CN107645393A - Determine the method, apparatus and system of the black-box system input and output degree of association - Google Patents
Determine the method, apparatus and system of the black-box system input and output degree of association Download PDFInfo
- Publication number
- CN107645393A CN107645393A CN201610579303.2A CN201610579303A CN107645393A CN 107645393 A CN107645393 A CN 107645393A CN 201610579303 A CN201610579303 A CN 201610579303A CN 107645393 A CN107645393 A CN 107645393A
- Authority
- CN
- China
- Prior art keywords
- data
- kqi
- kpi
- association
- degree
- 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.)
- Withdrawn
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a kind of method, apparatus and system for determining the black-box system input and output degree of association, this method includes：Quality of service index KQI data in black-box system are matched with Key Performance Indicator KPI data and form data vector space；KPI data is clustered according to the type of service of KQI data；Data vector space is decomposed, to isolate linked character of the KPI data to KQI data, and calculates the normalization degree of association of the KPI data to KQI data；The KPI data associated with each KQI data is determined according to the normalization degree of association respectively, and calculates associated weights of the associated KPI data to KQI data；Associated KPI data and associated weights are defined as to the input and output degree of association of black-box system.By the present invention, solve the problems, such as in correlation technique it is determined that accuracy is too low during the black-box system input and output degree of association and looks for not fully closed copula.
Description
Technical field
The present invention relates to the communications field, in particular to a kind of method for determining the black-box system input and output degree of association,
Device and system.
Background technology
When a certain quality of service index KQI in service network (including communication system) (Key Quality Indicators,
Also it is Key Quality Indicator) deteriorate, it is desirable to find out the chief performance index KPI (Key for causing KQI to deteriorate
Performance Indicators), to carry out parameter adjustment or the network optimization.And between KQI and KPI it is a black box system
System, the schematic diagram that it is black-box system between KQI and KPI according to correlation technique of the present invention that Fig. 1, which is, as shown in Figure 1.Tradition solves
Method relies primarily on artificial experience, in face of numerous systems or local KPI, reports parameter, warning information, assistance messages etc. (generally
Up to hundreds of), it is extremely difficult timely and accurately to investigate problem.Therefore, black box incidence relation between KQI and KPI is cracked always
It is the focus that industry is concerned about.
The theoretical foundation of correlation technique is classical statistics, it is believed that the degree of association between KQI and KPI has fixed function.Examination
Figure carries out Function Fitting (i.e. KQI=f (KPI)) using modes such as parameter Estimation, regression analyses to the degree of association, but this method is not
Correct conclusion can be obtained, can not also solve the present situation of " the good KQI of KPI are poor ".Fig. 2 be according in correlation technique of the present invention with dissipate
Point diagram signal " the good KQI of KPI are poor " phenomenon schematic diagram, and existing method be fitted/return can not correctly reflect KQI with
Incidence relation between KPI.As shown in Figure 2.
It is limited to theoretical foundation, existing method rule of thumb can only artificially select the KPI items, artificial of a small amount of " should be related "
Weight between specified associations model and given KPI, then it is associated degree calculating.Cause problems with：
1) it is, not comprehensive.Existing method has no ability to calculate the degree of association between all KPI to be assessed and KQI simultaneously, also can not
Tackle the variable situation of KPI lists.Typical case is presented as the KPI items for omitting strong correlation.2) it is, inaccurate.Existing method can not be accurate
All KPI to be assessed of the ground quantum chemical method and current KQI degree of association.Typical case is presented as the degree of association conclusion of mistake.3), based on 1)
With 2), existing method can not find out KQI Complete Orthogonal KPI set, and then can not accurately guide parameters optimization or business increment
Excavate.4), the method in correlation technique carries out being divided into two kinds during KPI clusters：One kind is artificial division, and another kind is only based on Europe
The simple clustering algorithm (such as K-Means) of formula distance, is required to that cluster number is manually specified, and not in combination with service feature
With engineering significance.5), based on 1) to 4), the method for correlation technique is using multiple KPI thresholding cascading judgement KQI without standard measure
It is no to transfinite or provide the confidence level that KQI transfinites.And then can not be in the case of vacant KQI data, according to multidimensional KPI thresholdings to net
Network parameter optimizes or pre-optimized.
For above mentioned problem present in correlation technique, at present it is not yet found that the solution of effect.
The content of the invention
The embodiments of the invention provide it is a kind of determine the black-box system input and output degree of association method, apparatus and system,
At least to solve in correlation technique it is determined that accuracy is too low during the black-box system input and output degree of association and looks for not fully closed copula
Problem.
According to one embodiment of present invention, there is provided a kind of method for determining the black-box system input and output degree of association, bag
Include：Quality of service index KQI data in black-box system are matched with Key Performance Indicator KPI data and form data vector sky
Between；The KPI data is clustered according to the type of service of the KQI data, wherein, cluster result is orthogonal for selecting
Strong correlation KPI items, and auxiliary judgement achievement data health degree；The data vector space is decomposed, it is described to isolate
KPI data calculates the normalization degree of association of the KPI data to the KQI data to the linked characters of the KQI data；
The KPI data associated with each KQI data is determined according to the normalization degree of association respectively, and calculates the phase
Associated weights of the KPI data of association to the KQI data；The associated KPI data and the associated weights are determined
For the input and output degree of association of the black-box system.
Alternatively, carrying out cluster to the KPI data according to the type of service of the KQI data includes：By the KQI numbers
It is divided into KQI data Layers and KPI data layer according to the KPI data；Add between the KQI data Layers and the KPI data layer
Enter to the KQI data corresponding to the related abstract layer parameter of type of service, wherein, level of abstraction KPI data is carried out it is regular or
Mapping transformation, to be adapted to corresponding mining algorithm；KPI data is clustered using the abstract layer parameter.
Alternatively, carrying out decomposition to the data vector space includes：To the data vector space in the following manner
At least one decomposed, isolate linked character of the KPI data to the KQI data：In Spatial Dimension dimensionality reduction, in sky
Between dimension directly split, Spatial Dimension rise tie up.
Alternatively, quality of service index KQI data are matched with Key Performance Indicator KPI data and forms data vector space
The degree of association of the KQI data is included with isolating the KPI data：KQI data are matched with KPI data form data to
Quantity space is to isolate the degree of association of the KPI data to the KQI data.
Alternatively, quality of service index KQI data are matched with Key Performance Indicator KPI data and forms data vector space
The degree of association of the KQI data is included with isolating the KPI data：According to the traffic performance information pair of the KQI data
The KPI data carries out fitting of distribution and figure displaying, judges the significant degree of KPI data, lack as needed, extremely or
Mapping is handled；KQI data are matched with the effectively KPI data and form data vector space to isolate the KPI data pair
The degree of association of the KQI data.
Alternatively, composition data vector is matched with KPI data from KQI data and is spatially separating out the KPI data to described
The method of the degree of association of KQI data includes：Dimension-reduction treatment, directly decompose, extract validity feature again after liter dimension.
Alternatively, the traffic performance information obtains according to matching presetting database and/or according to business demand.
Alternatively, dimensionality reduction operation is carried out to the data vector space includes：Decision tree beta pruning, merging is returned, is clustered, specially
Family's auxiliary judgement etc..
