Private network quality evaluation method, device, equipment, medium and computer program product

Abstract

The application provides a private network quality evaluation method, device, equipment, medium and computer program product, the method comprises: data cleaning is carried out to the private network data collected to obtain a private network data table;The basic index set is generated based on the private network data table;Based on the maximum mutual information coefficient of each pair of index sequence in the basic index set, the clustering distance of each pair of index sequence is determined;The quality of the target private network is evaluated based on the optimal index set;The optimal index set is determined based on the clustering distance.The application comprehensively evaluates the private network through the basic index set, and evaluates the quality of different private networks through the optimal index, thereby improving the accuracy of private network quality evaluation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of network quality evaluation, and in particular to a private network quality evaluation method, device, equipment, medium and computer program product. BACKGROUND

[0002] The current private network quality evaluation scheme usually relies on a series of key performance indicators formulated by experts according to public network experience, but the private network form has diversity, and the business types of the private network and the public network also have large differences. Direct use of the key performance indicators of the public network cannot comprehensively and accurately reflect the quality of the private network, and cannot be suitable for complex and variable business scenarios. Therefore, in the face of various private network scenarios that need to formulate indicators, how to improve the comprehensiveness and accuracy of private network quality evaluation has become a technical problem to be solved. SUMMARY

[0003] The present application provides a private network quality evaluation method, device, equipment, medium and computer program product, which solves the defects of not being comprehensive and accurate in the existing private network quality evaluation technology, and realizes more comprehensive and accurate private network quality evaluation.

[0004] The present application provides a private network quality evaluation method, which comprises the following steps.

[0005] Data cleaning is performed on the collected private network data to obtain a private network data table;

[0006] A basic indicator set is generated based on the private network data table;

[0007] The clustering distance of each pair of indicator sequences is determined based on the maximum mutual information coefficient of each pair of indicator sequences in the basic indicator set.

[0008] The target private network is quality evaluated based on the optimal indicator set; the optimal indicator set is determined based on the clustering distance.

[0009] According to the private network quality evaluation method provided by the present application, the generation of the basic indicator set based on the private network data table comprises:

[0010] Based on a preset sliding window, the statistical class indicators, the change condition class indicators and the ratio class indicators of each private network data in the private network data table are determined;

[0011] Based on the statistical class indicators, the change condition class indicators and the ratio class indicators, a basic indicator set is generated.

[0012] According to the private network quality evaluation method provided by the present application, the generation of the basic indicator set based on the private network data table comprises:

[0013] determine the information entropy and the coefficient of variation of each index in the basic index set;

[0014] filter each index in the basic index set based on the information entropy and the coefficient of variation;

[0015] adjust the step length of the preset sliding window based on the number of indexes in the filtered index set to obtain an updated index set;

[0016] determine the updated index set as the basic index set if the number of indexes in the updated index set after filtering is greater than a preset threshold.

[0017] According to the special network quality evaluation method provided by the application, the determination of the clustering distance of each pair of index sequences in the basic index set based on the maximum mutual information coefficient of each pair of index sequences comprises:

[0018] determine the joint probability distribution of the first index and the second index, the first marginal probability distribution of the first index, and the second marginal probability distribution of the second index; the first index belongs to the first index sequence, the second index belongs to the second index sequence, and the first index sequence and the second index sequence are any two index sequences in the basic index set;

[0019] determine the mutual information value of the first index and the second index based on the joint probability distribution, the first marginal probability distribution, and the second marginal probability distribution;

[0020] construct a mutual information matrix of the first index sequence and the second index sequence based on the mutual information value of the first index and the second index;

[0021] determine the maximum mutual information coefficient of the first index sequence and the second index sequence based on the maximum mutual information value in the mutual information matrix.

[0022] According to the special network quality evaluation method provided by the application, the determination of the clustering distance of each pair of index sequences in the basic index set based on the maximum mutual information coefficient of each pair of index sequences further comprises:

[0023] determine the clustering distance of the first index sequence and the second index sequence based on the maximum mutual information coefficient of the first index sequence and the second index sequence;

[0024] cluster and merge the index sequences in the basic index set based on the clustering distance to obtain the clustering distance of each pair of index sequences in the updated basic index set.

