CN105873113A - Method and system for positioning wireless quality problem - Google Patents
Method and system for positioning wireless quality problem Download PDFInfo
- Publication number
- CN105873113A CN105873113A CN201510030063.6A CN201510030063A CN105873113A CN 105873113 A CN105873113 A CN 105873113A CN 201510030063 A CN201510030063 A CN 201510030063A CN 105873113 A CN105873113 A CN 105873113A
- Authority
- CN
- China
- Prior art keywords
- user
- data
- abnormal
- index data
- level
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 24
- 230000002159 abnormal effect Effects 0.000 claims abstract description 116
- 230000008447 perception Effects 0.000 claims abstract description 26
- 238000012545 processing Methods 0.000 claims abstract description 9
- 238000012216 screening Methods 0.000 claims description 17
- 238000004364 calculation method Methods 0.000 claims description 16
- 230000004913 activation Effects 0.000 claims description 13
- 230000011664 signaling Effects 0.000 claims description 9
- 230000004931 aggregating effect Effects 0.000 claims 1
- 230000002547 anomalous effect Effects 0.000 claims 1
- 239000013598 vector Substances 0.000 description 10
- 238000012360 testing method Methods 0.000 description 9
- 238000003324 Six Sigma (6σ) Methods 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000005457 optimization Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 239000006185 dispersion Substances 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000013441 quality evaluation Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000010267 cellular communication Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- GVVPGTZRZFNKDS-JXMROGBWSA-N geranyl diphosphate Chemical compound CC(C)=CCC\C(C)=C\CO[P@](O)(=O)OP(O)(O)=O GVVPGTZRZFNKDS-JXMROGBWSA-N 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Landscapes
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses a method and a system for positioning a wireless quality problem. The method comprises steps: call bill data of users are acquired, the call bill data of the users are summarized through a user dimension and a time dimension, and user level indicator data are outputted; based on the user level indicator data, an abnormal threshold is determined, and abnormal users are selected from users based on the abnormal threshold; the user level indicator data of the abnormal users are scored according to different dimensions, user level scoring data corresponding to the abnormal users are obtained, the abnormal users are subjected to clustering processing based on the user level scoring data, and a clustering result is obtained; and based on the clustering result, perception information of the abnormal users is determined, and based on the perception information, a cell to which the abnormal users, whose perception is lower than a perception threshold, belong is determined to be a problem cell. Thus, the positioning efficiency of the wireless quality problem can be enhanced.
Description
Technical Field
The present invention relates to communications technologies, and in particular, to a method and a system for positioning a wireless quality problem.
Background
For a mobile communication network, the uncontrollable property of wireless transmission is inevitable, so that the guarantee optimization of the wireless quality of the network is always a key point for the mobile network, and compared with the hardware fault of network equipment, the network equipment has the characteristics of large user influence area and continuous occurrence of the problem that a fault unit is not repaired, so that various alarms can be designed for the network equipment, and the normal operation of the network equipment is guaranteed to prevent the fault; the wireless problem of the network has a relatively small influence on the user, and the user mobility is not continuous, so that the wireless quality of the network is more hidden from the hardware failure of the network equipment.
In view of the related technical situation, when the wireless quality problem of the network is located, the method mainly focuses on the collection of various types of data, including automatic measurement, storage of various types of large amounts of test data, and unified uploading of the test data to the background control center, and the background control center performs automatic centralized analysis on the test data and performs early warning on abnormal indexes of the wireless quality.
In the related technology, from a purely objective angle, Key Performance Indicators (KPIs) at a user level and a network element level are counted through the acquisition of basic signaling of different interfaces, the analysis of the signaling is emphasized, and the difference from the traditional network Performance optimization is that the KPI is counted to be capable of correlating the user level and the network element level through the signaling acquisition of the whole network, namely, the KPI is completely used as the basis to evaluate the network element with poor wireless quality index in the network, so as to provide reference for the network Performance analysis, and the difference from the wireless quality problem positioning of the related technology is only that the granularity of analyzing test data is reduced to the user level.
The related art has the problem of low efficiency for positioning the wireless quality problem of the network, which is shown in the following aspects:
1) the granularity of network management analysis is too large, so that the user cannot be minimized, and the improvement of user perception cannot be close to the statistics of a user level;
2) the test data is obtained through actual measurement from the perspective of the user, a large amount of labor and time are consumed, the cost is high, the test data is only the sampling result of part of users, all the users cannot be completely covered, and the wireless quality evaluation result of the network is inaccurate;
3) the wireless quality problem of the network is positioned only from the existing user angle, and the wireless quality is evaluated from the user-level indexes such as KPI angle, so that the improvement effect of the wireless instruction evaluation result for network optimization on user perception is limited.
Disclosure of Invention
The embodiment of the invention provides a method and a system for positioning a wireless quality problem, which can improve the positioning efficiency of the wireless quality.
The technical scheme of the embodiment of the invention is realized as follows:
a wireless quality issue location system, the system comprising:
the index calculation module is used for acquiring the call ticket data of the user, collecting the call ticket data of the user through user dimension and time dimension and outputting user-level index data;
an anomaly threshold calculation module for determining an anomaly threshold based on the user-level indicator data;
the abnormal user screening module is used for screening abnormal users from the users based on the abnormal threshold value;
the single-dimension scoring module is used for scoring the user-level index data of the abnormal user in different dimensions and outputting user-level scoring data corresponding to the abnormal user;
the multidimensional clustering module is used for clustering the abnormal users based on the grading data of the user level and outputting clustering results;
the low-perceptibility user screening module is used for determining the perceptibility information of the abnormal user based on the clustering result;
and the problem cell convergence module is used for determining a problem cell to which the abnormal user with the perception degree lower than the perception degree threshold belongs based on the perception degree information and outputting a list of the problem cell.
