate time series features, and use the long short-term memory network to process real-time data, and combine the time series features to generate real-time behavior features;

[0031] Update the user behavior model according to the real-time behavior features to generate an updated user behavior model, classify the user behavior to generate a user behavior classification result;

[0032] When there are anomalies in the user behavior classification result, use the random forest algorithm to correct the abnormal behavior to generate a corrected classification result, and update the user behavior model again according to the corrected classification result to generate the final user behavior model;

[0033] Predict the user behavior through the final user behavior model to generate an accurate user behavior prediction result.

[0034] Furthermore, based on the user behavior prediction model, use the isolation forest algorithm to perform anomaly detection on real-time data. When detecting abnormal data, mark it as a potential security threat and record the relevant features, which specifically includes:

[0035] Obtain the time series features and input them into the isolation forest model. By constructing multiple random trees, perform isolation operations on each data point and calculate its average path length in all trees; the shorter the path length, the easier it is for the data point to be isolated, and it is an abnormal point; when detecting abnormal data, mark it as a potential security threat and record the relevant features of the abnormal data to generate an abnormal feature set.

[0036] Furthermore, different push strategies are adopted according to the urgency of the warning information, specifically including:

[0037] Use a classification algorithm to classify the warning information according to the urgency level, preset the thresholds for high-risk and general warnings, and select the push strategy according to the classification results;

[0038] Analyze user preferences from the user behavior logs, establish a user preference model, determine the order of the push methods. When the urgency level is high risk, for the real-time push scenario, use multi-threading technology to synchronously call the SMS, email, and App notification interfaces;

[0039] When the urgency level is a general warning, for the sequential push scenario, use a queue mechanism to sequentially call the push interfaces, and by monitoring the push result logs, update the user preference model in real time to optimize the push order.

[0040] Furthermore, establish a device parameter optimization knowledge base in the cloud. For the detected abnormal situations, automatically match the optimal parameter adjustment plan, and optimize the parameters of the portable WiFi device through the remote control interface, specifically including:

[0041] Pre-establish a device parameter knowledge base in the cloud, use a data acquisition interface to obtain the real-time operating parameters of the portable WiFi device, and judge whether there are abnormal situations in the device parameters through a preset anomaly detection algorithm;

[0042] When an anomaly is detected, automatically match the optimal parameter adjustment plan from the knowledge base, and according to the matching result, call the remote control interface to optimize the parameters of the portable WiFi device;

[0043] Continuously update the device parameter knowledge base, use machine learning algorithms to optimize the anomaly detection model and the parameter adjustment plan, and through the remote control interface, monitor the operating status of the device in real time to obtain the execution effect of the optimized parameters.

[0044] The beneficial effects of the present invention are:

[0045] Through the intelligent sensing chip to dynamically adjust the sampling frequency, the system can flexibly switch the sampling mode according to the device power state, adopt low-frequency sampling to save energy when the power is low, and adopt high-frequency sampling to ensure data accuracy when the power is high. Combined with the adaptive edge computing framework, the system dynamically allocates computing resources according to the real-time data volume, preferentially processes data at the edge node, reduces the dependence on the cloud, thereby effectively reducing latency and improving the response speed, solving the problem that it is difficult to balance energy consumption and data acquisition accuracy in traditional methods, and at the same time improving the overall performance and efficiency of the system;

[0046] Based on a deep learning model, user behavior features are extracted through a convolutional neural network and a long short-term memory network, and an isolation forest algorithm is combined for anomaly detection. The CNN is used to extract spatial features, and the LSTM is used to process time series data. Finally, an accurate user behavior prediction model is generated. The isolation forest algorithm identifies anomaly points by calculating the average path length of data points. The shorter the path of a data point, the more likely it is to be an anomaly point, effectively solving the accuracy problem of anomaly detection in a complex data environment and improving the reliability and real-time performance of the early warning system;