[0025] According to the special network quality evaluation method provided by the application, the quality evaluation of the target special network based on the preferred index set comprises:

[0026] The termination condition of the cluster merging is that the basic index set obtained after the cluster merging contains only one index sequence, or the maximum mutual information coefficient of any pair of index sequences in the basic index set obtained after the cluster merging is greater than a target threshold value;

[0027] The basic index set after the cluster merging termination is screened based on the cluster distance to obtain a preferred index set and a candidate index set;

[0028] The target index set is obtained by selecting the indexes in the candidate index set;

[0029] The preferred index set and the target index set are merged to obtain a selected index set.

[0030] The application further provides a special network quality evaluation device, comprising the following modules:

[0031] A data cleaning module is configured to clean the collected special network data to obtain a special network data table;

[0032] A basic index set generation module is configured to generate a basic index set based on the special network data table;

[0033] A cluster distance determination module is configured to determine the cluster distance of each pair of index sequences based on the maximum mutual information coefficient of each pair of index sequences in the basic index set;

[0034] A special network quality evaluation module is configured to evaluate the quality of a target special network based on a selected index set; the selected index set is determined based on the cluster distance.

[0035] The application further provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the special network quality evaluation method.

[0036] The application further provides a non-transitory computer readable storage medium having a computer program stored thereon, wherein the computer program is executable by a processor to implement the special network quality evaluation method.

[0037] The application further provides a computer program product comprising a computer program, wherein the computer program is executable by a processor to implement the special network quality evaluation method.

[0038] The application provides a private network quality evaluation method, device, equipment, medium and computer program product. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the application or prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative effort on the basis of these drawings.

[0040] Figure 1 Fig. 1 is one of the flow diagrams of the private network quality evaluation method provided by the application.

[0041] Figure 2 Fig. 2 is another of the flow diagrams of the private network quality evaluation method provided by the application.

[0042] Figure 3 Fig. 3 is a structural diagram of the private network quality evaluation device provided by the application.

[0043] Figure 4 Fig. 4 is a structural diagram of the electronic equipment provided by the application. DETAILED DESCRIPTION

[0044] In order to make the objects, technical solutions and advantages of the application clearer, the technical solutions in the application will be described clearly and completely below with reference to the drawings in the application. Obviously, the described embodiments are some of the embodiments of the application, but not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative effort belong to the protection scope of the application.

[0045] The private network quality evaluation method, device, equipment, medium and computer program product of the application will be described below. Figures 1-4

[0046] Figure 1 Fig. 1 is one of the flow diagrams of the private network quality evaluation method provided by the application, as shown in the figure, the method comprises the following: Figure 1

[0047] ​​Step 100, data cleaning is performed on the collected private network data to obtain a private network data table;

[0048] For a private network, such as a ToB private network, as much basic data as possible is collected through the 5G Operation and Maintenance Center (OMC) and signaling systems, sensors and Internet of Things data, and questionnaire surveys, and non-numeric data is encoded and cleaned into numeric data, and finally a data table with all data being numeric is generated. The types of data collected include network data, business perception data, sensor and Internet of Things data, and questionnaire data.

[0049] Network data: Collect basic network performance data, mainly from the network quality evaluation provided by the OMC. For example, the number of Quality of Service Flow (QoS Flow) establishment requests, the number of QoS Flow establishment successes, the number of QoS Flow abnormal disconnections, the number of successful inter-gNB NG handovers, the number of successful inter-gNB Xn handovers, the number of bearer establishment requests, the number of bearer establishment successes, the bearer establishment duration, signal strength, and signal-to-noise ratio, etc.

[0050] Business perception data: Collect all private network business perception performance data, which is mainly from the signaling system, providing evaluation basis for the main private network bearer business. For example, in a video monitoring private network, the collected data includes the number of handshake requests, the number of handshake successes, the number of handshake failures, the number of retransmissions, and the establishment delay in Transmission Control Protocol (TCP), Hypertext Transfer Protocol (HTTP), and streaming media playback business.

