CN105873105B - A kind of mobile radio communication abnormality detection and localization method based on network Quality of experience - Google Patents
A kind of mobile radio communication abnormality detection and localization method based on network Quality of experience Download PDFInfo
- Publication number
- CN105873105B CN105873105B CN201610262009.9A CN201610262009A CN105873105B CN 105873105 B CN105873105 B CN 105873105B CN 201610262009 A CN201610262009 A CN 201610262009A CN 105873105 B CN105873105 B CN 105873105B
- Authority
- CN
- China
- Prior art keywords
- abnormal
- point
- network
- experience
- key performance
- 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.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/02—Arrangements for optimising operational condition
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses a kind of mobile radio communication abnormality detections and localization method based on network Quality of experience, are characterized in through the abnormality detection subsystem based on network Quality of experience and abnormal root because of positioning subsystem, real-time exception monitoring and abnormal positioning to real-time performance；In abnormity diagnosis part, from the angle of network synthesis user experience, three Key Performance Indicators is selected to be divided using coarseness threshold value as network diagnosis feature and the method for fine granularity cluster carries out anomaly classification；In abnormal root because of position portion, abnormal symptom feature is obtained using the matched mode of Cumulative Distribution Function, cluster analysis is carried out in each type, different abnormal roots can be realized because of positioning；Cellular Networks abnormality detection and abnormal root are finally constituted because of positioning subsystem.The present invention can not only detect apparent exception, additionally it is possible to detect potential exception, and can be directed to different Exception Types carry out abnormal root because positioning；Can also autonomous learning, constantly improve abnormality detection and Gen Yin positioning accuracy.
Description
Technical field
The invention belongs to mobile communications network monitorings and optimisation technique field, and in particular to cellular mobile communications networks are experienced
Quality (Quality of Experience, QoE) carries out monitoring in real time and to abnormal root because of the method and system positioned,
And the long term evolution (LTE) of monitoring 4G universal mobile communications technologies now net and future mobile communications network operation quality and service water
Flat processing method.
Background technology
The network optimization for cellular mobile communications networks is a system engineering through whole network development overall process,
And mobile communication network operator is improving network investment benefit, running quality and the important technical of service level.
Abnormality detection mechanism is an important research content in network optimisation techniques, available for finding and positioning honeycomb mobile radio communication
Abnormal behaviour in network.Mobile communication network operators come into effect abnormality detection to cellular mobile communications networks at present,
Used method is mainly monitored Key Performance Indicator (Key Performance Indicator, KPI), then
Use experience threshold value carries out hard decision to Key Performance Indicator KPI.All or part is only worked as in being limited in that for this method
Interested KPI indexs are fallen when except normality threshold range, which just assert that network exception occurs and alarms.Analytical table
Bright, this method for detecting abnormality based on KPI threshold values, detection probability and threshold value setting relevance are very big, and to falling in threshold value
The abnormal behaviour at edge is often difficult to really reflect network quality there are larger false-alarm and false dismissal probability.Existing base
It, also can not be accurately and automatic in the general lack of ability of discovery to other potential abnormal behaviours of the method for detecting abnormality of KPI threshold values
The abnormal root of positioning because.This mode often relies on the experience of skilled engineer to position the root of the abnormal behaviour detected
Cause.
Invention content
The purpose of the present invention is to propose to a kind of mobile radio communication abnormality detection and localization method based on network Quality of experience,
By carrying out cluster analysis to Key Performance Indicator (KPI), realizing the abnormal behaviour diagnosis based on network Quality of experience and classifying,
Be automatically completed different abnormal roots because positioning.
