Multi-dimensional dynamic monitoring platform and method for network service quality

Abstract

The present invention discloses a multi-dimensional dynamic monitoring platform and method for network service quality, which belongs to the field of network communication technology, wherein the platform includes: a delay fluctuation sequence acquisition module, which is used to monitor the network delay and packet loss rate within a duration, and perform delay fluctuation analysis; an implicit packet loss rate sequence acquisition module, which is used to perform packet loss prediction, obtain a delay packet loss rate sequence and a fluctuation packet loss rate sequence, calculate and obtain a predicted packet loss rate sequence, and calculate and obtain an implicit packet loss rate sequence; an implicit delay fluctuation sequence acquisition module, which is used to perform retransmission delay prediction, and calculate and obtain an implicit delay fluctuation sequence; a quality level acquisition module, which is used to perform implicit fault stability calculation, and classify and obtain the quality level of the network service as a monitoring result of the network service quality. The present application solves the technical problem that the prior art lacks multi-dimensional dynamic analysis capabilities for network service quality monitoring, which leads to inaccurate network performance evaluation and untimely fault warning.

Description

Technical Field

[0001] The present invention relates to the field of network communication technology, and in particular to a multi-dimensional dynamic monitoring platform and method for network service quality. Background Art

[0002] With the rapid development and widespread popularity of Internet technology, network services have become an important infrastructure in modern society. From video conferencing to e-commerce, from cloud computing to the Internet of Things, network stability and quality are crucial to user experience and business operations. Realizing multi-dimensional dynamic monitoring and analysis of network performance has become an important technical requirement for ensuring efficient network services.

[0003] At present, network service quality monitoring mainly relies on the collection and analysis of basic performance indicators. However, these indicators can only reflect surface problems and cannot deeply reveal potential faults hidden in fluctuations. Some studies have tried to use prediction models to infer network risks, but the lack of fusion analysis of multi-dimensional data has resulted in incomplete results. In addition, when evaluating the network service quality level, existing systems pay more attention to the discrete values ​​of actual indicators and ignore the stability analysis of hidden faults, making it difficult to deal with dynamic problems in complex network environments. Summary of the invention

[0004] This application provides a multi-dimensional dynamic monitoring platform and method for network service quality, aiming to solve the technical problems in the prior art of lacking multi-dimensional dynamic analysis capabilities for network service quality monitoring, which leads to inaccurate network performance evaluation and untimely fault warning.

[0005] In view of the above problems, the present application provides a multi-dimensional dynamic monitoring platform and method for network service quality.

[0006] The first aspect disclosed in the present application provides a multi-dimensional dynamic monitoring platform for network service quality, which includes a delay fluctuation sequence acquisition module, which is used to monitor the network delay and packet loss rate within a duration during the network service process, obtain the network delay sequence and the packet loss rate sequence, and perform delay fluctuation analysis based on the network delay sequence to obtain the delay fluctuation sequence; an implicit packet loss rate sequence acquisition module, which is used to perform packet loss prediction based on the network delay sequence and the delay fluctuation sequence, respectively, to obtain the delay packet loss rate sequence and the fluctuation packet loss rate sequence, calculate to obtain the predicted packet loss rate sequence, and calculate to obtain the implicit packet loss rate sequence in combination with the packet loss rate sequence; an implicit delay fluctuation sequence acquisition module, which is used to perform retransmission delay prediction based on the packet loss rate sequence, obtain the predicted delay fluctuation sequence, and calculate to obtain the implicit delay fluctuation sequence in combination with the delay fluctuation sequence; a quality level acquisition module, which is used to perform implicit fault stability calculation based on the implicit packet loss rate sequence and the implicit delay fluctuation sequence, obtain the implicit fault stability coefficient, and classify and obtain the quality level of the network service as the monitoring result of the network service quality.

[0007] Another aspect disclosed in the present application provides a multi-dimensional dynamic monitoring method for network service quality, the method comprising: monitoring network delay and packet loss rate within a duration during a network service process to obtain a network delay sequence and a packet loss rate sequence; performing delay fluctuation analysis based on the network delay sequence to obtain a delay fluctuation sequence; performing packet loss prediction based on the network delay sequence and the delay fluctuation sequence to obtain a delay packet loss rate sequence and a fluctuation packet loss rate sequence, calculating a predicted packet loss rate sequence, and combining the packet loss rate sequence to obtain an implicit packet loss rate sequence; performing retransmission delay prediction based on the packet loss rate sequence to obtain a predicted delay fluctuation sequence, and combining the delay fluctuation sequence to obtain an implicit delay fluctuation sequence; performing implicit fault stability calculation based on the implicit packet loss rate sequence and the implicit delay fluctuation sequence to obtain an implicit fault stability coefficient, and classifying to obtain a quality level of the network service as a monitoring result of the network service quality.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] Due to the technical solution of a multi-dimensional dynamic monitoring platform for network service quality composed of a delay fluctuation sequence acquisition module, an implicit packet loss rate sequence acquisition module, an implicit delay fluctuation sequence acquisition module and a quality grade acquisition module, the network delay sequence and packet loss rate sequence are acquired by monitoring the network delay and packet loss rate, and the delay fluctuation sequence, the implicit packet loss rate sequence and the implicit delay fluctuation sequence are further calculated, and finally the implicit fault stability is calculated and the quality grade is output. This solves the technical problem of the lack of multi-dimensional dynamic analysis capability for network service quality monitoring in the prior art, which leads to inaccurate network performance evaluation and untimely fault warning, and achieves the technical effect of comprehensively and accurately reflecting the changes in network service quality, providing reliable service quality evaluation and timely fault warning.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 A structural diagram of a multi-dimensional dynamic monitoring platform for network service quality is provided for an embodiment of the present application.