[0047] An equipment parameter optimization knowledge base is established in the cloud. The system can obtain the equipment operation parameters in real time and judge whether there are any anomalies. Once an anomaly is detected, the system automatically matches the optimal parameter adjustment plan from the knowledge base and optimizes the equipment through the remote control interface. At the same time, machine learning algorithms are used to continuously update the knowledge base, optimize the anomaly detection model and the parameter adjustment plan, solve the performance problems caused by abnormal parameters during the equipment operation process, improve the intelligent management level of the equipment, and ensure the efficient operation of the equipment under the optimized parameters. Brief Description of the Drawings

[0048] Figure 1 It is a schematic flow chart of a monitoring and early warning system for portable WiFi access devices based on big data analysis provided by this application;

[0049] Figure 2 It is a schematic flow chart of obtaining a historical data set by a monitoring and early warning system for portable WiFi access devices based on big data analysis provided by this application;

[0050] Figure 3 It is a schematic flow chart of an accurate user behavior prediction model of a monitoring and early warning system for portable WiFi access devices based on big data analysis provided by this application. Detailed Embodiments

[0051] To further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, exemplary embodiments will be described in detail here, and the examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. On the contrary, they are only examples of methods and systems consistent with some aspects of this application as detailed in the appended claims.

[0052] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms "a", "the", and "said" used in this application and the appended claims are also intended to include the plural forms unless the context clearly dictates otherwise. It should also be understood that the term "and / or" used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0053] The following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation manners, features, and effects of the present invention.

[0054] Embodiment 1

[0055] Please refer to Figures 1 - 3 , this embodiment provides a monitoring and early warning system for portable WiFi access devices based on big data analysis, including:

[0056] An intelligent perception and dynamic sampling module, responsible for the collection of device data and the dynamic adjustment of the sampling frequency. By embedding an intelligent perception chip, it can collect the data of portable WiFi devices in real time and dynamically adjust the sampling frequency according to the preset data accuracy requirements and the device power status.

[0057] Furthermore, dynamically adjusting the sampling frequency according to the preset data accuracy requirements and the device power status specifically includes:

[0058] The intelligent perception chip is embedded in the portable WiFi device to collect the connection status, traffic usage, and signal strength data of the device in real time, obtain the current power status of the device, and determine whether the power is lower than the preset threshold. When the power is lower than the threshold, a low-frequency sampling mode is adopted; when the power is higher than the threshold, a high-frequency sampling mode is adopted. Adjust the data collection accuracy according to the sampling frequency, store and analyze the collected data, and form an operation status report of the device.

[0059] Specifically, by dynamically adjusting the sampling frequency, while ensuring the data collection accuracy, it significantly reduces the device energy consumption, optimizes the battery life, and at the same time avoids the collection of invalid data, improves the data processing efficiency, and provides efficient support for the intelligent management and early warning of the device.

[0060] An adaptive edge computing module, responsible for local data processing and the dynamic allocation of computing resources. It adopts an adaptive edge computing framework to dynamically adjust the computing resource allocation according to the real-time data volume; reduces the dependence on the cloud and improves the system response speed;

[0061] Furthermore, adopting an adaptive edge computing framework to dynamically adjust the computing resource allocation according to the real-time data volume specifically includes:

[0062] An adaptive edge computing framework is adopted to obtain the real-time data volume, and it is judged whether the data volume exceeds the preset local processing capacity threshold. When it exceeds the threshold, part of the data is uploaded to the cloud for distributed processing. When it does not exceed the threshold, the processing is preferentially completed at the edge node;

[0063] According to the real-time data volume and the status of local computing resources, the computing resource allocation strategy is dynamically adjusted. When the data volume continues to increase, the allocation ratio of computing resources for the edge node is increased. When the data volume decreases, the allocation ratio of computing resources for the edge node is decreased;

[0064] Through a preset load balancing algorithm, computing tasks are allocated between the edge node and the cloud. When the load of the edge node is high, part of the computing tasks are migrated to the cloud. When the load of the cloud is high, part of the computing tasks are migrated to the edge node;

[0065] Through a preset resource monitoring mechanism, the usage status of computing resources of the edge node and the cloud is obtained in real time. When the resource utilization rate of the edge node reaches the preset threshold, a resource allocation adjustment strategy is triggered. When the resource utilization rate of the cloud reaches the preset threshold, a resource allocation adjustment strategy is triggered;