Alternatively, the data vector space directly decompose includes：Based on Bayesian statistics algorithm to the number
Decomposed according to vector space, equivalent numerical value computational methods based on singular value decomposition thinking etc..
Alternatively, extracting validity feature again after being augmented to the data vector space includes：Based on SVMs
SVM algorithm decomposes again after being augmented to the data vector space；Processing is augmented based on neural network algorithm, i.e.,
Hidden unit number is higher than input dimension.
Alternatively, the KPI number associated with each KQI data is being determined according to the normalization degree of association respectively
According to afterwards, methods described also includes：The peacekeeping of quantization one quantization for calculating the KPI data associated with each KQI data is more
Tie up thresholding；The False Rate and/or misdetection rate of the KQI data over run are obtained according to the quantization multidimensional thresholding；According to the leakage
Sentence rate and/or misdetection rate analyzes the complete the base whether associated KPI data includes the KQI data over run space.
Alternatively, whether the associated KPI data is being analyzed comprising described according to the misdetection rate and/or misdetection rate
After the complete base in KQI data over run space, methods described also includes：
The associated KPI data is analyzed according to the misdetection rate and/or misdetection rate and does not include the KQI data over run
The probability in space.
Alternatively, the KPI number associated with each KQI data is being determined according to the normalization degree of association respectively
According to afterwards, methods described also includes：Judge whether the KQI data lack；In the case where judging the KQI shortage of data,
The probability that the KQI data judge by accident is reversely inferred according to the quantization multidimensional thresholding of history KPI data, and carry out system pre-optimized with
Parameter and adjustment.
Alternatively, the KPI data includes at least one of：Wireless heterogeneous networks (Radio Resource
Control, referred to as RRC) connection be created as power, RAB (the Evolved Radio Access of evolution
Bearer, referred to as E-RAB) it is created as switching between power, wireless percent of call completed, E-RAB drop rates, base station ENB (evolution NodeB)
Success rate, the up packet loss in community user face, the descending packet loss in community user face, the descending average delay in community user face, cell
User plane is descending to abandon bag rate, cell downlink bag number, the up Block Error Rate of MAC layer, medium education (Media Access
Control, referred to as MAC) the descending Block Error Rate of layer, up initial mixing HARQ HARQ retransmit ratio, down initial
(Hybrid Automatic Repeat Request, referred to as HARQ) retransmit ratio, descending double-current flow accounting, it is up just
It is phase-shift keying (PSK) (Quadrature Phase Shift Keying, the referred to as QPSK) ratio of friendship, up 16QAM ratios, descending
QPSK ratios, descending 16QAM ratios, descending 64 quadrature amplitude modulation (Quadrature Amplitude Modulation, letter
Referred to as QAM) ratio, uplink service byte number of eating dishes without rice or wine, downlink business byte number of eating dishes without rice or wine, ascending physical signal resource block Physical
Resource Block, referred to as PRB) average utilization, descending PRB average utilizations, it is up per PRB average throughputs, under
Row is per PRB average throughputs, -110dBm coverage rates, average signal and interference plus noise ratio (Signal to Interence
Plus Noise Ratio, referred to as SINR), the average channel quality of subband 0 instruction CQI (Channel Quality
Indicator), the average activation equipment UE numbers of user plane.
Alternatively, the KQI data include HTTP (Hypertext Transfer Protocol, abbreviation
For HTTP) response delay.
Alternatively, the cluster includes at least one of：Capacity performance index cluster, access index cluster, efficiency index gather
Class, complete holding index cluster.
Alternatively, the complete holding index cluster also includes at least one of：Packet Service cluster, uplink complete guarantor
Hold cluster, descending complete holding cluster.
According to another embodiment of the invention, there is provided a kind of device for determining the black-box system input and output degree of association,
Including：Separation module, for the quality of service index KQI data in black-box system to be matched with Key Performance Indicator KPI data
Data vector space is formed to isolate the degree of association of the KPI data to the KQI data；Cluster module, for according to institute
The type of service for stating KQI data clusters to the KPI data；First computing module, for the data vector space
Decomposed, and calculate the normalization degree of association of the KPI data to the KQI data；Second computing module, for according to institute
State the normalization degree of association and determine the KPI data associated with each KQI data respectively, and calculate described associated
Associated weights of the KPI data to the KQI data.Determining module, for the associated KPI data to be associated into power with described
It is defined as the input and output degree of association of the black-box system again.
According to still another embodiment of the invention, there is provided a kind of correlation analysis system, including：Memory cell, for depositing
Store up the KQI data and KPI data in service network；Data pre-processing unit, for the KQI data and the KPI data
Pre-processed, wherein, the pretreatment includes：Data Matching, data cleansing, statistical nature extraction and statistics are in
It is existing.Cluster cell, for carrying out intelligent clustering to the KPI data, and export cluster table；Vector space resolving cell, with institute
Data pre-processing unit connection is stated, for being decomposed to the vector space of pretreated KQI data and KPI data composition,
Extraction can association component of the KQI data to the KPI data that quantifies of normalizing.Quantify association computing unit, with it is described to
Quantity space resolving cell connects, and for carrying out normalizing quantum chemical method to the association component, obtains the KQI data to described
The quantization degree of association of KPI data, its total weight order is calculated, and export the quantization incidence matrix for including the weight；Multidimensional door
Computing unit is limited, for calculating the multidimensional quantization threshold of continuous item KPI data according to the incidence matrix, counter pushing away what KQI transfinited
False Rate and/or misdetection rate, output multidimensional quantization threshold matrix transfinites with KQI assesses data；Optimize unit, for according to
Multidimensional quantization threshold matrix transfinites with the KQI assesses the service network progress network optimization of data institute.
Alternatively, the system also includes：Business datum interface, comprising interface is presented, for receiving external command to institute
The output data for stating system carries out auxiliary judgement.
Alternatively, the system also includes：Data mining analysis algorithm pond, the data mining for storing the system are calculated
Method.Database, for storing the data analysis of the system with excavating conclusion, and pilot process information.
According to still another embodiment of the invention, a kind of storage medium is additionally provided.The storage medium is arranged to storage and used
In the program code for performing following steps：
Quality of service index KQI data in black-box system are matched with Key Performance Indicator KPI data form data to
Quantity space；
The KPI data is clustered according to the type of service of the KQI data, wherein, cluster result is used to select
Orthogonal strong correlation KPI items, and auxiliary judgement achievement data health degree；
The data vector space is decomposed, it is special to isolate association of the KPI data to the KQI data
Sign, and calculate the normalization degree of association of the KPI data to the KQI data；
The KPI data associated with each KQI data is determined according to the normalization degree of association respectively, and calculated
Go out associated weights of the associated KPI data to the KQI data；
The input and output that the associated KPI data is defined as to the black-box system with the associated weights associate
Degree.
