CN108563770B - Scene-based KPI and multi-dimensional network data cleaning method - Google Patents
Scene-based KPI and multi-dimensional network data cleaning method Download PDFInfo
- Publication number
- CN108563770B CN108563770B CN201810360670.2A CN201810360670A CN108563770B CN 108563770 B CN108563770 B CN 108563770B CN 201810360670 A CN201810360670 A CN 201810360670A CN 108563770 B CN108563770 B CN 108563770B
- Authority
- CN
- China
- Prior art keywords
- data
- cell
- attribute
- missing
- attributes
- 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
Landscapes
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses a scene-based KPI and multidimensional network data cleaning method, firstly, importing the collected data into a data structure; normalizing the imported data format, dividing the resource data according to scenes, combining the resource data of the same scene to obtain resource subdata, and detecting and processing conflicts generated by data values; correlating data of a plurality of data sources or files, and judging and processing the condition that the redundancy and the mode of the data are not matched; processing data which cannot be directly subjected to data mining; checking the deletion rate of each attribute, and determining a processing mode according to the deletion rate, wherein the processing mode comprises discarding and filling by adopting a K-NN regression method; backing up original data in a data storage module and storing the cleaned data; the invention realizes effective cleaning of data and solves the technical problem that deep mining of data cannot be carried out due to high complexity of the data.
Description
Technical Field
The invention belongs to the field of data cleaning, and particularly relates to a scene-based KPI and multidimensional network data cleaning method.
Background
In the operation management of the mobile communication network, some Key Performance Indicators (KPIs for short) need to be concerned, such as call drop rate, call loss, and the like, besides the daily maintenance, an operator wants to master the factors affecting the KPIs, obtain the association between the KPIs and the network, and facilitate the later-stage network optimization task allocation and guarantee.
Before deep analysis and mining are carried out on the association degree between the KPI and the network, effective cleaning is required to be carried out on data, and the complexity degree of the data is reduced.
Disclosure of Invention
The invention aims to optimize network data, provides a scene-based KPI and multi-dimensional network data cleaning method, realizes effective cleaning of data, and solves the technical problem that deep mining of the data cannot be performed due to high complexity of the data.
The invention adopts the following technical scheme, a scene-based KPI and multidimensional network data cleaning method comprises the following specific steps:
1) dividing the acquired original data into cell basic dimension data and a problem cell list, wherein the cell basic dimension data are directly acquired cell data, namely data which are not subjected to KPI (Key performance indicator) threshold judgment; the problem cell list is cell data with obvious degradation of a certain KPI, namely data judged by a KPI index threshold; importing data into a data structure of a DataFrame in a Pandas package by using a data import module;
2) normalizing the imported data format in a data processing module, dividing resource data in the cell basic dimension data according to scenes, calling a concat function to merge the resource data of the same scene to obtain resource subdata, and detecting and processing conflicts generated by data values in the merging process;
3) associating data of a plurality of data sources or files in a data cleaning module, namely selecting associated primary keywords, combining the resource subdata divided according to scenes and basic dimensional data of other cells according to the primary keywords, and unifying data modes in combination; judging the condition that the redundancy and the mode of the data are not matched in the merging process, deleting the redundancy attribute and unifying the expression mode of the attribute mode; only numerical values are reserved or replaced by numerical values for data which cannot be directly subjected to data mining;
4) checking the missing rate of each attribute in a missing processing module, primarily filling missing values, then determining a processing mode according to the missing rate, discarding the attributes with the missing rate more than 50%, redundant attributes or attributes irrelevant to the analysis subject, and filling the missing values of other attributes by adopting a K-NN regression method;
5) and backing up the original data in the data storage module and storing the cleaned data.
Preferably, the cell basic dimension data in step 1) includes: resource data, performance data, work parameter data, neighbor cell data and measurement data; the problem cell list includes: a long Term evolution lte (long Term evolution) high load cell, a fourth generation mobile communication technology 4G zero flow cell, a long Term evolution lte (long Term evolution) high flow problem serious cell, a high definition voice voltage (voice over lte) high drop call cell, a low measurement report mr (measurement report) coverage cell, and a high definition voice voltage (voice over lte) high drop call cell.
Preferably, the Pandas package of Python is called in step 1) to store data in the data structure of the DataFrame.
Preferably, the normalizing the imported data format in step 2) is to modify the format of Chinese and English and special characters in the data, and uniformly change the format into a lower case and half-corner format and remove the blank space.
