CN109936857B - Intelligent identification method for wireless perceptibility - Google Patents
Intelligent identification method for wireless perceptibility Download PDFInfo
- Publication number
- CN109936857B CN109936857B CN201910051747.2A CN201910051747A CN109936857B CN 109936857 B CN109936857 B CN 109936857B CN 201910051747 A CN201910051747 A CN 201910051747A CN 109936857 B CN109936857 B CN 109936857B
- Authority
- CN
- China
- Prior art keywords
- wireless
- perceptibility
- data
- sample data
- index data
- 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
Images
Abstract
The invention discloses an intelligent identification method of wireless perceptibility, which comprises the following steps: collecting wireless perceptibility index data of a region to be detected; inputting the index data into a self-encoder neural network, and performing dimensionality reduction operation on the index data; performing clustering analysis on the index data subjected to dimensionality reduction, and calculating the distance between the index data and the center points of two predetermined effective clusters; according to the distance calculation result, the detected area is classified into a class with a short distance, and the wireless perception degree identification result of the detected area is output.
Description
Technical Field
The invention belongs to the field of wireless communication, and particularly relates to an intelligent identification method for wireless perceptibility.
Background art:
with the continuous iterative update of communication technology, especially the rapid popularization of network technology and internet technology in China, various network applications, video software, social software and electronic commerce have already entered thousands of households. And also because of the increase of network services, the demands of users on communication rate and wireless perceived quality are higher and higher.
The wireless perceptibility is the only standard for measuring the wireless signal quality of a region. The standard is reflected on a plurality of indexes, including 24 indexes such as page response success rate, page response delay, page display success rate, page display delay millisecond (ms), page download rate kilobits per second (kbps), initial playing success rate of mobile video and the like. Because each index has different influence on the wireless perceptibility, the areas cannot be screened and discriminated by a conventional rule means (such as Key Performance Index (KPI) threshold filtering or Key Quality Index (KQI) single index filtering). Therefore, the wireless perception of the problem area is found and maintained in time, complaints of owners are reduced, and user experience is improved.
Disclosure of Invention
The invention aims to provide an intelligent identification method of wireless perceptibility, which aims to solve the defects caused in the prior art.
A wireless perceptibility intelligent identification method comprises the following steps:
collecting wireless perceptibility index data of a region to be detected;
inputting the index data into a self-encoder neural network, and performing dimensionality reduction operation on the index data;
performing clustering analysis on the index data subjected to dimensionality reduction, and calculating the distance between the index data and the center points of two predetermined effective clusters;
and according to the distance calculation result, classifying the detected region into a class with a closer distance, and outputting a wireless perception degree identification result of the detected region.
Preferably, the wireless perceptibility index data includes: the system comprises a page response success rate, a page response time delay, a page downloading rate, a mobile video initial playing success rate, a mobile video per minute pause number, a pause duration ratio, an initial cache time delay, a streaming media rate, an instant messaging response success rate, an instant messaging response time delay, a mobile service game response success rate and a mobile service game response time delay.
Preferably, the method for determining the center point of the effective cluster includes the following steps:
collecting multi-period wireless perception index data of a plurality of known problem areas;
merging multi-period wireless perception index data of the same problem area into a multi-dimensional vector with day as granularity, and taking the multi-dimensional vector as sample data;
carrying out dimensionality reduction operation on the sample data by utilizing a self-encoder neural network;
performing cluster analysis on the sample data subjected to dimensionality reduction, outputting two cluster center points and determining the category represented by the two cluster center points;
and if the sample data of more than half of the problem areas are correctly classified, determining that the current cluster center point is the effective cluster center point.
Preferably, the method for determining the center point of the effective cluster further includes:
preprocessing collected multi-period wireless perceptibility index data of a plurality of known problem areas;
the pretreatment comprises the following steps:
converting table data corresponding to multi-period wireless perceptibility index data of a plurality of known problem areas into comma files, and cleaning the comma files;
and carrying out normalization processing on the cleaned data.
Preferably, the data cleansing method includes:
deleting redundant and repeated sample data;
and filling the null value by adopting a mean filling method.
Preferably, the normalization processing method includes:
the following operations are performed for the same type of sample data for each known problem region:
defining a difference value variable used for storing the difference between the maximum value and the minimum value of sample data of the same type;
and traversing each sample data in the same type, replacing the sample data with the difference between the original value and the minimum value, and dividing the difference by the difference variable.