The present invention is based on the mobile radio communication abnormality detection and localization method of network Quality of experience, from network parameter,
Whether judging Network Abnormal, it is characterised in that：Using the abnormality detection subsystem based on network Quality of experience and based on network body
The abnormal root for the amount of checking the quality divides the anomaly classification that network entirety is carried out with feature clustering using threshold value, simultaneously because of positioning subsystem
To remaining symptom of network -- remaining Key Performance Indicator carries out cluster analysis；Concrete operations are：
The abnormality detection subsystem based on network Quality of experience, input is from communication network radio resource controller first
The network key performance indicator (KPI) of (Radio Network Controller, RNC) acquisition, then from access property, complete
Property, the aspect of retentivity three select corresponding network key performance indicator as network overall experience quality, for reaction network
In all users average user perceive；Again by the way of threshold value division, according to the above-mentioned three classes key performance being previously set
The decision threshold of index marks off apparent exception, slight abnormality and apparent normal three classes point；It is mapped using self-organizing nerve
(Self-Organizing Map, SOM) and two kinds of clustering algorithms of K central point algorithms (K-mediods) are to slight abnormality point and bright
Aobvious abnormal point carries out fine granularity analysis again：Self-organizing nerve mapping SOM is a kind of unsupervised learning algorithm, comprising input layer and
Output layer：Input layer corresponds to the input vector of a higher-dimension, and output layer is had by a series of M of the tissue on two-dimensional grid × N number of
Sequence node is formed, and input node is connect with output node by weight vectors；In learning process, find therewith apart from shortest defeated
Go out layer unit i.e. winning unit, be updated；Meanwhile by the right value update of adjacent domain, output node is made to keep input vector
Topological characteristic, ultimately form M × N number of group, correspond to different types of exception respectively；And K central point algorithms are a kind of classics
Clustering algorithm, randomly selects point set centered on a group cluster sample first, and each central point corresponds to a cluster；Then it calculates each
Sample point is put into that shortest cluster of distance center point, calculates in each cluster to the distance of each central point by sample point, will be away from
The minimum point of the exhausted degree error of each sample point distance is as new central point in cluster；If last new center point set and former center
Point set is identical, then algorithm terminates；Using the mapping of self-organizing nerve and K central point algorithms to slight abnormality point and apparent abnormal point pair
The abnormal data set answered is clustered, and is finally obtained K Key Performance Indicator set, is exactly Exception Type；
It is described based on the abnormal root of network Quality of experience because of positioning subsystem, closed between normal point and abnormal point by finding
The difference of keyness energy index parameter positions the possible cause of abnormal point；Then the abnormal point class obtained from abnormality detection subsystem
Type sets out, and compares the cumulative distribution function (Cumulative of remaining Key Performance Indicator between normal point and abnormal point
Distribution Function, CDF) curve graph, corresponding abnormal symptom Key Performance Indicator is extracted, builds abnormal symptom
Key Performance Indicator library；It is special to the symptom Key Performance Indicator of above-mentioned acquisition using K central point algorithms again in each type
Sign is clustered, and then obtain different abnormal symptom lists, realizes root because of positioning by searching for the list.
Mobile radio communication abnormality detection and localization method the present invention is based on network Quality of experience are taken by being based on net
The abnormality detection subsystem of network Quality of experience and based on the abnormal root of network Quality of experience because of positioning subsystem, to real-time performance reality
When exception monitoring and root cause analysis；In abnormity diagnosis part, due to the angle from network Quality of experience, three classes key performance is selected
Index is as network holistic diagnosis feature, the method progress anomaly classification clustered using the division of coarseness threshold value and fine granularity,
False-alarm and false dismissal can be effectively avoided, is capable of detecting when apparent abnormal and potential exception；In abnormal root because of position portion, by
In using the matched mode of cumulative distribution function to remaining Key Performance Indicator to normal and abnormal point, structure abnormal symptom is crucial
Performance indicator library, and cluster analysis is carried out in each Exception Type, different abnormal roots can be realized because of positioning.Take this hair
The system that bright method is built can be with autonomous learning, so as to constantly improve the accurate of abnormality detection and Gen Yin positioning
Degree.