By the present invention, the quality of service index KQI data in black-box system are matched with Key Performance Indicator KPI data
Form data vector space；The KPI data is clustered according to the type of service of the KQI data, wherein, cluster result
For selecting orthogonal strong correlation KPI items, and auxiliary judgement achievement data health degree；The data vector space is divided
Solution, to isolate linked character of the KPI data to the KQI data, and calculates the KPI data to the KQI data
The normalization degree of association；The KPI number associated with each KQI data is determined according to the normalization degree of association respectively
According to, and calculate associated weights of the associated KPI data to the KQI data；By the associated KPI data and
The associated weights are defined as the input and output degree of association of the black-box system, due to can quantum chemical method KQI data to KPI data
The degree of association and being normalized compares, therefore can solve in correlation technique it is determined that essence during the black-box system input and output degree of association
Exactness is too low and the problem of looking for not fully closed copula, has reached lifting quality of service and has mitigated the effect of artificial burden.
Brief description of the drawings
Accompanying drawing described herein is used for providing a further understanding of the present invention, forms the part of the application, this hair
Bright schematic description and description is used to explain the present invention, does not form inappropriate limitation of the present invention.In the accompanying drawings：
The schematic diagram that it is black-box system between KQI and KPI according to correlation technique of the present invention that Fig. 1, which is,；
Fig. 2 is according to the phenomenon schematic diagram for illustrating " the good KQI of KPI are poor " in correlation technique of the present invention with scatter diagram；
Fig. 3 is the flow chart of the method for the determination black-box system input and output degree of association according to embodiments of the present invention；
Fig. 4 is the structured flowchart of the device of the determination black-box system input and output degree of association according to embodiments of the present invention；
Fig. 5 is the structured flowchart of correlation analysis system according to embodiments of the present invention；
Fig. 6 is that KQI-KPI multidimensional provided in an embodiment of the present invention quantifies association analysis protocol procedures schematic diagram；
Fig. 7 is that KQI-KPI multidimensional provided in an embodiment of the present invention quantifies correlation analysis system block diagram；
Fig. 8 is initial hierarchical cluster schematic diagram in wisdom cluster process provided in an embodiment of the present invention；
Fig. 9 is the parameter frequency matched curve figure that the embodiment of the present invention is examined by normal distribution and data source health degree；
Figure 10 is the embodiment of the present invention not by normal distribution-test but the parameter frequency examined by data source health degree
Matched curve figure；
Figure 11 is the parameter frequency that the embodiment of the present invention is not examined not by normal distribution-test and by data source health degree
Spend matched curve figure；
Figure 12 is that the embodiment of the present invention is associated a KPI weight meter using main quantization determination point, auxiliary quantization determination point
The schematic diagram of calculation.
Embodiment
Describe the present invention in detail below with reference to accompanying drawing and in conjunction with the embodiments.It should be noted that do not conflicting
In the case of, the feature in embodiment and embodiment in the application can be mutually combined.
It should be noted that term " first " in description and claims of this specification and above-mentioned accompanying drawing, "
Two " etc. be for distinguishing similar object, without for describing specific order or precedence.
Embodiment 1
A kind of method for determining the black-box system input and output degree of association is provided in the present embodiment, and Fig. 3 is according to this hair
The flow chart of the method for the determination black-box system input and output degree of association of bright embodiment, as shown in figure 3, the flow includes following step
Suddenly：
Step S302, the quality of service index KQI data in black-box system are matched into structure with Key Performance Indicator KPI data
Into data vector space to isolate the degree of association of the KPI data to KQI data；
Step S304, KPI data is clustered according to the type of service of KQI data, wherein, cluster result is used to select
Orthogonal strong correlation KPI items, and auxiliary judgement achievement data health degree；
Step S306, data vector space is decomposed, to isolate linked character of the KPI data to KQI data, and
Calculate the normalization degree of association of the KPI data to KQI data；
Step S308, the KPI data associated with each KQI data is determined according to the normalization degree of association respectively, and counted
Calculate associated weights of the associated KPI data to KQI data；
Step S310, associated KPI data and associated weights are defined as to the input and output degree of association of black-box system.
Optionally, the scene of the present embodiment can be applied in the network optimization, user's portrait, comment, recommend etc., but be not limited to
This.
By above-mentioned steps, by the quality of service index KQI data in black-box system and Key Performance Indicator KPI data
With composition data vector space；The KPI data is clustered according to the type of service of the KQI data, wherein, cluster knot
Fruit is used to select orthogonal strong correlation KPI items, and auxiliary judgement achievement data health degree；The data vector space is divided
Solution, to isolate linked character of the KPI data to the KQI data, and calculates the KPI data to the KQI data
The normalization degree of association；The KPI number associated with each KQI data is determined according to the normalization degree of association respectively
According to, and calculate associated weights of the associated KPI data to the KQI data；By the associated KPI data and
The associated weights are defined as the input and output degree of association of the black-box system, due to can quantum chemical method KQI data to KPI data
The degree of association and being normalized compares, therefore can solve in correlation technique it is determined that essence during the black-box system input and output degree of association
Exactness is too low and the problem of looking for not fully closed copula, has reached lifting quality of service and has mitigated the effect of artificial burden.
Optionally, the degree of association between KQI and KPI is regarded as stochastic variable rather than fixed function, preferably Bayesian statistics is managed
By method flow is：
By the KQI of network/communication system both ends collection, temporally granularity is alignd with spatial granularity matching with KPI data.
With reference to agreement, specification and business actual demand etc., the KQI matched after aliging is further cleaned with KPI data.Number
According to having in cleaning：Rejecting abnormalities value is with filling up missing values.
The feature of KQI and KPI data after statistics cleaning, and presented with a variety of graph modes.Optionally, by business datum
Expert is analyzed with reference to statistical indicator, judges that data source health degree (is misrepresented deliberately/failed to report with the presence or absence of what numerical method can be sentenced
Deng), data distribution type etc., with more preferable adaptation data mining algorithm.
Intelligent hierarchical cluster is carried out to the KPI data after cleaning and obtains cluster table.Optionally, tied by business datum expert
Close business demand and be divided into several classes with engineering significance auxiliary judgement, and finely tune the fuzzy KPI classification of boundary.
The vector space formed to KQI-KPI data is decomposed, and that isolates each KPI-KQI quantifies the degree of association,
And it is normalized.Optionally, vector space is augmented or Coordinate Conversion can quantifies to close to obtain apparent KQI-KPI
Connection degree.
Judgement is ranked up to the KPI-KQI normalization degree of association, according to degree of association size collating sort.
Check and clustered belonging to continuous item KPI, rule out final continuous item KPI.Optionally, by business datum expert according to system
Meter information, business demand, cluster table and engineering significance specify final continuous item KPI.
The degree of association is normalized according to final continuous item KPI KPI-KQI, calculates final continuous item KPI normalized weight.
The degree of association, and KQI preset warning thresholding are normalized according to final continuous item KPI KPI-KQI, calculated each
Final continuous item KPI one-dimensional thresholding.
According to KQI preset warning thresholding, and erroneous judgement/misdetection rate accuracy requirement, calculate multiple final continuous item KPI's
Joint thresholding.Optionally, thresholding dimension is combined between 2 and final continuous item KPI quantity.
According to all final continuous item KPI one-dimensional thresholding with combining thresholding, final continuous item KPI multidimensional thresholding is calculated
Calculated using final continuous item KPI multidimensional thresholding and it is expected False Rate (KPI->KQI the KQI under) transfinites misdetection rate
(KQI->KPI).Wherein, False Rate defines：In the KQI data that KPI multidimensional thresholdings filter out, ratio that KQI does not transfinite；Leakage
Sentence rate definition：In the data that KQI transfinites, not KPI multidimensional thresholding screening in the range of ratio.