[0012] Figure 2 A flow chart of a multi-dimensional dynamic monitoring method for network service quality is provided for an embodiment of the present application.

[0013] Explanation of the reference numerals: delay fluctuation sequence obtaining module 11 , implicit packet loss rate sequence obtaining module 12 , implicit delay fluctuation sequence obtaining module 13 , quality level obtaining module 14 . DETAILED DESCRIPTION

[0014] The overall idea of ​​the technical solution provided by this application is as follows:

[0015] The embodiment of the present application provides a multi-dimensional dynamic monitoring platform and method for network service quality. By monitoring network delay and packet loss rate, delay fluctuation sequence and packet loss rate sequence are generated, and the implicit packet loss rate and implicit delay fluctuation sequence are calculated based on the deviation between theoretical prediction and actual data to identify potential faults. Through implicit fault stability analysis, the quality level of network services is further evaluated.

[0016] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically described below in conjunction with the drawings in the specification.

[0017] Embodiment 1, as Figure 1 As shown, the embodiment of the present application provides a multi-dimensional dynamic monitoring platform for network service quality, which includes:

[0018] The delay fluctuation sequence acquisition module 11 is used to monitor the network delay and packet loss rate within a duration during the network service process, obtain the network delay sequence and packet loss rate sequence, and perform delay fluctuation analysis based on the network delay sequence to obtain the delay fluctuation sequence.

[0019] Specifically, the network delay series is a collection of delay values ​​collected during the monitoring process, usually indexed by timestamps to form a time series. Delay fluctuation refers to the degree of change in network delay at a certain point in time compared to the delay at other points in time. The packet loss rate series refers to a data series that records the change in packet loss rate over a period of time, which is used to analyze the dynamic trend of packet loss.

[0020] In the process of network service, the network delay and packet loss rate of multiple data transmissions within a certain period of time are collected through network monitoring tools to generate two sequences: network delay sequence and packet loss rate sequence. Specifically, the network traffic is monitored and the network delay values ​​at different time points are recorded to form a time series. Based on the network delay sequence, the delay value changes at any two time points are compared to calculate the fluctuation size.

[0021] Through delay fluctuation analysis, the performance changes of network services at different time points can be accurately quantified. This fine-grained fluctuation capture method can promptly discover hidden problems such as network jitter and packet retransmission, and is particularly suitable for scenarios with high stability requirements. Compared with single delay monitoring, fluctuation analysis can more comprehensively reflect the dynamic characteristics of network quality and provide data support for further optimizing network performance.

[0022] The implicit packet loss rate sequence acquisition module 12 is used to perform packet loss prediction according to the network delay sequence and the delay fluctuation sequence, obtain a delay packet loss rate sequence and a fluctuation packet loss rate sequence, calculate and obtain a predicted packet loss rate sequence, and calculate and obtain an implicit packet loss rate sequence in combination with the packet loss rate sequence.

[0023] Specifically, the delay packet loss rate sequence refers to the set of packet loss rates predicted based on the network delay sequence, reflecting the impact of delay changes on packet loss. The fluctuation packet loss rate sequence refers to the set of packet loss rates predicted based on the delay fluctuation sequence, reflecting the impact of delay fluctuations on packet loss. The predicted packet loss rate sequence is a set of packet loss rate prediction values ​​calculated by combining the delay packet loss rate sequence and the fluctuation packet loss rate sequence. The implicit packet loss rate sequence refers to the deviation value sequence calculated by combining the actual packet loss rate sequence with the predicted packet loss rate sequence, which is used to further analyze the cause of packet loss.

[0024] First, a prediction model is trained based on historical network data (such as delay sequence, delay fluctuation sequence, and actual packet loss rate sequence) to construct a delay packet loss prediction path and a fluctuation packet loss prediction path. Common models include linear regression, support vector machine, or deep learning model. Then, the network delay sequence and delay fluctuation sequence are input into the packet loss prediction model respectively, and the delay packet loss rate sequence and the fluctuation packet loss rate sequence are calculated respectively. Specifically, the delay packet loss rate sequence and the fluctuation packet loss rate sequence are weighted averaged or directly averaged to obtain the predicted packet loss rate sequence. The actual packet loss rate sequence and the predicted packet loss rate sequence are calculated by difference to obtain the implicit packet loss rate sequence.

[0025] By calculating the implicit packet loss rate sequence, we can effectively separate the systematic and random factors in the packet loss rate, thus helping to locate the network performance bottleneck. Compared with analyzing only the actual packet loss rate, the implicit packet loss rate sequence more accurately reflects the combined impact of delay and delay fluctuation.

[0026] The implicit delay fluctuation sequence acquisition module 13 is used to perform retransmission delay prediction according to the packet loss rate sequence to obtain a predicted delay fluctuation sequence, and calculate and obtain an implicit delay fluctuation sequence in combination with the delay fluctuation sequence.

[0027] Specifically, the predicted delay fluctuation sequence refers to a series of predicted values ​​output by the retransmission delay prediction channel, which represents the predicted delay fluctuation at different time points. The implicit delay fluctuation sequence refers to a sequence calculated by combining the actual measured delay fluctuation sequence with the predicted delay fluctuation sequence, which is used to quantify the implicit delay fluctuation that is not directly measured.