[0051] Sensor and Internet of Things data: Use a dedicated data collection system or Internet of Things gateway to connect sensors through standardized interfaces, read sensors in real time or periodically, and output data. For example, sound pressure, frequency spectrum, tone, loudness, and duration of a sound sensor.

[0052] Questionnaire data: Obtain supervisor perception data through a questionnaire survey and numerize the results. For example, the stability and perception satisfaction of each device, etc. The results are numerized according to the number of grades, for example, stability is divided into high, medium and low, with scores of 80%, 60% and 40%, and satisfaction is divided into 1-5 grades, etc.

[0053] The above collected data is cleaned with the collection time point as the row and the data name as the column to obtain a private network data table.

[0054] Step 200, generating a basic index set based on the private network data table;

[0055] Through the collected data, in a sliding window with a step of N, multi-class statistical indicators and multi-class change indicators of each collected data are calculated, and a count ratio of the same service in different stages is further calculated, so as to construct a complete ratio indicator. After the indicators are combined, the basic index set is formed.

[0056] Step 300, determining a clustering distance of each pair of index sequences in the basic index set based on a maximum mutual information coefficient of each pair of index sequences;

[0057] In order to reduce the correlation between the indexes in the basic index set, the indexes are optimized by hierarchical clustering, and then the key indexes are extracted. Since the indexes are often not linearly correlated, the correlation between the indexes can be measured by the maximum mutual information coefficient. In order to better determine which indexes to optimize, hierarchical clustering can be used to optimize the indexes in the same cluster by using the clustering structure of the data. The optimization basis is the clustering distance of the index sequence, and the clustering distance of the index sequence is determined based on the maximum mutual information coefficient of the index sequence.

[0058] Step 400, performing quality evaluation on the target private network based on the optimized index set; the optimized index set is determined based on the clustering distance.

[0059] The process of optimizing the indexes based on the clustering distance includes: all indexes in an index cluster with a clustering distance greater than a threshold value are reserved and put into an optimized index set S1, and the elements in the S1 set are indexes. An index cluster with a clustering distance less than a certain threshold value is put into a candidate index cluster set, and the elements in the candidate index cluster set are index clusters. The index clusters in the candidate index cluster set are further screened, the screening result is combined with the indexes in the above-mentioned optimized index set, and an optimized index set is obtained. Finally, the target private network is evaluated in quality by the optimized index set.

[0060] In this embodiment, the collected private network data is cleaned to obtain a private network data table, and a basic index set is generated based on the private network data table. The clustering distance of each pair of index sequences in the basic index set is determined based on the maximum mutual information coefficient of each pair of index sequences in the basic index set. The optimized index set is determined based on the clustering distance, and finally the target private network is evaluated in quality by the indexes in the optimized index set. The private network is comprehensively evaluated by the basic index set, and the quality of the private network is evaluated by the optimized indexes for different private networks, thereby improving the accuracy of the private network quality evaluation.

[0061] In one embodiment, the private network quality evaluation method provided by the embodiment of the present application can further include:

[0062] Step 210, based on the preset sliding window, determine the statistical class index, the change condition class index and the ratio class index of each private network data in the private network data table;

[0063] Step 220, based on the statistical class index, the change condition class index and the ratio class index, generate a basic index set.

[0064] Statistical class index: in a sliding window with a step of N, calculate multiple statistical values of each collected data point as statistical indicators, including mean value (arithmetic mean value of data in the window); median (the value in the middle after sorting the data in the window); mode (the value with the highest frequency of occurrence in the window); standard deviation (square root of the average of the square of the deviation of the data from its mean value, reflecting the dispersion of the data); quartile (three partition points dividing the data in the window into four equal parts); skewness (skewness of data distribution); kurtosis (sharpness of data distribution).

[0065] Change condition class index: in a sliding window with a step of N, calculate the change condition of each collected data between different collection points, and construct cumulative change rate, average change rate, cumulative difference value and average difference value.

[0066] Cumulative change rate: calculate the quantity ratio between each collected data point in the window, and sum all the change rates at different time points. For example, the number of lost packets at time point t1 is n1, and at time point t2 is n2, then the change rate is n2 / n1. The sum of all change rates in the window is the cumulative change rate.