[0066] According to the preset resource allocation strategy, the computing resource allocation ratios of the edge node and the cloud are dynamically adjusted. When the resource utilization rate of the edge node continues to be higher than the preset threshold, the allocation ratio of computing resources for the edge node is increased. When the resource utilization rate of the cloud continues to be higher than the preset threshold, the allocation ratio of computing resources for the cloud is increased;

[0067] Through a preset resource optimization algorithm, resource optimization allocation is carried out between the edge node and the cloud. When the resource utilization rate of the edge node is lower than the preset threshold, the allocation ratio of computing resources for the edge node is decreased. When the resource utilization rate of the cloud is lower than the preset threshold, the allocation ratio of computing resources for the cloud is decreased.

[0068] Specifically, adopting an adaptive edge computing framework to dynamically adjust the computing resource allocation can flexibly allocate computing tasks according to the real-time data volume and resource status, effectively solving the problems of load imbalance and low efficiency caused by fixed resource allocation in edge computing; through the load balancing algorithm and resource optimization strategy, the system intelligently migrates tasks between the edge node and the cloud, avoiding resource overload or idleness, significantly reducing latency and improving data processing efficiency, and enhancing the flexibility and scalability of the system.

[0069] The data preprocessing and correction module obtains historical data from the cloud big data platform, preprocesses the data, uses the interpolation filling method for missing data, and corrects deviation data to obtain a complete and accurate historical data set;

[0070] Further, obtain historical data from the cloud big data platform, and preprocess the data. For missing data, use the interpolation filling method, and correct the deviation data to obtain a complete and accurate historical data set, specifically including:

[0071] S11. Obtain historical data from the cloud big data platform, identify the missing values and deviation values in the data, use the linear interpolation method to fill the missing part of the missing values, and generate a data sequence without missing values;

[0072] S12. Determine the abnormal range of the deviation values through the standard deviation and the mean, use the median substitution method to correct the deviation data, and merge the filled missing data and the corrected deviation data to generate a complete data set;

[0073] S13. Perform standardization processing on the complete data set to eliminate the dimension difference, generate a standardized data set, use the principal component analysis method to reduce the dimension of the standardized data set, extract the main feature data, and use the K-means clustering algorithm to perform clustering analysis on the feature data to generate a historical data set with accurate classification.

[0074] Specifically, by obtaining historical data from the cloud and preprocessing it, the analysis problems caused by data missing, deviation, and high-dimensional features are solved. By using linear interpolation to fill the missing values and the median substitution method to correct the deviation data, a complete and accurate data set is generated. Further, through standardization, principal component analysis for dimension reduction, and K-means clustering, key features are extracted and classified. This process effectively improves the data quality, reduces the data dimension, and enhances the training efficiency and accuracy of the subsequent deep learning model, providing a high-quality data foundation for accurate user behavior analysis and anomaly detection.

[0075] The deep learning model construction module constructs a deep learning model based on the preprocessed historical data. First, use a convolutional neural network to extract user behavior features to obtain a user behavior model; then introduce a long short-term memory network for time series analysis, and iteratively update the model in combination with real-time data to finally obtain an accurate user behavior prediction model;

[0076] Further, based on the preprocessed historical data, construct a deep learning model. First, use a convolutional neural network to extract user behavior features to obtain a user behavior model, specifically including:

[0077] Obtain the preprocessed historical data, input it into the convolutional neural network for feature extraction, perform feature extraction on the input data through multi-layer convolution and pooling operations, automatically learn the spatial and temporal correlations in the data, and generate a preliminary feature data set;

[0078] The convolutional neural network is trained based on the preliminary feature dataset, and the network parameters are continuously optimized through the backpropagation algorithm, enabling the network to better capture the key features of user behavior. The user behavior model is updated according to the feature dataset to obtain the user behavior model, thereby achieving accurate modeling and prediction of user behavior.

[0079] Specifically, by using a convolutional neural network (CNN) to perform feature extraction and model training on the preprocessed historical data, the technical problem that traditional methods are difficult to effectively capture the complex features of user behavior is solved. CNN automatically learns the spatial and temporal correlations in the data through multiple layers of convolution and pooling operations, generates a high-quality feature dataset, and optimizes the network parameters through backpropagation to accurately model user behavior. This process significantly improves the accuracy and prediction ability of the user behavior model, provides strong technical support for subsequent anomaly detection and behavior analysis, and enhances the intelligence level of the system.