Preferably, the method of cluster analysis comprises the following steps:
randomly selecting two sample data as cluster center points;
respectively calculating the distance between the rest sample data and the center point of the selected two clusters, and classifying the sample data into the same cluster with the closer distance;
and respectively selecting a new cluster center point in the two clusters, recalculating the distance from all sample data to the two new cluster center points, and reclassifying.
The invention has the advantages that: the intelligent identification method for the wireless perceptibility can be used for carrying out screening without using a conventional rule means, and carrying out intelligent prediction by carrying out dimension reduction analysis on data characteristics and clustering distinguishing data categories according to historical data.
Drawings
FIG. 1 is a schematic flow chart of the present invention.
Fig. 2 is a table corresponding to the wireless perceptibility index data of the present invention.
Fig. 3 is a problem area sample data table.
Detailed Description
In order to make the technical means, the creation characteristics, the achievement purposes and the effects of the invention easy to understand, the invention is further described with the specific embodiments.
As shown in fig. 1 and fig. 3, an intelligent recognition method for wireless perceptibility comprises the following steps:
collecting wireless perceptibility index data of a region to be detected;
inputting the index data into a self-encoder neural network, and performing dimensionality reduction operation on the index data;
performing clustering analysis on the index data subjected to dimensionality reduction, and calculating the distance between the index data and the center points of two predetermined effective clusters;
and according to the distance calculation result, classifying the detected region into a class with a closer distance, and outputting a wireless perception degree identification result of the detected region.
In this embodiment, the wireless perceptibility index data includes: page response success rate, page response delay millisecond (ms), page download rate kilobit per second (kbps), mobile video initial play success rate, mobile video per minute pause times, pause duration ratio, initial cache delay millisecond (ms), streaming media rate kilobit per second (kbps), instant messaging response success rate, instant messaging response delay, mobile service game response success rate, mobile service game response delay millisecond (ms).
In this embodiment, the method for determining the center point of the effective cluster includes the following steps:
collecting multi-period wireless perception index data of a plurality of known problem areas;
merging multi-period wireless perception index data of the same problem area into a multi-dimensional vector with day as granularity, and taking the multi-dimensional vector as sample data;
carrying out dimensionality reduction operation on the sample data by utilizing a self-encoder neural network;
performing cluster analysis on the sample data subjected to dimensionality reduction, outputting two cluster center points and determining the category represented by the two cluster center points;
and if the sample data of more than half of the problem areas are correctly classified, determining that the current cluster center point is the effective cluster center point. Wherein the problem area refers to: the method comprises the steps that an area with poor wireless signal quality and frequently complained by a user appears for many times in a week, data sample collection is carried out on a problem area, and the problem area sample data table is provided in the prior art. In the problem area sample table, a threshold value is set for each index, data which is not in the threshold value range is marked out to be used as a problem index, statistics is carried out on the problem index in the data per hour, and a predicted value with the number of the problem indexes being more than 2 is marked as 1, namely the problem data. For example: statistical analysis is performed on 168 pieces of data in a certain area, and if the predicted value of 100 pieces of data exceeds 60%, the area is a problem area. FIG. 3 shows a table of data samples labeled as problem areas after measurement.
In this embodiment, the method for determining the center point of the effective cluster further includes:
preprocessing collected multi-period wireless perceptibility index data of a plurality of known problem areas;
the pretreatment comprises the following steps:
converting table data corresponding to the multi-period wireless perceptibility index data of a plurality of known problem areas into comma files, and cleaning the comma files;
and carrying out normalization processing on the cleaned data.
In this embodiment, the data cleansing method includes:
deleting redundant and repeated sample data;
and filling the null value by adopting a mean filling method.
In this embodiment, the normalization processing method includes:
the following operations are performed on the same type of sample data of each known problem area:
defining a difference value variable used for storing the difference between the maximum value and the minimum value of sample data of the same type;
and traversing each sample data in the same type, replacing the sample data with the difference between the original value and the minimum value, and dividing the difference by the difference variable.
In this embodiment, the method for cluster analysis includes the following steps:
randomly selecting two sample data as cluster center points;
respectively calculating the distance between the rest sample data and the center point of the selected two clusters, and classifying the sample data into the same cluster with the closer distance;
and respectively selecting a new cluster center point in the two clusters, recalculating the distance from all sample data to the two new cluster center points, reclassifying, performing iteration for multiple times, stopping iteration until the center points are not changed any more, and outputting the center points of the two clusters.