The incidence matrix and KPI multidimensional threshold matrix between KQI and its final continuous item KPI are presented at interface, and will association
Extract feature deposit backstage expert's wisdom database.Optionally, the whole middle output result calculated in judgement flow can be by industry
Being presented at interface for data craft selectivity of being engaged in, carries out auxiliary judgement and artificial fine setting.
After new KQI arrives with KPI data, carrying out KQI with existing multidimensional thresholding transfinites judgement, if False Rate rising scale
More than skew thresholding, then gather all data and recalculate；Otherwise it is micro- to be continuing with existing association analysis conclusion combining adaptive
Adjust, instruct KQI optimizations and parameter adjustment.
Pass through the present embodiment so that operator, networking business, network operation business can use the mode quantified quickly comprehensive
A certain KQI strong correlation KPI is found out exactly, and calculates corresponding KPI individual events and multidimensional thresholding, and then counter whether pushes away KQI
Transfinite, accurately easily instructed for offers such as network evaluation, performance optimization, parameter adjustments, lift quality of service, and greatly mitigate
Artificial burden.
Through the above description of the embodiments, those skilled in the art can be understood that according to above-mentioned implementation
The method of example can add the mode of required general hardware platform to realize by software, naturally it is also possible to by hardware, but a lot
In the case of the former be more preferably embodiment.Based on such understanding, technical scheme is substantially in other words to existing
The part that technology contributes can be embodied in the form of software product, and the computer software product is stored in a storage
In medium (such as ROM/RAM, magnetic disc, CD), including some instructions to cause a station terminal equipment (can be mobile phone, calculate
Machine, server, or network equipment etc.) method that performs each embodiment of the present invention.
Embodiment 2
A kind of device for determining the black-box system input and output degree of association, system, the device are additionally provided in the present embodiment
It is used to realize above-described embodiment and preferred embodiment with system, had carried out repeating no more for explanation.Used as following
, term " module " can realize the combination of the software and/or hardware of predetermined function.Although the device described by following examples
Preferably realized with software, but hardware, or software and hardware combination realization and may and be contemplated.
Fig. 4 is the structured flowchart of the device of the determination black-box system input and output degree of association according to embodiments of the present invention, such as
Shown in Fig. 4, the device includes：
Separation module 40, for by the quality of service index KQI data in black-box system and Key Performance Indicator KPI data
Matching forms data vector space to isolate the degree of association of the KPI data to KQI data；
Cluster module 42, for being clustered according to the type of service of KQI data to KPI data；
First computing module 44, for being decomposed to data vector space, and calculate KPI data and KQI data are returned
One changes the degree of association；
Second computing module 46, for determining the KPI associated with each KQI data respectively according to the normalization degree of association
Data, and calculate associated weights of the associated KPI data to KQI data.
Determining module 48, the input and output for associated KPI data and associated weights to be defined as to black-box system are closed
Connection degree.
Fig. 5 is the structured flowchart of correlation analysis system according to embodiments of the present invention, as shown in figure 5, the system includes：
Memory cell 50, for the KQI data and KPI data in Networks for Storage Services；
Data pre-processing unit 52, for being pre-processed to KQI data and KPI data, wherein, pretreatment includes：Number
Presented according to matching, data cleansing, statistical nature extraction and statistics.
Cluster cell 54, for carrying out intelligent clustering to KPI data, and export cluster table；
Vector space resolving cell 56, is connected with data pre-processing unit, for pretreated KQI data and KPI
The vector space that data are formed is decomposed, and extraction can association component of the KQI data to KPI data that quantifies of normalizing.
Quantify association computing unit 58, be connected with vector space resolving cell, for carrying out normalizing quantization to association component
Calculate, obtain the quantization degree of association of the KQI data to KPI data, calculate its total weight order, and export the quantization comprising weight and close
Join matrix；
Multidimensional threshold computation unit 60, for the multidimensional quantization threshold according to incidence matrix calculating continuous item KPI data, instead
False Rate and/or the misdetection rate that KQI transfinites are pushed away, output multidimensional quantization threshold matrix transfinites with KQI assesses data；
Optimize unit 32, assess data institute service network for being transfinited according to multidimensional quantization threshold matrix and KQI and carry out net
Network optimizes.
Optionally, system also includes：Business datum interface, comprising interface is presented, for receiving external command to system
Output data carries out auxiliary judgement；According to mining analysis algorithm pond, the data mining algorithm for storage system；Database, it is used for
The data analysis of storage system is with excavating conclusion, and pilot process information.
According to the system of the present embodiment, optionally, basic database, for storing all indexs and operational administrative data.
Business datum expert's interface, interface is presented comprising data and conclusion, expert checks statistical analysis for business datum
Information, and input subsidiary conditions or judgement.
KQI/KPI memory cell, for storing all KQI/KPI items, include type of service, customer demand, expert's auxiliary
Etc. information.
Data pre-processing unit, according to information such as distribute leaflets, type of service and customer demands, from memory cell extraction KQI with
KPI items complete or collected works or subset, Data Matching, data cleansing, statistical nature extraction and statistics presentation are carried out successively, for associating
Analysis calculates.
Intelligent clustering unit, according to the type of business, distribution and engineering significance, adjudicated with reference to business datum expert, to KPI
Item carries out intelligent clustering, exports intelligent clustering table.Cluster process is not limited to single clustering algorithm, does not also preassign fixed cluster
Quantity, completely based on data and business.
Vector space resolving cell, for being decomposed to pretreated KQI and KPI data composition vector space,
Extraction can normalizing quantify KQI-KPI association component.Include the modes such as decomposition after directly decomposing and be augmented.
Quantify association computing unit, the KQI-KPI association components for being extracted to vector space resolving cell are returned
One quantum chemical method, obtain KQI-KPI and quantify the degree of association.And according to the degree of association and intelligent clustering table is quantified, entirety is participated in associating
The KPI items that analysis calculates are divided into four kinds：Final continuous item KPI, related similar terms KPI, reminder item KPI nothing to do with items KPI.So
Afterwards according to all final continuous item KPI quantization degree of association, its total weight order is calculated.Interface output quantization pass finally is being presented
Join matrix (including weight).
Multidimensional threshold computation unit, for calculating final continuous item KPI multidimensional quantization threshold, then calculate KPI multidimensional
Quantization threshold is counter to push away False Rate and the misdetection rate that KQI transfinites.Finally surpass in presentation interface output multidimensional quantization threshold matrix with KQI
Limit False Rate and misdetection rate.
Data mining analysis algorithm pond, dug for storing the total data that may be used in whole association analysis calculation process
Dig algorithm.Each computing module is according to data and service feature are automatic or the suitable algorithm of expert's assisted Selection.
Data craft wisdom storehouse, for storing the association analysis conclusion based on business and pilot process information, available for industry
Business association analysis, network/system parameter optimization and other value-added services.
It should be noted that above-mentioned modules can be realized by software or hardware, for the latter, Ke Yitong
Cross in the following manner realization, but not limited to this：Above-mentioned module is respectively positioned in same processor；Or above-mentioned modules are with any
The form of combination is located in different processors respectively.