[0080] Furthermore, based on the preprocessed historical data, a deep learning model is constructed. After first using a convolutional neural network to extract user behavior features and obtaining the user behavior model, it further includes:

[0081] S21. Obtain the time series data of user behavior, input it into the long short-term memory network for time series analysis to generate time series features, and use the long short-term memory network to process the real-time data to generate real-time behavior features in combination with the time series features;

[0082] S22. Update the user behavior model according to the real-time behavior features to generate an updated user behavior model, classify the user behavior to generate a user behavior classification result;

[0083] S23. When there are anomalies in the user behavior classification result, use the random forest algorithm to correct the abnormal behavior to generate a corrected classification result, and update the user behavior model again according to the corrected classification result to generate the final user behavior model;

[0084] S24. Predict the user behavior through the final user behavior model to generate an accurate user behavior prediction result.

[0085] Specifically, by combining a Convolutional Neural Network (CNN), a Long Short-Term Memory Network (LSTM), and a Random Forest algorithm, the technical problems of a single model being difficult to comprehensively capture the spatio-temporal characteristics of user behavior and misclassifying abnormal behavior are solved. First, the LSTM is used to analyze time series data to generate real-time behavior features, making up for the deficiencies of the CNN in the time dimension; subsequently, the user behavior model is updated and classified through the real-time behavior features, further enhancing the dynamic adaptability of the model. When an abnormal classification result is detected, the Random Forest algorithm is used for correction to ensure the accuracy of classification, and the model is updated again based on the correction result to generate the final user behavior model. This process realizes the accurate modeling, classification, and prediction of user behavior, significantly improving the model's recognition ability and prediction accuracy for complex user behavior.

[0086] Anomaly detection module, based on the user behavior prediction model, uses the Isolation Forest algorithm to detect anomalies in real-time data. When abnormal data is detected, it is marked as a potential security threat and relevant features are recorded;

[0087] Furthermore, based on the user behavior prediction model, the Isolation Forest algorithm is used to detect anomalies in real-time data. When abnormal data is detected, it is marked as a potential security threat and relevant features are recorded, specifically including:

[0088] Obtain time series features and input them into the Isolation Forest model. By constructing multiple random trees, isolate each data point and calculate its average path length in all trees; the shorter the path length, the easier it is for the data point to be isolated, and thus it is an abnormal point. When abnormal data is detected, it is marked as a potential security threat and relevant features of the abnormal data are recorded to generate an abnormal feature set.

[0089] Specifically, by using the Isolation Forest algorithm to detect anomalies in real-time data, the technical problem of efficiently and accurately identifying abnormal behavior in a complex data environment is solved. The Isolation Forest quickly identifies data points with shorter path lengths as abnormal points by constructing multiple random trees and calculating the average path length of data points, without relying on prior knowledge or complex parameter adjustment, effectively reducing the false alarm rate. Its technical effect is reflected in being able to quickly mark potential security threats, record relevant features of abnormal data, generate an abnormal feature set, thereby providing a reliable basis for real-time monitoring and early warning, and enhancing the anomaly detection efficiency and security of the system.

[0090] Multi-channel early warning module, responsible for notifying users of the detected abnormal information in the form of an early warning, obtaining the user's receiving channel information from the user preference database, matching the optimal transmission method for different types of early warning information, generating a multi-channel early warning plan, and adopting different push strategies according to the urgency of the early warning information;

[0091] Among them, the generation of the multi-channel early warning plan is specifically as follows: obtain the user's receiving channel preference information from the user preference database through a database query statement, clarify the priority settings of the user for different channels such as text messages, emails, and App notifications, and then classify and evaluate the priority according to the type (such as high risk, medium risk, low risk) and urgency of the early warning information. For high-risk early warnings, the system preferentially selects text messages and App notifications with strong real-time performance for pushing; for medium- and low-risk early warnings, the delivery methods are selected in turn according to the priority order of the user's preferences; then, the system uses a rule-based algorithm or a machine learning model to dynamically adjust the delivery strategy in combination with the user's historical behavior and preferences, and generate a multi-channel early warning plan; finally, the system sends early warning information through multiple channels to ensure that users can receive notifications in a timely and accurate manner; at the same time, the system will also dynamically optimize the early warning strategy according to user feedback and log data to improve the delivery efficiency of early warning information and user satisfaction.