A wireless quality problem location method, the method comprising:
acquiring call ticket data of a user, and collecting the call ticket data of the user through user dimension and time dimension to obtain user-level index data;
determining an abnormal threshold value based on the user-level index data, and screening abnormal users from the users based on the abnormal threshold value;
grading the user-level index data of the abnormal user in different dimensions, outputting user-level grading data corresponding to the abnormal user, and clustering the abnormal user based on the user-level grading data to obtain a clustering result of the abnormal user;
and determining the perception degree information of the abnormal user based on the clustering result, determining a problem cell to which the abnormal user with the perception degree lower than a perception degree threshold belongs, and outputting a list of the problem cell.
In the embodiment of the invention, the user-level test data (with finer granularity than that of the prior art) is obtained through the ticket data collected by the big data platform, and the wireless quality of the network is analyzed from the perspective of the existing user in comparison with the related art.
Drawings
Fig. 1 is a definition of 2G and 3G core network interfaces by 3 GPP;
FIG. 2 is a first schematic structural diagram of a wireless problem location system according to an embodiment of the present invention;
FIG. 3 is a second schematic structural diagram of a wireless problem location system according to an embodiment of the present invention;
fig. 4 is a flowchart of an implementation of a method for locating a radio problem according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The test data source of the wireless quality problem positioning system described in the embodiment of the present invention is a data acquisition system based on a Gn interface, and as shown in fig. 1, is a definition of a 3GPP to 2G and 3G core network interfaces, where the Gn interface is located between a Serving General Packet Radio Service (GPRS) Support Node (SGSN) and a Gateway GPRS Support Node (GGSN).
The definition of the interface can be seen in: <3GPP TS 29.060V4.11.0>, Figure 1: GPRS logical architecture with interface names dentations.
As shown in fig. 2, the wireless quality problem location system obtains test data from a data acquisition system based on Gn interface and performs wireless quality evaluation, where the data acquisition system supports Packet Data Protocol (PDP) information data and service information data for outputting fields such as hour-cell-user, based on the existing infrastructure of the data acquisition system.
The system for positioning the wireless quality problem described in the embodiment relates to processing three types of data, and the processing is completed through seven modules in the system: wherein,
three types of data include: user-level index data, user-level scoring data and a problem cell list;
the system comprises an index calculation module 10, an abnormal threshold calculation module 20, an abnormal user screening module 30, a single-dimensional degree scoring module 40, a multi-dimensional clustering module 50, a low-perceptibility user screening module 60 and a problem cell convergence module 70.
Three types of data are explained below.
1) User level indicator data
Different from the wireless quality index adopted by the related technology, the index adopted in the embodiment of the invention is an index which can effectively embody the perception of the user and is selected by the inventor in practice from a plurality of indexes;
the indicator mainly relates to a signaling initiated when a user uses a data service, and comprises the following steps: the frequency of PDP creation failure, the frequency of PDP creation success and the frequency of no downlink flow after PDP activation; the user-level index data is obtained by collecting data corresponding to indexes of a single user dimension, directly reflects the condition of a user level and is closest to user perception; furthermore, to assist in analysing the radio quality of the network, the location information from which the user-level indicator data is generated may also be used to identify the user-level indicator data, i.e. the CELL code CELLID of the CELL (CELL) of the cellular communication network (other fields may be added to ensure uniqueness).
In the aspect of data collection, the user-level index data is obtained by collecting data corresponding to three indexes (including PDP creation failure frequency, PDP creation success frequency, and frequency of no downlink traffic after PDP activation) with a user identifier and CELLID as packet fields, and one example of the user-level index data is shown in table 1;
TABLE 1
2) User-level scoring data
The data corresponding to the three indexes of the PDP creation failure frequency, the PDP creation success frequency and the frequency of the non-downlink traffic after PDP activation in table 1, that is, the three index data are scored one by one, so as to obtain three scoring data of the PDP creation failure frequency score, the PDP creation success frequency score and the frequency score of the non-downlink traffic after PDP activation, and an example of the user-level scoring data obtained by using the three scoring data is shown in table 2:
TABLE 2
3) An example of the problem cell list is shown in table 3:
| cell Coding (CELLID) | Number of users with poor perception |
| 10001 | 910 |
| 10002 | 813 |
| 10003 | 734 |
| 10004 | 566 |
| 10005 | 480 |
| 10006 | 386 |
| 10007 | 370 |
| 10008 | 360 |
| 10009 | 358 |
| 10010 | 357 |
| 10011 | 349 |
| 10012 | 343 |
| 10013 | 340 |
| 10014 | 338 |
TABLE 3
The modules of the wireless quality problem location system shown in fig. 2 are explained below.
1) Index calculation module 10
The input of the index calculation module 10 is data of a session level signaling plane and a user plane output by a Gn signaling acquisition platform (corresponding to a data acquisition system), that is, call ticket data of a user, the index calculation module 10 collects and outputs user level index data of the data through user and time dimensions, for example, the user level index data of each user is output in a specific time dimension (one month or one day), and the user level index data of each user includes index data of at least three dimensions (that is, data corresponding to three indexes, namely PDP creation failure frequency, PDP creation success frequency and frequency of no downlink flow after PDP activation); it should be noted that, in the embodiment of the present invention, the problem cell is determined by using the index data of the three dimensions, and in practical applications, other index data based on the PDP, such as a PDP creation failure ratio, a PDP creation failure/success ratio, and the like, may also be used.
The definition and calculation formula of the above index are shown in table 4:
TABLE 4
2) Anomaly threshold calculation module 20
The abnormal threshold calculation module 20 is configured to determine an abnormal threshold according to the abnormal frequency of the normal event, and determine whether the index data is normal according to the abnormal threshold.