Description of the drawings
Fig. 1 is based on the mobile radio communication abnormality detection of network Quality of experience and the overall structure diagram of localization method.
Fig. 2 is the structure diagram of the abnormality detection subsystem B based on network Quality of experience.
Fig. 3 is because of the structure diagram of positioning subsystem C based on the abnormal root of network Quality of experience.
Specific embodiment
Embodiment 1：
Mobile radio communication abnormality detection and localization method of the present embodiment based on network Quality of experience, including using based on net
The abnormality detection subsystem of network Quality of experience and based on the abnormal root of network Quality of experience because of positioning subsystem, divided using threshold value
The anomaly classification of network overall experience quality is carried out with feature clustering, while remaining symptom Key Performance Indicator gathers to network
Alanysis；
Fig. 1 gives the overall structure signal of mobile radio communication abnormality detection and localization method based on network Quality of experience
Figure.Mobile radio communication abnormality detection and localization method of the present embodiment based on network Quality of experience specifically include following steps：From
Key Performance Indicator (the Key of communication network radio resource controller (Radio Network Controller, RNC) acquisition
Performance Indicator, KPI), form Key Performance Indicator library A, the input as whole system.In the present embodiment 1
In, Key Performance Indicator library A is 62 of 27 days of 2659 cells of Chengde City, Hebei Province movement 3G RNC acquisitions key
It can index.First in terms of access property, integrality, retentivity three, corresponding Key Performance Indicator is selected, whole as network
Quality of experience (Quality ofExperience, QoE) perceives for the average user of users all in reaction network；This reality
It is wireless access, voice quality and TCH cutting off rate three classes Key Performance Indicators A1 to apply selected in example 1.Then, by this three classes
Key Performance Indicator A1 inputs abnormality detection subsystem B, obtains different Exception Type B4；Finally, by these Exception Types B4
With the abnormal root of remaining Key Performance Indicator A2 inputs in addition to above-mentioned three classes Key Performance Indicator A1 because of positioning subsystem C, obtain
Abnormal symptom list D1 can be obtained by abnormal root because of D by searching for these lists.
Fig. 2 gives the structure diagram of the abnormality detection subsystem B based on network Quality of experience：In the present embodiment 1,
Abnormality detection subsystem B inputs are Key Performance Indicator library A, first in terms of access property, integrality, retentivity three, respectively
Select the Quality of experience of wireless access, voice quality and TCH cutting off rates three classes Key Performance Indicator A1 as network entirety
QoE perceives for the average user of users all in reaction network.Since abnormal degree is different, while abnormal points are very
It is few, it is easily submerged in normal point, therefore chromatographic analysis is carried out to data.B1 is divided to Key Performance Indicator first with threshold value
Preceding 10000 data point in the A of library carries out coarseness classification.So-called coarseness refers to the levels of precision that problem analysis is reached,
Following fine granularity also means this.Coarseness analysis specifically refers to according to the above-mentioned three classes key performance being previously set
The decision threshold of A1 is marked, obtains apparent exception, slight abnormality and apparent normal three classes point.This coarseness analysis and operator
Hard decision it is similar, when certain Key Performance Indicator indexs below or above setting threshold value when, it is assumed that be network occur it is different
It often and alarms, apparent and slight abnormality point is sorted out in the present embodiment 1 and adds up to 4079.Then, it is different to slight and apparent two class
Normal point B2 carries out fine granularity analysis again.This step mainly maps (Self-Organizing Map, SOM) using self-organizing nerve
Data set is clustered with the hybrid algorithm of K central point algorithms (K-mediods), obtains different Exception Types.It is so-called
Cluster refers to the method classified to the group of unknown classification, is a kind of important method of data mining, is used here from group
Knit nerve mapping (Self-Organizing Map, SOM) and two kinds of clustering algorithms of K central point algorithms (K-mediods).SOM is
A kind of unsupervised learning algorithm, includes input layer and output layer.Input layer correspond to a higher-dimension input vector, output layer by
A series of M × N number of ordered nodes of tissues on two-dimensional grid are formed, and input node is connected with output node by weight vectors
It connects；In learning process, find therewith apart from shortest output layer unit i.e. winning unit, be updated；Meanwhile by adjacent domain
Right value update, make output node keep input vector topological characteristic, ultimately form M × N number of group, correspond to inhomogeneity respectively
The exception of type.And K central point algorithm K-mediods algorithms are a kind of Classic Clustering Algorithms, randomly select a group cluster sample first
Point set centered on this, each central point correspond to a cluster；Then calculating each sample point, (such as Europe is several to the distance of each central point
Reed distance), sample point is put into that shortest cluster of distance center point, by each cluster middle-range cluster each sample point distance it is exhausted
The minimum point of error is spent as new central point；If last new center point set is identical with former center point set, algorithm terminates.