[0092] Furthermore, according to the urgency of the early warning information, different push strategies are adopted, specifically including:

[0093] Use a classification algorithm to classify the urgency of the early warning information, preset the thresholds for high-risk and general early warnings, and select the push strategy according to the classification results;

[0094] Analyze the user's preferences from the user behavior log, establish a user preference model, determine the order of the push methods. When the urgency is high risk, in the real-time push scenario, use multi-threaded technology to synchronously call the text message, email, and App notification interfaces to ensure the push efficiency;

[0095] When the urgency is a general early warning, in the sequential push scenario, use a queue mechanism to call the push interface in turn to avoid resource conflicts, and optimize the push order by monitoring the push result log and updating the user preference model in real time.

[0096] Among them, the general early warning includes medium- and low-risk early warnings.

[0097] Specifically, through the classification algorithm based on urgency and the user preference model, the balance problem between efficiency and personalization in multi-channel early warning push is solved, the push efficiency and accuracy of early warning information are improved, ensuring that users can receive important information in a timely manner, while taking into account the user experience and the reasonable utilization of system resources.

[0098] The device parameter optimization module establishes a device parameter optimization knowledge base in the cloud, automatically matches the optimal parameter adjustment plan for the detected abnormal situation, and optimizes the parameters of the portable WiFi device through the remote control interface to ensure that the device works in a safe and efficient operating state.

[0099] Furthermore, establish a device parameter optimization knowledge base in the cloud. For the detected abnormal situations, automatically match the optimal parameter adjustment plan, and optimize the parameters of the portable Wi-Fi device through the remote control interface. Specifically, it includes:

[0100] Establish a device parameter knowledge base in the cloud in advance, use the data acquisition interface to obtain the real-time operation parameters of the portable Wi-Fi device, and judge whether there are abnormal situations in the device parameters through the preset anomaly detection algorithm;

[0101] When an anomaly is detected, automatically match the optimal parameter adjustment plan from the knowledge base, and according to the matching result, call the remote control interface to optimize the parameters of the portable Wi-Fi device;

[0102] Continuously update the device parameter knowledge base, use machine learning algorithms to optimize the anomaly detection model and parameter adjustment plan, and monitor the device operation status in real time through the remote control interface to obtain the execution effect of the optimized parameters.

[0103] Among them, the device parameter knowledge base iteratively updates the parameter adjustment plan library according to the optimization effect data to form a closed-loop optimization system.

[0104] Specifically, by establishing a device parameter optimization knowledge base in the cloud and combining the anomaly detection algorithm and the remote control interface, it solves the performance problems caused by parameter anomalies during the operation of the portable Wi-Fi device, and at the same time improves the intelligent management level of the device. It not only improves the operation efficiency and stability of the device, but also reduces the manual intervention cost and enhances the self-adaptability and intelligent level of the system.

[0105] In this embodiment, in the portable Wi-Fi device monitoring and early warning system, the intelligent perception and dynamic sampling module collects key data of the device in real time through an embedded intelligent perception chip, such as connection status, traffic usage, and signal strength, and dynamically adjusts the sampling frequency according to the device power to balance data accuracy and energy consumption. The collected data is first transmitted to the adaptive edge computing module, which dynamically allocates computing resources according to the data volume and processes the data locally. Data beyond the local processing capacity is uploaded to the cloud for distributed processing to improve the response speed. The data preprocessing and correction module in the cloud fills and corrects historical data to generate a complete and accurate data set, providing a basis for subsequent analysis. The deep learning model construction module uses this data to construct a user behavior prediction model, extracts features through a convolutional neural network and combines a long short-term memory network for time series analysis to achieve iterative update of the model. The anomaly detection module based on this model uses the isolation forest algorithm to detect real-time data, mark potential security threats, and record features. Once an anomaly is detected, the multi-channel early warning module pushes the early warning information to the user through SMS, email, or App notifications according to user preferences and the urgency of the early warning. At the same time, the device parameter optimization module matches the optimal parameter adjustment plan in the cloud and optimizes the portable Wi-Fi device through a remote control interface to ensure that the device operates in a safe and efficient state. Through the close cooperation of each module, the entire system realizes the full-process intelligent management from data collection, processing, analysis to early warning and optimization, ensuring the safety and efficiency of users' network usage.