Embodiment 3
Be described in further detail with reference to implementation of the concrete scene to technical scheme, but illustrated embodiment not as
For limitation of the invention the embodiments of the invention provide a kind of KQI-KPI multidimensional quantization association analysis scheme, Fig. 6 is of the invention real
The KQI-KPI multidimensional for applying example offer quantifies association analysis protocol procedures schematic diagram, and as shown in Figure 6, Fig. 7 is the embodiment of the present invention
The KQI-KPI multidimensional of offer quantifies correlation analysis system block diagram, and system module structure corresponding to the program is as shown in Figure 7.Should
Scheme comprises the following steps：
S101, main corresponding function unit are module 40.According to conditions such as type of service, protocol specifications, module 40 is from mould
Block 30 reads the KQI items and KPI items of analysis to be associated；Then in the slave module 10 of module 40, corresponding KQI and KPI data is read, is entered
The pretreatment operations such as row Data Matching, data cleansing, statistical analysis, fitting, feature extraction., must in process of data preprocessing
Choosing, the calling module 90 of module 40；Optionally, the calling module 100 of module 40.
S102, main corresponding function unit are module 50.The calling module 90 of module 50, is obtained with module 40 in S101
Part index number characteristic value, is not limited the initial hierarchical cluster of categorical measure, and Fig. 8 is that wisdom provided in an embodiment of the present invention is gathered
Initial hierarchical cluster schematic diagram in class process, as shown in Figure 8.Then the calling module 100 of module 50 carries out final cluster judgement,
And export cluster table.Optionally, the calling module 20 of module 40 carries out live expert and aids in cluster judgement.
S103, main corresponding function unit are module 60.The KQI and KPI data of the calling module 40 of module 60 output, are formed
KQI-KPI data vectors space, calling module 90 carry out vector space decomposition operation, isolate quantifiable KQI-KPI associations
Component.
S104, main corresponding function unit are module 70.Module 70 carries out normalizing to KQI-KPI association components and quantifies fortune
Calculate, obtain the degrees of association of each KPI with corresponding KQI.KQI-KPI degrees of association scope is [0,1], sets unrelated thresholding and associated gate
Limit, you can rule out continuous item KPI, reminder item KPI nothing to do with items KPI.To reduce the overlapping journey between continuous item KPI as far as possible
Degree, the cluster table of the calling module 50 of module 70 output, rules out final continuous item KPI, other are same from same cluster correlation item KPI
Continuous item KPI in cluster is then referred to as related similar terms KPI.After module 70 calculates final continuous item KPI normalized weight,
Each KQI related KPI matrixes are exported, include weight.
S105, main corresponding function unit are module 80.The final continuous item for the KQI that module 80 rules out according to module 70
KPI and weight, final continuous item KPI individual event thresholding and multidimensional thresholding are calculated successively.It is more according to business and system accuracy requirement
The dimension of thresholding is tieed up from two dimension, and the quantity no more than final continuous item KPI.
S106, main corresponding function unit are module 80.Final continuous item KPI individual events and multidimensional with step S105 outputs
Thresholding, data after the KQI obtained with step S101 and KPI is cleaned, the False Rate and misdetection rate that checking KQI transfinites.Then with by mistake
Sentence whether the limit that rate is inspection by variable misdetection rate is equal or close to zero, "Yes" then shows that current final KPI items contain KQI and surpassed
The complete base of limit, "No" then show that the still old related KPI of the KQI are not covered by module 30.
The present embodiment also includes following example, and certain provincial capital's main city zone online low rate is found for mobile communication carrier
Main related KPI.Using selected date, 24 hours granularity cell DBMSs of 4000 cells, KQI is investigated：HTTP
The correlation degree of response delay and 30 wireless side KPI.
Step 1, S101 is belonged to.According to abnormal traffic distribute leaflets, it is wireless with corresponding 30 that slave module 30 reads KQI to be assessed
Side KPI item lists.As shown in table 1.
Table 1
Step 2, S101 is belonged to.Slave module 10 reads the KQI and KPI of analysis to be associated 24 hours granularity cell series
According to, and alignd according to acquisition time with cell number matching.
Step 3, S101 is belonged to.Outlier processing, statistical analysis, fitting of distribution are carried out to the KQI after matching and KPI data
Deng, and extract statistical characteristics.The presentation example of module 20 is as follows
Fig. 9 is the parameter frequency matched curve figure that the embodiment of the present invention is examined by normal distribution and data source health degree,
As shown in Figure 9, the index parameter has passed through the inspection of logarithm normal distribution；Health degree judgement is passed through simultaneously.Available for mould
All statistics and mining algorithm of block 90.
Figure 10 is the embodiment of the present invention not by normal distribution-test but the parameter frequency examined by data source health degree
Matched curve figure, as shown in Figure 10, the index parameter by the inspection of normal distribution/logarithm normal distribution, do not automatically attempt to
Match other common distributions；Health degree judgement is passed through；The statistics for meeting normal distribution premise is needed in unusable module 90
Or mining algorithm.
Figure 11 is the parameter frequency that the embodiment of the present invention is not examined not by normal distribution-test and by data source health degree
Matched curve figure is spent, as shown in Figure 11, the index parameter fails not by the inspection of normal distribution/logarithm normal distribution
With other common distributions；Do not judged by health degree；Follow-up calculating is not participated in, and submits malfunction elimination immediately.On inspection, this is small
Area has that failing to report for this parameter repeats to connect report with burst.
Step 4, S102 is belonged to.The KPI data and statistical characteristics obtained using step 1 to step 3,30 KPI are entered
The capable initial hierarchical cluster for not limiting categorical measure, as shown in Figure 8.
Step 5, S102 is belonged to.Based on step 4, expert's clustering information in binding modules 100,30 KPI are gathered for 6
Class.Example such as table 2.
Table 2
Step 6, S103 is belonged to.This example carries out KQI-KPI data vector spatial decompositions using Bayesian statistics principle, with
Conditional probability describes KPI->KQI mapping association component.Because KQI-KPI data have spatially strictly matched with time dimension
And numerical value is within the module 10, it is known that then according to Bayesian formula, for given KQI numerical value (generally taking warning thresholding), KPI->
KQI association component can be drawn by numerical calculations.
Step 7, S104 is belonged to.Based on step 6, to KPI->KQI association components carry out quantum chemical method and normalized, and obtain
KPI-KQI normalizings quantify degree of association ∈ [0,1].The thresholding for being used to adjudicate continuous item KPI in this example is set higher than 0.7, judgement
The thresholding of outlier is set below 0.5, and the judgement between 0.5 and 0.7 is reminder item.Decision method is according to business demand and numerical value
Feature determines, is not limited to thresholding and sentences firmly.