Slight abnormality point and the corresponding abnormal data set of apparent abnormal point are clustered using SOM and K-mediods algorithms, final
To K class Key Performance Indicator set, here it is Exception Type B4.The Hybrid Clustering Algorithm B2 of joint SOM and K-mediods, can
So that high dimensional data is transformed on two dimensional surface, there is typical effect of visualization, while also reduce the dimension sum number of data
According to amount.The recommendation 10 × 10 of SOM algorithms is used in the present embodiment 1, the type center after being clustered to SOM selects suitable K values
(most suitable according to operating experience 10 to 20) with K-mediods algorithms, merges the type with larger similitude, most
K big types (combinations of the different values of above-mentioned three classes Key Performance Indicator A1), that is, Exception Type B4 are obtained eventually.Newly
The above-mentioned three classes Key Performance Indicator A1 of subzone network input this module, it is by the matched mode of similitude, i.e., logical first
Coarseness division is crossed, then is matched with corresponding Exception Type central point, affiliated type may finally be found from Exception Type B4.
Fig. 3 is because of the structure diagram of positioning subsystem based on the abnormal root of network Quality of experience.
Exception root shown in Fig. 3 is because in positioning subsystem structure chart, it can be seen that entire exception root is because of positioning subsystem
Structure.First, to divided in abnormality detection subsystem B by threshold value the apparent normal and apparent abnormal point that B1 obtains remaining
Key Performance Indicator A2 (i.e. 59 KPI of residue) carries out abnormal symptom extraction C1.Method is its that calculate normal point and abnormal point
The Cumulative Distribution Function (CDF) of remaining Key Performance Indicator, compares difference between the two.When certain class Key Performance Indicator (KPI)
CDF there are during notable difference, it may be considered that the KPI may be abnormal Producing reason.The side compared using this CDF
Formula, structure abnormal symptom KPI libraries.Then, each Exception Type B4 obtained to abnormality detection subsystem B, using K-mediods
The KPI that algorithm appears in it in abnormal symptom KPI libraries carries out abnormal symptom cluster C2, and then it is different to obtain such abnormal difference
Normal symptom checklist D1, that is, the possible KPI set that such is caused to generate extremely.Exception is can be obtained by by searching for these lists
Root is because of D.In embodiment 1, the wireless access of each cell, voice quality, TCH call drop three classes in test data are detected first
KPI matches it with the output abnormality type B 4 of abnormality detection subsystem, obtains affiliated Exception Type.It extracts again small
Area remaining Key Performance Indicator A2 searches the abnormal symptom list D1 under the abnormal class, finally analyzes abnormal root because of D.
From description and the analytic process of example above operating process it is found that the input of the present invention is the network acquired from network element
Parameter, calculating process are realized on the hardware such as server, and output can then be exported different in a manner of text or screen display etc.
Whether often, Exception Type and abnormal root be because of these results.