[0067] Average change rate: calculate the quantity ratio between each collected data point in the window, and average all the change rates. For example, the average change rate is calculated by averaging all the change rates of the number of lost packets.

[0068] Cumulative difference value: calculate the quantity difference between each collected data point in the window, and sum all the difference values at different time points. For example, the time delay at time point t1 is n1, and at time point t2 is n2, then the time delay difference value is n2-n1. The sum of all difference values in the window is the cumulative difference value.

[0069] Average difference value: calculate the quantity difference between each collected data point in the window, and average all the difference values. For example, the average difference value is calculated by averaging all the difference values of the time delay.

[0070] Ratio type index: in the sliding window with a step of N, the count ratio of the same service in different stages is further calculated to construct the ratio index. These count items need to be determined based on the prior knowledge of the service. Common ratio indexes include success rate and failure rate, which can be obtained by calculating the ratio of the number of successes or failures to the total number of times in the window.

[0071] In this embodiment, the collected data is calculated by a sliding window to obtain statistical indexes, change condition indexes and ratio indexes of each data, and a basic index set is constructed and generated.

[0072] Figure 2 Figure 2 is a flowchart of the method for evaluating the quality of a private network provided by the present application, as shown in Figure 2 The method can further include the following steps.

[0073] Step 500: determining the information entropy and the coefficient of variation of each index in the basic index set;

[0074] Step 600: filtering each index in the basic index set based on the information entropy and the coefficient of variation;

[0075] Step 700: adjusting the step of the preset sliding window based on the number of indexes in the filtered index set to obtain an updated index set;

[0076] Step 800: in the case where the number of indexes after filtering the updated index set is greater than a preset threshold, determining the updated index set as the basic index set.

[0077] The generated basic index set needs to be filtered and recursively optimized.

[0078] Index filtering: filtering the basic index set to filter out non-compliant indexes. The filtering method includes information entropy index filtering and coefficient of variation index filtering. Information entropy index filtering: calculating the information entropy of each index in the basic index set, and filtering out indexes with information entropy lower than a preset threshold; coefficient of variation index filtering: calculating the coefficient of variation of each index in the basic index set, and filtering out indexes with a coefficient of variation lower than a preset threshold.

[0079] Recursive optimization: if the number of indexes after filtering is small, the step of the sliding window is adjusted, the basic index set is regenerated and filtered, and the process is repeated until the number of indexes after filtering meets the requirements.

[0080] In this embodiment, the index filtering and recursive optimization filter out non-compliant indexes, further improving the accuracy of the basic index set.

[0081] In one embodiment, the method for evaluating the quality of a private network provided by the present application can further include the following steps.

[0082] Step 310, determining a joint probability distribution of the first indicator and the second indicator, a first marginal probability distribution of the first indicator, and a second marginal probability distribution of the second indicator; the first indicator belongs to a first indicator sequence, the second indicator belongs to a second indicator sequence, and the first indicator sequence and the second indicator sequence are any two indicator sequences in the basic indicator set;

[0083] Step 320, determining a mutual information value of the first indicator and the second indicator based on the joint probability distribution, the first marginal probability distribution, and the second marginal probability distribution;

[0084] Step 330, constructing a mutual information matrix of the first indicator sequence and the second indicator sequence based on the mutual information value of the first indicator and the second indicator;

[0085] Step 340, determining a maximal mutual information coefficient of the first indicator sequence and the second indicator sequence based on a maximal mutual information value in the mutual information matrix.

[0086] Calculation of the maximal mutual information coefficient matrix: for each pair of indicators in the basic indicator set, calculate their maximal mutual information coefficient (MIC), which can measure the correlation between each pair of indicators. The specific steps include:

[0087] 1. Define an indicator sequence: record each pair of indicators as a continuous numerical sequence.

[0088] 2. Calculate MIC:

[0089] Step 1, data normalization and discretization: since the indicators formed by the collected data are continuous and have large differences in scale, the indicator sequence is first normalized and then discretized. The values of the indicator sequence (not in the 0-1 interval) are normalized by the minimum-maximum normalization method to scale the indicator values to 0-1. All single-indicator data sets have the same scale.