Step 8, S104 is belonged to.To the continuous item KPI ruled out, the cluster table that query steps 5 obtain, similar middle association is taken
The maximum KPI judgements of degree are final continuous item KPI, then the quantity of final continuous item is not over 6 classes.Figure 12 is of the invention real
Apply example and a schematic diagram for KPI weight calculations is associated using main quantization determination point, auxiliary quantization determination point, this example is in step 7
The middle auxiliary judgement point (in accompanying drawing 12 " quantify determination point 2 ") that adds (" quantifies determination point 1 ") with main determination point in accompanying drawing 12
Degree of association numerical value calculate the normalized weight of final continuous item jointly.If for example, rule out 4 final continuous item KPIF1Arrive
KPIF4Afterwards, refer to the attached drawing 12, each final continuous item KPIiThe non-normalized weight W of (i=1to 4)Oi=(quantify determination point 1
Locate probable value at probable value+quantization determination point 2)/2-0.5；Then normalized weight WNi=WOi/Σi(WOi), i=1to 4.
It is as shown in table 3 that example is presented in the conclusion output of module 70.
Table 3
Step 9, S105 is belonged to.This example, with it is expected False Rate, calculates final continuous item KPI list according to Bayesian formula
Item False Rate combines False Rate desired distribution with two dimension.According to the final continuous item KPI weight proportions that step 8 obtains and two-dimentional phase
False Rate distribution is hoped, the adaptive numerical index numbering that should determine that the final continuous item KPI matched two-by-two, two-dimentional joint is obtained with this
Threshold matrix.
Step 10, S106 is belonged to.Thresholding is combined with two dimension according to final continuous item KPI individual event thresholding, obtained with reference to step 2
The KQI-KPI matched datas space arrived, calculate KQI：The final False Rate of http response time delay and misdetection rate.After step 10
It is as shown in table 4 that example is presented in final KPI multidimensional thresholding output.
Table 4
Step 11, S106 is belonged to.It is expected False Rate as variable, under given KQI warning thresholdings, carry out step 9 with
The calculating of step 10, obtains KQI:The misdetection rate curve of http response time delay.Sentence according to whether misdetection rate can reach or approach 0
Whether the final continuous item KPI that breaks is complete.In this example, when http response time delay takes 90ms, 95ms, 100ms and 105ms,
Reach zero to fail to judge, such as table 5.Thus the association analysis scheme and system that the present invention describes are proved, have found from 30 KPI items
Standby KQI continuous items, meet the description in beneficial effect.
Table 5
Obviously, those skilled in the art should be understood that above-mentioned each module of the invention or each step can be with general
Computing device realize that they can be concentrated on single computing device, or be distributed in multiple computing devices and formed
Network on, alternatively, they can be realized with the program code that computing device can perform, it is thus possible to they are stored
Performed in the storage device by computing device, and in some cases, can be with different from shown in order execution herein
The step of going out or describing, they are either fabricated to each integrated circuit modules respectively or by multiple modules in them or
Step is fabricated to single integrated circuit module to realize.So, the present invention is not restricted to any specific hardware and software combination.
The preferred embodiments of the present invention are the foregoing is only, are not intended to limit the invention, for the skill of this area
For art personnel, the present invention can have various modifications and variations.Within the spirit and principles of the invention, that is made any repaiies
Change, equivalent substitution, improvement etc., should be included in the scope of the protection.
Claims (21)
- A kind of 1. method for determining the black-box system input and output degree of association, it is characterised in that including：Quality of service index KQI data in black-box system are matched with Key Performance Indicator KPI data and form data vector sky Between；The KPI data is clustered according to the type of service of the KQI data, wherein, cluster result is orthogonal for selecting Strong correlation KPI items, and auxiliary judgement achievement data health degree；The data vector space is decomposed, to isolate linked character of the KPI data to the KQI data, and Calculate the normalization degree of association of the KPI data to the KQI data；The KPI data associated with each KQI data is determined according to the normalization degree of association respectively, and calculates institute State associated weights of the associated KPI data to the KQI data；The associated KPI data and the associated weights are defined as to the input and output degree of association of the black-box system.
- 2. according to the method for claim 1, it is characterised in that by quality of service index KQI data and Key Performance Indicator KPI data matching forms data vector space to be included with isolating the KPI data to the degree of association of the KQI data：KQI data are matched with KPI data in one or more dimensions and form data vector space to isolate the KPI data To the degree of association of the KQI data.
- 3. according to the method for claim 1, it is characterised in that according to the type of service of the KQI data to the KPI numbers Include according to cluster is carried out：The KQI data and the KPI data are divided into KQI data Layers and KPI data layer；Added between the KQI data Layers and the KPI data layer to the KQI data corresponding to type of service is related takes out As layer parameter, wherein, level of abstraction carries out regular or mapping transformation to KPI data, to be adapted to corresponding mining algorithm；KPI data is clustered using the abstract layer parameter.
- 4. according to the method for claim 1, it is characterised in that carrying out decomposition to the data vector space includes：The data vector space is decomposed at least one in the following manner, isolates the KPI data to the KQI The linked character of data：Directly split, in Spatial Dimension liter dimension in Spatial Dimension dimensionality reduction, in Spatial Dimension.
- 5. according to the method for claim 1, it is characterised in that by quality of service index KQI data and Key Performance Indicator KPI data matching forms data vector space to be included with isolating the KPI data to the degree of association of the KQI data：Fitting of distribution is carried out to the KPI data according to the traffic performance information of the KQI data and figure is shown, and determines to close Manage KPI data；KQI data are matched with the rationally KPI data and form data vector space to isolate the KPI data to described The degree of association of KQI data.
- 6. according to the method for claim 5, it is characterised in that the traffic performance information is according to matching presetting database And/or obtained according to business demand.
- 7. according to the method for claim 1, it is characterised in that carrying out decomposition to the data vector space includes：Dimension-reduction treatment is carried out to the data vector space；The data vector space is directly decomposed；Validity feature value is extracted again after being augmented to the data vector space.
- 8. according to the method for claim 7, it is characterised in that data vector space dimension-reduction treatment is included with lower section At least one method：Decision tree beta pruning, return merging, cluster, expert's auxiliary judgement.
- 9. according to the method for claim 7, it is characterised in that the data vector space, which is directly decomposed, to be included：The data vector space is decomposed based on Bayesian statistics algorithm；Equivalent numerical value based on singular value decomposition thinking calculates.
- 10. according to the method for claim 7, it is characterised in that carried out again after being augmented to the data vector space Decomposition includes：Decomposed again after being augmented based on the algorithm of support vector machines to the data vector space；Processing is augmented based on neural network algorithm, i.e., Hidden unit number is higher than input dimension.
- 11. according to the method described in claim 1 to 10 any one, it is characterised in that according to the normalization degree of association After determining the KPI data associated with each KQI data respectively, methods described also includes：Calculate the one-dimensional thresholding of quantization or multidimensional thresholding of the KPI data associated with each KQI data；The False Rate and/or misdetection rate of the KQI data over run are obtained according to the one-dimensional thresholding of the quantization or multidimensional thresholding；It is whether empty comprising the KQI data over run that the associated KPI data is analyzed according to the misdetection rate and/or misdetection rate Between complete base.
- 12. according to the method for claim 11, it is characterised in that institute is being analyzed according to the misdetection rate and/or misdetection rate After stating the complete base whether associated KPI data includes the KQI data over run space, methods described also includes：The associated KPI data is analyzed according to the misdetection rate and/or misdetection rate and does not include the KQI data over run space Probability.