Based on the input user-level indicator data, the anomaly threshold calculation module 20 determines the anomaly threshold using the six sigma principle: theoretically and practically speaking, the index data of each dimension are subject to normal distribution, so whether the index data are abnormal or not can be judged by the six sigma principle, i.e. the value of the index data is within plus or minus three sigma (corresponding to the anomaly threshold) of the mean value of the index data, the index data is normal, otherwise, the index data is abnormal, the user corresponding to the index data is an abnormal user, since the value of the abnormal index data has a negative value in the value space (i.e. index data mean-3 standard deviation) outside of the negative three sigma (sigma corresponds to the standard deviation statistic of the index data) of the mean value, therefore, the value space of the abnormal index data is determined as a value space (i.e., a value space larger than the index data mean +3 × standard deviation) other than the positive three sigma (sigma corresponds to the standard deviation statistic of the index data) of the index data mean.
Let the user-level indicator data be as shown in table 5,
TABLE 5
The calculated mean, standard deviation, and anomaly thresholds for the user-level data of table 5 are shown in table 6:
| PDP creation failure frequency | PDP creation success frequency | Frequency of no downlink traffic after PDP activation | |
| Mean value | 6 | 2 | 3 |
| Standard deviation of | 5.26 | 0.74 | 1.95 |
| Anomaly threshold | 21.64 | 3.88 | 9.28 |
TABLE 6
Taking the PDP creation number as an example, if the average value is 6 and the standard deviation is 5.26, the corresponding anomaly threshold is 6+5.26 × 3 — 21.64.
3) Abnormal user filtering module 30
Screening out high-frequency users with abnormal index data according to an abnormal threshold value; namely, when the value of the user index data is in the range of plus and minus three sigma (corresponding to abnormal threshold) of the average value of the index data, the index data of the user is judged to be normal, otherwise, the user is judged to be a high-frequency user with abnormal index data, namely, the user is judged to be an abnormal user.
Let the user-level indicator data be as shown in table 7,
TABLE 7
The anomaly thresholds for the three indices are shown in table 8:
| PDP creation failure frequency | PDP creation success frequency | Frequency of no downlink traffic after PDP activation | |
| Threshold value | 21.64 | 3.88 | 9.28 |
TABLE 8
The results of the abnormal user screening are shown in table 9:
TABLE 9
4) One-dimensional scoring module 40
The single-dimension scoring module 40 scores three index data of the abnormal user in three dimensions based on the abnormal user screened by the abnormal user screening module 30, thereby obtaining user-level scoring data.
Discretizing the index data of each dimension (i.e. each index) respectively, for example, dividing the frequency count corresponding to the index data of each user into 10 groups, and generating a score of the index data of each dimension of each user based on the grouping result; for example, for the index of the PDP creation success frequency, the scores of the index data of different users are shown in table 10:
| subscriber identity (IMSI code) | Cell Coding (CELLID) | PDP creation success frequency | PDP creation success frequency score |
| 460000******1 | 10001 | 3 | 10 |
| 460000******10 | 10010 | 1 | 8 |
| 460000******11 | 10011 | 1 | 5 |
| 460000******12 | 10012 | 1 | 8 |
| 460000******13 | 10013 | 2 | 5 |
| 460000******14 | 10014 | 1 | 5 |
| 460000******2 | 10002 | 2 | 8 |
| 460000******3 | 10003 | 1 | 10 |
| 460000******4 | 10004 | 2 | 8 |
| 460000******5 | 10005 | 1 | 5 |
| 460000******6 | 10006 | 1 | 5 |
| 460000******7 | 10007 | 2 | 5 |
| 460000******8 | 10008 | 3 | 8 |
| 460000******9 | 10009 | 2 | 5 |
Watch 10
5) Multidimensional clustering module 50
The scoring data of the abnormal users in the three dimensions are associated by using a clustering method based on the scoring data of the single-dimension scoring module 40, and the abnormal users are clustered, so that the perceptibility of each user in the finally obtained cluster tends to be consistent, thereby realizing a mode of determining the perceptibility of the user based on the scoring data of the three dimensions comprehensively evaluated, and clustering the users based on the perceptibility.
The specific processing steps of the multidimensional clustering module 50 include:
step 1), randomly selecting 5 (or other values larger than 2 depending on the fineness of the clusters, wherein the higher the fineness requirement of the clusters, the larger the number of initial clustering centers) objects from all objects (including the scores of index data of all abnormal users, each object corresponding to user-level score data of one abnormal user, including the scores of three index data of users, and each object corresponding to user-level index data of one user) as 5 initial clustering centers (corresponding to 5 clusters); for other objects, allocating the object to be processed to the cluster to which the cluster center with the minimum Euclidean distance to the object to be processed belongs to obtain a new cluster according to the Euclidean distance between the object to be processed (the object except the cluster center in all the objects) and the cluster center (namely, the Euclidean distance between the vector formed by the score data of the index data of three dimensions in other objects and the vector formed by the score data of the index data of three dimensions in the cluster center;
step 2), calculating a mean vector of the new cluster (namely forming score data of three index data of each object in the cluster into a three-dimensional vector, and solving an average value of the three-dimensional vectors corresponding to all the objects in the cluster); and continuously repeating the step 1) and the step 2) until the mean square error between the cluster centers is converged, and stopping the processing.
And 3) determining the cluster number to which each user belongs, wherein at the moment, the value of the score data of the index data of the object in each cluster reaches the maximum compactness degree, and the score data of the index data of the objects in different clusters reaches the maximum dispersion degree.