The present invention is based on the mobile radio communication abnormality detections and localization method of network Quality of experience, are experienced by being based on network
The abnormality detection subsystem of quality and based on the abnormal root of network Quality of experience because of positioning subsystem, it is abnormal in real time to real-time performance
Monitoring；In abnormity diagnosis part, from the angle of network synthesis user experience, three classes Key Performance Indicator is selected as network diagnosis
Feature, is divided using coarseness threshold value and the method for fine granularity cluster carries out anomaly classification；In Network Abnormal root because of positioning region
Point, abnormal symptom KPI libraries are built using the matched mode of Cumulative Distribution Function, cluster analysis is carried out in each Exception Type,
Abnormal symptom list is obtained, for positioning different abnormal roots because finally constituting Cellular Networks abnormality detection and abnormal root because of positioning
Subsystem.Thick fine-grained chromatographic analysis is carried out to Exception Type since the method for the present invention takes, is capable of detecting when apparent different
Normal and potential exception, and for different Exception Type carry out abnormal root because positioning.In addition, the method for the present invention is taken to be taken
The clustering algorithm that the system built uses can with autonomous learning, so as to constantly improve abnormality detection and Gen Yin positioning it is accurate
Degree.
Claims (1)
1. a kind of mobile radio communication abnormality detection and localization method based on network Quality of experience from network parameter, judge
Whether Network Abnormal, it is characterised in that：It checks the quality using the abnormality detection subsystem based on network Quality of experience and based on network body
The abnormal root of amount divides the anomaly classification that network entirety is carried out with feature clustering using threshold value, while to net because of positioning subsystem
Remaining Key Performance Indicator of network carries out cluster analysis；Concrete operations are：
The abnormality detection subsystem based on network Quality of experience, first input are acquired from communication network radio resource controller
Network key performance indicator, then from access property, corresponding network key performance is selected to refer in terms of integrality, retentivity three
It is denoted as network overall experience quality, is perceived for the average user of users all in reaction network；It is divided again using threshold value
Mode, according to the decision threshold of above-mentioned three classes Key Performance Indicator being previously set, mark off apparent exception, slight abnormality and
Apparent normal three classes point；Using the mapping of self-organizing nerve and two kinds of clustering algorithms of K central point algorithms to slight abnormality point and significantly
Abnormal point carries out fine granularity analysis again：The mapping of self-organizing nerve is a kind of unsupervised learning algorithm, includes input layer and output
Layer：Input layer corresponds to the input vector of a higher-dimension, a series of M × N number of orderly section of the output layer by tissues on two-dimensional grid
Point is formed, and input node is connect with output node by weight vectors；In learning process, find therewith apart from shortest output layer
Unit, that is, winning unit is updated；Meanwhile by the right value update of adjacent domain, output node is made to keep the topology of input vector
Feature ultimately forms M × N number of group, corresponds to different types of exception respectively；And K central point algorithms are a kind of classical cluster calculations
Method, randomly selects point set centered on a group cluster sample first, and each central point corresponds to a cluster；Then each sample point is calculated
To the distance of each central point, sample point is put into that shortest cluster of distance center point, is calculated in each cluster, it will be away from each in cluster
The exhausted degree error of sample point distance is minimum to be put as new central point；If last new center point set and former center point set phase