[0106] Embodiment 2

[0107] This embodiment demonstrates the high efficiency and intelligence level of a portable Wi-Fi access device monitoring and early warning system based on big data analysis in practical applications, verifying the feasibility of its technical solution.

[0108] Specifically, a user uses a portable Wi-Fi device for daily network connection, and the device collects connection status, traffic usage, and signal strength data in real time through the intelligent perception and dynamic sampling module. Set the current power of the device to 30%, and the preset low power threshold to 20%. Therefore, the device adopts a high-frequency sampling mode and collects data every 5 seconds. When the device power drops to 18%, the system automatically switches to a low-frequency sampling mode and collects data every 30 seconds to extend the device's battery life. Through dynamically adjusting the sampling frequency, the device collected approximately 10,000 pieces of data within 24 hours, including detailed information on connection status, traffic usage, and signal strength. These data are transmitted to the adaptive edge computing module, which dynamically allocates computing resources according to the real-time data volume. Set the current data volume to 8,000 pieces, which does not exceed the local processing capacity threshold (preset to 10,000 pieces). Therefore, all data is processed at the edge node, and the processing time is 2 seconds.

[0109] In the data preprocessing and correction module, the system obtains historical data from the cloud big data platform. It is set that there are 5% missing values and 3% deviation values in the historical dataset. The missing values are filled by the linear interpolation method, and the deviation data is corrected using the median substitution method. Finally, a complete dataset containing 100,000 data records is generated. After standardization processing and dimensionality reduction by principal component analysis, 10 main features are extracted, and 5 accurately classified datasets are generated through the K-means clustering algorithm. The deep learning model construction module uses this data to train a convolutional neural network (CNN) and a long short-term memory network (LSTM). The accuracy of the finally generated user behavior prediction model reaches 95% on the test set. Based on this model, the anomaly detection module uses the Isolation Forest algorithm to detect real-time data. It is set that 2 pieces of abnormal data are detected. The system marks them as potential security threats and records the relevant features. The multi-channel warning module pushes warning information to users via App notifications according to user preferences (prefer to receive App notifications) and warning urgency (high risk), and the push success rate is 98%. At the same time, the device parameter optimization module matches the optimal parameter adjustment plan in the cloud and optimizes the portable WiFi device through the remote control interface. It is set that the signal strength of the device after optimization is increased by 15%, and the traffic usage efficiency is increased by 10%. Through the close cooperation of each module, the whole system realizes the full-process intelligent management from data collection, processing, analysis to warning and optimization, significantly improving the security and efficiency of users' network usage.

[0110] The above description is only a preferred embodiment of the present invention and does not impose any form of limitation on the present invention. Although the present invention has been disclosed as above with the preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to the equivalent embodiments with equivalent changes within the scope of the technical solution of the present invention. However, as long as it does not deviate from the content of the technical solution of the present invention, any brief modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A monitoring and early warning system for portable WiFi access devices based on big data analysis, characterized in that: Including: The intelligent perception and dynamic sampling module is responsible for the acquisition of device data and the dynamic adjustment of the sampling frequency. By embedding an intelligent perception chip, it can collect the data of the portable Wi-Fi device in real time, and dynamically adjust the sampling frequency according to the preset data accuracy requirements and the device power status; The adaptive edge computing module is responsible for local data processing and the dynamic allocation of computing resources. By adopting an adaptive edge computing framework, it can dynamically adjust the computing resource allocation according to the real-time data volume, reducing the dependence on the cloud; The data preprocessing and correction module obtains historical data from the cloud big data platform, preprocesses the data, fills in the missing data using the interpolation method, and corrects the deviation data to obtain a complete and accurate historical data set; The deep learning model construction module constructs a deep learning model based on the preprocessed historical data. First, it uses a convolutional neural network to extract user behavior features to obtain a user behavior model; then it introduces a long short-term memory network for time series analysis, and iteratively updates the model in combination with real-time data to finally obtain an accurate user behavior prediction model; The anomaly detection module uses the isolation forest algorithm to detect anomalies in real-time data based on the user behavior prediction model. When detecting abnormal data, it marks it as a potential security threat and records the relevant features; The multi-channel warning module is responsible for notifying users of the detected abnormal information in the form of a warning. It obtains the user's receiving channel information from the user preference database, matches the optimal transmission method for different types of warning information, generates a multi-channel warning plan, and adopts different push strategies according to the urgency of the warning information; The device parameter optimization module establishes a device parameter optimization knowledge base in the cloud. For the detected abnormal situations, it automatically matches the optimal parameter adjustment plan and optimizes the parameters of the portable Wi-Fi device through the remote control interface.