[0090] Data discretization is achieved by binning the indicator sequence. The method of binning is to divide the continuous numerical distribution in the original indicator sequence into multiple intervals or "bins", and the mean value of all values falling into the same bin is regarded as the value of the bin. This application uses the method of equal-width binning, for example, for packet loss, each 0.1 packet loss can be an interval, i.e., [0-0.1), [0.1-0.2), and [0.2-0.3), etc. The specific bin width is selected according to the number of collected data and the computing power, but the interval cannot be too small, and the number of bins is generally greater than 50.

[0091] Step 2, calculate mutual information: for any two indicator sequences X and Y in the first and second indicators in this embodiment, calculate the joint probability distribution p(x, y) of the first and second indicators and the respective marginal probability distributions p(x) and p(y), and calculate the mutual information of the first and second indicators by the joint probability distribution and the respective marginal probability distributions.

[0092] Step 3, construct the indicator sequence MI matrix: for each bin pair (Xb, Yc) of the discretization of each indicator sequence, calculate its mutual information I(Xb; Yc), and construct an N x N MI matrix, where N is the number of bins of the discretization of the indicator sequence.

[0093] Step 4, normalize the MI matrix: find the maximum mutual information value max_MI in the MI matrix. Subtract the expected mutual information E[I(Xb; Yc)] in the case of no correlation from each element in the MI matrix, and then normalize by (max_MI - E[I(Xb; Yc)]) to ensure that the maximum value of the normalized matrix is 1.

[0094] Step 5, calculate the MIC matrix of the indicator sequence pairs: combine all the calculated MIC values to obtain the MIC matrix, where the element [i, j] of the MIC matrix represents the MIC value between the indicator sequence i and the indicator sequence j. A higher mutual information value indicates a stronger correlation between the two indicators.

[0095] The embodiment calculates the mutual information values between the indicators to obtain the maximum mutual information coefficient of each pair of indicator sequences in the basic indicator set, providing a data basis for the calculation of the clustering distance between the indicator sequences.

[0096] In one embodiment, the private network quality evaluation method provided by the embodiment of the application can further include:

[0097] Step 350, determining the clustering distance of the first indicator sequence and the second indicator sequence based on the maximum mutual information coefficient of the first indicator sequence and the second indicator sequence;

[0098] Step 360, clustering and merging the indicator sequences in the basic indicator set based on the clustering distance to obtain the clustering distance of each pair of indicator sequences in the updated basic indicator set.

[0099] Initialization of indicator clustering:

[0100] 1. Calculate the indicator clustering matrix: based on the indicator sequence MIC matrix, obtain a clustering matrix, the rows and columns of the matrix are indicators, and the element [i, j] of the matrix represents the clustering distance between indicator i and indicator j. The calculation method of the clustering distance is the reciprocal of the corresponding element in the MIC matrix.

[0101] 2. Initialization of index clustering: In the initial stage of hierarchical clustering, each data point is considered as a separate cluster, i.e. each cluster contains only one data point. If there are N data points, there are initially N clusters.

[0102] 3. Merging of index clusters: The method of calculating the index cluster can choose average linkage, i.e. when two index clusters are merged, the average of all possible distances between the two index clusters is taken. The two index cluster clusters with the smallest average distance are merged into a new cluster. The members of this new cluster include all data points of the original two clusters.

[0103] 4. Updating of the index cluster matrix: After merging the clusters, the cluster matrix between the clusters needs to be updated. The distance between the newly formed cluster and other clusters needs to be recalculated. First, for the new cluster and other clusters, the pairwise distances between all data points in the new cluster and all data points in other clusters are calculated. Then, the average of these pairwise distances is calculated, which is the distance between the new cluster and other clusters.

[0104] 5. Recursive index clustering: Using the updated distance matrix, repeat the above content until the stopping condition is met. The stopping condition can be that all data points in the basic index set are merged into one cluster, or the maximum mutual information coefficient between any two index clusters is less than a specified threshold.