Together, then algorithm terminates；It is mapped using self-organizing nerve corresponding different to slight abnormality point and apparent abnormal point with K central point algorithms
Regular data collection is clustered, and is finally obtained K Key Performance Indicator set, is exactly Exception Type；
It is described based on the abnormal root of network Quality of experience because of positioning subsystem, it is key between normal point and abnormal point by finding
The difference of energy index parameter positions the possible cause of abnormal point；Then the abnormal vertex type obtained from abnormality detection subsystem goes out
Hair compares the cumulative distribution function curve graph of remaining Key Performance Indicator between normal point and abnormal point, extracts corresponding exception
Symptom Key Performance Indicator, structure abnormal symptom Key Performance Indicator library；In each type, again using K central point algorithms
The symptom Key Performance Indicator feature of above-mentioned acquisition is clustered, and then obtains different abnormal symptom lists, by searching for
The list realizes root because of positioning.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610262009.9A CN105873105B (en) | 2016-04-22 | 2016-04-22 | A kind of mobile radio communication abnormality detection and localization method based on network Quality of experience |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610262009.9A CN105873105B (en) | 2016-04-22 | 2016-04-22 | A kind of mobile radio communication abnormality detection and localization method based on network Quality of experience |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105873105A CN105873105A (en) | 2016-08-17 |
| CN105873105B true CN105873105B (en) | 2018-07-03 |
Family
ID=56629062
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610262009.9A Active CN105873105B (en) | 2016-04-22 | 2016-04-22 | A kind of mobile radio communication abnormality detection and localization method based on network Quality of experience |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105873105B (en) |
Families Citing this family (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107809780A (en) * | 2016-09-09 | 2018-03-16 | 中兴通讯股份有限公司 | It is a kind of to the active process method, apparatus of user equipment access failure and base station |
| CN106254153B (en) * | 2016-09-19 | 2019-12-10 | 腾讯科技（深圳）有限公司 | Network anomaly monitoring method and device |
| CN108076197B (en) * | 2016-11-14 | 2020-10-30 | 中移(苏州)软件技术有限公司 | Method and device for detecting terminal network performance degradation |
| CN108260148B (en) * | 2016-12-28 | 2021-02-09 | 华为技术服务有限公司 | Fault detection method and device |
| CN108271194B (en) * | 2016-12-31 | 2021-09-03 | 中国移动通信集团北京有限公司 | Information processing method and device |
| US11132620B2 (en) | 2017-04-20 | 2021-09-28 | Cisco Technology, Inc. | Root cause discovery engine |
| CN107105432A (en) * | 2017-06-12 | 2017-08-29 | 北京奇虎科技有限公司 | The detection method and device of WLAN |
| CN107395450B (en) * | 2017-06-15 | 2018-11-06 | 腾讯科技（深圳）有限公司 | Using the monitoring method and device, storage medium, electronic device for logging in situation |
| CN109981326B (en) * | 2017-12-28 | 2022-01-25 | 中国移动通信集团山东有限公司 | Method and device for positioning household broadband sensing fault |
| CN109634819B (en) * | 2018-10-26 | 2022-02-01 | 创新先进技术有限公司 | Alarm root cause positioning method and device and electronic equipment |
| CN109547248A (en) * | 2018-11-20 | 2019-03-29 | 上海航天测控通信研究所 | Based on artificial intelligence in orbit aerocraft ad hoc network method for diagnosing faults and device |
| CN111327435B (en) * | 2018-12-13 | 2022-07-05 | 中兴通讯股份有限公司 | Root cause positioning method, server and storage medium |
| CN111628901B (en) * | 2019-02-28 | 2022-11-18 | 华为技术有限公司 | Index anomaly detection method and related device |
| CN111698700B (en) * | 2019-03-15 | 2021-08-27 | 大唐移动通信设备有限公司 | Method and device for judging working state of cell |
| CN112543465B (en) * | 2019-09-23 | 2022-04-29 | 中兴通讯股份有限公司 | Abnormity detection method, abnormity detection device, terminal and storage medium |
| CN111064614B (en) * | 2019-12-17 | 2020-12-08 | 腾讯科技（深圳）有限公司 | Fault root cause positioning method, device, equipment and storage medium |
| CN111200530A (en) * | 2019-12-31 | 2020-05-26 | 畅捷通信息技术股份有限公司 | Method and device for performing root cause analysis based on KPI (Key performance indicator) |
| CN111405605B (en) * | 2020-03-24 | 2023-03-24 | 东南大学 | Wireless network interruption detection method based on self-organizing mapping |