[0028] The packet loss rate sequence within a period of time is obtained from the network monitoring data as the basic input for predicting the retransmission delay. Using the pre-trained retransmission delay prediction channel, the packet loss rate sequence is input into the model one by one, and the corresponding predicted delay fluctuation sequence is output. The delay fluctuation sequence obtained from real-time monitoring is used as the actual measured network delay fluctuation value. By subtracting the predicted delay fluctuation sequence from the actual delay fluctuation sequence, the implicit delay fluctuation sequence is calculated to reflect the potential delay fluctuation that the prediction model has not fully captured. Specifically, implicit delay fluctuation = actual delay fluctuation - predicted delay fluctuation.

[0029] By quantifying the implicit delay fluctuation, we can distinguish the actual delay fluctuation caused by packet loss and retransmission from the other potential factors. The implicit delay fluctuation sequence provides network administrators with a clear optimization direction.

[0030] The quality level acquisition module 14 is used to perform implicit fault stability calculation according to the implicit packet loss rate sequence and the implicit delay fluctuation sequence, obtain implicit fault stability coefficient, and classify and obtain the quality level of the network service as the monitoring result of the network service quality.

[0031] Specifically, implicit fault stability refers to the quantification of the stability level of network faults by analyzing the changes in implicit packet loss rate and implicit delay fluctuations. The implicit fault stability coefficient is a numerical indicator used to indicate the stability of network services, which is calculated based on the statistical results of implicit packet loss rate and implicit delay fluctuations. The network service quality level refers to the service quality evaluation level obtained by classifying the implicit fault stability coefficient, which is used to reflect the reliability and stability of network services in different time periods.

[0032] To calculate the implicit fault stability based on the implicit packet loss rate sequence and the implicit delay fluctuation sequence, it is necessary to first extract the data points in each sequence, such as the implicit packet loss rate and implicit delay fluctuation at a specific timestamp. By randomly sampling a number of implicit packet loss rate values ​​and implicit delay fluctuation values, the difference between each timestamp data point and the random sampling mean is calculated to obtain the implicit stability difference. Furthermore, the ratio of the difference to the current time point value is used as the implicit stability coefficient, forming the implicit packet loss rate stability coefficient sequence and the implicit delay fluctuation stability coefficient sequence respectively.

[0033] Subsequently, the implicit fault stability coefficient is obtained by calculating the mean of the two sets of sequences at all timestamps. Finally, the coefficient is input into a pre-trained classification model (such as a decision tree or support vector machine), and the quality level of the network service is output based on the mapping relationship between the implicit fault stability coefficient and the service quality level in the historical data. For example, when the implicit fault stability coefficient is high, the classification model determines that the service is at a "poor" level, while a low stability coefficient corresponds to an "excellent" level.

[0034] This step can accurately identify the stability changes of hidden network faults, thereby making a more scientific classification of service quality. This not only improves the accuracy of network monitoring, but also helps operators discover potential network problems earlier and develop effective optimization plans.

[0035] Furthermore, the delay fluctuation sequence acquisition module 11 is also used to perform the following steps: during the network service process, monitor the network delay and packet loss rate at multiple timestamps within the duration, and construct a network delay sequence and a packet loss rate sequence; extract the first network delay at the first timestamp in the network delay sequence, perform delay fluctuation analysis, and obtain the first delay fluctuation; continue to perform delay fluctuation analysis on other network delays to obtain a delay fluctuation sequence.

[0036] Specifically, the delay fluctuation series is a record of the network delay fluctuation values ​​over a period of time, which is used to reflect the fluctuation trend of network delay.

[0037] First, collect network delay and packet loss rate data during the monitoring time. You can use Wireshark or PingPlotter to record at multiple timestamps. Then, in the network delay sequence, select the delay value at a certain timestamp and use this value as the benchmark for delay fluctuation analysis. Specifically, calculate the difference between the delay value at the first time point and the delay value at the next time point to obtain the first fluctuation value. Randomly select several points from the sequence and calculate the difference between these points and the benchmark value. The above method processes the delay values ​​of all timestamps in the sequence in turn to obtain a complete delay fluctuation sequence.

[0038] By monitoring network delay and packet loss rate, the generated network delay series and packet loss rate series can reflect the dynamic changes of network performance, while the delay fluctuation series further quantifies the stability of network delay. Compared with single delay or packet loss rate data, the delay fluctuation series provides more comprehensive information and reveals hidden network instability problems.

[0039] Furthermore, a first network delay at a first timestamp is extracted from the network delay sequence, and a delay fluctuation analysis is performed to obtain a first delay fluctuation, including: randomly extracting a number of random network delays from the network delay sequence; and calculating a delay difference between an average of the number of random network delays and the first network delay as the first delay fluctuation.

[0040] Specifically, the first timestamp is the first recorded time point in the network delay sequence. For example, in the above sequence, the recording point corresponding to the first timestamp may be the 1st second. The first network delay refers to the network delay value corresponding to the first timestamp, which is called the first network delay. Random network delay refers to a number of delay data points randomly selected from the network delay sequence.