- 13. according to the method described in claim 1 to 10 any one, it is characterised in that according to the normalization degree of association After determining the KPI data associated with each KQI data respectively, methods described also includes：Judge whether the KQI data lack；In the case where judging the KQI shortage of data, according to the reversely deduction of the quantization multidimensional thresholding of history KPI data The probability of KQI data erroneous judgement, and carry out system pre-optimized and parameter and adjustment.
- 14. according to the method for claim 1, it is characterised in that the KPI data includes at least one of：Radio resource control RRC connection is created as power, the RAB E-RAB of evolution is created as power, wireless connection Handover success rate between rate, E-RAB drop rates, base station ENB, the up packet loss in community user face, the descending packet loss in community user face, The descending average delay in community user face, community user face is descending abandons bag rate, cell downlink bag number, the up Block Error Rate of MAC layer, media The descending Block Error Rate of access control MAC layer, up initial mixing HARQ HARQ retransmit ratio, down initial HARQ is retransmitted Ratio, descending double-current flow accounting, uplink orthogonal phase-shift keying (PSK) QPSK ratios, up 16QAM ratios, descending QPSK ratios, under Row 16QAM ratios, descending 64 quadrature amplitude modulation QAM ratios, uplink service byte number of eating dishes without rice or wine, downlink business byte number of eating dishes without rice or wine, Ascending physical signal resource block PRB average utilizations, descending PRB average utilizations, up every PRB average throughputs, descending every PRB are put down Equal handling capacity, -110dBm coverage rates, average signal and interference plus noise are than SINR, the average channel quality of subband 0 instruction CQI, use The average activation equipment UE numbers in family face.
- 15. according to the method for claim 1, it is characterised in that the KQI data are rung including HTTP HTTP Answer time delay.
- 16. according to the method for claim 1, it is characterised in that the cluster includes at least one of：Capacity performance index is gathered Class, access index cluster, efficiency index cluster, complete holding index cluster.
- 17. according to the method for claim 16, it is characterised in that it is described it is complete keep index cluster also include it is following at least One of：Packet Service cluster, uplink complete holding cluster, descending complete holding cluster.
- A kind of 18. correlation analysis system, it is characterised in that including：Memory cell, for the KQI data and KPI data in Networks for Storage Services；Data pre-processing unit, for being pre-processed to the KQI data and the KPI data, wherein, the pretreatment bag Include：Data Matching, data cleansing, statistical nature extraction and statistics are presented；Cluster cell, for carrying out intelligent clustering to the KPI data, and export cluster table；Vector space resolving cell, it is connected with the data pre-processing unit, for pretreated KQI data and KPI numbers Decomposed according to the vector space of composition, extraction can association component of the KQI data to the KPI data that quantifies of normalizing；Quantify association computing unit, be connected with the vector space resolving cell, for carrying out normalizing amount to the association component Change and calculate, obtain the quantization degree of association of the KQI data to the KPI data, calculate its total weight order, and export and include institute State the quantization incidence matrix of weight；Multidimensional threshold computation unit, for calculating the multidimensional quantization threshold of continuous item KPI data according to the incidence matrix, counter pushing away The False Rate and/or misdetection rate that KQI transfinites, output multidimensional quantization threshold matrix transfinites with KQI assesses data；Optimize unit, the data institute service network is assessed for being transfinited according to the multidimensional quantization threshold matrix and the KQI Carry out performance optimization.
- 19. system according to claim 18, it is characterised in that the system also includes：Business datum interface, comprising interface is presented, carrying out auxiliary to the output data of the system for receiving external command sentences Certainly.
- 20. system according to claim 18, it is characterised in that the system also includes：Data mining analysis algorithm pond, for storing the data mining algorithm of the system；Database, for storing the data analysis of the system with excavating conclusion, and pilot process information.
- A kind of 21. device for determining the black-box system input and output degree of association, it is characterised in that including：Separation module, for the quality of service index KQI data in black-box system to be matched into structure with Key Performance Indicator KPI data Into data vector space to isolate the degree of association of the KPI data to the KQI data；Cluster module, for being clustered according to the type of service of the KQI data to the KPI data；First computing module, for being decomposed to the data vector space, and the KPI data is calculated to the KQI numbers According to the normalization degree of association；Second computing module, it is associated with each KQI data for being determined respectively according to the normalization degree of association KPI data, and calculate associated weights of the associated KPI data to the KQI data；Determining module, for the associated KPI data and the associated weights to be defined as to the input of the black-box system Export the degree of association.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610579303.2A CN107645393A (en) | 2016-07-20 | 2016-07-20 | Determine the method, apparatus and system of the black-box system input and output degree of association |
| PCT/CN2017/087940 WO2018014674A1 (en) | 2016-07-20 | 2017-06-12 | Method, apparatus, and system for determining degree of association of input and output of black box system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610579303.2A CN107645393A (en) | 2016-07-20 | 2016-07-20 | Determine the method, apparatus and system of the black-box system input and output degree of association |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107645393A true CN107645393A (en) | 2018-01-30 |
Family
ID=60991875
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610579303.2A Withdrawn CN107645393A (en) | 2016-07-20 | 2016-07-20 | Determine the method, apparatus and system of the black-box system input and output degree of association |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN107645393A (en) |
| WO (1) | WO2018014674A1 (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108875365A (en) * | 2018-04-22 | 2018-11-23 | 北京光宇之勋科技有限公司 | A kind of intrusion detection method and intrusion detection detection device |
| CN109729540A (en) * | 2019-01-18 | 2019-05-07 | 福建福诺移动通信技术有限公司 | A kind of base station parameter automatic optimization method neural network based |
| WO2019184640A1 (en) * | 2018-03-26 | 2019-10-03 | 华为技术有限公司 | Indicator determination method and related device thereto |
| CN110493803A (en) * | 2019-09-17 | 2019-11-22 | 南京邮电大学 | A kind of cell scenario division methods based on machine learning |
| CN110659731A (en) * | 2018-06-30 | 2020-01-07 | 华为技术有限公司 | Neural network training method and device |
| CN111327450A (en) * | 2018-12-17 | 2020-06-23 | 中国移动通信集团北京有限公司 | Method, device, equipment and medium for determining quality difference reason |
| CN112153663A (en) * | 2019-06-26 | 2020-12-29 | 大唐移动通信设备有限公司 | Wireless network evaluation method and device |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110837841B (en) * | 2018-08-17 | 2024-05-21 | 北京亿阳信通科技有限公司 | KPI degradation root cause identification method and device based on random forest |
| CN110288467B (en) * | 2019-04-19 | 2023-07-25 | 平安科技（深圳）有限公司 | Data mining method and device, electronic equipment and storage medium |
| US11816542B2 (en) * | 2019-09-18 | 2023-11-14 | International Business Machines Corporation | Finding root cause for low key performance indicators |
| CN114386728A (en) * | 2020-10-19 | 2022-04-22 | 中国移动通信集团北京有限公司 | KQI perception limited condition determining method, device, equipment and computer storage medium |
| CN112950908B (en) * | 2021-02-03 | 2022-12-30 | 重庆川仪自动化股份有限公司 | Data monitoring and early warning method, system, medium and electronic terminal |
| CN117596133B (en) * | 2024-01-18 | 2024-04-05 | 山东中测信息技术有限公司 | Service portrayal and anomaly monitoring system and monitoring method based on multidimensional data |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102098719A (en) * | 2011-01-11 | 2011-06-15 | 大唐移动通信设备有限公司 | Method and device for determining network quality |
| CN102104900A (en) * | 2011-01-27 | 2011-06-22 | 大唐移动通信设备有限公司 | Method and equipment for analyzing user perception |
| CN102625344A (en) * | 2012-03-13 | 2012-08-01 | 重庆信科设计有限公司 | Model and method for evaluating user experience quality of mobile terminal |
| CN102685791A (en) * | 2012-05-22 | 2012-09-19 | 北京东方文骏软件科技有限责任公司 | Method for evaluating user quality of experience (QoE) of WAP (Wireless Application Protocol) services by simulating user behavior |
| CN102685789A (en) * | 2012-05-22 | 2012-09-19 | 北京东方文骏软件科技有限责任公司 | Method for evaluating QoE (Quality Of Experience) of voice service user perception experience by simulating user behaviors |
| CN103138963A (en) * | 2011-11-25 | 2013-06-05 | 华为技术有限公司 | Method and device for positioning network problems based on user perception |
| CN104994133A (en) * | 2015-05-22 | 2015-10-21 | 华中科技大学 | Mobile Web webpage access user experience perception evaluating method based on network KPI |
| 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 |
| US20160162346A1 (en) * | 2014-12-08 | 2016-06-09 | Alcatel-Lucent Usa, Inc. | Root cause analysis for service degradation in computer networks |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102612060B (en) * | 2012-03-31 | 2014-06-04 | 西安交通大学 | Evaluation method based on entropy value calculation and used for compatibility of cross-layer design |
| CN102685717B (en) * | 2012-05-08 | 2014-10-08 | 中国联合网络通信集团有限公司 | network service quality parameter identification method and device |
-
2016
- 2016-07-20 CN CN201610579303.2A patent/CN107645393A/en not_active Withdrawn
-
2017
- 2017-06-12 WO PCT/CN2017/087940 patent/WO2018014674A1/en active Application Filing
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102098719A (en) * | 2011-01-11 | 2011-06-15 | 大唐移动通信设备有限公司 | Method and device for determining network quality |
| CN102104900A (en) * | 2011-01-27 | 2011-06-22 | 大唐移动通信设备有限公司 | Method and equipment for analyzing user perception |
| CN103138963A (en) * | 2011-11-25 | 2013-06-05 | 华为技术有限公司 | Method and device for positioning network problems based on user perception |
| CN102625344A (en) * | 2012-03-13 | 2012-08-01 | 重庆信科设计有限公司 | Model and method for evaluating user experience quality of mobile terminal |
| CN102685791A (en) * | 2012-05-22 | 2012-09-19 | 北京东方文骏软件科技有限责任公司 | Method for evaluating user quality of experience (QoE) of WAP (Wireless Application Protocol) services by simulating user behavior |
| CN102685789A (en) * | 2012-05-22 | 2012-09-19 | 北京东方文骏软件科技有限责任公司 | Method for evaluating QoE (Quality Of Experience) of voice service user perception experience by simulating user behaviors |
| US20160162346A1 (en) * | 2014-12-08 | 2016-06-09 | Alcatel-Lucent Usa, Inc. | Root cause analysis for service degradation in computer networks |
| CN104994133A (en) * | 2015-05-22 | 2015-10-21 | 华中科技大学 | Mobile Web webpage access user experience perception evaluating method based on network KPI |
| 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 |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019184640A1 (en) * | 2018-03-26 | 2019-10-03 | 华为技术有限公司 | Indicator determination method and related device thereto |
| CN108875365A (en) * | 2018-04-22 | 2018-11-23 | 北京光宇之勋科技有限公司 | A kind of intrusion detection method and intrusion detection detection device |
| CN108875365B (en) * | 2018-04-22 | 2023-04-07 | 湖南省金盾信息安全等级保护评估中心有限公司 | Intrusion detection method and intrusion detection device |
| CN110659731A (en) * | 2018-06-30 | 2020-01-07 | 华为技术有限公司 | Neural network training method and device |
| CN110659731B (en) * | 2018-06-30 | 2022-05-17 | 华为技术有限公司 | Neural network training method and device |
| CN111327450A (en) * | 2018-12-17 | 2020-06-23 | 中国移动通信集团北京有限公司 | Method, device, equipment and medium for determining quality difference reason |
| CN109729540A (en) * | 2019-01-18 | 2019-05-07 | 福建福诺移动通信技术有限公司 | A kind of base station parameter automatic optimization method neural network based |
| CN109729540B (en) * | 2019-01-18 | 2022-05-17 | 福建福诺移动通信技术有限公司 | Base station parameter automatic optimization method based on neural network |
| CN112153663A (en) * | 2019-06-26 | 2020-12-29 | 大唐移动通信设备有限公司 | Wireless network evaluation method and device |
| CN110493803A (en) * | 2019-09-17 | 2019-11-22 | 南京邮电大学 | A kind of cell scenario division methods based on machine learning |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2018014674A1 (en) | 2018-01-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107645393A (en) | Determine the method, apparatus and system of the black-box system input and output degree of association | |
| US11496353B2 (en) | Root cause analysis and automation using machine learning | |
| CN107295594B (en) | Method for switching network, device and mobile terminal | |
| CN106534333A (en) | Bidirectional selection computing unloading method based on MEC and MCC | |
| CN102149171B (en) | Method and device for selecting network based on converged heterogeneous radio access network | |
| CN102098684B (en) | System and method for allocating cross-layer resources in cognitive radio network | |
| CN110662245B (en) | Base station load early warning method and device based on deep learning | |
| CN111132190A (en) | Base station load early warning method and device | |
| CN108063676A (en) | Communication network failure method for early warning and device | |
| CN104462115A (en) | Spam message identifying method and device | |
| Kiran et al. | Resource allocation optimization in LTE-A/5G networks using big data analytics | |
| CN107943697A (en) | Problem distribution method, device, system, server and computer-readable storage medium | |
| CN106922028A (en) | A kind of method of discrimination and device of wireless telecommunication system User Priority | |
| CN106453143A (en) | Bandwidth setting method, device and system | |
| CN109960839B (en) | Service link discovery method and system of service support system based on machine learning | |
| D’andreagiovanni | On improving the capacity of solving large-scale wireless network design problems by genetic algorithms | |
| CN108833227A (en) | A kind of smart home communication optimization scheduling system and method based on edge calculations | |
| CN107371179A (en) | Measurement result report method, measurement result method of reseptance, relevant device and system | |
| CN107015993A (en) | A kind of user type recognition methods and device | |
| Sangaiah et al. | Automatic fault detection and diagnosis in cellular networks and beyond 5g: intelligent network management | |
| CN103957268B (en) | Rule-driven data transmission method | |
| CN113660687B (en) | Network difference cell processing method, device, equipment and storage medium | |
| CN103889005B (en) | A kind of cooperative multipoint transmission dispatching method and device | |
| CN107155192A (en) | A kind of user experience quality appraisal procedure and device | |
| CN112035286A (en) | Method and device for determining fault cause, storage medium and electronic device |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20180130 |
|
| WW01 | Invention patent application withdrawn after publication |