Preferably, in the step 2), a conflict generated by the data values in the merging process is detected and processed, the conflict generated by the data values includes that different values or data types occur to the same cell data from different data sources, and when different values occur, data with a low missing rate is retained; when the data types are different, the character type data is replaced by numerical type data.
Preferably, cgi is selected as a primary keyword in the step 3), if cgi is missing, the public land mobile network PLMN and the cell identifier eCI are used for synthesis, otherwise, the cell chinese name is selected as a keyword;
and calling a merge function by the resource subdata and the basic dimensional data of other cells after scene division according to keywords in the resource subdata and combining the resource subdata and the basic dimensional data of other cells in an internal connection inner mode to form the basic dimensional subdata of the cells under the scene division.
Preferably, the determining the redundancy of the data in the step 3) specifically includes: if the attribute can be derived from other attributes, the other attributes used for the derivation are redundant attributes, the redundant attributes are deleted, and the attributes of the derivation are retained as result attributes.
Preferably, the K-nearest neighbor (K-NN) regression method is adopted in the step 4) to fill up, find K similar samples of the samples with the missing attribute, and replace the missing value of the sample attribute with the average value of the similar samples.
Preferably, the k similar samples of the samples lacking the attribute are the k samples having the smallest distance to the samples lacking the attribute, and the distance between two samples is measured using the euclidean distance: two sample points X ═ X1,x2,…xn) And Y ═ Y1,y2,…yn) The euclidean distance between them is:
wherein, the sample refers to a cell, x in the samplei、yi、x1、x2、xn、y1、y2And ynThe values respectively corresponding to the same attributes of the two cells are indicated, m is the distribution number of the two samples without attribute missing values, and n is the attribute number of the two samples; if xiIs equal to yiThen (x)i-yi) Set to 0 if xiIs not equal to yiThen is (x)i-yi) Is set to 1.
Preferably, the method further includes merging the basic dimension data and the problem cell list data, that is, detecting whether the cgi of the cell in the basic dimension data exists in the cgi list of the cell in the problem cell list data, if so, constructing a new attribute, setting the attribute of the cell existing in both the basic dimension data and the problem cell list data to 1, and setting the attribute of the rest cells to 0.
The invention has the following beneficial effects: the invention relates to a scene-based KPI and multi-dimensional network data cleaning method, which realizes effective cleaning of data and solves the technical problem that deep excavation of the data cannot be carried out due to high complexity of the data; the method and the device clean the collected KPI and the multi-dimensional network data, so that the data is standard and uniform, and the later correlation analysis is facilitated.
Drawings
FIG. 1 is a flow diagram of a scenario-based KPI and multidimensional network data cleansing method of the present invention;
FIG. 2 is a flow diagram of a data processing module of the present invention;
FIG. 3 is a flow diagram of a data cleansing module of the present invention;
FIG. 4 is a flow diagram of a miss processing module of the present invention.
Detailed Description
The technical solution of the present invention is further explained with reference to the embodiments according to the drawings.
The invention adopts the following technical scheme, a scene-based KPI and multidimensional network data cleaning method, experimental data come from some data of some areas of an operator, the data are all given in a form Excel mode, the invention is realized by utilizing Python, and FIG. 1 is a flow chart of the scene-based KPI and multidimensional network data cleaning method, and the specific steps are as follows:
1) dividing the acquired original data into cell basic dimension data and a problem cell list, wherein the cell basic dimension data are directly acquired cell data, namely data which are not subjected to KPI (Key performance indicator) threshold judgment; the problem cell list is cell data with obvious degradation of a certain KPI, namely data judged by a KPI index threshold; and importing the data into a data structure of the DataFrame in the Pandas package by using a data import module. The key performance indicator KPI has respective index requirements, namely, the threshold. Such as successful establishment of an RRC (radio resource control protocol) connection, means that the user equipment has established a signalling connection with the network, which is considered satisfactory if the RRC connection success rate > 95%, and is considered to be degraded if less than 95%. In the same way, the wireless disconnection rate is less than 3%, the wireless connection rate is more than 95%, and the like are all threshold judgment bases.
The cell basic dimension data comprises: resource data, performance data, work parameter data, neighbor cell data and measurement data; the problem cell list includes: the problem cell list includes: the system comprises a long term evolution LTE high load cell, a 4G zero flow cell, a long term evolution LTE high flow problem serious cell, a high definition voice voltage high drop call cell, a low measurement report MR coverage rate cell and a high definition voice voltage high drop call cell.
And calling a Pandas package of Python by using a data import module to import the acquired data into a data structure of the DataFrame. Pandas is a data analysis package of python, and DataFrame is the most common data structure in Pandas package, which is a two-dimensional tabular data structure, similar to a table in a database, and the processing of tabular data is very advantageous.
2) FIG. 2 is a flow chart of a data processing module of the present invention, which standardizes the imported data format in the data processing module, specifically, modifies the Chinese and English and special characters existing in the data into lowercase and half-angle formats and removes spaces;
dividing resource data in basic dimension data of a cell according to scenes, dividing the acquired resource data into different coverage scenes such as residential areas, universities, business centers, scenic areas, subways, high-speed rails, ordinary irons, national roads, provinces and the like according to the attributes of the coverage scenes of the resource data, merging the resource data of the same scene by using a concat function to form a plurality of resource subdata divided according to the scenes, wherein the problem of data value conflict can be encountered in the process of synthesizing the resource subdata, and detecting and processing the conflict generated by the data values in the merging process;
the data of the same cell from different data sources can have different values or data types, and when the different values appear, the data with less missing rate is reserved; the data types are different, such as: the support switch attribute is expressed by yes or no in one data source and by 0 and 1 in the other data source, but actually refers to the same concept, which may be of a character type or a numerical type, with the numerical type replacing the character type.
3) FIG. 3 is a flow chart of a data cleansing module according to the present invention, which associates data of multiple data sources or files in the data cleansing module, i.e., selects associated primary keys, merges resource sub-data divided according to scenes with other basic dimension data according to the primary keys, and unifies data modes in the merging;
selecting the associated keyword: data are often associated with each other through keywords, and data can be merged through the keywords, and generally speaking, the keywords have no practical significance for data mining;
aiming at the communication field, in order to obtain complete data of the same cell, various data of the cell can be associated and combined through a keyword cgi or a cell Chinese name, wherein the cgi is used as a main keyword and has a corresponding relation with the cell Chinese name;
cgi is called a universal mobile telecommunications system terrestrial radio access network cell global identifier for globally identifying a cell in a public land mobile network, PLMN, each cell having a unique cgi consisting of PLMN and eCI, e.g. 460-00-725834-2; wherein PLMN is formed by Mobile Network Code (MNC) and country code (MCC) to which the mobile subscriber belongs, eCI is cell identity, formed by eNodeB Id and Localcell Id.
Preferentially selecting cgi as a primary keyword, if the cgi is missing, synthesizing by using a Public Land Mobile Network (PLMN) and a cell identifier eCI, otherwise, selecting a cell Chinese name as the keyword;
and calling merge functions according to keywords in the resource subdata and combining the resource subdata and other cell basic dimension data (including performance data, engineering parameter data, adjacent cell data, measurement data and the like) divided according to the scene to form cell basic dimension subdata under the sub-scene.
In the data merging process, the problem of data redundancy and mode mismatch may be encountered, and the data redundancy judgment specifically includes: if the attribute can be deduced from other attributes, the other attributes used for deduction are redundant attributes, the redundant attributes are deleted, and the attributes of the deduction are reserved as result attributes; when the modes are not matched, unifying the expression modes of the attributes;
for example: uplink packet loss rates Q00_ [0, 1%), Q01_ [1, 2 ]) … uplink packet loss rates Q18_ [18, 19%), Q19_ [19, 20 ], Q20_ [20, 30 ], Q21_ [30, 40 ]) … uplink packet loss rates Q27_ [90, 100], where 28 items total, the average uplink QCI packet loss rate can be calculated, the uplink packet loss rate is probability, 99 represents that the packet loss rate is 99%, 100 represents that the packet loss rate is 100%, and the packet loss rate is basically impossible to reach 20%, and for the completeness of data, the interval is written to [90, 100 ]. Then 28 attributes such as uplink packet loss rate Q00_ [0, 1), uplink packet loss rate Q01_ [1, 2) and the like can be deleted as redundancy attributes, and the average uplink QCI packet loss rate is reserved as result attributes;
the mode does not match: the same attribute of the same cell of different data sources may have different expressions, for example, there may be an attribute of cgi in some data, and there may be data named by a cell global identifier, which actually refers to the same attribute. In view of this, such attributes are named uniformly.
Data that cannot be directly data mined either retains only values or replaces them with values:
simultaneously comprises Chinese, English and number: for example, data appearing in the DRX short cycle length is in the format of SFN, where N is a number, this attribute represents a short cycle duration adopted by Discontinuous Reception (DRX), and in units of subframes, SFN represents that the short cycle duration (including on-duration time) is N subframes, so that only the following number N is reserved;
similarly, the same processing mode is adopted for the attributes of the DRX long period, the DRX duration timer, the DRX inactive timer, the DRX retransmission waiting timer, the maximum transmission times of the lead code, the initial receiving target power, the default paging period, the power climbing step length and the like;
processing according to a method only containing Chinese and English without the above conditions; for example, the device type (major class), the local city, the county, the coverage type, the operating frequency band, the device manufacturer, etc. may be replaced by numbers, for example, 0 represents a 'macro base station' and 1 represents a 'micro base station' in the device type (major class); the working frequency band can be replaced by 'D frequency band', 1 represents 'E frequency band' and 2 represents 'F frequency band'.
4) Fig. 4 is a flowchart of a deletion processing module of the present invention, in which the deletion rate of each attribute is checked in the deletion processing module, the deletion rate of each KPI and the dimension data is calculated, and a deletion value is initially filled through a calculation formula, for example, a certain attribute is missing, and the attribute is obtained by other attributes and can be calculated through a formula;
for example, in the case that the average value of UE transmission power headroom in the measurement data is missing, the power headroom is the difference between the maximum transmission power allowed by the UE and the currently estimated PUSCH transmission power. The unit of the method is dB, the range is [ -23dB, +40dB ], the method is divided into 64 grade values of 0-63 in total, and one actual dB value corresponds to one grade value. The average value of the ue transmission power margin can be obtained by dividing the sum of the 64 levels of the power margin by the number of the ue transmission power margin sampling points to obtain an estimated value. Since the power headroom is a range [ a, b ], ((a + b))/2 is taken as an estimation value, only missing values are padded up to ensure the reliability of data.
Then, a processing mode is determined according to the missing rate, and attributes with the missing rate more than 50%, redundant attributes or attributes irrelevant to the analysis subject are discarded, so that the scale of data processing is reduced; filling missing values of other attributes by adopting a K-NN regression method;
and filling by adopting a K-NN regression method, finding out K similar samples of the samples with the missing attributes, and replacing the missing values of the sample attributes by the average values of the similar samples.
The k similar samples of the missing-attribute sample are the k samples with the smallest distance to the missing-attribute sample, and the euclidean distance is used to measure the distance between the two samples: two sample points X ═ X1,x2,…xn) And Y ═ Y1,y2,…yn) The euclidean distance between them is:
wherein, the sample refers to a cell, x in the samplei、yi、x1、x2、xn、y1、y2And ynThe values respectively corresponding to the same attributes of the two cells are indicated, m is the distribution number of the two samples without attribute missing values, and n is the number of the attributes in the two samples; if xiIs equal to yiThen (x)i-yi) Is set to 0, ifxiIs not equal to yiThen is (x)i-yi) Is set to 1.
5) And backing up the original data in the data storage module and storing the cleaned data.
The method also comprises the steps of merging the basic dimension data and the problem cell list data, namely detecting whether the cgi of the cell in the basic dimension data exists in a cgi list of the cell in the problem cell list data, if so, constructing a new attribute, setting the attribute of the cell existing in the basic dimension data and the problem cell list data to be 1, and setting the attribute of the rest cells to be 0.
Taking an LTE (Long Term Evolution ) high-load cell as an example, detecting whether cgi of a cell in basic dimension data exists in a cell cgi list of the LTE high-load cell; if so, constructing a new attribute: the attribute of the high-load cell is set to be 1, and the high-load cell is represented as the high-load cell; the attribute is set to 0 for the remaining cells, indicating that the cell is not a high load cell.
The above description is only for the specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention.
Claims (8)
1. A scene-based KPI and multi-dimensional network data cleaning method is characterized by comprising the following steps:
1) dividing the acquired original data into cell basic dimension data and a problem cell list, wherein the cell basic dimension data are directly acquired cell data, namely data which are not subjected to KPI (Key performance indicator) threshold judgment; the problem cell list is cell data with obvious degradation of a certain KPI, namely data judged by a KPI index threshold; importing data into a data structure of a DataFrame in a Pandas package by using a data import module; the cell basic dimension data comprises: resource data, performance data, work parameter data, neighbor cell data and measurement data; the problem cell list includes: a long term evolution LTE high load cell, a 4G zero flow cell, a long term evolution LTE high flow problem serious cell, a high definition voice voltage high drop call cell, a low measurement report MR coverage cell and a high definition voice voltage high drop call cell;
2) normalizing the imported data format in a data processing module, dividing resource data in cell basic dimension data according to scenes, calling a concat function for merging the resource data in the same scene to obtain resource subdata, and detecting and processing conflicts generated by data values in the merging process, wherein the conflicts generated by the data values comprise different values or data types of the data in the same cell from different data sources, and when the different values occur, the data with less missing rate is reserved; when the data types are different, replacing character data with numerical data;
3) associating data of a plurality of data sources or files in a data cleaning module, namely selecting associated primary keywords, combining the resource subdata divided according to the scene with basic dimension data of other cells according to the primary keywords, and unifying data modes in combination; judging the condition that the redundancy and the mode of the data are not matched in the merging process, deleting the redundancy attribute and unifying the expression mode of the attribute mode; only numerical values are reserved or replaced by numerical values for data which cannot be directly subjected to data mining;
4) checking the missing rate of each attribute in a missing processing module, primarily filling missing values, then determining a processing mode according to the missing rate, discarding the attributes with the missing rate more than 50%, redundant attributes or attributes irrelevant to the analysis subject, and filling the missing values of other attributes by adopting a K-NN regression method;
5) and backing up the original data in the data storage module and storing the cleaned data.
2. A method as claimed in claim 1, wherein the Pandas package of Python is called in step 1) to store data in a data structure of a DataFrame.
3. A method for cleaning KPI and multidimensional network data based on scenes as claimed in claim 1, wherein the normalizing of the imported data format in step 2) is specifically to modify the format of chinese and english and special characters existing in the data into lowercase and half-corner formats uniformly and to remove spaces.
4. A method as claimed in claim 1, wherein cgi is selected as a primary key in step 3), if cgi is missing, then the method is synthesized using a public land mobile network PLMN and a cell identity eCI, otherwise, a cell chinese name is selected as a key;
and calling merge functions by the resource subdata and the basic dimensional data of other cells after scene division according to keywords in the resource subdata and combining in an inner mode to form the basic dimensional subdata of the cells under the scene division.
5. A method for cleaning KPI and multidimensional network data based on a scenario as claimed in claim 1, wherein the determining of redundancy of data in step 3) is specifically: if the attribute can be derived from other attributes, the other attributes used for the derivation are redundant attributes; and deleting the redundant attribute, and keeping the attribute of the inferred performance as a result attribute.
6. A method as claimed in claim 1, wherein K similar samples, specifically K samples for finding the missing attribute, are filled by K-NN regression in step 4), and the missing attribute value of the sample is replaced by the average value of the similar samples.
7. A scenario-based KPI and multidimensional network data washing method according to claim 5, wherein the k similar samples of the missing attribute samples are the k samples with the smallest distance to the missing attribute samples, and Euclidean distance is used to measure the distance between two samples: two sample points X ═ X1，x2，…xn) And Y ═ Y1，y2，…yn) The euclidean distance between them is:
wherein, the sample refers to a cell, x in the samplei、yi、x1、x2、xn、y1、y2And ynThe values respectively corresponding to the same attributes of the two cells are indicated, m is the distribution number of the two samples without attribute missing values, and n is the attribute number; if xiIs equal to yiThen (x)i-yi) Set to 0 if xiIs not equal to yiThen is (x)i-yi) Is set to 1.
8. A scene-based KPI and multidimensional network data cleansing method according to claim 1, characterized by further comprising merging cell basic dimension data with problem cell list data, i.e. detecting whether cgi of a cell in the cell basic dimension data exists in the cgi list of the cell in the problem cell list data, if so, constructing a new attribute, setting the attribute of the cell existing in both the basic dimension data and the problem cell list data to 1, and setting the attributes of the remaining cells to 0.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810360670.2A CN108563770B (en) | 2018-04-20 | 2018-04-20 | Scene-based KPI and multi-dimensional network data cleaning method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810360670.2A CN108563770B (en) | 2018-04-20 | 2018-04-20 | Scene-based KPI and multi-dimensional network data cleaning method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108563770A CN108563770A (en) | 2018-09-21 |
| CN108563770B true CN108563770B (en) | 2022-05-17 |
Family
ID=63536023
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810360670.2A Active CN108563770B (en) | 2018-04-20 | 2018-04-20 | Scene-based KPI and multi-dimensional network data cleaning method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108563770B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109359850A (en) * | 2018-10-10 | 2019-02-19 | 大连诺道认知医学技术有限公司 | A kind of method and device generating risk assessment scale |
| CN109783481A (en) * | 2018-12-19 | 2019-05-21 | 新华三大数据技术有限公司 | Data processing method and device |
| CN109782722A (en) * | 2019-01-31 | 2019-05-21 | 吕梁市军民融合协同创新研究院 | One kind is towards the quick processing system of industrial scene large-scale production line multidimensional data |
| CN110097920B (en) * | 2019-04-10 | 2022-09-20 | 大连理工大学 | Metabonomics data missing value filling method based on neighbor stability |
| CN111125194B (en) * | 2019-12-25 | 2023-04-11 | 中国建筑科学研究院有限公司 | Data construction method and device applied to urban-level clean heating |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7610299B2 (en) * | 2006-11-30 | 2009-10-27 | International Business Machines Corporation | Method of processing data |
| CN106709035A (en) * | 2016-12-29 | 2017-05-24 | 贵州电网有限责任公司电力科学研究院 | Preprocessing system for electric power multi-dimensional panoramic data |
| CN107766541A (en) * | 2017-10-30 | 2018-03-06 | 北京国电通网络技术有限公司 | With electricity consumption overall situation full dose data transfer and storage method, device, electronic equipment |
-
2018
- 2018-04-20 CN CN201810360670.2A patent/CN108563770B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7610299B2 (en) * | 2006-11-30 | 2009-10-27 | International Business Machines Corporation | Method of processing data |
| CN106709035A (en) * | 2016-12-29 | 2017-05-24 | 贵州电网有限责任公司电力科学研究院 | Preprocessing system for electric power multi-dimensional panoramic data |
| CN107766541A (en) * | 2017-10-30 | 2018-03-06 | 北京国电通网络技术有限公司 | With electricity consumption overall situation full dose data transfer and storage method, device, electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108563770A (en) | 2018-09-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108563770B (en) | Scene-based KPI and multi-dimensional network data cleaning method | |
| CN108923900B (en) | Transmission method, device and the computer-readable medium of reference signal | |
| US20170374491A1 (en) | Method for instructing update of system information, base station and user equipment | |
| CN103313301A (en) | Method and device for realizing report measurement and report measurement result | |
| EP2136483A1 (en) | A method, system and device of mapping system information block | |
| US9572032B2 (en) | Method and system for terminal to access network and terminal | |
| CN109792712A (en) | Paging method, the network equipment and terminal device | |
| US10306643B2 (en) | System and method for improved band-channel scanning and network switching | |
| CN103634881B (en) | Frequency searching method and device, terminal | |
| CN109885636B (en) | User portrait drawing method and server | |
| CN111866799B (en) | Resource request method, resource allocation method, device and medium | |
| CN103326925B (en) | A kind of information push method and device | |
| CN110362598A (en) | Method, apparatus, storage medium and the electronic equipment of data query | |
| CN104244223A (en) | Method and device for querying contact people | |
| CN104348941B (en) | A kind of method and device for searching expectation user using cell phone address book | |
| CN106060847A (en) | Signal blind area determining method and signal blind area determining system | |
| CN102378263B (en) | Method and system for equalizing distribution of physical downlink control channels | |
| RU2590907C1 (en) | Method and device for obtaining information of neighbouring cells | |
| Bejarano-Luque et al. | A context-aware data-driven algorithm for small cell site selection in cellular networks | |
| CN113329094B (en) | Information push time determining method, device, equipment, medium and product | |
| CN103581953A (en) | Association method for wireless drive test data and signaling monitoring data | |
| CN101208876A (en) | Noise estimation in wireless communication system | |
| CN103096358B (en) | A kind ofly realize method that multicarrier measurement result reports and device | |
| CN109756956A (en) | Communication means and communication device, access point apparatus and site apparatus | |
| CN114222284A (en) | Terminal equipment identification method and device and readable 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 |