[0041] First, within a set time period, network delay data is collected through network monitoring tools (such as traffic capture devices or server-side delay records) to form a delay sequence. For example, the delay sequence is obtained by sampling within 10 seconds. The delay value corresponding to the first timestamp is extracted from the delay sequence, that is, the first value in the sequence. Several data points are randomly selected from the delay sequence, which do not include the first network delay, for further analysis. The randomly extracted delay values ​​are averaged. The delay fluctuation value is obtained by calculating the difference between the first network delay and the mean of the random delay. The above process is repeatedly applied to other timestamps in the delay sequence to form a complete delay fluctuation sequence, reflecting the dynamic change trend of network delay.

[0042] Through random sampling and delay fluctuation calculation, this method can comprehensively evaluate the dynamic characteristics of network delay. Compared with using only the average delay value, this method pays more attention to reflecting the volatility and transient changes of delay. Using the delay fluctuation index, potential network problems can be better discovered and located, and the accuracy and reliability of network optimization and service quality evaluation can be improved.

[0043] Furthermore, the implicit packet loss rate sequence acquisition module 12 is also used to perform the following steps: pre-training a packet loss prediction channel, wherein the packet loss prediction channel includes a delayed packet loss prediction path and a fluctuating packet loss prediction path; inputting multiple network delays in the network delay sequence into the delayed packet loss prediction path, and predicting and outputting multiple delayed packet loss rates to construct a delayed packet loss rate sequence; inputting multiple delay fluctuations in the delay fluctuation sequence into the fluctuating packet loss prediction path, and predicting and outputting multiple fluctuating packet loss rates to construct a fluctuating packet loss rate sequence; calculating the mean of each delayed packet loss rate and the corresponding fluctuating packet loss rate in the delayed packet loss rate sequence and the fluctuating packet loss rate sequence to obtain a predicted packet loss rate sequence; and according to the mapping relationship between the packet loss rate sequence and the predicted packet loss rate sequence, subtracting the corresponding predicted packet loss rate from each packet loss rate to obtain an implicit packet loss rate sequence.

[0044] Specifically, the packet loss prediction channel includes a delay packet loss prediction path and a fluctuation packet loss prediction path, which are used to predict the contribution of delay and delay fluctuation to the packet loss rate. The delay packet loss prediction path refers to predicting the packet loss rate using the network delay sequence. The fluctuation packet loss prediction path refers to predicting the packet loss rate using the delay fluctuation sequence.

[0045] First, historical network log data, including delay sequence, delay fluctuation sequence and their corresponding packet loss rate data, are used to train the delay packet loss prediction path and the fluctuation packet loss prediction path respectively. The network delay sequence collected in real time is input into the delay packet loss prediction path to obtain the delay packet loss rate sequence; the delay fluctuation sequence is input into the fluctuation packet loss prediction path to obtain the fluctuation packet loss rate sequence. The delay packet loss rate and the fluctuation packet loss rate are weighted or directly averaged to obtain the predicted packet loss rate sequence. The actual packet loss rate sequence is calculated by difference with the predicted packet loss rate sequence to obtain the implicit packet loss rate sequence.

[0046] Through the packet loss prediction channel, we can efficiently analyze the combined impact of network delay and delay fluctuation on packet loss, and provide prediction results to assist network optimization. At the same time, the calculation of implicit packet loss rate series can reveal potential reasons for packet loss that are not captured by existing models, such as physical link failures or device performance bottlenecks.

[0047] Furthermore, the pre-training of the packet loss prediction channel includes: collecting a sample network delay set and a sample delay fluctuation set according to the network log data in the historical time, respectively as input data, and collecting a sample packet loss rate set as supervision data; respectively using the sample network delay set and the sample delay fluctuation set as input data, using the sample packet loss rate set as supervision data, and using machine learning to construct and train a delay packet loss prediction path and a fluctuation packet loss prediction path; combining the trained delay packet loss prediction path and fluctuation packet loss prediction path to obtain a packet loss prediction channel.

[0048] Specifically, the sample network delay set refers to the network delay data set collected in historical time, which is used to train the delay packet loss prediction path. The sample delay fluctuation set refers to the delay fluctuation data set calculated in historical time, which is used to train the fluctuation packet loss prediction path. The sample packet loss rate set corresponds to the historical packet loss rate data set of network delay and fluctuation, which is used as the supervision label for model training.

[0049] Firstly, network delay, delay fluctuation and packet loss rate data are extracted from historical network logs, and a sample network delay set, a sample delay fluctuation set and a sample packet loss rate set are constructed respectively.

[0050] Using the sample network delay set as input and the sample packet loss rate set as supervision data, a machine learning algorithm is used to train the delay packet loss prediction path. Specifically, the delay value and packet loss rate are normalized to ensure that the input features and labels are within a uniform range. Select a support vector machine or multi-layer perceptron as the model architecture. Optimize the model parameters based on the training data set so that it can accurately predict the packet loss rate. Using the sample delay fluctuation set as input data and the sample packet loss rate set as supervision data, machine learning is used to train the fluctuation packet loss prediction path. Similar preprocessing and training process as the delay packet loss prediction path.

[0051] The trained delay packet loss prediction path and fluctuation packet loss prediction path are combined to form a packet loss prediction channel. This channel can simultaneously receive inputs of network delay and delay fluctuation, and predict the impact of both on the packet loss rate, providing a basis for further calculation of the implicit packet loss rate.