2. The monitoring and early warning system for portable WiFi access devices based on big data analysis according to claim 1, wherein: Dynamically adjust the sampling frequency according to the preset data accuracy requirements and the device power status, specifically including: The intelligent perception chip is embedded in the portable Wi-Fi device to collect the connection status, traffic usage, and signal strength data of the device in real time, obtain the current power status of the device, and judge whether the power is lower than the preset threshold. When the power is lower than the threshold, the low-frequency sampling mode is adopted; when the power is higher than the threshold, the high-frequency sampling mode is adopted. Adjust the data acquisition accuracy according to the sampling frequency, store and analyze the collected data, and form a device operation status report.

3. The monitoring and early warning system for portable WiFi access devices based on big data analysis according to claim 1, characterized in that: Adopt an adaptive edge computing framework to dynamically adjust the computing resource allocation according to the real-time data volume, specifically including: Adopt an adaptive edge computing framework to obtain the real-time data volume, and judge whether the data volume exceeds the preset local processing capacity threshold. When it exceeds the threshold, part of the data is uploaded to the cloud for distributed processing; when it does not exceed the threshold, the processing is preferentially completed at the edge node; Dynamically adjust the computing resource allocation strategy according to the real-time data volume and the local computing resource status. When the data volume continues to increase, increase the computing resource allocation ratio of the edge node; when the data volume decreases, reduce the computing resource allocation ratio of the edge node; Allocate computing tasks between edge nodes and the cloud through a preset load balancing algorithm. When the load on an edge node is high, migrate some computing tasks to the cloud; when the load on the cloud is high, migrate some computing tasks to the edge node. Obtain the computing resource usage status of edge nodes and the cloud in real time through a preset resource monitoring mechanism. When the resource utilization rate of an edge node reaches a preset threshold, trigger a resource allocation adjustment strategy; when the resource utilization rate of the cloud reaches a preset threshold, trigger a resource allocation adjustment strategy. Dynamically adjust the computing resource allocation ratio between edge nodes and the cloud according to a preset resource allocation strategy. When the resource utilization rate of an edge node continuously exceeds the preset threshold, increase the computing resource allocation ratio of the edge node; when the resource utilization rate of the cloud continuously exceeds the preset threshold, increase the computing resource allocation ratio of the cloud. Conduct resource optimization allocation between edge nodes and the cloud through a preset resource optimization algorithm. When the resource utilization rate of an edge node is lower than the preset threshold, reduce the computing resource allocation ratio of the edge node; when the resource utilization rate of the cloud is lower than the preset threshold, reduce the computing resource allocation ratio of the cloud.

4. The monitoring and early warning system for portable WiFi access devices based on big data analysis according to claim 1, characterized in that: Obtain historical data from the cloud big data platform and preprocess the data. For missing data, use the interpolation filling method, and for deviated data, perform correction processing to obtain a complete and accurate historical data set, specifically including: Obtain historical data from the cloud big data platform, identify missing values and deviated values in the data, and use the linear interpolation method to fill in the missing parts of the missing values to generate a data sequence without missing values. Determine the abnormal range of deviated values through standard deviation and mean, use the median substitution method to correct the deviated data, and merge the filled missing data and corrected deviated data to generate a complete data set. Perform standardization processing on the complete data set to eliminate the dimension difference, generate a standardized data set, use the principal component analysis method to reduce the dimension of the standardized data set, extract the main feature data, and use the K-means clustering algorithm to perform clustering analysis on the feature data to generate a classified historical data set.