[0105] The embodiment calculates the clustering between indexes by the maximum mutual information coefficient between indexes, and provides a data basis for the selection of indexes.

[0106] In one embodiment, the special network quality evaluation method provided by the embodiment of the application can further include:

[0107] Step 410, determining that the termination condition of cluster merging is that the basic index set obtained after cluster merging contains only one index sequence, or the maximum mutual information coefficient of any pair of index sequences in the basic index set obtained after cluster merging is greater than a target threshold;

[0108] Step 420, filtering the basic index set after cluster merging based on the cluster distance to obtain a preferred index set and a candidate index set;

[0109] Step 430, selecting the target index set by selecting the indexes in the candidate index set;

[0110] Step 440, merging the preferred index set and the target index set to obtain a selected index set.

[0111] Index selection:

[0112] 1, select the index to be preferred: all the indexes in the index cluster with a clustering distance greater than a threshold value are reserved as a preferred index set S1, and the elements in the set are indexes. The index cluster with a distance less than a certain threshold value is recorded as a selected index cluster set Ω, and the elements in the set are index clusters.

[0113] For example, select the index cluster with a clustering distance greater than 6 as S1, and the elements of S1 are the 8 elements BGHJEFLM. Select the index cluster with a clustering distance less than 6 to join the selected index cluster set Ω, which has 2 elements, namely index cluster 1 composed of A and K, and index cluster 2 composed of L and M.

[0114] 2, selection method: select the index in each cluster in the selected index cluster set Ω, and form an index set S2 after selection. The selection method is as follows: select the index with the maximum entropy in the cluster. For example, index clusters A and K, assuming that the entropy of A is greater than that of K, then select A and discard K; for index clusters C and D, assuming that the entropy of C is greater than that of D, then select C and discard D; the elements of the selected index set S2 are A and C.

[0115] 3, preferred index set generation: combine S1 and S2 to generate a preferred index set S.

[0116] The embodiment improves the accuracy of the quality evaluation of the private network by selecting the preferred index for quality evaluation.

[0117] The private network quality evaluation device provided by the application is described below. The private network quality evaluation device described below can be correspondingly referred to the private network quality evaluation method described above.

[0118] Please refer to Figure 3 The application also provides a private network quality evaluation device, which comprises:

[0119] The data cleaning module 301 is used for cleaning the collected private network data to obtain a private network data table.

[0120] The basic index set generation module 302 is used for generating a basic index set based on the private network data table.

[0121] The clustering distance determination module 303 is used for determining the clustering distance of each pair of index sequences based on the maximum mutual information coefficient of each pair of index sequences in the basic index set.

[0122] The private network quality evaluation module 304 is used for performing quality evaluation on a target private network based on a preferred index set; and the preferred index set is determined based on the clustering distance.

[0123] Optionally, the basic index set generation module comprises:

[0124] The data index determination unit is configured to determine, based on a preset sliding window, a statistical index, a change condition index, and a ratio index of each special network data in the special network data table.

[0125] The basic index set generation unit is configured to generate a basic index set based on the statistical index, the change condition index, and the ratio index.

[0126] Optionally, the special network quality evaluation device further comprises:

[0127] The information entropy and coefficient of variation determination module is configured to determine information entropy and a coefficient of variation of each index in the basic index set.

[0128] The index filtering module is configured to filter each index in the basic index set based on the information entropy and the coefficient of variation.

[0129] The sliding window step adjustment module is configured to adjust a step of the preset sliding window based on a number of indexes in the filtered index set to obtain an updated index set.

[0130] The basic index set determination module is configured to determine the updated index set as the basic index set if a number of indexes after filtering the updated index set is greater than a preset threshold.

[0131] Optionally, the clustering distance determination module comprises:

[0132] The index probability distribution determination unit is configured to determine a joint probability distribution of a first index and a second index, a first marginal probability distribution of the first index, and a second marginal probability distribution of the second index; the first index belongs to a first index sequence, the second index belongs to a second index sequence, and the first index sequence and the second index sequence are any two index sequences in the basic index set.

[0133] The mutual information value determination unit is configured to determine a mutual information value of the first index and the second index based on the joint probability distribution, the first marginal probability distribution, and the second marginal probability distribution.