Carrying out dimensionality reduction on the sample data by using a self-encoder neural network, carrying out clustering analysis on the dimensionality reduced data and outputting a cluster center point coordinate; this step is the core step of the present invention. Strictly speaking, before the dimension reduction operation is performed on the data, the neural network needs to be trained, and when the loss function is reduced to be almost unchanged, namely converged, the neural network can be considered to be used for the dimension reduction of the data. When the cluster center point is found, if more than half of the labeled problem area data are gathered into a certain cluster, the cluster can be regarded as a problem area cluster, and the next prediction can be carried out.
And according to the cluster center point output by clustering, classifying the data to be detected into the class with the closer distance by calculating the Euclidean distance between the data to be detected and the cluster center point, and completing prediction.
Based on the above, the intelligent identification method for the wireless perceptibility can perform screening without using a conventional rule means, and perform intelligent prediction by performing dimension reduction analysis on data characteristics and clustering to distinguish data types according to historical data. By adopting the method, the category of the unknown cell data to be tested can be effectively identified, a solution is formed in a targeted manner, the problem is found and solved before the complaint of the user, and the user experience is greatly improved.
It will be appreciated by those skilled in the art that the invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The embodiments disclosed above are therefore to be considered in all respects as illustrative and not restrictive. All changes which come within the scope of or equivalence to the invention are intended to be embraced therein.
Claims (6)
1. A wireless perceptibility intelligent identification method is characterized by comprising the following steps:
collecting wireless perceptibility index data of a region to be detected;
inputting the index data into a self-encoder neural network, and performing dimensionality reduction operation on the index data;
performing clustering analysis on the index data subjected to dimensionality reduction, and calculating the distance between the index data and the center points of two predetermined effective clusters;
according to the distance calculation result, the detected region is classified into a class with a closer distance, and the wireless perception degree identification result of the detected region is output;
the method for determining the center point of the effective cluster comprises the following steps:
collecting multi-period wireless perception index data of a plurality of known problem areas;
merging multi-period wireless perception index data of the same problem area into a multi-dimensional vector with day as granularity, and taking the multi-dimensional vector as sample data;
carrying out dimensionality reduction operation on the sample data by utilizing a self-encoder neural network;
performing cluster analysis on the sample data subjected to dimensionality reduction, outputting two cluster center points and determining the category represented by the two cluster center points;
and if the sample data of more than half of the problem areas are correctly classified, determining that the current cluster center point is the effective cluster center point.
2. The intelligent recognition method for wireless perceptibility according to claim 1, characterized in that: the wireless perceptibility index data comprises: the system comprises a page response success rate, a page response time delay, a page downloading rate, a mobile video initial playing success rate, a mobile video per minute pause number, a pause duration ratio, an initial cache time delay, a streaming media rate, an instant messaging response success rate, an instant messaging response time delay, a mobile service game response success rate and a mobile service game response time delay.
3. The intelligent recognition method for wireless perceptibility according to claim 1, characterized in that: the method for determining the center point of the effective cluster further comprises the following steps:
preprocessing collected multi-period wireless perceptibility index data of a plurality of known problem areas;
the pretreatment comprises the following steps:
converting table data corresponding to multi-period wireless perceptibility index data of a plurality of known problem areas into comma files, and cleaning the comma files;
and carrying out normalization processing on the cleaned data.
4. The intelligent recognition method for wireless perceptibility according to claim 3, characterized in that: the data cleaning method comprises the following steps:
deleting redundant and repeated sample data;
and filling the null value by adopting a mean filling method.
5. The intelligent recognition method for wireless perceptibility according to claim 3, characterized in that: the normalization processing method comprises the following steps:
the following operations are performed on the same type of sample data of each known problem area:
defining a difference value variable used for storing the difference between the maximum value and the minimum value of sample data of the same type;
and traversing each sample data in the same type, replacing the sample data with the difference between the original value and the minimum value, and dividing the difference by the difference variable.
6. The intelligent recognition method for wireless perceptibility according to claim 1, characterized in that: the method for cluster analysis comprises the following steps:
randomly selecting two sample data as cluster center points;
respectively calculating the distance between the rest sample data and the center point of the selected two clusters, and classifying the sample data into the same cluster with the closer distance;
and respectively selecting a new cluster center point in the two clusters, recalculating the distance from all sample data to the two new cluster center points, and reclassifying.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910051747.2A CN109936857B (en) | 2019-01-21 | 2019-01-21 | Intelligent identification method for wireless perceptibility |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910051747.2A CN109936857B (en) | 2019-01-21 | 2019-01-21 | Intelligent identification method for wireless perceptibility |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109936857A CN109936857A (en) | 2019-06-25 |
CN109936857B true CN109936857B (en) | 2022-09-16 |
Family
ID=66985039
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910051747.2A Active CN109936857B (en) | 2019-01-21 | 2019-01-21 | Intelligent identification method for wireless perceptibility |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109936857B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110493803B (en) * | 2019-09-17 | 2023-07-11 | 南京邮电大学 | Cell scene division method based on machine learning |
CN110704780A (en) * | 2019-09-30 | 2020-01-17 | 南京邮电大学 | Method for accurately positioning perception difference area |
CN110636528A (en) * | 2019-09-30 | 2019-12-31 | 南京邮电大学 | Regional wireless perception problem feature identification method |
CN114079957A (en) * | 2020-08-10 | 2022-02-22 | 中国移动通信有限公司研究院 | Method and equipment for detecting abnormal state of cell |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103310233A (en) * | 2013-06-28 | 2013-09-18 | 青岛科技大学 | Similarity mining method of similar behaviors between multiple views and behavior recognition method |
KR101414045B1 (en) * | 2013-06-24 | 2014-07-02 | 인하대학교 산학협력단 | Method for detecting target using hyperspectral image using inputted spectral libray data |
CN105050125A (en) * | 2015-06-23 | 2015-11-11 | 武汉虹信通信技术有限责任公司 | Method and device for evaluating mobile data service quality oriented to user experience |
-
2019
- 2019-01-21 CN CN201910051747.2A patent/CN109936857B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101414045B1 (en) * | 2013-06-24 | 2014-07-02 | 인하대학교 산학협력단 | Method for detecting target using hyperspectral image using inputted spectral libray data |
CN103310233A (en) * | 2013-06-28 | 2013-09-18 | 青岛科技大学 | Similarity mining method of similar behaviors between multiple views and behavior recognition method |
CN105050125A (en) * | 2015-06-23 | 2015-11-11 | 武汉虹信通信技术有限责任公司 | Method and device for evaluating mobile data service quality oriented to user experience |
Non-Patent Citations (1)
Title |
---|
基于特性指标降维的日负荷曲线聚类分析;刘思等;《电网技术》;20160331;第40卷(第3期);正文2.1-2.2节 * |
Also Published As
Publication number | Publication date |
---|---|
CN109936857A (en) | 2019-06-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109936857B (en) | Intelligent identification method for wireless perceptibility | |
CN103117903B (en) | Surfing flow method for detecting abnormality and device | |
CN107248082B (en) | Card maintenance identification method and device | |
US8861691B1 (en) | Methods for managing telecommunication service and devices thereof | |
CN106649831B (en) | Data filtering method and device | |
CN107967488B (en) | Server classification method and classification system | |
CN108768695B (en) | KQI problem positioning method and device | |
CN109978575B (en) | Method and device for mining user flow operation scene | |
CN111652661B (en) | Mobile phone client user loss early warning processing method | |
CN104246713A (en) | User sensory quality estimation device, terminal bottleneck determination device, similar operation extraction device, and methods and programs therefor | |
CN112488716A (en) | Abnormal event detection system | |
CN107358346B (en) | Evaluation information processing method and device for communication quality | |
CN107784511A (en) | A kind of customer loss Forecasting Methodology and device | |
CN107222319B (en) | Communication operation analysis method and device | |
CN109194622B (en) | Encrypted flow analysis feature selection method based on feature efficiency | |
CN112101692A (en) | Method and device for identifying poor-quality users of mobile Internet | |
CN109274834B (en) | Express number identification method based on call behavior | |
CN113660147B (en) | IP session sequence periodicity evaluation method based on fuzzy entropy | |
CN111080362A (en) | Advertisement monitoring system and method | |
CN110677269B (en) | Method and device for determining communication user relationship and computer readable storage medium | |
CN110968570A (en) | Distributed big data mining system facing E-commerce platform | |
CN113517990B (en) | Method and device for predicting net recommendation value NPS (network performance indicator) | |
CN111930808B (en) | Method and system for improving blacklist accuracy by using key value matching model | |
CN108880945A (en) | A kind of cloud monitoring system and method | |
CN111581508B (en) | Service monitoring method, device, equipment and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |