CN106487571A - A kind of method and device of assessment network performance index variation tendency - Google Patents

Abstract

The invention provides a kind of method and device of assessment network performance index variation tendency, solves existing manual observation index variation tendency labor intensive and time, impact analysis is comprehensive, the problem of accuracy and extensibility.The present invention includes：Obtain the desired value of multiple network performance indexes of network to be detected；According to the desired value of multiple network performance indexes, the coefficient correlation between the moving average of multiple network performance indexes is obtained；According to the coefficient correlation between the moving average of multiple network performance indexes, in multiple network performance indexes one similarity for presetting the variation tendency of network performance index and the variation tendency of other non-default network performance indexes is obtained.The solution of the present invention has been automatically performed the assessment to index variation tendency, improves the analysis efficiency of the network optimization, and the comprehensive and extensibility for realizing analyzing.

Description

A kind of method and device of assessment network performance index variation tendency

Technical field

The present invention relates to the technical field of communications applications, particularly relates to a kind of assessment network performance index change and becomes The method and device of gesture.

Background technology

After communication network builds up, need by continuing to optimize the normal operation that just can guarantee that network, so that net Network quality meets user's request.The network optimization be by the existing network for having run is carried out data acquisition, The means such as data analysis, Parameter analysis, hardware check, find out the reason for affecting network quality, and pass through Parameter adjustment, network structure regulation, device configuration adjustment and other technological means, it is ensured that network is high-quality Operation, makes conventional network resources obtain optimum efficiency.

During the existing network optimization, network optimization personnel are needed according to substantial amounts of network performance index report Table, by way of drawing index change curve, the variation tendency of manual observation indices.Simultaneously as The index of at present existing network definition is a lot, and network optimization personnel simultaneously can not possibly draw the change song of each index successively Line chart carries out case study, can only choose its some key for thinking and refer to according to existing network optimization experience Mark carries out observation analysis, forms initial optimization scheme.Similarly, network optimization people after prioritization scheme enforcement Impossible impact of the peep optimization scheme to all indexs of member, can only observe which according to previous experiences and think possibility The affected index variation tendency of meeting, draws assessment result.The major defect of this process is：

1st, a large amount of manpowers and time are expended：Due to the network optimization be an automatic network build up operation after start hold Continue continuous process, network optimization personnel need constantly to carry out selecting index, index observing, index evaluation mistake Journey, if rely solely on eye-observation index, it will waste time and energy；

2nd, lack the comprehensive of analysis：Due to network performance index enormous amount, network optimization personnel simultaneously can not The variation tendency of all indexs all can be analyzed, also be impossible to fully understand that prioritization scheme is right after implementing The impact of existing network quality.It is possible to the index set Long-term change trend as expected of network optimization personnel care, but other Index occurs in that deterioration is but and undiscovered, and these situations will the accuracy of impact analysis and comprehensive；

3rd, lack the extensibility of analysis：During the network optimization, network optimization personnel are basis products in the past Tired a large amount of Optimization Experiences are chosen the index set of its care and are observed.With the new demand of the existing network network optimization, More thinner network performance indexes will constantly be defined, at this moment network optimization personnel for new define index with The relation of original index does not have experience follow, and can largely effect on the efficiency of analysis.

Content of the invention

It is an object of the invention to provide a kind of method and device of assessment network performance index variation tendency, uses To solve existing manual observation index variation tendency labor intensive and time, impact analysis is comprehensive, accuracy Problem with extensibility.

To achieve these goals, the invention provides a kind of assessment network performance index variation tendency method, Including：

Obtain the desired value of multiple network performance indexes of network to be detected；

According to the desired value of multiple network performance indexes, the movement of multiple network performance indexes is obtained Coefficient correlation between mean value；

According to the coefficient correlation between the moving average of multiple network performance indexes, obtain multiple described In network performance index one presets the variation tendency of network performance index and other non-default network performance indexes Variation tendency similarity.

Wherein, include the step of the desired value of the multiple network performance indexes for obtaining network to be detected：

Obtain multiple network performance indexes of the network to be detected multiple times respectively within a predetermined period of time The desired value of point.

Wherein, the multiple network performance indexes for obtaining the network to be detected are distinguished within a predetermined period of time Multiple time points desired value the step of include：

Obtain N number of network performance index of the network to be detected T respectively in the predetermined amount of time The desired value of time point, obtains original performance index matrix P：

Wherein, Pi is all of desired value of i-th network performance index, Pi_jExist for i-th network performance index The desired value of j-th time point in the predetermined amount of time, N and T are the integer more than 1, and i is 1 to the integer between N, and j is 1 to the integer between T.

Wherein, the desired value according to multiple network performance indexes, obtains multiple network performances The step of coefficient correlation between the moving average of index, includes：

According to the desired value of multiple time points of each described network performance index in the predetermined amount of time, Obtain the moving average of each network performance index；

According to the moving average of each network performance index, multiple network performance indexes are obtained Coefficient correlation between moving average.

Wherein, the multiple time points according to each described network performance index in the predetermined amount of time Desired value, include the step of the moving average for obtaining each network performance index：

Moving average Ai of each network performance index is obtained by equation below_j：

Wherein, the network to be detected includes N number of network performance index, each described network performance index bag Include the desired value of T time point within a predetermined period of time, Pi_nFor i-th network performance index described pre- The desired value of n-th time point that fixes time in section, Ai_jRepresent j-th movement of i-th network performance index Mean value, i are that j is that mt represents each described network 1 to the integer between T 1 to the integer between N The corresponding time span of the moving average of performance indications, and mt is the integer less than T, N and T are greatly In 1 integer.

Wherein, the moving average according to each network performance index, obtains multiple networks The step of coefficient correlation between the moving average of performance indications, includes：

Correlation coefficient r between the moving average of multiple network performance indexes is obtained by equation below_ij：

Wherein, the network to be detected includes N number of network performance index, each described network performance index bag Include the desired value of T time point within a predetermined period of time, r_ijFor i-th network performance index and j-th net Coefficient correlation between network performance indications, Ai_kRepresent k-th moving average of i-th network performance index,Represent the mean value of all desired values of i-th network performance index, N and T is the integer more than 1, k For 1 to the integer between T-1.

Wherein, the coefficient correlation between the moving average according to multiple network performance indexes, obtains Take in multiple network performance indexes one variation tendency for presetting network performance index and other non-default nets The step of similarity of the variation tendency of network performance indications, includes：

According to the default network performance index, the movement of non-default network performance index is put down with each described respectively The size of the coefficient correlation between average, to the default network performance index and each described non-default network The similarity of the variation tendency of performance indications is ranked up, wherein, the bigger non-default internetworking of coefficient correlation Can index bigger with the similarity of the variation tendency of the default network performance index, less non-of coefficient correlation Default network performance index is less with the similarity of the variation tendency of the default network performance index.

Embodiments of the invention additionally provide a kind of device of assessment network performance index variation tendency, including：

First acquisition module, for obtaining the desired value of multiple network performance indexes of network to be detected；

Second acquisition module, for the desired value according to multiple network performance indexes, obtains multiple described Coefficient correlation between the moving average of network performance index；

3rd acquisition module, for the correlation between the moving average according to multiple network performance indexes Coefficient, obtain in multiple network performance indexes one variation tendency for presetting network performance index and other The similarity of the variation tendency of non-default network performance index.

Wherein, first acquisition module includes：

First acquisition unit, for obtaining multiple network performance indexes of the network to be detected respectively predetermined The desired value of the multiple time points in the time period.

Wherein, the first acquisition unit refers to specifically for N number of network performance of the acquisition network to be detected The desired value of mark T time point respectively in the predetermined amount of time, obtains original performance index matrix P：

Wherein, Pi is all of desired value of i-th network performance index, Pi_jExist for i-th network performance index The desired value of j-th time point in the predetermined amount of time, N and T are the integer more than 1, and i is 1 to the integer between N, and j is 1 to the integer between T.

Wherein, second acquisition module includes：

Second acquisition unit, for many in the predetermined amount of time according to each described network performance index The desired value of individual time point, obtains the moving average of each network performance index；

3rd acquiring unit, for the moving average according to each network performance index, obtains multiple Coefficient correlation between the moving average of the network performance index.

Wherein, the second acquisition unit refers to specifically for obtaining each described network performance by equation below Target moving average Ai_j：

Wherein, the network to be detected includes N number of network performance index, each described network performance index bag Include the desired value of T time point within a predetermined period of time, Pi_nFor i-th network performance index described pre- The desired value of n-th time point that fixes time in section, Ai_jRepresent j-th movement of i-th network performance index Mean value, i are that j is that mt represents each described network 1 to the integer between T 1 to the integer between N The corresponding time span of the moving average of performance indications, and mt is the integer less than T, N and T are greatly In 1 integer.

Wherein, the 3rd acquiring unit refers to specifically for obtaining multiple network performances by equation below Correlation coefficient r between target moving average_ij：

Wherein, the network to be detected includes N number of network performance index, each described network performance index bag Include the desired value of T time point within a predetermined period of time, r_ijFor i-th network performance index and j-th net Coefficient correlation between network performance indications, Ai_kRepresent k-th moving average of i-th network performance index,Represent the mean value of all desired values of i-th network performance index, N and T is the integer more than 1, k For 1 to the integer between T-1.

Wherein, the 3rd acquisition module specifically for according to the default network performance index respectively with each The size of the coefficient correlation between the moving average of the non-default network performance index, to the default net The similarity of the variation tendency of non-default network performance index is ranked up network performance indications with each described, its In, the variation tendency of the bigger non-default network performance index of coefficient correlation and the default network performance index Similarity bigger, the less non-default network performance index of coefficient correlation and the default network performance index Variation tendency similarity less.

The embodiment of the present invention has the advantages that：

The method of the assessment network performance index variation tendency of the embodiment of the present invention, obtains many of network to be detected The desired value of individual network performance index；According to the desired value of multiple network performance indexes, multiple institutes are obtained State the coefficient correlation between the moving average of network performance index；According to multiple network performance indexes Coefficient correlation between moving average, obtained in multiple network performance indexes preset network performance The similarity of the variation tendency of the variation tendency of index and other non-default network performance indexes.The present invention passes through Similitude to network performance index variation tendency is analyzed, change that can quickly to a large amount of network performance indexes Change trend is analyzed, and substantially increases analysis efficiency, breaches the limited index of dependence previous experiences selection and enters The limitation of row analysis, it is ensured that the comprehensive, accuracy of analysis and extensibility.

Description of the drawings

Fig. 1 is the first workflow diagram of the method for present invention assessment network performance index variation tendency；

Fig. 2 is the second workflow diagram of the method for present invention assessment network performance index variation tendency；

Fig. 3 is the 3rd workflow diagram of the method for present invention assessment network performance index variation tendency；

Fig. 4 is the 4th workflow diagram of the method for present invention assessment network performance index variation tendency

Fig. 5 is the structural representation of the device of present invention assessment network performance index variation tendency.

Specific embodiment

For making the technical problem to be solved in the present invention, technical scheme and advantage clearer, below in conjunction with tool Body embodiment and accompanying drawing are described in detail.

In prior art, generally comprise the step of the network optimization：

1st, data acquisition：Network performance index is gathered by test, webmaster or monitoring system；

2nd, data analysis：Network performance index to gathering is analyzed, such as index absolute value and its change Deng；

3rd, positioning problems：By index analysis result, existing network is carried out in conjunction with the Optimization Experience of network optimization personnel Problem judgment；

4th, scheme is implemented：According to the existing network possibility problem that judges, prioritization scheme is formulated, and carries out existing network reality Apply；

5th, recruitment evaluation：Before and after contrasting scheme enforcement, the situation of change of network performance index, assess excellent Whether effectively change scheme, if unsolved problem, needs suboptimization again.

In above-mentioned network optimization step, network performance index is the basis of optimization personnel analysis, optimizes personnel and leads to Crossing the mutation analysis to network performance index carries out network problem positioning, by network performance after scheme enforcement The analysis again of index is optimized the recruitment evaluation of scheme.

When network optimization personnel carry out network performance index analysis, being will be defeated for predefined network performance index Go out into excel form, changing trend diagram is drawn as to its index set of concern, is entered according to index changing trend diagram Row case study.After preliminary solution being drawn and implement by analysis, optimize personnel again to interested in which Index set draw changing trend diagram, and assess effect of optimization further.Due to the network optimization, personnel need people The variation tendency of work observation index, consumes a large amount of manpowers and time, and as network optimization personnel can only root According to this toward the limited index of empiric observation, comprehensive, accuracy and the extensibility of analysis is have impact on.

Based on the problems referred to above, a kind of side of assessment network performance index variation tendency is embodiments provided Method, solves existing manual observation index variation tendency labor intensive and time, and impact analysis is comprehensive, accurate True property and the problem of extensibility.

First embodiment：

As shown in figure 1, the method for the assessment network performance index variation tendency, including：

Step 1：Obtain the desired value of multiple network performance indexes of network to be detected；

Step 2：According to the desired value of multiple network performance indexes, multiple network performance indexes are obtained Moving average between coefficient correlation；

Step 3：According to the coefficient correlation between the moving average of multiple network performance indexes, obtain many In the individual network performance index one presets the variation tendency of network performance index and other non-default internetworkings The similarity of the variation tendency of energy index.

Specifically, above-mentioned default network performance index can be any one index in multiple network performance indexes, Change feelings that can be to any two index in network performance index in preset time period by above-mentioned steps 3 The similarity of condition is analyzed.

The method of the assessment network performance index variation tendency of the embodiment of the present invention, obtains many of network to be detected The desired value of individual network performance index；According to the desired value of multiple network performance indexes, multiple institutes are obtained State the coefficient correlation between the moving average of network performance index；According to multiple network performance indexes Coefficient correlation between moving average, obtained in multiple network performance indexes preset network performance The similarity of the variation tendency of the variation tendency of index and other non-default network performance indexes.The present invention passes through Similitude to network performance index variation tendency is analyzed, change that can quickly to a large amount of network performance indexes Change trend is analyzed, and substantially increases analysis efficiency, breaches the limited index of dependence previous experiences selection and enters The limitation of row analysis, it is ensured that the comprehensive, accuracy of analysis and extensibility.

Second embodiment：

As shown in Fig. 2 the method for the assessment network performance index variation tendency, including：

Step 11：Obtain multiple network performance indexes of network to be detected respectively within a predetermined period of time multiple The desired value of time point；

In a particular embodiment of the present invention, it is assumed that have N number of network performance index, then can in advance statistically State N number of network performance index to connect in certain geography granularity (cell/region/the whole network etc.) and predetermined amount of time The desired value of continuous T time point (hour/day etc.), N and T are the integer more than 1.

Step 2：According to the desired value of multiple network performance indexes, multiple network performance indexes are obtained Moving average between coefficient correlation；

Step 3：According to the coefficient correlation between the moving average of multiple network performance indexes, obtain many In the individual network performance index one presets the variation tendency of network performance index and other non-default internetworkings The similarity of the variation tendency of energy index.

Preferably, above-mentioned steps 11 may particularly include：

Obtain the T time of N number of network performance index of network to be detected respectively in the predetermined amount of time The desired value of point, obtains original performance index matrix P：

Wherein, Pi is all of desired value of i-th network performance index, Pi_jExist for i-th network performance index The desired value of j-th time point in the predetermined amount of time, N and T are the integer more than 1, and i is 1 to the integer between N, and j is 1 to the integer between T.

In the embodiment of the present invention, by obtaining N number of network performance index T respectively within a predetermined period of time Desired value Pi of time point_j, and using the achievement data of each time point as an analysis sample, N can be obtained T sample of individual index, is the phase of situation of change next to any two index in T time point Strong data are provided like degree analysis to support, and can be quickly and easily by original performance index matrix P Related data is obtained, effectively increases treatment effeciency.

In addition, the time point in predetermined amount of time can be hour, day or week etc., network optimization personnel can basis Demand is set.Such as assume that network optimization personnel wish daily to analyze certain cell network performance of continuous 30 days Index, then obtain N number of network performance index index sampled value of continuous 30 days of the cell (as per 15 points One sampled point of clock), and substantial amounts of for each network performance index index sampled value is daily processed, obtain The original performance index matrix P of the cell, matrix size are arranged for N row 30.

3rd embodiment：

As shown in figure 3, the method for the assessment network performance index variation tendency, including：

Step 11：Obtain multiple network performance indexes of network to be detected respectively within a predetermined period of time multiple The desired value of time point

Step 21：According to multiple time points of each described network performance index in the predetermined amount of time Desired value, obtains the moving average of each network performance index；

Step 22：According to the moving average of each network performance index, multiple internetworkings are obtained Coefficient correlation between the moving average of energy index；

Step 3：According to the coefficient correlation between the moving average of multiple network performance indexes, obtain many In the individual network performance index one presets the variation tendency of network performance index and other non-default internetworkings The similarity of the variation tendency of energy index.

Preferably, above-mentioned steps 21 are specifically included：

Moving average Ai of each network performance index is obtained by equation below_j：

Wherein, the network to be detected includes N number of network performance index, each described network performance index bag Include the desired value of T time point within a predetermined period of time, Pi_nFor i-th network performance index described pre- The desired value of n-th time point that fixes time in section, Ai_jRepresent j-th movement of i-th network performance index Mean value, i are that j is that mt represents each described network 1 to the integer between T 1 to the integer between N The corresponding time span of the moving average of performance indications, and mt is the integer less than T, N and T are greatly In 1 integer.

Preferably, above-mentioned steps 22 are specifically included：

Correlation coefficient r between the moving average of multiple network performance indexes is obtained by equation below_ij：

Wherein, the network to be detected includes N number of network performance index, each described network performance index bag Include the desired value of T time point within a predetermined period of time, r_ijFor i-th network performance index and j-th net Coefficient correlation between network performance indications, Ai_kRepresent k-th moving average of i-th network performance index,Represent the mean value of all desired values of i-th network performance index, N and T is the integer more than 1, k For 1 to the integer between T-1.

In the embodiment of the present invention, network optimization personnel are in problem analysis or assessment effect of optimization, it is assumed that select Index i as preset reference index, then index i with the coefficient correlation of the variation tendency of other remaining indexs is By r_ijTo represent, j is 1 to the integer between N, r here_ijValue wherein, is more connect between -1 and 1 Nearly 1 characteristic index i is more identical with the variation tendency of index j, closer to the change of -1 characteristic index i and index j Change trend is not more conversely, have significant relation closer to 0 characteristic index i with the variation tendency of index j.

The variation tendency of hypothesis index i is to decline in observation cycle, and network optimization personnel are during analysis It should be understood which index is also declined, and which index can rise.Can be according to r_ijTo all N-1 Network performance index is classified, and is such as divided into, variation tendency identical with index i variation tendency contrary or no aobvious Three class such as work relation, sorting technique can adopt clustering algorithm (as K-means algorithm), it is also possible to according to need Interval range to be delimited is classifying.

Preferably, above-mentioned steps 3 may particularly include：

According to the default network performance index, the movement of non-default network performance index is put down with each described respectively The size of the coefficient correlation between average, to the default network performance index and each described non-default network The similarity of the variation tendency of performance indications is ranked up, wherein, the bigger non-default internetworking of coefficient correlation Can index bigger with the similarity of the variation tendency of the default network performance index, less non-of coefficient correlation Default network performance index is less with the similarity of the variation tendency of the default network performance index.

The method of the assessment network performance index variation tendency of the embodiment of the present invention, by multiple internetworkings Coefficient correlation between the moving average of energy index is analyzed, and obtains other non-default network performance indexes Similarity with default network performance index variation tendency such that it is able to according to the change of default network performance index Change trend, obtains the variation tendency of other non-default network performance indexes, substantially increases treatment effeciency.

One that the embodiment of the present invention is exemplified below implements process.

Technical scheme, as shown in figure 4, including：

Step S1：Obtain T respectively within a predetermined period of time of N number of network performance index of network to be detected The desired value of time point, obtains original performance index matrix P：

Wherein, Pi is all of desired value of i-th network performance index, Pi_jExist for i-th network performance index The desired value of j-th time point in the predetermined amount of time, N and T are the integer more than 1, and i is 1 to the integer between N, and j is 1 to the integer between T；

Step S2：Assume network performance index mean value of concern time span be mt, according to primitiveness Energy index matrix P, obtains rolling average matrix avg：

Wherein, avg is changed into the matrix of T-mt+1 row N row, Ai_jRepresent the jth of i-th network performance index Individual moving average, circular are as follows：

Wherein, the network to be detected includes N number of network performance index, each described network performance index bag Include the desired value of T time point within a predetermined period of time, Pi_nFor i-th network performance index described pre- The desired value of n-th time point that fixes time in section, Ai_jRepresent j-th movement of i-th network performance index Mean value, i are that j is that mt represents each described network 1 to the integer between T 1 to the integer between N The corresponding time span of the moving average of performance indications, and mt is the integer less than T, N and T are greatly In 1 integer；

Step S3：The coefficient correlation between each column vector of rolling average matrix avg is calculated, is obtained change and becomes Gesture correlation matrix R_avg:

Wherein,

Wherein, the network to be detected includes N number of network performance index, each described network performance index bag Include the desired value of T time point within a predetermined period of time, r_ijFor i-th network performance index and j-th net Coefficient correlation between network performance indications, Ai_kRepresent k-th moving average of i-th network performance index,Represent the mean value of all desired values of i-th network performance index, N and T is the integer more than 1, k For 1 to the integer between T-1；

Step S4：Network optimization personnel are in problem analysis or assessment effect of optimization, it is assumed that have selected index i As preset reference index, then pass through variation tendency correlation matrix R_avgIn i-th row, obtain other internetworkings Energy index and the similarity of index i variation tendency.

The method of the assessment network performance index variation tendency of the embodiment of the present invention, by the shifting in data mining Dynamic mean value feedback method, by the network performance index of heterogeneous networks parameter by means such as association, clusters, Grope the variation tendency similarity between index, obtain the performance issue being likely to occur in the network optimization；And with net The moving average of network performance indications is characterized value, carries out data mining, and emphasis is entered with variation tendency as target Row is investigated.

The method of the assessment network performance index variation tendency of the embodiment of the present invention, can be by network performance The analysis of index variation tendency relevance, gropes and finds unknown problem, it is contemplated that solve potential network associate The problems such as performance, fault；Network optimization personnel are in selecting index, index observing, index evaluation process every time In when needing analysis indexes variation tendency, by the scheme of the embodiment of the present invention, you can be rapidly completed, need not Through picture tendency chart and eye-observation, analysis efficiency is substantially increased；According to the solution of the present invention, network is excellent Change personnel can be quickly to all indexs variation tendency be all analyzed, breach dependence conventional Optimization Experience choosing The limitation that limited index is analyzed is taken, accomplishes to analyze network index situation and overall understanding optimization side comprehensively Impact after case enforcement to existing network quality.

Fourth embodiment：

The present invention provides a kind of device of assessment network performance index variation tendency, as shown in figure 5, including：

First acquisition module 51, for obtaining the desired value of multiple network performance indexes of network to be detected；

Second acquisition module 52, for the desired value according to multiple network performance indexes, obtains multiple institutes State the coefficient correlation between the moving average of network performance index；

3rd acquisition module 53, for the phase between the moving average according to multiple network performance indexes Close coefficient, obtain in multiple network performance indexes one variation tendency for presetting network performance index and its The similarity of the variation tendency of his non-default network performance index.

The present invention, can be quickly to a large amount of nets by being analyzed to the similitude of network performance index variation tendency The variation tendency of network performance indications is analyzed, and substantially increases analysis efficiency, breaches dependence previous experiences Choose the limitation that limited index is analyzed, it is ensured that the comprehensive, accuracy of analysis and extensibility.

Wherein, first acquisition module 51 includes：

First acquisition unit, for obtaining multiple network performance indexes of network to be detected respectively in the scheduled time The desired value of the multiple time points in section.

Wherein, the first acquisition unit is divided specifically for N number of network performance index of acquisition network to be detected The desired value of other T time point in the predetermined amount of time, obtains original performance index matrix P：

Wherein, Pi is all of desired value of i-th network performance index, Pi_jExist for i-th network performance index The desired value of j-th time point in the predetermined amount of time, N and T are the integer more than 1, and i is 1 to the integer between N, and j is 1 to the integer between T.

In a particular embodiment of the present invention, above-mentioned first acquisition unit may particularly include data preprocessing module, The data preprocessing module is used for existing network all-network performance indications being actually needed by network optimization personnel, It is processed into the original index matrix in certain particle size and cycle.The input of the module needs to divide for network optimization personnel The geographical granularity (such as cell/region/the whole network etc.) of analysis, predetermined amount of time (such as hour/day/week etc.), with And all-network performance indications sampled value.Geography granularity needed for being output as and the network performance in predetermined amount of time Index original matrix.For example, network optimization personnel wish daily to analyze the network performance index of certain cell, then The all N number of network performance index index sampled values of continuous 30 days of certain cell need to be input into (as per 15 minutes one Individual sampled point), substantial amounts of for each network performance index index sampled value is daily processed by data preprocessing module, The cell all-network performance indications desired value original matrix of continuous 30 days is exported, matrix size is 30 row N Row.

Wherein, second acquisition module 52 includes：

Second acquisition unit, for many in the predetermined amount of time according to each described network performance index The desired value of individual time point, obtains the moving average of each network performance index；

3rd acquiring unit, for the moving average according to each network performance index, obtains multiple Coefficient correlation between the moving average of the network performance index.

In a particular embodiment of the present invention, above-mentioned second acquisition module can be specially rolling average computing module, The rolling average computing module is used for calculating the phase relation of the moving average between all-network performance indications Number.The all-network performance index value original matrix being input into after processing for data preprocessing module, by calculating Coefficient correlation between each column vector of the rolling average matrix of original matrix, the correlation of exportable moving average Matrix.

Wherein, the second acquisition unit refers to specifically for obtaining each described network performance by equation below Target moving average Ai_j：

Wherein, the network to be detected includes N number of network performance index, each described network performance index bag Include the desired value of T time point within a predetermined period of time, Pi_nFor i-th network performance index described pre- The desired value of n-th time point that fixes time in section, Ai_jRepresent j-th movement of i-th network performance index Mean value, i are that j is that mt represents each described network 1 to the integer between T 1 to the integer between N The corresponding time span of the moving average of performance indications, and mt is the integer less than T, N and T are greatly In 1 integer.

Wherein, the 3rd acquiring unit refers to specifically for obtaining multiple network performances by equation below Correlation coefficient r between target moving average_ij：

Wherein, the network to be detected includes N number of network performance index, each described network performance index bag Include the desired value of T time point within a predetermined period of time, r_ijFor i-th network performance index and j-th net Coefficient correlation between network performance indications, Ai_kRepresent k-th moving average of i-th network performance index,Represent the mean value of all desired values of i-th network performance index, N and T is the integer more than 1, k For 1 to the integer between T-1.

Wherein, the 3rd acquisition module 53 is specifically for according to the default network performance index respectively with often The size of the coefficient correlation between the moving average of the individual non-default network performance index, to described default The similarity of the variation tendency of non-default network performance index is ranked up network performance index with each described, Wherein, the bigger non-default network performance index of coefficient correlation is become with the change of the default network performance index The similarity of gesture is bigger, and the less non-default network performance index of coefficient correlation is referred to the default network performance The similarity of target variation tendency is less.

In a particular embodiment of the present invention, above-mentioned 3rd acquisition module 53 may particularly include sort module and comment Estimate module, the sort module is specifically for classifying to network performance index variation tendency.It is input into as network Optimization personnel choose analysis reference index, trend correlation computing module export variation tendency correlation matrix, The classification number of needs, all kinds of index sets needed for being output as.Sorting technique can be using clustering algorithm or people For definition.For example, network optimization personnel wish to be divided into and reference index variation tendency phase network performance index On the contrary and nothing three class of significant relation, sort module is exportable three classes index set to same, trend.

All kinds of indexs that the input of above-mentioned evaluation module is exported for sort module, by all kinds of phase relation of comparison Number, is output as the sequence of all kinds of indexs and reference index variation tendency correlation.For example, network optimization personnel Wish to analyze the index maximum with amplitude of variation in reference index variation tendency identical index set, you can it is right to take Answer the maximum index of coefficient correlation in index set.

The device of the assessment network performance index variation tendency of the embodiment of the present invention, by network performance index The similitude of variation tendency is analyzed, and variation tendency that can be quickly to a large amount of network performance indexes is analyzed, Analysis efficiency is substantially increased, is breached and relies on previous experiences that the limitation that limited index is analyzed is chosen, Ensure that comprehensive, accuracy and the extensibility of analysis.

It should be noted that the device is device corresponding with said method embodiment, said method embodiment In all implementations can reach identical technique effect all suitable for the embodiment of the device, also.

Presently preferred embodiments of the present invention is the foregoing is only, not in order to limit the present invention, all at this Within bright spirit and principle, any modification, equivalent substitution and improvement that is made etc., should be included in this Within bright protection domain.

Claims (14)

1. a kind of assessment network performance index variation tendency method, it is characterised in that include：

Obtain the desired value of multiple network performance indexes of network to be detected；

According to the desired value of multiple network performance indexes, the movement of multiple network performance indexes is obtained Coefficient correlation between mean value；

According to the coefficient correlation between the moving average of multiple network performance indexes, obtain multiple described In network performance index one presets the variation tendency of network performance index and other non-default network performance indexes Variation tendency similarity.

2. method according to claim 1, it is characterised in that acquisition network to be detected multiple The step of desired value of network performance index, includes：

Obtain multiple network performance indexes of the network to be detected multiple times respectively within a predetermined period of time The desired value of point.

3. method according to claim 2, it is characterised in that the acquisition network to be detected The step of desired value of multiple network performance indexes multiple time points respectively within a predetermined period of time, includes：

Obtain N number of network performance index of the network to be detected T respectively in the predetermined amount of time The desired value of time point, obtains original performance index matrix P：

$P=\left[\begin{array}{ccccc}P1& ...& Pi& ...& PN\end{array}\right]=\left[\begin{array}{ccccc}P{1}_1& ...& {\mathrm{Pi}}_1& ...& {\mathrm{PN}}_1\\ .& .& .& .& .\\ .& .& .& .& .\\ .& .& .& .& .\\ P{1}_j& ...& {\mathrm{Pi}}_j& ...& {\mathrm{PN}}_j\\ .& .& .& .& .\\ .& .& .& .& .\\ .& .& .& .& .\\ P{1}_T& ...& {\mathrm{Pi}}_T& ...& {\mathrm{PN}}_T\end{array}\right];$

Wherein, Pi is all of desired value of i-th network performance index, Pi_jExist for i-th network performance index The desired value of j-th time point in the predetermined amount of time, N and T are the integer more than 1, and i is 1 to the integer between N, and j is 1 to the integer between T.

4. method according to claim 2, it is characterised in that described according to multiple network performances The desired value of index, obtains the step of the coefficient correlation between the moving average of multiple network performance indexes Suddenly include：

According to the desired value of multiple time points of each described network performance index in the predetermined amount of time, Obtain the moving average of each network performance index；

According to the moving average of each network performance index, multiple network performance indexes are obtained Coefficient correlation between moving average.

5. method according to claim 4, it is characterised in that described according to each network performance The desired value of multiple time points of the index in the predetermined amount of time, obtains each described network performance index Moving average the step of include：

Moving average Ai of each network performance index is obtained by equation below_j：

${\mathrm{Ai}}_j=\frac{\underset{n=j}{\overset{j+mt-1}{\Σ}}{\mathrm{Pi}}_n}{mt};$

Wherein, the network to be detected includes N number of network performance index, each described network performance index bag Include the desired value of T time point within a predetermined period of time, Pi_nFor i-th network performance index described pre- The desired value of n-th time point that fixes time in section, Ai_jRepresent j-th movement of i-th network performance index Mean value, i are that j is that mt represents each described network 1 to the integer between T 1 to the integer between N The corresponding time span of the moving average of performance indications, and mt is the integer less than T, N and T are greatly In 1 integer.

6. method according to claim 5, it is characterised in that described according to each network performance The moving average of index, obtains the coefficient correlation between the moving average of multiple network performance indexes The step of include：

Correlation coefficient r between the moving average of multiple network performance indexes is obtained by equation below_ij：

$r_{ij}=\frac{\underset{k=1}{\overset{T-1}{\Σ}}({\mathrm{Ai}}_k-\stackrel{\‾}{Ai})({\mathrm{Aj}}_k-\stackrel{\‾}{Aj})}{\sqrt{\underset{k=1}{\overset{T-1}{\Σ}}{({\mathrm{Ai}}_k-\stackrel{\‾}{Ai})}^2}\sqrt{\underset{k=1}{\overset{T-1}{\Σ}}{({\mathrm{Aj}}_k-\stackrel{\‾}{Aj})}^2}};$

Wherein, the network to be detected includes N number of network performance index, each described network performance index bag Include the desired value of T time point within a predetermined period of time, r_ijFor i-th network performance index and j-th net Coefficient correlation between network performance indications, Ai_kRepresent k-th moving average of i-th network performance index,Represent the mean value of all desired values of i-th network performance index, N and T is the integer more than 1, k For 1 to the integer between T-1.

7. method according to claim 1, it is characterised in that described according to multiple network performances Coefficient correlation between the moving average of index, obtained in multiple network performance indexes preset net The step of similarity of the variation tendency of network performance indications and the variation tendency of other network performance indexes, includes：

According to the default network performance index, the movement of non-default network performance index is put down with each described respectively The size of the coefficient correlation between average, to the default network performance index and each described non-default network The similarity of the variation tendency of performance indications is ranked up, wherein, the bigger non-default internetworking of coefficient correlation Can index bigger with the similarity of the variation tendency of the default network performance index, less non-of coefficient correlation Default network performance index is less with the similarity of the variation tendency of the default network performance index.

8. a kind of assessment network performance index variation tendency device, it is characterised in that include：

First acquisition module, for obtaining the desired value of multiple network performance indexes of network to be detected；

Second acquisition module, for the desired value according to multiple network performance indexes, obtains multiple described Coefficient correlation between the moving average of network performance index；

3rd acquisition module, for the correlation between the moving average according to multiple network performance indexes Coefficient, obtain in multiple network performance indexes one variation tendency for presetting network performance index and other The similarity of the variation tendency of non-default network performance index.

9. device according to claim 8, it is characterised in that first acquisition module includes：

First acquisition unit, for obtaining multiple network performance indexes of the network to be detected respectively predetermined The desired value of the multiple time points in the time period.

10. method according to claim 9, it is characterised in that the first acquisition unit is specifically used When N number of network performance index of the network to be detected T in the predetermined amount of time respectively is obtained Between put desired value, obtain original performance index matrix P：

$P=\left[\begin{array}{ccccc}P1& ...& Pi& ...& PN\end{array}\right]=\left[\begin{array}{ccccc}P{1}_1& ...& {\mathrm{Pi}}_1& ...& {\mathrm{PN}}_1\\ .& .& .& .& .\\ .& .& .& .& .\\ .& .& .& .& .\\ P{1}_j& ...& {\mathrm{Pi}}_j& ...& {\mathrm{PN}}_j\\ .& .& .& .& .\\ .& .& .& .& .\\ .& .& .& .& .\\ P{1}_T& ...& {\mathrm{Pi}}_T& ...& {\mathrm{PN}}_T\end{array}\right];$

Wherein, Pi is all of desired value of i-th network performance index, Pi_jExist for i-th network performance index The desired value of j-th time point in the predetermined amount of time, N and T are the integer more than 1, and i is 1 to the integer between N, and j is 1 to the integer between T.

11. devices according to claim 9, it is characterised in that second acquisition module includes：

Second acquisition unit, for many in the predetermined amount of time according to each described network performance index The desired value of individual time point, obtains the moving average of each network performance index；

3rd acquiring unit, for the moving average according to each network performance index, obtains multiple Coefficient correlation between the moving average of the network performance index.

12. devices according to claim 11, it is characterised in that the second acquisition unit is specifically used In moving average Ai for obtaining each network performance index by equation below_j：

${\mathrm{Ai}}_j=\frac{\underset{n=j}{\overset{j+mt-1}{\Σ}}{\mathrm{Pi}}_n}{mt};$

Wherein, the network to be detected includes N number of network performance index, each described network performance index bag Include the desired value of T time point within a predetermined period of time, Pi_nFor i-th network performance index described pre- The desired value of n-th time point that fixes time in section, Ai_jRepresent j-th movement of i-th network performance index Mean value, i are that j is that mt represents each described network 1 to the integer between T 1 to the integer between N The corresponding time span of the moving average of performance indications, and mt is the integer less than T, N and T are greatly In 1 integer.

13. devices according to claim 12, it is characterised in that the 3rd acquiring unit is specifically used Correlation coefficient r between the moving average for obtaining multiple network performance indexes by equation below_ij：

$r_{ij}=\frac{\underset{k=1}{\overset{T-1}{\Σ}}({\mathrm{Ai}}_k-\stackrel{\‾}{Ai})({\mathrm{Aj}}_k-\stackrel{\‾}{Aj})}{\sqrt{\underset{k=1}{\overset{T-1}{\Σ}}{({\mathrm{Ai}}_k-\stackrel{\‾}{Ai})}^2}\sqrt{\underset{k=1}{\overset{T-1}{\Σ}}{({\mathrm{Aj}}_k-\stackrel{\‾}{Aj})}^2}};$

Wherein, the network to be detected includes N number of network performance index, each described network performance index bag Include the desired value of T time point within a predetermined period of time, r_ijFor i-th network performance index and j-th net Coefficient correlation between network performance indications, Ai_kRepresent k-th moving average of i-th network performance index,Represent the mean value of all desired values of i-th network performance index, N and T is the integer more than 1, k For 1 to the integer between T-1.

14. devices according to claim 8, it is characterised in that the 3rd acquisition module is specifically used According to the default network performance index respectively with each described non-default network performance index rolling average The size of the coefficient correlation between value, to the default network performance index and each described non-default internetworking The similarity of the variation tendency of energy index is ranked up, wherein, the bigger non-default network performance of coefficient correlation Index is bigger with the similarity of the variation tendency of the default network performance index, less non-pre- of coefficient correlation If network performance index is less with the similarity of the variation tendency of the default network performance index.