[0134] The mutual information matrix construction unit is configured to construct a mutual information matrix of the first index sequence and the second index sequence based on the mutual information value of the first index and the second index.

[0135] The maximum mutual information coefficient determination unit is configured to determine a maximum mutual information coefficient of the first index sequence and the second index sequence based on a maximum mutual information value in the mutual information matrix.

[0136] Optionally, the clustering distance determination module further comprises:

[0137] The clustering distance determination unit is used to determine the clustering distance between the first index sequence and the second index sequence based on the maximum mutual information coefficient between the first index sequence and the second index sequence.

[0138] The clustering distance update unit is used to cluster and merge the index sequences in the basic index set based on the clustering distance, so as to obtain the clustering distance of each pair of index sequences in the updated basic index set.

[0139] Optionally, the private network quality assessment module includes:

[0140] The cluster merging termination condition determination unit is used to determine the termination condition of the cluster merging as follows: the basic index set obtained after cluster merging contains only one index sequence, or the maximum mutual information coefficient of any pair of index sequences in the basic index set obtained after cluster merging is greater than the target threshold.

[0141] The basic indicator set screening unit is used to screen the basic indicator set after the clustering and merging is terminated based on the clustering distance, so as to obtain the preferred indicator set and the candidate indicator set.

[0142] The indicator selection unit is used to select the best indicators from the candidate indicator set to obtain the target indicator set.

[0143] The optimal index set determination unit is used to merge the preferred index set and the target index set to obtain the optimal index set.

[0144] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4 As shown, the electronic device may include a processor 410, a communications interface 420, a memory 430, and a communication bus 440. The processor 410, communications interface 420, and memory 430 communicate with each other via the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a private network quality assessment method. This method includes: cleaning the collected private network data to obtain a private network data table; generating a basic index set based on the private network data table; determining the clustering distance of each pair of index sequences based on the maximum mutual information coefficient of each pair of index sequences in the basic index set; and performing a quality assessment of the target private network based on the optimal index set, wherein the optimal index set is determined based on the clustering distance.

[0145] In addition, the logic instructions in the memory 430 described above can be implemented in the form of a software function unit and sold or used as an independent product, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0146] In another aspect, the present application also provides a computer program product, which comprises a computer program, the computer program can be stored on a non-transitory computer readable storage medium, and the computer program can be executed by a processor to enable a computer to execute the private network quality evaluation method provided by the above-mentioned methods. The method comprises: performing data cleaning on the collected private network data to obtain a private network data table; generating a basic index set based on the private network data table; determining the clustering distance of each pair of index sequences in the basic index set based on the maximum mutual information coefficient of each pair of index sequences; and performing quality evaluation on the target private network based on the optimal index set; and the optimal index set is determined based on the clustering distance.

[0147] In another aspect, the present application also provides a computer program product, which comprises a computer program, the computer program can be stored on a non-transitory computer readable storage medium, and the computer program can be executed by a processor to enable a computer to execute the private network quality evaluation method provided by the above-mentioned methods. The method comprises: performing data cleaning on the collected private network data to obtain a private network data table; generating a basic index set based on the private network data table; determining the clustering distance of each pair of index sequences in the basic index set based on the maximum mutual information coefficient of each pair of index sequences; and performing quality evaluation on the target private network based on the optimal index set; and the optimal index set is determined based on the clustering distance.

[0148] The device embodiments described above are only schematic, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the present embodiment scheme according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0149] Those skilled in the art can clearly understand the implementation of the embodiments by means of software and necessary general hardware platforms through the description of the above embodiments, and the embodiments can also be implemented by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, and the computer software product can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the method described in each embodiment or some parts of the embodiment.