An example of the clustering result after the data object clustering process is shown in table 11:
TABLE 11
6) Low-perceptibility user screening module 60
According to the clustering result, the perceptibility information of the users is evaluated as shown in table 12, and users outputting low perceptibility, such as users in class 1, class 2, and class 3 in table 10, all belong to users with low perceptibility (i.e. the perceptibility is lower than the perceptibility threshold).
| Subscriber identity (IMSI code) | Cell Coding (CELLID) | Clustering results | Sensing a condition |
| 460000******1 | 10001 | Class 1 | Is very poor |
| 460000******2 | 10002 | Class 1 | Is very poor |
| 460000******4 | 10004 | Class 2 | Is poor |
| 460000******7 | 10007 | Class 3 | Difference (D) |
| 460000******8 | 10008 | Class 2 | Is poor |
| 460000******9 | 10009 | Class 3 | Difference (D) |
| 460000******12 | 10012 | Class 3 | Difference (D) |
| 460000******3 | 10003 | Class 4 | In general |
| 460000******10 | 10010 | Class 4 | In general |
| 460000******11 | 10011 | Class 4 | In general |
| 460000******13 | 10013 | Class 3 | Difference (D) |
| 460000******6 | 10006 | Class 5 | In general |
| 460000******5 | 10005 | Class 5 | In general |
| 460000******14 | 10014 | Class 5 | In general |
TABLE 12
7) Problem cell convergence module 70
Analyzing the concentration degree of the users with low perceptibility, converging the users with low perceptibility to the problem cell (namely, positioning the cell with the distribution degree density of the users with low perceptibility exceeding the density threshold), and outputting a cell list, wherein the cell list can comprise the total number of the users in the cell and the proportion of the users with low perceptibility in the cell as shown in table 13, thereby improving the efficiency of positioning and optimizing the wireless quality problem; the low perceptibility is a user with some indexes lower than the threshold, the indexes of the perceptibility are different under different conditions, in practical application, the mapping relation between the value of the scoring data and different perceptibility can be preset, and the perceptibility of the abnormal user in each cluster is determined based on the scoring data of the abnormal user in the cluster; or, for different perceptibility, presetting value spaces of corresponding scoring data (value spaces can be respectively set according to index data of different dimensions), and determining the perceptibility of the abnormal user in each cluster according to the value space corresponding to the scoring data of the abnormal user in each cluster; for example, if an abnormal user in a cluster fails to activate and create the PDP 10 times in the same cell (more than 10 times corresponding to a value space with a poor perceptibility), the perceptibility of the user is poor; for example, in the same cell, the download rate of a certain user is very low, the TCP delay is very long, and the perception of the user is also poor.
| 25673 | 261 | 2313 | 11.28％ |
| 5722 | 261 | 2785 | 9.37％ |
| 13184 | 173 | 2156 | 8.02％ |
| 3511 | 125 | 2019 | 6.19％ |
| 25032 | 122 | 1991 | 6.13％ |
| 171 | 153 | 2537 | 6.03％ |
| 3442 | 126 | 2093 | 6.02％ |
| 2061 | 156 | 2681 | 5.82％ |
| 3673 | 121 | 2215 | 5.46％ |
| 17532 | 152 | 2788 | 5.45％ |
| 501 | 111 | 2095 | 5.30％ |
| 20583 | 170 | 4079 | 4.17％ |
| 35001 | 179 | 4498 | 3.98％ |
| 35003 | 201 | 5066 | 3.97％ |
Watch 13
In practical applications, the wireless quality problem location system may be implemented by a Microprocessor (MCU) or a logic programmable gate array (FPGA) in a server carrying the wireless quality problem location system.
The method for locating a radio quality problem according to the embodiment of the present invention corresponds to the processing of each module in the above-described system for locating a radio quality problem, and as shown in fig. 3 and 4, includes the following steps:
step 1, obtaining call ticket data of a user from a data acquisition system, and outputting user-level index data through the collection of user dimension and time dimension, namely data corresponding to three indexes (including frequency of PDP creation failure, frequency of PDP creation success and frequency of non-downlink flow after PDP activation).
The ticket data includes session-level PDP signaling data and user service data under a Gn interface GPRS Tunnel Protocol (GTP), and the ticket data includes a user identifier, a cell identifier, a PDP creation request, a PDP creation reply, a service flow generated after the PDP creation, and the like.
The Call ticket Data may be Call detail Record Database (CDR) Data or External Data Representation (XDR) Data, and a Data structure of the Call ticket Data is shown in table 14:
TABLE 14
Collecting through session level data, calculating data (namely index data) corresponding to three indexes of PDP creation failure frequency, PDP creation success frequency and PDP activated non-downlink flow frequency within a period of time to output user level index data, wherein the user level index data of each user comprises three-dimensional index data (data respectively corresponding to three indexes of PDP creation failure frequency, PDP creation success frequency and PDP activated non-downlink flow frequency), the user level index data is identified by a user identifier and a cell identifier, and one example of the user level index data is shown in Table 15;
watch 15
And 2, calculating an abnormal threshold according to the user-level index data.
Determining an abnormal threshold value according to the abnormal frequency of the normal event, and judging whether the index data is normal or not according to the abnormal threshold value;
determining an abnormal threshold value by adopting a six-sigma principle based on input user-level index data: theoretically and practically speaking, the index data of each dimension are subject to normal distribution, so whether the index data are abnormal or not can be judged by the six sigma principle, i.e. the value of the index data is within plus or minus three sigma (corresponding to the anomaly threshold) of the mean value of the index data, the index data is normal, otherwise, the index data is abnormal, the user corresponding to the index data is an abnormal user, since the value of the abnormal index data has a negative value in the value space (i.e. index data mean-3 standard deviation) outside of the negative three sigma (sigma corresponds to the standard deviation statistic of the index data) of the mean value, therefore, the value space of the abnormal index data is determined as a value space (i.e., a value space larger than the index data mean +3 × standard deviation) other than the positive three sigma (sigma corresponds to the standard deviation statistic of the index data) of the index data mean.
Let the user-level metric data be as shown in table 16,
TABLE 16
The anomaly threshold obtained according to the anomaly threshold calculation method is shown in table 17:
| PDP creation failure frequency | PDP creation success frequency | Frequency of no downlink traffic after PDP activation | |
| Mean value | 6 | 2 | 3 |
| Standard deviation of | 5.26 | 0.74 | 1.95 |
| Frequency anomaly threshold | 21.64 | 3.88 | 9.28 |
TABLE 17
And 3, screening out users with abnormal index data, namely abnormal users according to the abnormal threshold value.
Theoretically and practically, the index data should be normally distributed, so that whether the index data is abnormal or not can be judged by the six-sigma principle, that is, the index data is within the range of plus or minus three sigma (corresponding to an abnormal threshold) of the mean value of the index data, the index data is normal, otherwise, the index data is abnormal, and since the value space of the abnormal index data is a negative value outside the minus three sigma (sigma corresponds to the standard deviation statistic of the index data) of the mean value (that is, the average value-3 standard deviation of the index data), the value space of the abnormal index data is the value space (that is, the value space is larger than the mean value +3 standard deviation of the index data) outside the plus three sigma (sigma corresponds to the standard deviation statistic of the index data) of the mean value of the index data.
Let the user-level indicator data be as shown in table 18:
watch 18
By calculating the mean and standard deviation of the data of the three indexes, it can be determined that the abnormal threshold values corresponding to the three indexes are shown in table 19:
| PDP creation failure frequency | PDP creation success frequency | Frequency of no downlink traffic after PDP activation | |
| Threshold value | 21.64 | 3.88 | 9.28 |
Watch 19
The results of the abnormal user screening are shown in table 20:
watch 20
And 4, grading the abnormal users in three dimensions based on the screened users with abnormal index data, namely respectively grading the index data of the three dimensions of the abnormal users, so as to obtain user-level grading data.
Discretizing index data of each dimension in the index data of the three dimensions respectively, dividing frequency numbers corresponding to the index data of each user into 10 groups and generating a score of the index data of each user; for example, for the index of the PDP creation success frequency, the scores of the individual index data (taking the PDP creation success frequency index data as an example) of different users are shown in table 21:
TABLE 21
And 5, correlating the scoring data of the three dimensions of the user-level scoring data by using a clustering method, and clustering abnormal users to enable the perceptibility of each user in the finally obtained cluster to be consistent, so that a mode of determining the perceptibility of the user based on the scoring data of the three dimensions is realized, and clustering processing is performed on the user based on the perceptibility.
The specific treatment steps comprise:
step 1), randomly selecting 5 (or other numerical values larger than 2 depending on the fineness of the clusters, wherein the higher the fineness requirement of the clusters is, the larger the number of initial clustering centers is), as 5 initial clustering centers (corresponding to 5 clusters), from all the objects (including the scores of index data of three dimensions of all abnormal users, and each object corresponds to user-level score data of one abnormal user); for other objects, allocating the object to be processed to the cluster to which the cluster center with the minimum Euclidean distance to the object to be processed belongs to obtain a new cluster according to the Euclidean distance between the object to be processed (the object except the cluster center in all the objects) and the cluster center (namely, the Euclidean distance between the vector formed by the score data of the index data of three dimensions in other objects and the vector formed by the score data of the index data of three dimensions in the cluster center;
step 2), calculating the mean vector of each new cluster (namely forming the grading data of the three index data of each object in the cluster into a three-dimensional vector and solving the mean value of the three-dimensional vectors corresponding to all the objects in the cluster); and continuously repeating the step 1) and the step 2) until the mean square error between the cluster centers is converged, and stopping the processing.
And 3) determining the cluster number to which each user belongs, wherein at the moment, the value of the score data of the index data of the object in each cluster reaches the maximum compactness degree, and the score data of the index data of the objects in different clusters reaches the maximum dispersion degree.
An example of the clustering result after the data object clustering process is shown in table 22:
TABLE 22
And 6, screening out users with low perceptibility based on the clustering result.
According to the clustering result, the information of the perceptibility of the user is evaluated as shown in table 23, and users outputting low perceptibility, such as users in class 1, class 2, and class 3 in table 10, all belong to users with low perceptibility.
| Subscriber identity (IMSI code) | Cell Coding (CELLID) | Clustering results | Sensing a condition |
| 460000******1 | 10001 | Class 1 | Is very poor |
| 460000******2 | 10002 | Class 1 | Is very poor |
| 460000******4 | 10004 | Class 2 | Is poor |
| 460000******7 | 10007 | Class 3 | Difference (D) |
| 460000******8 | 10008 | Class 2 | Is poor |
| 460000******9 | 10009 | Class 3 | Difference (D) |
| 460000******12 | 10012 | Class 3 | Difference (D) |
| 460000******3 | 10003 | Class 4 | In general |
| 460000******10 | 10010 | Class 4 | In general |
| 460000******11 | 10011 | Class 4 | In general |
| 460000******13 | 10013 | Class 3 | Difference (D) |
| 460000******6 | 10006 | Class 5 | In general |
| 460000******5 | 10005 | Class 5 | In general |
| 460000******14 | 10014 | Class 5 | In general |
TABLE 23
And 7, positioning the user with low perceptibility to the problem cell, and outputting a list of the problem cell.
Based on the users with low awareness obtained in step 6, the users with low awareness of each cell can be obtained, and the ratio of the users with low awareness of the cells is obtained by combining the total number of the users of the cells, as shown in table 24, and then used as the basis for whether the cells have problems and need to be optimized; the low perceptibility is a user with some indexes lower than the threshold, and under different conditions, the indexes of the perceptibility are different, for example, if a certain user fails to activate and create the PDP in the same cell for 10 times, the perceptibility of the user is poor; for example, in the same cell, the download rate of a certain user is very low, the TCP delay is very long, and the perception of the user is also poor.
| Cell Coding (CELLID) | Number of users with poor perception | Total user | Abnormal user proportion |
| 25673 | 261 | 2313 | 11.28％ |
| 5722 | 261 | 2785 | 9.37％ |
| 13184 | 173 | 2156 | 8.02％ |
| 3511 | 125 | 2019 | 6.19％ |
| 25032 | 122 | 1991 | 6.13％ |
| 171 | 153 | 2537 | 6.03％ |
| 3442 | 126 | 2093 | 6.02％ |
| 2061 | 156 | 2681 | 5.82％ |
| 3673 | 121 | 2215 | 5.46％ |
| 17532 | 152 | 2788 | 5.45％ |
| 501 | 111 | 2095 | 5.30％ |
| 20583 | 170 | 4079 | 4.17％ |
| 35001 | 179 | 4498 | 3.98％ |
| 35003 | 201 | 5066 | 3.97％ |
Watch 24
In summary, the embodiments of the present invention collect data based on the Gn interface, and are based on three indexes: the PDP establishing failure frequency, the PDP establishing success frequency and the frequency index without downlink flow after PDP activation are processed by utilizing a six-sigma statistical method to screen out abnormal users, and then clustering is carried out on the abnormal users by utilizing a clustering algorithm, so that the user perception of the abnormal users is comprehensively evaluated;
the three indexes used in the embodiment of the invention are indexes which are checked and screened to confirm that abnormal users are screened most effectively so as to evaluate the perception of the abnormal users;
the embodiment of the invention obtains the ticket data by the Gn interface-based data system, and the Gn interface-based data system is basic in GPRS, TD-SCDMA and WCDMA networks, thereby being suitable for different networks.
Those of ordinary skill in the art will understand that: all or part of the steps for implementing the method embodiments may be implemented by hardware related to program instructions, and the program may be stored in a computer readable storage medium, and when executed, the program performs the steps including the method embodiments; and the aforementioned storage medium includes: various media that can store program codes, such as a removable Memory device, a Random Access Memory (RAM), a Read-Only Memory (ROM), a magnetic disk, and an optical disk.
Alternatively, the integrated unit of the present invention may be stored in a computer-readable storage medium if it is implemented in the form of a software functional module and sold or used as a separate product. Based on such understanding, the technical solutions of the embodiments of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the methods described in the embodiments of the present invention. And the aforementioned storage medium includes: a removable storage device, a RAM, a ROM, a magnetic or optical disk, or various other media that can store program code.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.
Claims (10)
1. A wireless quality issue location system, the system comprising:
the index calculation module is used for acquiring call ticket data of a user, collecting user dimension and time dimension of the user call ticket data and outputting user-level index data;
an anomaly threshold calculation module for determining an anomaly threshold based on the user-level indicator data;
the abnormal user screening module is used for screening abnormal users from the users based on the abnormal threshold value;
the single-dimension scoring module is used for scoring the user-level index data of the abnormal user in different dimensions and outputting user-level scoring data corresponding to the abnormal user;
the multidimensional clustering module is used for clustering the abnormal users based on the grading data of the user level and outputting clustering results of the abnormal users;
the low-perceptibility user screening module is used for determining the perceptibility information of the abnormal user based on the clustering result;
and the problem cell convergence module is used for determining a cell to which the abnormal user with the perception degree lower than the perception degree threshold belongs as a problem cell based on the perception degree information.
2. The system of claim 1,
the index calculation module is also used for acquiring the ticket data of the user from a Gn interface-based data acquisition system, and the ticket data of the user comprises session level signaling plane data and user plane data;
and collecting the user call ticket data through user dimension and time dimension to obtain index data corresponding to indexes of at least three dimensions as follows: the frequency of packet data protocol PDP creation failure, the frequency of PDP creation success and the frequency of no downlink flow after PDP activation.
3. The system of claim 2,
the abnormal threshold calculation module is further used for determining the mean value and the standard deviation of the index data of each dimension;
determining the sum of the determined mean and three times the standard deviation of each of the dimensions as an anomaly threshold of the index data of the corresponding dimension.
4. The system of claim 2,
the single-dimension scoring module is further configured to perform discretization processing on the index data corresponding to each dimension in the user-level index data of the abnormal user, and group the discretized index data;
and correspondingly determining the grade of the index data of each dimension of the abnormal user based on the grouping result.
5. The system of any one of claims 1 to 4,
the multi-dimension clustering module is further used for selecting N objects from all the objects as initial clustering centers, wherein N is an integer greater than or equal to 2, and distributing the objects to be processed to the cluster to which the clustering center with the minimum Euclidean distance to the objects to be processed belongs on the basis of the minimum Euclidean distance principle to obtain new clusters; until the time when the user wants to use the device,
convergence of mean square error between the cluster centers; the objects correspond to the user-level scoring data of the abnormal users one by one, and the objects to be processed are all the objects except the N objects.
6. A method for locating a wireless quality problem, the method comprising:
acquiring call ticket data of a user, and collecting user dimension and time dimension of the call ticket data to obtain user-level index data;
determining an abnormal threshold value based on the user-level index data, and screening abnormal users from the users based on the abnormal threshold value;
grading the user-level index data of the abnormal user in different dimensions to obtain user-level grading data of the abnormal user, and clustering the abnormal user based on the user-level grading data to obtain a clustering result of the abnormal user;
and determining the perception degree information of the abnormal user based on the clustering result, and determining the cell to which the abnormal user with the perception degree lower than the perception degree threshold belongs as a problem cell.