5. The monitoring and early warning system for portable WiFi access devices based on big data analysis according to claim 1, characterized in that: Based on the preprocessed historical data, construct a deep learning model. First, use a convolutional neural network to extract user behavior features to obtain a user behavior model, specifically including: Obtain the preprocessed historical data, input it into the convolutional neural network for feature extraction, perform feature extraction on the input data through multiple convolutional and pooling operations, automatically learn the spatial and temporal correlations in the data, and generate a preliminary feature data set. Train the convolutional neural network according to the preliminary feature data set, continuously optimize the network parameters through the backpropagation algorithm to enable the network to better capture the key features of user behavior, and update the user behavior model according to the feature data set to obtain the user behavior model.

6. The monitoring and early warning system for portable WiFi access devices based on big data analysis according to claim 5, wherein: After constructing a deep learning model based on the preprocessed historical data and first using a convolutional neural network to extract user behavior features to obtain a user behavior model, it also includes: Obtain the time series data of user behavior, input it into a long short-term memory network for time series analysis to generate time series features, use the long short-term memory network to process real-time data, and combine the time series features to generate real-time behavior features. Update the user behavior model according to real-time behavior characteristics to generate an updated user behavior model, classify user behaviors to generate user behavior classification results; When there are anomalies in the user behavior classification results, use the random forest algorithm to correct the abnormal behaviors to generate corrected classification results, and update the user behavior model again according to the corrected classification results to generate the final user behavior model; Predict user behaviors through the final user behavior model to generate accurate user behavior prediction results.

7. The monitoring and early warning system for portable WiFi access devices based on big data analysis according to claim 1, characterized in that: Based on the user behavior prediction model, use the isolation forest algorithm to detect anomalies in real-time data. When abnormal data is detected, mark it as a potential security threat and record relevant features, specifically including: Obtain time series features and input them into the isolation forest model. By constructing multiple random trees, isolate each data point and calculate its average path length in all trees; the shorter the path length, the easier it is for the data point to be isolated, and thus it is an abnormal point. When abnormal data is detected, mark it as a potential security threat and record the relevant features of the abnormal data to generate an abnormal feature set.

8. The monitoring and early warning system for a portable WiFi access device based on big data analysis according to claim 1, characterized in that: According to the urgency of the warning information, adopt different push strategies, specifically including: Use a classification algorithm to classify the urgency of the warning information, preset the thresholds for high-risk and general warnings, and select a push strategy according to the classification results; Analyze user preferences from the user behavior log, establish a user preference model, and determine the order of push methods. When the urgency is high risk, perform a real-time push scenario and synchronously call SMS, email, and App notification interfaces using multi-threading technology; When the urgency is a general warning, perform a sequential push scenario, use a queue mechanism to sequentially call the push interface, and optimize the push order by monitoring the push result log and updating the user preference model in real-time.

9. A monitoring and warning system for a portable WiFi access device based on big data analysis according to claim 1, characterized in that: Establish a device parameter optimization knowledge base in the cloud. For the detected abnormal situations, automatically match the optimal parameter adjustment plan, and optimize the parameters of the portable WiFi device through the remote control interface, specifically including: Pre-establish a device parameter knowledge base in the cloud, use a data acquisition interface to obtain the real-time operating parameters of the portable WiFi device, and use a preset anomaly detection algorithm to determine whether there are abnormal situations in the device parameters; When an anomaly is detected, automatically match the optimal parameter adjustment plan from the knowledge base, and according to the matching result, call the remote control interface to optimize the parameters of the portable WiFi device; Continuously update the device parameter knowledge base, use machine learning algorithms to optimize the anomaly detection model and parameter adjustment plan, and monitor the device operating status in real-time through the remote control interface to obtain the execution effect of the optimized parameters.

Images