[0150] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacement for some technical features thereof; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for evaluating the quality of a private network, characterized in that, The method comprises the following steps: data cleaning is performed on the collected private network data to obtain a private network data table; a basic index set is generated based on the private network data table; a clustering distance of each pair of index sequences in the basic index set is determined based on a maximum mutual information coefficient of each pair of index sequences in the basic index set; a target private network is quality evaluated based on a preferred index set; the preferred index set is determined based on the clustering distance; the determination of the clustering distance of each pair of index sequences in the basic index set based on the maximum mutual information coefficient of each pair of index sequences in the basic index set comprises the following steps: a joint probability distribution of a first index and a second index, a first marginal probability distribution of the first index, and a second marginal probability distribution of the second index are determined; the first index belongs to a first index sequence, the second index belongs to a second index sequence, and the first index sequence and the second index sequence are any two index sequences in the basic index set; a mutual information value of the first index and the second index is determined based on the joint probability distribution, the first marginal probability distribution, and the second marginal probability distribution; a mutual information matrix of the first index sequence and the second index sequence is constructed based on the mutual information value of the first index and the second index; a maximum mutual information coefficient of the first index sequence and the second index sequence is determined based on a maximum mutual information value in the mutual information matrix.

2. The private network quality assessment method of claim 1, wherein, the generation of the basic index set based on the private network data table comprises the following steps: statistical class indexes, change condition class indexes, and ratio class indexes of each private network data in the private network data table are determined based on a preset sliding window; a basic index set is generated based on the statistical class indexes, the change condition class indexes, and the ratio class indexes.

3. The private network quality assessment method of claim 2, wherein, after the generation of the basic index set based on the private network data table, the following steps are included: information entropy and a coefficient of variation of each index in the basic index set are determined; each index in the basic index set is filtered based on the information entropy and the coefficient of variation; an updated index set is obtained by adjusting a step length of the preset sliding window based on a number of indexes in the filtered index set; in a case where the number of indexes in the updated index set after filtering is greater than a preset threshold, the updated index set is determined as the basic index set.

4. The private network quality assessment method of claim 1, wherein, the determination of the clustering distance of each pair of index sequences in the basic index set based on the maximum mutual information coefficient of each pair of index sequences in the basic index set further comprises the following steps: the clustering distance of the first index sequence and the second index sequence is determined based on the maximum mutual information coefficient of the first index sequence and the second index sequence; index sequences in the basic index set are clustered and merged based on the clustering distance to obtain a clustering distance of each pair of index sequences in an updated basic index set.

5. The private network quality assessment method of claim 4, wherein, the quality evaluation of the target private network based on the preferred index set comprises the following steps: the termination condition of the clustering and merging is that the basic index set obtained after the clustering and merging contains only one index sequence, or a maximum mutual information coefficient of any pair of index sequences in the basic index set obtained after the clustering and merging is greater than a target threshold. screening the base index set after the clustering merging termination based on the clustering distance, to obtain a preferred index set and a candidate index set; optimizing the indexes in the candidate index set to obtain a target index set; merging the preferred index set and the target index set to obtain an optimized index set.

6. A private network quality evaluation device characterized by comprising: comprise: a data cleaning module, configured to clean the collected private network data to obtain a private network data table; a base index set generation module, configured to generate a base index set based on the private network data table; a clustering distance determination module, configured to determine the clustering distance of each pair of index sequences in the base index set based on the maximum mutual information coefficient of each pair of index sequences in the base index set; a private network quality evaluation module, configured to evaluate the quality of a target private network based on an optimized index set; the optimized index set is determined based on the clustering distance; the determination of the clustering distance of each pair of index sequences in the base index set based on the maximum mutual information coefficient of each pair of index sequences in the base index set comprises: determining the joint probability distribution of a first index and a second index, the first edge probability distribution of the first index, and the second edge probability distribution of the second index; the first index belongs to a first index sequence, the second index belongs to a second index sequence, and the first index sequence and the second index sequence are any two index sequences in the base index set; determining the mutual information value of the first index and the second index based on the joint probability distribution, the first edge probability distribution, and the second edge probability distribution; constructing a mutual information matrix of the first index sequence and the second index sequence based on the mutual information value of the first index and the second index; determining the maximum mutual information coefficient of the first index sequence and the second index sequence based on the maximum mutual information value in the mutual information matrix.

7. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to realize the private network quality evaluation method of any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the private network quality evaluation method of any one of claims 1 to 5.

9. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to realize the private network quality evaluation method of any one of claims 1 to 5.

Images