7. The method of claim 6, wherein the step of obtaining the user-level index data by aggregating the call ticket data of the user through a user dimension and a time dimension comprises:
acquiring call ticket data of the user from a data acquisition system based on a Gn interface, wherein the call ticket data of the user comprises session level signaling plane data and user plane data;
collecting the user dimension and the time dimension of the call ticket data of the user to obtain index data corresponding to the indexes of at least three dimensions as follows: the frequency of packet data protocol PDP creation failure, the frequency of PDP creation success and the frequency of no downlink flow after PDP activation.
8. The method of claim 7, wherein determining an anomaly threshold based on user-level metric data comprises:
respectively calculating the mean value and the standard deviation of the index data of each dimension in the user-level index data; and determining the sum of the determined mean and three times of the standard deviation as the abnormal threshold of the index data of the corresponding dimension.
9. The method of claim 7, wherein the obtaining user-level score data corresponding to the abnormal user from scoring the user-level index data of the abnormal user in different dimensions comprises:
discretizing the index data corresponding to each dimension in the user-level index data of the abnormal user, and grouping the discretized index data;
determining a score for the metric data for each of the dimensions of the anomalous user based on the grouping results.
10. The method according to any one of claims 6 to 9, wherein the clustering the abnormal users based on the user-level scoring data to obtain a clustering result comprises:
selecting N objects from all the objects as initial clustering centers, wherein N is an integer greater than or equal to 2, and distributing the objects to be processed to the cluster to which the clustering center with the minimum Euclidean distance to the objects to be processed belongs on the basis of the minimum Euclidean distance principle to obtain a new cluster; until the time when the user wants to use the device,
convergence of mean square error between the cluster centers;
the objects correspond to the user-level scoring data of the abnormal users one by one, and the objects to be processed are all the objects except the N objects.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510030063.6A CN105873113B (en) | 2015-01-21 | 2015-01-21 | Wireless quality positioning problems method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510030063.6A CN105873113B (en) | 2015-01-21 | 2015-01-21 | Wireless quality positioning problems method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105873113A true CN105873113A (en) | 2016-08-17 |
| CN105873113B CN105873113B (en) | 2019-05-28 |
Family
ID=56623017
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510030063.6A Active CN105873113B (en) | 2015-01-21 | 2015-01-21 | Wireless quality positioning problems method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105873113B (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106326461A (en) * | 2016-08-30 | 2017-01-11 | 杭州东方通信软件技术有限公司 | Real time processing guarantee method and system based on network signaling record |
| CN108108743A (en) * | 2016-11-24 | 2018-06-01 | 百度在线网络技术（北京）有限公司 | Abnormal user recognition methods and the device for identifying abnormal user |
| CN108322363A (en) * | 2018-02-12 | 2018-07-24 | 腾讯科技（深圳）有限公司 | Propelling data abnormality monitoring method, device, computer equipment and storage medium |
| CN109994128A (en) * | 2017-12-30 | 2019-07-09 | 中国移动通信集团贵州有限公司 | Voice quality problem localization method, device, equipment and medium |
| CN110163457A (en) * | 2018-02-14 | 2019-08-23 | 北京京东尚科信息技术有限公司 | A kind of abnormal localization method and device of operational indicator |
| CN110633388A (en) * | 2019-08-15 | 2019-12-31 | 宜通世纪物联网研究院(广州)有限公司 | Real-time index generation method, system and storage medium based on communication XDR |
| CN110636528A (en) * | 2019-09-30 | 2019-12-31 | 南京邮电大学 | Regional wireless perception problem feature identification method |
| CN110728525A (en) * | 2018-07-16 | 2020-01-24 | 中国移动通信集团浙江有限公司 | Delimitation processing method and device for network batch user complaints |
| CN110868732A (en) * | 2018-08-27 | 2020-03-06 | 中国移动通信集团广东有限公司 | VoLTE radio access failure problem positioning method, system and equipment |
| CN111030876A (en) * | 2019-12-25 | 2020-04-17 | 武汉绿色网络信息服务有限责任公司 | NB-IoT terminal fault positioning method and device based on DPI |
| CN111372073A (en) * | 2018-12-26 | 2020-07-03 | 中国移动通信集团四川有限公司 | Video quality evaluation method, device, equipment and medium |
| CN113219504A (en) * | 2017-02-28 | 2021-08-06 | 荣耀终端有限公司 | Positioning information determining method and device |
| CN114071520A (en) * | 2020-08-06 | 2022-02-18 | 中国移动通信集团广东有限公司 | LTE network problem positioning method and device and electronic equipment |
| CN115843046A (en) * | 2021-10-15 | 2023-03-24 | 中国联合网络通信集团有限公司 | Elevated road data processing method and device, electronic equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102014408A (en) * | 2010-12-16 | 2011-04-13 | 北京拓明科技有限公司 | Data service flow and cell reelection-based mobile network analysis method |
| CN102075978A (en) * | 2011-01-28 | 2011-05-25 | 浪潮通信信息系统有限公司 | Voice service user negative perception-based network problem analysis method |
| CN103138963A (en) * | 2011-11-25 | 2013-06-05 | 华为技术有限公司 | Method and device for positioning network problems based on user perception |
| JP2014123885A (en) * | 2012-12-21 | 2014-07-03 | Kddi Corp | Radio wave propagation characteristic estimation device, radio wave propagation characteristic estimation method, and computer program |
-
2015
- 2015-01-21 CN CN201510030063.6A patent/CN105873113B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102014408A (en) * | 2010-12-16 | 2011-04-13 | 北京拓明科技有限公司 | Data service flow and cell reelection-based mobile network analysis method |