[0052] By pre-training the packet loss prediction channel, we can achieve quantitative prediction of the combined impact of network delay and delay fluctuation on packet loss rate, and improve the accuracy of packet loss rate prediction by modeling independent paths of delay and fluctuation. We can identify the impact mode of delay and fluctuation on packet loss under different network conditions, and provide a more refined decision-making basis for network management.

[0053] Furthermore, the implicit delay fluctuation sequence acquisition module 13 is also used to perform the following steps: input multiple packet loss rates in the packet loss rate sequence into a pre-trained retransmission delay prediction channel, predict and output multiple predicted delay fluctuations, and construct a predicted delay fluctuation sequence, wherein the retransmission delay prediction channel is obtained by training with a sample packet loss rate set and a sample delay fluctuation set; according to the mapping relationship between the predicted delay fluctuation sequence and the delay fluctuation sequence, subtract the corresponding predicted delay fluctuation from each delay fluctuation to obtain an implicit delay fluctuation sequence.

[0054] Specifically, the retransmission delay prediction channel refers to a model pre-trained by machine learning technology, which is used to input packet loss rate data and output predicted delay fluctuations. The predicted delay fluctuation sequence refers to a series of predicted delay fluctuation values ​​generated by the retransmission delay prediction channel according to the packet loss rate sequence.

[0055] The packet loss rate values ​​at each time point are obtained from the network monitoring data to form a packet loss rate sequence. The retransmission delay prediction channel is constructed by using the historical network log data through the machine learning model. Specifically, the sample packet loss rate set and the sample delay fluctuation set in the historical network log are collected as input data and supervision data, respectively. Then, the prediction model is trained based on these data using supervised learning techniques, such as regression analysis or deep learning models (such as LSTM or GRU). By optimizing the loss function (such as mean square error), the model can accurately predict the impact of packet loss rate on delay fluctuation. Finally, the trained models are combined to form an end-to-end retransmission delay prediction channel, which is used to input the real-time packet loss rate sequence and output the corresponding predicted delay fluctuation sequence. The packet loss rate sequence is input into the trained retransmission delay prediction channel, and the corresponding delay fluctuation value is predicted and output to construct the predicted delay fluctuation sequence. The delay fluctuation sequence is extracted from the real-time network monitoring data. By subtracting the predicted delay fluctuation from the actual delay fluctuation at each time point, the implicit delay fluctuation sequence is obtained to quantify the potential factors not captured by the prediction model.

[0056] By analyzing the difference between the actual delay and the predicted value, the accuracy of the retransmission delay prediction model is improved, making subsequent service optimization more targeted. The analysis of implicit delay fluctuations enables the monitoring platform to provide more detailed network performance diagnosis reports.

[0057] Further, the quality level acquisition module 14 is further configured to perform the following steps: extracting a first implied packet loss rate at a first timestamp in the implied packet loss rate sequence, and randomly extracting a plurality of random implied packet loss rates in the implied packet loss rate sequence; calculating an implied packet loss rate difference between an average of the plurality of random implied packet loss rates and the first implied packet loss rate, and calculating a ratio of the implied packet loss rate difference to the first implied packet loss rate as a first implied packet loss rate stability coefficient; continuing to calculate to obtain an implied packet loss rate stability coefficient sequence of the implied packet loss rate sequence, and calculating to obtain the implied packet loss rate stability coefficient sequence; An implicit fluctuation stability coefficient sequence containing a delay fluctuation sequence; calculating the implicit packet loss rate stability coefficient sequence and the mean of the implicit packet loss rate stability coefficient and the implicit fluctuation stability coefficient at each timestamp in the implicit fluctuation stability coefficient sequence to obtain an implicit fault stability coefficient sequence, and calculating the mean to obtain an implicit fault stability coefficient; inputting the implicit fault stability coefficient into a network service quality classifier, classifying and mapping to obtain a network service quality level, wherein the network service quality classifier is constructed based on a mapping relationship between a sample implicit fault stability coefficient set and a sample network service quality level set.

[0058] Specifically, the implicit packet loss rate series refers to a set of time series calculated by the difference between the real-time packet loss rate and the predicted packet loss rate, reflecting the potential packet loss phenomenon. The implicit packet loss rate series can reveal potential network problems other than explicit packet loss. The implicit packet loss rate stability coefficient refers to an indicator to measure the volatility of the implicit packet loss rate, which is calculated by the difference and ratio between the implicit packet loss rate at a certain moment and the random sampling mean. The network service quality classifier is a model trained based on historical sample data using the implicit fault stability coefficient as input, which is used to classify the current network service quality into "excellent", "medium" or "poor" levels.

[0059] Extract the implicit packet loss rate value of a timestamp (such as time T1) from the implicit packet loss rate sequence, and randomly extract several other implicit packet loss rate values. Calculate the mean of these values, find the difference with the current value, and standardize the difference as a ratio to obtain the first implicit packet loss rate stability coefficient. Repeat this step for other timestamps in the sequence to form an implicit packet loss rate stability coefficient sequence. For the implicit delay fluctuation sequence, use a method similar to the packet loss rate stability coefficient calculation. Extract the implicit delay fluctuation value at a certain time point, randomly sample the mean of other values, calculate the difference and standardize it as a ratio to obtain the implicit delay fluctuation stability coefficient sequence. Calculate the mean of the implicit packet loss rate stability coefficient and the implicit delay fluctuation stability coefficient at each timestamp as the implicit fault stability coefficient sequence. Calculate the mean of the entire sequence to obtain the overall implicit fault stability coefficient.

[0060] Input the implicit fault stability coefficient into the network service quality classifier. The classifier maps the input to the quality level based on the pre-trained model. Specifically, a large amount of network operation data is collected, and the implicit packet loss rate stability coefficient and the implicit delay fluctuation stability coefficient are extracted as input features, and the corresponding service quality levels (such as "excellent", "medium", and "poor") are marked. Then, the sample data is trained using a machine learning algorithm (such as decision tree, random forest, or deep learning) to optimize the classification boundary. After training, the classifier can quickly predict the network service quality level based on the current implicit fault stability coefficient, and realize accurate classification and real-time monitoring.

[0061] By calculating implicit fault stability and classifying quality levels, we can effectively capture hidden fault tendencies in the network and provide quantitative indicators and early warning capabilities for network operation and maintenance.

[0062] In summary, the multi-dimensional dynamic monitoring platform for network service quality provided by the embodiments of the present application has the following technical effects:

[0063] 1. By establishing a multi-dimensional dynamic monitoring platform for network service quality, it is possible to comprehensively monitor various indicators such as network delay and packet loss rate, analyze network performance changes in real time, generate implicit fault-related data, improve the accuracy and reliability of network quality assessment, and provide an accurate basis for operation, maintenance and optimization.

[0064] 2. By monitoring network delay and packet loss rate within a specific duration to form serialized data, combined with delay fluctuation analysis, it is possible to capture the trend of network performance changes, reflect network stability and abnormal fluctuations, and provide support for network fault diagnosis.

[0065] 3. By constructing a retransmission delay prediction channel and combining it with delay fluctuation data, an implicit delay fluctuation sequence is obtained, which can more accurately reflect the retransmission delay caused by network failures and improve the detection and diagnosis efficiency of network delay-related problems.

[0066] 4. Through the calculation and classification of implicit fault stability coefficients and combined with the training results of the network service quality classifier, a hierarchical evaluation of network service quality can be achieved, which can intuitively reflect the network status, provide data support for network optimization and decision-making, and improve management efficiency.

[0067] Embodiment 2 is based on the same inventive concept as the multi-dimensional dynamic monitoring platform for network service quality in the previous embodiment. Figure 2 As shown, the embodiment of the present application provides a multi-dimensional dynamic monitoring method for network service quality, the method comprising:

[0068] Step S100: During the network service process, monitor the network delay and packet loss rate within the duration, obtain the network delay sequence and the packet loss rate sequence, perform delay fluctuation analysis according to the network delay sequence, and obtain the delay fluctuation sequence; Step S200: Perform packet loss prediction according to the network delay sequence and the delay fluctuation sequence, respectively, obtain the delay packet loss rate sequence and the fluctuation packet loss rate sequence, calculate and obtain the predicted packet loss rate sequence, and calculate and obtain the implicit packet loss rate sequence in combination with the packet loss rate sequence; Step S300: Perform retransmission delay prediction according to the packet loss rate sequence, obtain the predicted delay fluctuation sequence, and calculate and obtain the implicit delay fluctuation sequence in combination with the delay fluctuation sequence; Step S400: Perform implicit fault stability calculation according to the implicit packet loss rate sequence and the implicit delay fluctuation sequence, obtain the implicit fault stability coefficient, and classify and obtain the quality level of the network service as the monitoring result of the network service quality.

[0069] Furthermore, during the network service process, the network delay and packet loss rate within the duration are monitored to obtain a network delay sequence and a packet loss rate sequence, and a delay fluctuation analysis is performed based on the network delay sequence to obtain a delay fluctuation sequence, including: during the network service process, the network delay and packet loss rate at multiple timestamps within the duration are monitored to construct a network delay sequence and a packet loss rate sequence; within the network delay sequence, a first network delay at a first timestamp is extracted to perform a delay fluctuation analysis to obtain a first delay fluctuation; and other network delays are continued to perform delay fluctuation analysis to obtain a delay fluctuation sequence.

[0070] Furthermore, a first network delay at a first timestamp is extracted from the network delay sequence, and a delay fluctuation analysis is performed to obtain a first delay fluctuation, including: randomly extracting a number of random network delays from the network delay sequence; and calculating a delay difference between an average of the number of random network delays and the first network delay as the first delay fluctuation.

[0071] Furthermore, according to the network delay sequence and the delay fluctuation sequence, packet loss prediction is performed respectively to obtain a delay packet loss rate sequence and a fluctuation packet loss rate sequence, a predicted packet loss rate sequence is calculated, and an implicit packet loss rate sequence is calculated in combination with the packet loss rate sequence, including: pre-training a packet loss prediction channel, wherein the packet loss prediction channel includes a delay packet loss prediction path and a fluctuation packet loss prediction path; inputting multiple network delays in the network delay sequence into the delay packet loss prediction path, and obtaining multiple delay packet loss rates as prediction outputs to construct a delay packet loss rate sequence; inputting multiple delay fluctuations in the delay fluctuation sequence into the fluctuation packet loss prediction path, and obtaining multiple fluctuation packet loss rates as prediction outputs to construct a fluctuation packet loss rate sequence; calculating the mean of each delay packet loss rate and the corresponding fluctuation packet loss rate in the delay packet loss rate sequence and the fluctuation packet loss rate sequence to obtain a predicted packet loss rate sequence; and according to the mapping relationship between the packet loss rate sequence and the predicted packet loss rate sequence, subtracting the corresponding predicted packet loss rate from each packet loss rate to obtain an implicit packet loss rate sequence.

[0072] Furthermore, the pre-training of the packet loss prediction channel includes: collecting a sample network delay set and a sample delay fluctuation set according to the network log data in the historical time, respectively as input data, and collecting a sample packet loss rate set as supervision data; respectively using the sample network delay set and the sample delay fluctuation set as input data, using the sample packet loss rate set as supervision data, and using machine learning to construct and train a delay packet loss prediction path and a fluctuation packet loss prediction path; combining the trained delay packet loss prediction path and fluctuation packet loss prediction path to obtain a packet loss prediction channel.

[0073] Furthermore, based on the packet loss rate sequence, retransmission delay prediction is performed to obtain a predicted delay fluctuation sequence, and an implicit delay fluctuation sequence is calculated in combination with the delay fluctuation sequence, including: inputting multiple packet loss rates in the packet loss rate sequence into a pre-trained retransmission delay prediction channel, predicting and outputting multiple predicted delay fluctuations, and constructing a predicted delay fluctuation sequence, wherein the retransmission delay prediction channel is trained using a sample packet loss rate set and a sample delay fluctuation set; and based on the mapping relationship between the predicted delay fluctuation sequence and the delay fluctuation sequence, each delay fluctuation is subtracted from the corresponding predicted delay fluctuation to obtain an implicit delay fluctuation sequence.

[0074] Further, according to the implicit packet loss rate sequence and the implicit delay fluctuation sequence, an implicit fault stability calculation is performed to obtain an implicit fault stability coefficient, and the quality level of the network service is obtained by classification, including: extracting a first implicit packet loss rate at a first timestamp in the implicit packet loss rate sequence, randomly extracting a plurality of random implicit packet loss rates in the implicit packet loss rate sequence; calculating an implicit packet loss rate difference between an average of the plurality of random implicit packet loss rates and the first implicit packet loss rate, and calculating a ratio of the implicit packet loss rate difference to the first implicit packet loss rate as a first implicit packet loss rate stability coefficient; continuing to calculate the implicit packet loss rate of the implicit packet loss rate sequence to obtain the first implicit packet loss rate stability coefficient. A packet loss rate stability coefficient sequence is included, and an implicit fluctuation stability coefficient sequence of the implicit delay fluctuation sequence is calculated; the mean of the implicit packet loss rate stability coefficient and the implicit fluctuation stability coefficient at each timestamp in the implicit packet loss rate stability coefficient sequence and the implicit fluctuation stability coefficient sequence is calculated to obtain an implicit fault stability coefficient sequence, and the mean is calculated to obtain an implicit fault stability coefficient; the implicit fault stability coefficient is input into a network service quality classifier, and the network service quality level is obtained by classification mapping, wherein the network service quality classifier is constructed based on the mapping relationship between a sample implicit fault stability coefficient set and a sample network service quality level set.

[0075] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0076] Furthermore, the first or second mentioned above not only represents an order relationship, but also represents a specific concept, and / or refers to the selection of multiple elements individually or in whole. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application intends to include these modifications and variations.

Claims

1. A multi-dimensional dynamic monitoring platform for network service quality, characterized by: The platform includes: A delay fluctuation sequence acquisition module is used to monitor the network delay and packet loss rate within a duration during the network service process, obtain a network delay sequence and a packet loss rate sequence, and perform delay fluctuation analysis based on the network delay sequence to obtain a delay fluctuation sequence; An implicit packet loss rate sequence acquisition module is used to perform packet loss prediction according to the network delay sequence and the delay fluctuation sequence, obtain a delay packet loss rate sequence and a fluctuation packet loss rate sequence, calculate and obtain a predicted packet loss rate sequence, and calculate and obtain an implicit packet loss rate sequence in combination with the packet loss rate sequence; An implicit delay fluctuation sequence acquisition module is used to perform retransmission delay prediction according to the packet loss rate sequence to obtain a predicted delay fluctuation sequence, and calculate an implicit delay fluctuation sequence in combination with the delay fluctuation sequence; The quality grade acquisition module is used to perform implicit fault stability calculation according to the implicit packet loss rate sequence and the implicit delay fluctuation sequence, obtain the implicit fault stability coefficient, and classify and obtain the quality grade of the network service as the monitoring result of the network service quality.

2. The multi-dimensional dynamic monitoring platform for network service quality according to claim 1, characterized in that: During the network service process, the network delay and packet loss rate within the duration are monitored to obtain a network delay sequence and a packet loss rate sequence, and a delay fluctuation analysis is performed according to the network delay sequence to obtain a delay fluctuation sequence, including: During the network service process, the network delay and packet loss rate at multiple time stamps within the duration are monitored to construct and obtain the network delay sequence and packet loss rate sequence; Extracting a first network delay at a first timestamp in the network delay sequence, performing delay fluctuation analysis, and obtaining a first delay fluctuation; Continue to perform delay fluctuation analysis on other network delays to obtain a delay fluctuation sequence.

3. The multi-dimensional dynamic monitoring platform for network service quality according to claim 2, characterized in that: Extracting a first network delay at a first timestamp in the network delay sequence, performing delay fluctuation analysis, and obtaining a first delay fluctuation, including: Randomly extracting a number of random network delays within the network delay sequence; A delay difference between an average of the plurality of random network delays and the first network delay is calculated as a first delay fluctuation.

4. The multi-dimensional dynamic monitoring platform for network service quality according to claim 1, characterized in that: According to the network delay sequence and the delay fluctuation sequence, packet loss prediction is performed respectively to obtain a delay packet loss rate sequence and a fluctuation packet loss rate sequence, and a predicted packet loss rate sequence is calculated and obtained, and combined with the packet loss rate sequence, an implicit packet loss rate sequence is calculated, including: Pre-training a packet loss prediction channel, wherein the packet loss prediction channel includes a delayed packet loss prediction path and a fluctuating packet loss prediction path; Inputting multiple network delays in the network delay sequence into the delay packet loss prediction path, predicting and outputting multiple delay packet loss rates, and constructing a delay packet loss rate sequence; Inputting multiple delay fluctuations in the delay fluctuation sequence into the fluctuation packet loss prediction path, predicting and outputting multiple fluctuation packet loss rates, and constructing a fluctuation packet loss rate sequence; Calculate the mean of each delayed packet loss rate and the corresponding fluctuating packet loss rate in the delayed packet loss rate sequence and the fluctuating packet loss rate sequence to obtain a predicted packet loss rate sequence; According to the mapping relationship between the packet loss rate sequence and the predicted packet loss rate sequence, the implicit packet loss rate sequence is obtained by subtracting the corresponding predicted packet loss rate from each packet loss rate.

5. The multi-dimensional dynamic monitoring platform for network service quality according to claim 4, characterized in that: Pre-trained packet loss prediction channel, including: According to the network log data in the historical time, the sample network delay set and the sample delay fluctuation set are collected as input data respectively, and the sample packet loss rate set is collected as supervision data; The sample network delay set and the sample delay fluctuation set are respectively used as input data, the sample packet loss rate set is used as supervision data, and machine learning is used to construct and train a delay packet loss prediction path and a fluctuation packet loss prediction path; The delayed packet loss prediction path and the fluctuating packet loss prediction path that have been trained are combined to obtain a packet loss prediction channel.

6. The multi-dimensional dynamic monitoring platform for network service quality according to claim 1, characterized in that: According to the packet loss rate sequence, retransmission delay prediction is performed to obtain a predicted delay fluctuation sequence, and combined with the delay fluctuation sequence, an implicit delay fluctuation sequence is calculated, including: Inputting multiple packet loss rates in the packet loss rate sequence into a pre-trained retransmission delay prediction channel, predicting and outputting multiple predicted delay fluctuations, and constructing a predicted delay fluctuation sequence, wherein the retransmission delay prediction channel is obtained by training with a sample packet loss rate set and a sample delay fluctuation set; According to the mapping relationship between the predicted delay fluctuation sequence and the delay fluctuation sequence, the implicit delay fluctuation sequence is obtained by subtracting the corresponding predicted delay fluctuation from each delay fluctuation.

7. The multi-dimensional dynamic monitoring platform for network service quality according to claim 1, characterized in that: According to the implicit packet loss rate sequence and the implicit delay fluctuation sequence, an implicit fault stability calculation is performed to obtain an implicit fault stability coefficient, and the quality level of the network service is obtained by classification, including: Extracting a first implicit packet loss rate at a first timestamp from the implicit packet loss rate sequence, and randomly extracting a plurality of random implicit packet loss rates from the implicit packet loss rate sequence; Calculating an implied packet loss rate difference between an average of the plurality of random implied packet loss rates and the first implied packet loss rate, and calculating a ratio of the implied packet loss rate difference to the first implied packet loss rate as a first implied packet loss rate stability coefficient; Continue to calculate and obtain an implicit packet loss rate stability coefficient sequence of the implicit packet loss rate sequence, and calculate and obtain an implicit fluctuation stability coefficient sequence of the implicit delay fluctuation sequence; Calculate the mean of the implicit packet loss rate stability coefficient and the implicit fluctuation stability coefficient at each time stamp in the implicit packet loss rate stability coefficient sequence and the implicit fluctuation stability coefficient sequence to obtain an implicit fault stability coefficient sequence, and calculate the mean to obtain an implicit fault stability coefficient; The implicit fault stability coefficient is input into a network service quality classifier, and the network service quality level is obtained by classification mapping, wherein the network service quality classifier is constructed based on the mapping relationship between a sample implicit fault stability coefficient set and a sample network service quality level set.

8. A multi-dimensional dynamic monitoring method for network service quality, characterized in that: A multi-dimensional dynamic monitoring platform for network service quality applied to any one of claims 1 to 7, the method comprising: During the network service process, the network delay and packet loss rate are monitored within a duration to obtain a network delay sequence and a packet loss rate sequence, and a delay fluctuation analysis is performed based on the network delay sequence to obtain a delay fluctuation sequence; According to the network delay sequence and the delay fluctuation sequence, packet loss prediction is performed respectively to obtain a delay packet loss rate sequence and a fluctuation packet loss rate sequence, and a predicted packet loss rate sequence is calculated and obtained, and combined with the packet loss rate sequence, an implicit packet loss rate sequence is calculated; According to the packet loss rate sequence, retransmission delay prediction is performed to obtain a predicted delay fluctuation sequence, and an implicit delay fluctuation sequence is calculated in combination with the delay fluctuation sequence; According to the implicit packet loss rate sequence and the implicit delay fluctuation sequence, an implicit fault stability calculation is performed to obtain an implicit fault stability coefficient, and the quality level of the network service is obtained by classification as a monitoring result of the network service quality.

Images