| CN111935761B (en) * | 2020-07-15 | 2024-04-09 | 山东建筑大学 | Automatic positioning method and system for quality difference root cause of mobile communication network |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102821002A (en) * | 2011-06-09 | 2012-12-12 | 中国移动通信集团河南有限公司信阳分公司 | Method and system for network flow anomaly detection |
| CN103581971A (en) * | 2012-07-23 | 2014-02-12 | 中兴通讯股份有限公司 | Method, device and access points for detecting abnormity of centralized wireless local area network |
| CN103945442A (en) * | 2014-05-07 | 2014-07-23 | 东南大学 | System anomaly detection method based on linear prediction principle in mobile communication system |
| CN104735710A (en) * | 2015-03-18 | 2015-06-24 | 大连理工大学 | Mobile network performance early warning pre-judging method based on trend extrapolation clustering |
-
2016
- 2016-04-22 CN CN201610262009.9A patent/CN105873105B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102821002A (en) * | 2011-06-09 | 2012-12-12 | 中国移动通信集团河南有限公司信阳分公司 | Method and system for network flow anomaly detection |
| CN103581971A (en) * | 2012-07-23 | 2014-02-12 | 中兴通讯股份有限公司 | Method, device and access points for detecting abnormity of centralized wireless local area network |
| CN103945442A (en) * | 2014-05-07 | 2014-07-23 | 东南大学 | System anomaly detection method based on linear prediction principle in mobile communication system |
| CN104735710A (en) * | 2015-03-18 | 2015-06-24 | 大连理工大学 | Mobile network performance early warning pre-judging method based on trend extrapolation clustering |
Non-Patent Citations (1)
| Title |
|---|
| Performance evaluation of anomaly detection in cellular core networks using self-organizing map;Pichanun Sukkhawatchani等;《Engineering/Electronics, Computer, Telecommunications and Information Technology，ECTI-CON》;20080831;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105873105A (en) | 2016-08-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105873105B (en) | A kind of mobile radio communication abnormality detection and localization method based on network Quality of experience | |
| CN109548041B (en) | Wireless coverage analysis method and system | |
| CN105577440B (en) | A kind of network downtime localization method and analytical equipment | |
| CN105871879B (en) | Network element abnormal behaviour automatic testing method and device | |
| CN112243249B (en) | LTE new access anchor point cell parameter configuration method and device under 5G NSA networking | |
| Sugiarto et al. | Data classification for air quality on wireless sensor network monitoring system using decision tree algorithm | |
| CN112911627B (en) | Wireless network performance detection method, device and storage medium | |
| CN115278741B (en) | Fault diagnosis method and device based on multi-mode data dependency relationship | |
| CN109872003A (en) | Obj State prediction technique, system, computer equipment and storage medium | |
| CN106651031B (en) | Lightning stroke flashover method for early warning and system based on historical information | |
| CN106612511B (en) | Wireless network throughput evaluation method and device based on support vector machine | |
| CN108271157B (en) | Pseudo base station identification method and device | |
| CN103728507A (en) | Grid fault diagnosis method based on data mining | |
| CN113422695B (en) | Optimization method for improving robustness of topological structure of Internet of things | |
| CN113032238A (en) | Real-time root cause analysis method based on application knowledge graph | |
| CN104039008B (en) | A kind of hybrid locating method | |
| CN109874104A (en) | User location localization method, device, equipment and medium | |
| CN116862081A (en) | Operation and maintenance method and system for pollution treatment equipment | |
| CN115689320A (en) | Health management method and device for base station operation and computer readable storage medium | |
| Palacios et al. | Combination of multiple diagnosis systems in self-healing networks | |
| WO2023093431A1 (en) | Model training method and apparatus, and device, storage medium and program product | |
| CN113255593B (en) | Sensor information anomaly detection method facing space-time analysis model | |
| CN107580329B (en) | Network analysis optimization method and device | |
| Rozaki | Network fault diagnosis using data mining classifiers | |
| CN116132081A (en) | Software defined network DDOS attack cooperative defense method based on ensemble learning |
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 |