| CN102075978A (en) * | 2011-01-28 | 2011-05-25 | 浪潮通信信息系统有限公司 | Voice service user negative perception-based network problem analysis method |
| CN103138963A (en) * | 2011-11-25 | 2013-06-05 | 华为技术有限公司 | Method and device for positioning network problems based on user perception |
| JP2014123885A (en) * | 2012-12-21 | 2014-07-03 | Kddi Corp | Radio wave propagation characteristic estimation device, radio wave propagation characteristic estimation method, and computer program |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106326461B (en) * | 2016-08-30 | 2019-07-30 | 杭州东方通信软件技术有限公司 | A kind of real-time processing support method and system based on network signal record |
| CN106326461A (en) * | 2016-08-30 | 2017-01-11 | 杭州东方通信软件技术有限公司 | Real time processing guarantee method and system based on network signaling record |
| CN108108743A (en) * | 2016-11-24 | 2018-06-01 | 百度在线网络技术（北京）有限公司 | Abnormal user recognition methods and the device for identifying abnormal user |
| CN113219504A (en) * | 2017-02-28 | 2021-08-06 | 荣耀终端有限公司 | Positioning information determining method and device |
| CN109994128A (en) * | 2017-12-30 | 2019-07-09 | 中国移动通信集团贵州有限公司 | Voice quality problem localization method, device, equipment and medium |
| CN109994128B (en) * | 2017-12-30 | 2021-06-04 | 中国移动通信集团贵州有限公司 | Voice quality problem positioning method, device, equipment and medium |
| CN108322363A (en) * | 2018-02-12 | 2018-07-24 | 腾讯科技（深圳）有限公司 | Propelling data abnormality monitoring method, device, computer equipment and storage medium |
| CN108322363B (en) * | 2018-02-12 | 2020-11-13 | 腾讯科技（深圳）有限公司 | Pushed data abnormity monitoring method and device, computer equipment and storage medium |
| CN110163457A (en) * | 2018-02-14 | 2019-08-23 | 北京京东尚科信息技术有限公司 | A kind of abnormal localization method and device of operational indicator |
| CN110728525B (en) * | 2018-07-16 | 2022-04-15 | 中国移动通信集团浙江有限公司 | Delimitation processing method and device for network batch user complaints |
| CN110728525A (en) * | 2018-07-16 | 2020-01-24 | 中国移动通信集团浙江有限公司 | Delimitation processing method and device for network batch user complaints |
| CN110868732A (en) * | 2018-08-27 | 2020-03-06 | 中国移动通信集团广东有限公司 | VoLTE radio access failure problem positioning method, system and equipment |
| CN111372073A (en) * | 2018-12-26 | 2020-07-03 | 中国移动通信集团四川有限公司 | Video quality evaluation method, device, equipment and medium |
| CN110633388B (en) * | 2019-08-15 | 2021-12-28 | 广东宜通衡睿科技有限公司 | Real-time index generation method, system and storage medium based on communication XDR |
| CN110633388A (en) * | 2019-08-15 | 2019-12-31 | 宜通世纪物联网研究院(广州)有限公司 | Real-time index generation method, system and storage medium based on communication XDR |
| CN110636528A (en) * | 2019-09-30 | 2019-12-31 | 南京邮电大学 | Regional wireless perception problem feature identification method |
| CN111030876A (en) * | 2019-12-25 | 2020-04-17 | 武汉绿色网络信息服务有限责任公司 | NB-IoT terminal fault positioning method and device based on DPI |
| CN111030876B (en) * | 2019-12-25 | 2022-05-06 | 武汉绿色网络信息服务有限责任公司 | NB-IoT terminal fault positioning method and device based on DPI |
| CN114071520A (en) * | 2020-08-06 | 2022-02-18 | 中国移动通信集团广东有限公司 | LTE network problem positioning method and device and electronic equipment |
| CN114071520B (en) * | 2020-08-06 | 2023-11-21 | 中国移动通信集团广东有限公司 | LTE network problem positioning method and device and electronic equipment |
| CN115843046A (en) * | 2021-10-15 | 2023-03-24 | 中国联合网络通信集团有限公司 | Elevated road data processing method and device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105873113B (en) | 2019-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105873113B (en) | Wireless quality positioning problems method and system | |
| CN104396188B (en) | System and method for carrying out basic reason analysis to mobile network property problem | |
| CN109548041B (en) | Wireless coverage analysis method and system | |
| CN109803295B (en) | Method and device for evaluating communication cell rectification priority | |
| US9491285B2 (en) | Technique for performance management in a mobile communications network | |
| CN106559803A (en) | A kind of base station construction appraisal procedure and device | |
| CN111294819B (en) | Network optimization method and device | |
| CN108243435B (en) | Parameter optimization method and device in LTE cell scene division | |
| CN104320795A (en) | Evaluation method for health degree of multidimensional wireless network | |
| US20120203788A1 (en) | Network management system and method for identifying and accessing quality of service issues within a communications network | |
| CN104123477A (en) | Group-oriented leasing analysis method based on life data | |
| CN112953738B (en) | Root cause alarm positioning system, method and device and computer equipment | |
| CN104901827B (en) | A kind of network resource evaluation method and device based on customer service structure | |
| CN108243429B (en) | Network coverage detection method and system for indoor distribution system | |
| CN107205245B (en) | Hot spot area automatic identification method and device | |
| CN101951623B (en) | User behavior statistical method and device based on user events | |
| CN106304085B (en) | Information processing method and device | |
| WO2022183729A1 (en) | User perception evaluation method and apparatus, and terminal and storage medium | |
| CN106304161A (en) | Data analysing method and device | |
| CN104486769A (en) | Method and device for selecting valuable cell | |
| CN106817710B (en) | Method and device for positioning network problem | |
| CN106488480B (en) | Work order engine implementation method and device | |
| WO2021008393A1 (en) | Method and system for identifying coverage black hole of wireless cell | |
| CN105050119B (en) | A kind of mobility statistical method and device | |
| CN104519511A (en) | Method and device for detecting scene breaks of communication network cells |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |