CN107483265A - A kind of network traffics dynamic prediction method and system based on wavelet analysis - Google Patents
A kind of network traffics dynamic prediction method and system based on wavelet analysis Download PDFInfo
- Publication number
- CN107483265A CN107483265A CN201710813447.4A CN201710813447A CN107483265A CN 107483265 A CN107483265 A CN 107483265A CN 201710813447 A CN201710813447 A CN 201710813447A CN 107483265 A CN107483265 A CN 107483265A
- Authority
- CN
- China
- Prior art keywords
- analyzed
- signal
- network traffics
- frequency
- low frequency
- 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.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/147—Network analysis or design for predicting network behaviour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
- G06F17/148—Wavelet transforms
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- General Physics & Mathematics (AREA)
- Computational Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Theoretical Computer Science (AREA)
- Pure & Applied Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
- Telephonic Communication Services (AREA)
Abstract
This application provides a kind of network traffics dynamic prediction method based on wavelet analysis, in network topology structure, the primary data of end to end network data on flows to be analyzed is normalized, obtains the time-domain signal of network traffics to be analyzed；Based on Wavelet Analysis Theory, the time-domain signal is converted into the time-frequency domain signal of network traffics to be analyzed；It is low frequency component, intermediate frequency component and high fdrequency component by the time-frequency domain signal decomposition, and according to the inverse transformation of small echo, obtains corresponding low frequency signal, intermediate-freuqncy signal and high-frequency signal respectively；The flux prediction model of low frequency signal, intermediate-freuqncy signal and high-frequency signal is built respectively, and respectively obtains the prediction result of low frequency signal, intermediate-freuqncy signal and high-frequency signal；The prediction result of low frequency signal, intermediate-freuqncy signal and high-frequency signal is synthesized, obtains the prediction result of network traffics to be analyzed.The complexity for reducing the error of network traffics dynamic prediction algorithm and calculating, improves precision of prediction.
Description
Technical field
The present invention relates to network traffics dynamic prediction technical field end to end, more particularly to a kind of be based on small echo
The network traffics dynamic prediction method and system of analysis.
Background technology
Network traffics are extremely important to network management, traffic engineering, network monitoring, routing optimality and network measure activity.
Network traffics, it is especially useful in end to end network flow, represent the behavioural characteristic of the network user and network equipment activity.Network traffics
Represent between node or original destination (OD) flow between.Especially, OD is to can be from network range or global point
The behavior of network activity is more effectively described.Therefore, network traffics play an important roll in proper network activity.It is however, right
Special Network is difficult to be estimated and predicted due to its height time-varying property in OD business.
Tebaldi et al. proposes a kind of new statistical method, and this method is built based on the statistical model of inversion method
Network source-destination traffic models and prediction network traffics.Tune et al. describes network traffics using theoretical information.
Takeda et al. carrys out estimating peer-to-peer traffic matrix using part metrical information, and proposes corresponding prediction algorithm.Zhang et al.
The prior information of traffic matrix is obtained based on Gravity Models management, then proposes traffic matrix estimation method.Jiang et al. is used
Neutral net is accurately estimated to simulate end-to-end network traffics and obtain.Chen et al. have studied predicting network flow in short-term.
Expectation-maximization algorithm is used to estimate traffic matrix, it is assumed that the business demand between OD pairs is independent of each other and obeys Gauss point
Cloth.The joint time-frequency that Jiang et al. also proposes network traffics using compressive sensing theory is estimated.Some in these methods have
Relatively large evaluated error, some methods are very sensitive to prior information.Therefore, above-mentioned model and method can not be caught exactly
Obtain network traffics.
The content of the invention
In view of this, the invention provides a kind of network traffics dynamic prediction method and system based on wavelet analysis, drop
The error of low network traffics dynamic prediction algorithm and the complexity calculated, improve precision of prediction.
In order to realize foregoing invention purpose, concrete technical scheme provided by the invention is as follows：
A kind of network traffics dynamic prediction method based on wavelet analysis, including：
In network topology structure, place is normalized to the primary data of end to end network data on flows to be analyzed
Reason, obtains the time-domain signal of network traffics to be analyzed；
Based on Wavelet Analysis Theory, the time-domain signal of the network traffics to be analyzed is converted into the network flow to be analyzed
The time-frequency domain signal of amount；
It is low frequency component, intermediate frequency component and high fdrequency component by the time-frequency domain signal decomposition of the network traffics to be analyzed, and
According to the inverse transformation of small echo, low frequency signal, intermediate-freuqncy signal and the high-frequency signal of the network traffics to be analyzed are obtained respectively；
The volume forecasting mould of the low frequency signals of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal is built respectively
Type, and respectively obtain the prediction result of the low frequency signals of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal；
The prediction result of the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal is synthesized,
Obtain the prediction result of the network traffics to be analyzed.
Preferably, it is described in network topology structure, the primary data of end to end network data on flows to be analyzed is entered
Row normalized, the time-domain signal of network traffics to be analyzed is obtained, including：
In network topology structure, the primary data for the end to end network data on flows being analysed to is entered according to preset rules
Row normalized, the time-domain signal of the network traffics to be analyzed is obtained, the time-domain signal value is within a preset range.
Preferably, the inverse transformation according to small echo, low frequency signal, the intermediate frequency of the network traffics to be analyzed are obtained respectively
Signal and high-frequency signal, including：
According to the inverse transformation of small echo, obtain the network traffics to be analyzed the low frequency coefficient of time-frequency domain, intermediate frequency coefficient and
High frequency coefficient；
According to the network traffics to be analyzed in the low frequency coefficient, intermediate frequency coefficient and high frequency coefficient of time-frequency domain, execution small echo
Inverse transformation, obtain low frequency signal, intermediate-freuqncy signal and the high-frequency signal of the network traffics to be analyzed respectively.
Preferably, the low frequency signal for building the network traffics to be analyzed respectively, intermediate-freuqncy signal and high-frequency signal
Flux prediction model, and respectively obtain the prediction of the low frequency signals of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal
As a result, including：
The low frequency signal is modeled using autoregression model, obtains the low frequency signal of the network traffics to be analyzed
Flux prediction model, and obtain the prediction result of the low frequency signal of the network traffics to be analyzed；
The intermediate-freuqncy signal is modeled using periodic signal model, obtains the intermediate frequency letter of the network traffics to be analyzed
Number flux prediction model, and obtain the prediction result of the intermediate-freuqncy signal of the network traffics to be analyzed；
The high-frequency signal is modeled using Gaussian noise model, obtains the high frequency letter of the network traffics to be analyzed
Number flux prediction model, and obtain the prediction result of the high-frequency signal of the network traffics to be analyzed.
Preferably, the prediction knot to the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal
Fruit is synthesized, and obtains the prediction result of the network traffics to be analyzed, including：
The prediction result of the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal is synthesized,
Obtain the time-domain signal of the network traffics to be analyzed；
Inverse normalized is carried out to the time-domain signal of the network traffics to be analyzed, obtains the network traffics to be analyzed
Estimate, and the prediction result using the estimate as the network traffics to be analyzed.
A kind of network traffics Dynamic Forecasting System based on wavelet analysis, including：
Normalized unit, in network topology structure, to end to end network data on flows to be analyzed just
Beginning data are normalized, and obtain the time-domain signal of network traffics to be analyzed；
Conversion unit, for based on Wavelet Analysis Theory, the time-domain signal of the network traffics to be analyzed to be converted into institute
State the time-frequency domain signal of network traffics to be analyzed；
Resolving cell, for being low frequency component, intermediate frequency component by the time-frequency domain signal decomposition of the network traffics to be analyzed
And high fdrequency component, and according to the inverse transformation of small echo, obtain respectively the low frequency signals of the network traffics to be analyzed, intermediate-freuqncy signal and
High-frequency signal；
Forecast model construction unit, for build respectively the low frequency signals of the network traffics to be analyzed, intermediate-freuqncy signal and
The flux prediction model of high-frequency signal, and respectively obtain low frequency signal, intermediate-freuqncy signal and the high frequency of the network traffics to be analyzed
The prediction result of signal；
Synthesis unit, for the prediction to the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal
As a result synthesized, obtain the prediction result of the network traffics to be analyzed.
Preferably, the normalized unit, specifically for the end-to-end net in network topology structure, being analysed to
The primary data of network data on flows is normalized according to preset rules, obtains the time domain of the network traffics to be analyzed
Signal, the time-domain signal value is within a preset range.
Preferably, the resolving cell includes：
Subelement is obtained, for the inverse transformation according to small echo, obtains low frequency of the network traffics to be analyzed in time-frequency domain
Coefficient, intermediate frequency coefficient and high frequency coefficient；
Perform subelement, for according to the network traffics to be analyzed time-frequency domain low frequency coefficient, intermediate frequency coefficient and height
Frequency coefficient, performs the inverse transformation of small echo, obtains low frequency signal, intermediate-freuqncy signal and the high frequency letter of the network traffics to be analyzed respectively
Number.
Preferably, the forecast model construction unit includes：
The forecast model structure subelement of low frequency signal, for being built to the low frequency signal using autoregression model
Mould, the flux prediction model of the low frequency signal of the network traffics to be analyzed is obtained, and obtain the network traffics to be analyzed
The prediction result of low frequency signal；
The forecast model structure subelement of intermediate-freuqncy signal, for being built to the intermediate-freuqncy signal using periodic signal model
Mould, the flux prediction model of the intermediate-freuqncy signal of the network traffics to be analyzed is obtained, and obtain the network traffics to be analyzed
The prediction result of intermediate-freuqncy signal；
The forecast model structure subelement of high-frequency signal, for being built to the high-frequency signal using Gaussian noise model
Mould, the flux prediction model of the high-frequency signal of the network traffics to be analyzed is obtained, and obtain the network traffics to be analyzed
The prediction result of high-frequency signal.
Preferably, the synthesis unit includes：
Subelement is synthesized, for the pre- of the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal
Survey result to be synthesized, obtain the time-domain signal of the network traffics to be analyzed；
Inverse normalized subelement, for being carried out to the time-domain signal of the network traffics to be analyzed at inverse normalization
Reason, obtains the estimate of the network traffics to be analyzed, and the prediction using the estimate as the network traffics to be analyzed
As a result.
It is as follows relative to prior art, beneficial effects of the present invention：
A kind of network traffics dynamic prediction method and system based on wavelet analysis provided by the invention, first, it will treat point
The time-domain signal of analysis network traffics is converted to time-frequency domain signal, and the intrinsic category of network traffics is characterized using Time-Frequency Analysis Method
Property.Then, in time-frequency domain, by the time-frequency domain signal decomposition of the network traffics to be analyzed be low frequency component, intermediate frequency component and
High fdrequency component, makes the variation tendency of low frequency part reflection network traffics, and intermediate-frequency section reflects the slow change of network traffics, radio-frequency head
Divide the quick change of reflection network traffics.Finally, build respectively the low frequency signals of the network traffics to be analyzed, intermediate-freuqncy signal and
The flux prediction model of high-frequency signal, and respectively obtain low frequency signal, intermediate-freuqncy signal and the high frequency of the network traffics to be analyzed
The prediction result of signal；The prediction result of the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal is entered
Row synthesis, obtain the prediction result of the network traffics to be analyzed.Relative to prior art, the present invention is not appointed using network traffics
The Statistical Distribution Characteristics of what prior information, but the inherent temporal correlation of network traffics is extracted by Time-Frequency Analysis, from
And can effectively avoid in existing statistical analysis technique to the intrinsic problem of network traffics prior information sensitiveness, so as to obtain essence
True predicting network flow result.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
There is the required accompanying drawing used in technology description to be briefly described, it should be apparent that, drawings in the following description are only this
The embodiment of invention, for those of ordinary skill in the art, on the premise of not paying creative work, can also basis
The accompanying drawing of offer obtains other accompanying drawings.
Fig. 1 is a kind of network traffics dynamic prediction method flow chart based on wavelet analysis disclosed in the embodiment of the present invention；
Fig. 2 is that a kind of structure of the network traffics Dynamic Forecasting System based on wavelet analysis disclosed in the embodiment of the present invention is shown
It is intended to.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, rather than whole embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other under the premise of creative work is not made
Embodiment, belong to the scope of protection of the invention.
Present embodiment discloses a kind of network traffics dynamic prediction method based on wavelet analysis, referring to Fig. 1, Fig. 1 is
The network traffics dynamic prediction method flow chart, is specifically included：
S101：In network topology structure, normalizing is carried out to the primary data of end to end network data on flows to be analyzed
Change is handled, and obtains the time-domain signal of network traffics to be analyzed；
In network topology structure, the primary data for the end to end network data on flows being analysed to is entered according to preset rules
Row normalized, the time-domain signal of the network traffics to be analyzed is obtained, the time-domain signal value is within a preset range.Subtract
The order of magnitude of small data, reduce the complexity that prediction algorithm calculates.
Preferably, the primary data of end to end network data on flows is x=(x (1), x (2) ..., x (m)), and it is carried out
Normalized obtains：
Wherein, x (t) be network traffics x in moment t value, -1≤zt (≤), 1 and
The time-domain signal of network traffics to be analyzed is z={ z (1), z (2) ..., z (m) }.
So, be advantageous to the setting of following prediction model parameterses, reduce the computation complexity of forecast model, to make
The result of prediction is closer to actual value, the precision of forecast model is further improved.
S102：Based on Wavelet Analysis Theory, the time-domain signal of the network traffics to be analyzed is converted into described to be analyzed
The time-frequency domain signal of network traffics；
Preferably, according to Wavelet Analysis Theory, the time-domain signal of network traffics to be analyzed is z={ z (1), z (2) ..., z
(m) the time-frequency domain signal } being converted into formula (2)：
Wherein, H is high-pass filter, and G represents low pass filter, ai(j+1)And di(j+1)It is z time-frequency coefficients respectively, I tables
Show the Decomposition order of wavelet transformation.
S103：It is low frequency component, intermediate frequency component and high frequency division by the time-frequency domain signal decomposition of the network traffics to be analyzed
Amount, and according to the inverse transformation of small echo, low frequency signal, intermediate-freuqncy signal and the high frequency for obtaining the network traffics to be analyzed respectively are believed
Number；
According to the inverse transformation of small echo, obtain the network traffics to be analyzed the low frequency coefficient of time-frequency domain, intermediate frequency coefficient and
High frequency coefficient；
According to the network traffics to be analyzed in the low frequency coefficient, intermediate frequency coefficient and high frequency coefficient of time-frequency domain, execution small echo
Inverse transformation, obtain low frequency signal, intermediate-freuqncy signal and the high-frequency signal of the network traffics to be analyzed respectively.
Preferably, it is low frequency component, middle frequency division by the time-frequency domain signal decomposition of the network traffics to be analyzed according to formula (2)
Amount and high fdrequency component, such as formula (3)：
Wl(f,t)、Wm(f,t)、Wh(f, t) each corresponds to the low frequency in time-frequency domain, intermediate frequency and high fdrequency component, W (f, t) generation
Time frequency signals of the table network traffics z in time t.
According to the inverse transformation of small echo, the low frequency coefficient that can obtain z in time-frequency domain is
Wl(f, t)=(aI,0,0,...,0) (4)
Intermediate frequency coefficient is
Wm(f, t)=(0, d1,d2,...,dk,0,...,0) (5)
High frequency coefficient is
Wh(f, t)=(0,0 ..., 0, dk+1,dk+2,...,dI) (6)
Then below equation can be obtained：
Wherein, zl(t), zm(t), zh(t) low frequency signal Ws of the network flow z (t) in time t is represented respectivelyl(f, t), Wm(f,
T), WhThe reconstruction signal of (f, t).
S104：The flow for building the low frequency signals of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal respectively is pre-
Model is surveyed, and respectively obtains the prediction result of the low frequency signals of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal；
The low frequency signal is modeled using autoregression model, obtains the low frequency signal of the network traffics to be analyzed
Flux prediction model, and obtain the prediction result of the low frequency signal of the network traffics to be analyzed；
Preferably, to low-frequency time-domain signal zl={ zl(1),zl(2),...,zl(m) } built using autoregression model
Mould：
zl(t)=b1zl(t-1)+b2zl(t-2)+...+bpzl(t-p)+ε (8)
Wherein, p is the exponent number of autoregression model, and ε represents error, b={ b1,b2,...,bpBe the low frequency signal stream
Measure the parameter of forecast model.Parameter b={ b1,b2,...,bpCan be determined by signal analysis.
The intermediate-freuqncy signal is modeled using periodic signal model, obtains the intermediate frequency letter of the network traffics to be analyzed
Number flux prediction model, and obtain the prediction result of the intermediate-freuqncy signal of the network traffics to be analyzed；
Preferably, to intermediate frequency time-domain signal zm={ zm(1),zm(2),...,zm(m) } built using periodic signal model
Mould：
zm(t)=usin (ω t+ α)+vsin (ω t+ β)+η (9)
Wherein, u and v is the coefficient of model；ω, α, β represent the frequency and initial phase of periodic signal respectively, and η represents mould
The error of type.The model of formula (9) can obtain in signal analysis theory.
The high-frequency signal is modeled using Gaussian noise model, obtains the high frequency letter of the network traffics to be analyzed
Number flux prediction model, and obtain the prediction result of the high-frequency signal of the network traffics to be analyzed.
Preferably, to high frequency time-domain signal zh={ zh(1),zh(2),...,zh(m) } built using Gaussian noise model
Mould：
Wherein, δ and μ represents the variance and average value of Gaussian noise, and λ represents error.The model of formula (10) can be in signal
Obtained in analysis theories.
S105：The prediction result of the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal is carried out
Synthesis, obtain the prediction result of the network traffics to be analyzed.
The prediction result of the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal is synthesized,
Obtain the time-domain signal of the network traffics to be analyzed；
Specifically, being synthesized to low frequency signal, intermediate-freuqncy signal and the high-frequency signal of the network traffics to be analyzed, obtain
The time-domain signal of the network traffics to be analyzed：
Inverse normalized is carried out to the time-domain signal of the network traffics to be analyzed, obtains the network traffics to be analyzed
Estimate, and the prediction result using the estimate as the network traffics to be analyzed.
Finally, according to formula (1), the estimation formulas of end-to-end flux is obtained：
A kind of network traffics dynamic prediction method based on wavelet analysis that the present embodiment provides, first, is analysed to net
The time-domain signal of network flow is converted to time-frequency domain signal, and the build-in attribute of network traffics is characterized using Time-Frequency Analysis Method.So
Afterwards, it is low frequency component, intermediate frequency component and high frequency division by the time-frequency domain signal decomposition of the network traffics to be analyzed in time-frequency domain
Amount, makes the variation tendency of low frequency part reflection network traffics, and intermediate-frequency section reflects the slow change of network traffics, HFS reflection
The quick change of network traffics.Finally, low frequency signal, intermediate-freuqncy signal and the high frequency letter of the network traffics to be analyzed are built respectively
Number flux prediction model, and respectively obtain the low frequency signals of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal
Prediction result；The prediction result of the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal is synthesized,
Obtain the prediction result of the network traffics to be analyzed.Relative to prior art, the present invention does not utilize any priori of network traffics
The Statistical Distribution Characteristics of information, but the inherent temporal correlation of network traffics is extracted by Time-Frequency Analysis, so as to have
Effect is avoided to the intrinsic problem of network traffics prior information sensitiveness in existing statistical analysis technique, so as to obtain accurate net
Network volume forecasting result.
Based on a kind of network traffics dynamic prediction method based on wavelet analysis, the present embodiment pair disclosed in above-described embodiment
A kind of network traffics Dynamic Forecasting System based on wavelet analysis should be disclosed, referring to Fig. 2, the network traffics dynamic prediction
System includes：
Normalized unit 101, in network topology structure, to end to end network data on flows to be analyzed
Primary data is normalized, and obtains the time-domain signal of network traffics to be analyzed；
Conversion unit 102, for based on Wavelet Analysis Theory, the time-domain signal of the network traffics to be analyzed to be converted into
The time-frequency domain signal of the network traffics to be analyzed；
Resolving cell 103, for being low frequency component, middle frequency division by the time-frequency domain signal decomposition of the network traffics to be analyzed
Amount and high fdrequency component, and according to the inverse transformation of small echo, low frequency signal, the intermediate-freuqncy signal of the network traffics to be analyzed are obtained respectively
And high-frequency signal；
Forecast model construction unit 104, for building low frequency signal, the intermediate-freuqncy signal of the network traffics to be analyzed respectively
With the flux prediction model of high-frequency signal, and low frequency signal, intermediate-freuqncy signal and the height of the network traffics to be analyzed are respectively obtained
The prediction result of frequency signal；
Synthesis unit 105, for the pre- of the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal
Survey result to be synthesized, obtain the prediction result of the network traffics to be analyzed.
Preferably, the normalized unit 101, is arrived specifically for the end in network topology structure, being analysed to
The primary data of end network flow data is normalized according to preset rules, obtains the network traffics to be analyzed
Time-domain signal, the time-domain signal value is within a preset range.
Preferably, the resolving cell 103 includes：
Subelement is obtained, for the inverse transformation according to small echo, obtains low frequency of the network traffics to be analyzed in time-frequency domain
Coefficient, intermediate frequency coefficient and high frequency coefficient；
Perform subelement, for according to the network traffics to be analyzed time-frequency domain low frequency coefficient, intermediate frequency coefficient and height
Frequency coefficient, performs the inverse transformation of small echo, obtains low frequency signal, intermediate-freuqncy signal and the high frequency letter of the network traffics to be analyzed respectively
Number.
Preferably, the forecast model construction unit 104 includes：
The forecast model structure subelement of low frequency signal, for being built to the low frequency signal using autoregression model
Mould, the flux prediction model of the low frequency signal of the network traffics to be analyzed is obtained, and obtain the network traffics to be analyzed
The prediction result of low frequency signal；
The forecast model structure subelement of intermediate-freuqncy signal, for being built to the intermediate-freuqncy signal using periodic signal model
Mould, the flux prediction model of the intermediate-freuqncy signal of the network traffics to be analyzed is obtained, and obtain the network traffics to be analyzed
The prediction result of intermediate-freuqncy signal；
The forecast model structure subelement of high-frequency signal, for being built to the high-frequency signal using Gaussian noise model
Mould, the flux prediction model of the high-frequency signal of the network traffics to be analyzed is obtained, and obtain the network traffics to be analyzed
The prediction result of high-frequency signal.
Preferably, the synthesis unit 105 includes：
Subelement is synthesized, for the pre- of the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal
Survey result to be synthesized, obtain the time-domain signal of the network traffics to be analyzed；
Inverse normalized subelement, for being carried out to the time-domain signal of the network traffics to be analyzed at inverse normalization
Reason, obtains the estimate of the network traffics to be analyzed, and the prediction using the estimate as the network traffics to be analyzed
As a result.
A kind of network traffics Dynamic Forecasting System based on wavelet analysis that the present embodiment provides, first, is analysed to net
The time-domain signal of network flow is converted to time-frequency domain signal, and the build-in attribute of network traffics is characterized using Time-Frequency Analysis Method.So
Afterwards, it is low frequency component, intermediate frequency component and high frequency division by the time-frequency domain signal decomposition of the network traffics to be analyzed in time-frequency domain
Amount, makes the variation tendency of low frequency part reflection network traffics, and intermediate-frequency section reflects the slow change of network traffics, HFS reflection
The quick change of network traffics.Finally, low frequency signal, intermediate-freuqncy signal and the high frequency letter of the network traffics to be analyzed are built respectively
Number flux prediction model, and respectively obtain the low frequency signals of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal
Prediction result；The prediction result of the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal is synthesized,
Obtain the prediction result of the network traffics to be analyzed.Relative to prior art, the present invention does not utilize any priori of network traffics
The Statistical Distribution Characteristics of information, but the inherent temporal correlation of network traffics is extracted by Time-Frequency Analysis, so as to have
Effect is avoided to the intrinsic problem of network traffics prior information sensitiveness in existing statistical analysis technique, so as to obtain accurate net
Network volume forecasting result.
The foregoing description of the disclosed embodiments, professional and technical personnel in the field are enable to realize or using the present invention.
A variety of modifications to these embodiments will be apparent for those skilled in the art, as defined herein
General Principle can be realized in other embodiments without departing from the spirit or scope of the present invention.Therefore, it is of the invention
The embodiments shown herein is not intended to be limited to, and is to fit to and principles disclosed herein and features of novelty phase one
The most wide scope caused.
Claims (10)
- A kind of 1. network traffics dynamic prediction method based on wavelet analysis, it is characterised in that including：In network topology structure, the primary data of end to end network data on flows to be analyzed is normalized, obtained To the time-domain signal of network traffics to be analyzed；Based on Wavelet Analysis Theory, the time-domain signal of the network traffics to be analyzed is converted into the network traffics to be analyzed Time-frequency domain signal；It is low frequency component, intermediate frequency component and high fdrequency component by the time-frequency domain signal decomposition of the network traffics to be analyzed, and according to The inverse transformation of small echo, low frequency signal, intermediate-freuqncy signal and the high-frequency signal of the network traffics to be analyzed are obtained respectively；The flux prediction model of the low frequency signals of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal is built respectively, and Respectively obtain the prediction result of the low frequency signals of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal；The prediction result of the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal is synthesized, obtained The prediction result of the network traffics to be analyzed.
- 2. according to the method for claim 1, it is characterised in that it is described in network topology structure, end to be analyzed is arrived The primary data of end network flow data is normalized, and obtains the time-domain signal of network traffics to be analyzed, including：In network topology structure, the primary data for the end to end network data on flows being analysed to is returned according to preset rules One change is handled, and obtains the time-domain signal of the network traffics to be analyzed, the time-domain signal value is within a preset range.
- 3. according to the method for claim 1, it is characterised in that the inverse transformation according to small echo, treated respectively described in acquisition Low frequency signal, intermediate-freuqncy signal and the high-frequency signal of network traffics are analyzed, including：According to the inverse transformation of small echo, low frequency coefficient, intermediate frequency coefficient and high frequency of the network traffics to be analyzed in time-frequency domain are obtained Coefficient；According to the network traffics to be analyzed in the low frequency coefficient, intermediate frequency coefficient and high frequency coefficient of time-frequency domain, the inverse of small echo is performed Conversion, low frequency signal, intermediate-freuqncy signal and the high-frequency signal of the network traffics to be analyzed are obtained respectively.
- 4. according to the method for claim 1, it is characterised in that the low frequency for building the network traffics to be analyzed respectively The flux prediction model of signal, intermediate-freuqncy signal and high-frequency signal, and respectively obtain the network traffics to be analyzed low frequency signal, The prediction result of intermediate-freuqncy signal and high-frequency signal, including：The low frequency signal is modeled using autoregression model, obtains the stream of the low frequency signal of the network traffics to be analyzed Forecast model is measured, and obtains the prediction result of the low frequency signal of the network traffics to be analyzed；The intermediate-freuqncy signal is modeled using periodic signal model, obtains the intermediate-freuqncy signal of the network traffics to be analyzed Flux prediction model, and obtain the prediction result of the intermediate-freuqncy signal of the network traffics to be analyzed；The high-frequency signal is modeled using Gaussian noise model, obtains the high-frequency signal of the network traffics to be analyzed Flux prediction model, and obtain the prediction result of the high-frequency signal of the network traffics to be analyzed.
- 5. according to the method for claim 1, it is characterised in that the low frequency signal to the network traffics to be analyzed, Intermediate-freuqncy signal and the prediction result of high-frequency signal are synthesized, and obtain the prediction result of the network traffics to be analyzed, including：The prediction result of the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal is synthesized, obtained The time-domain signal of the network traffics to be analyzed；Inverse normalized is carried out to the time-domain signal of the network traffics to be analyzed, obtains estimating for the network traffics to be analyzed Evaluation, and the prediction result using the estimate as the network traffics to be analyzed.
- A kind of 6. network traffics Dynamic Forecasting System based on wavelet analysis, it is characterised in that including：Normalized unit, in network topology structure, to the initial number of end to end network data on flows to be analyzed According to being normalized, the time-domain signal of network traffics to be analyzed is obtained；Conversion unit, for based on Wavelet Analysis Theory, the time-domain signal of the network traffics to be analyzed to be converted into described treat Analyze the time-frequency domain signal of network traffics；Resolving cell, for being low frequency component, intermediate frequency component and height by the time-frequency domain signal decomposition of the network traffics to be analyzed Frequency component, and according to the inverse transformation of small echo, low frequency signal, intermediate-freuqncy signal and the high frequency of the network traffics to be analyzed are obtained respectively Signal；Forecast model construction unit, for building low frequency signal, intermediate-freuqncy signal and the high frequency of the network traffics to be analyzed respectively The flux prediction model of signal, and respectively obtain low frequency signal, intermediate-freuqncy signal and the high-frequency signal of the network traffics to be analyzed Prediction result；Synthesis unit, for the prediction result to the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal Synthesized, obtain the prediction result of the network traffics to be analyzed.
- 7. system according to claim 6, it is characterised in that the normalized unit, specifically for being opened up in network Flutter in structure, the primary data for the end to end network data on flows being analysed to is normalized according to preset rules, obtains To the time-domain signal of the network traffics to be analyzed, the time-domain signal value is within a preset range.
- 8. system according to claim 6, it is characterised in that the resolving cell includes：Obtain subelement, for the inverse transformation according to small echo, obtain the network traffics to be analyzed time-frequency domain low frequency coefficient, Intermediate frequency coefficient and high frequency coefficient；Perform subelement, for according to the network traffics to be analyzed in the low frequency coefficient of time-frequency domain, intermediate frequency coefficient and high frequency system Number, performs the inverse transformation of small echo, obtains low frequency signal, intermediate-freuqncy signal and the high-frequency signal of the network traffics to be analyzed respectively.
- 9. system according to claim 6, it is characterised in that the forecast model construction unit includes：The forecast model structure subelement of low frequency signal, for being modeled to the low frequency signal using autoregression model, is obtained To the flux prediction model of the low frequency signal of the network traffics to be analyzed, and obtain the low frequency letter of the network traffics to be analyzed Number prediction result；The forecast model structure subelement of intermediate-freuqncy signal, for being modeled to the intermediate-freuqncy signal using periodic signal model, The flux prediction model of the intermediate-freuqncy signal of the network traffics to be analyzed is obtained, and obtains the intermediate frequency of the network traffics to be analyzed The prediction result of signal；The forecast model structure subelement of high-frequency signal, for being modeled to the high-frequency signal using Gaussian noise model, The flux prediction model of the high-frequency signal of the network traffics to be analyzed is obtained, and obtains the high frequency of the network traffics to be analyzed The prediction result of signal.
- 10. system according to claim 6, it is characterised in that the synthesis unit includes：Subelement is synthesized, for the prediction knot to the low frequency signal of the network traffics to be analyzed, intermediate-freuqncy signal and high-frequency signal Fruit is synthesized, and obtains the time-domain signal of the network traffics to be analyzed；Inverse normalized subelement, for carrying out inverse normalized to the time-domain signal of the network traffics to be analyzed, obtain To the estimate of the network traffics to be analyzed, and the prediction result using the estimate as the network traffics to be analyzed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710813447.4A CN107483265A (en) | 2017-09-11 | 2017-09-11 | A kind of network traffics dynamic prediction method and system based on wavelet analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710813447.4A CN107483265A (en) | 2017-09-11 | 2017-09-11 | A kind of network traffics dynamic prediction method and system based on wavelet analysis |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107483265A true CN107483265A (en) | 2017-12-15 |
Family
ID=60583839
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710813447.4A Pending CN107483265A (en) | 2017-09-11 | 2017-09-11 | A kind of network traffics dynamic prediction method and system based on wavelet analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107483265A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110839253A (en) * | 2019-11-08 | 2020-02-25 | 西北工业大学青岛研究院 | Method for determining wireless grid network flow |
| CN112994921A (en) * | 2019-12-17 | 2021-06-18 | 华为数字技术（苏州）有限公司 | Flow prediction method and related device |
| CN113141318A (en) * | 2021-03-29 | 2021-07-20 | 重庆邮电大学 | Improved TFRC congestion control system and method based on video traffic prediction model |
| CN115102869A (en) * | 2022-05-31 | 2022-09-23 | 上海云从汇临人工智能科技有限公司 | Network flow prediction method, system and computer storage medium |
| WO2022268030A1 (en) * | 2021-06-24 | 2022-12-29 | 中兴通讯股份有限公司 | Traffic prediction processing method, device and system, storage medium, and electronic device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101345664A (en) * | 2008-08-05 | 2009-01-14 | 成都市华为赛门铁克科技有限公司 | Detection method and apparatus for network flux exception |
| WO2011157781A1 (en) * | 2010-06-17 | 2011-12-22 | Step Sensortechnik Und Elektronik Pockau Gmbh | Method for ion mobility spectrometry |
| CN104168131A (en) * | 2014-06-05 | 2014-11-26 | 国家电网公司 | Flow generation method of power dispatching exchange network based on multicast communication |
-
2017
- 2017-09-11 CN CN201710813447.4A patent/CN107483265A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101345664A (en) * | 2008-08-05 | 2009-01-14 | 成都市华为赛门铁克科技有限公司 | Detection method and apparatus for network flux exception |
| WO2011157781A1 (en) * | 2010-06-17 | 2011-12-22 | Step Sensortechnik Und Elektronik Pockau Gmbh | Method for ion mobility spectrometry |
| CN104168131A (en) * | 2014-06-05 | 2014-11-26 | 国家电网公司 | Flow generation method of power dispatching exchange network based on multicast communication |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110839253A (en) * | 2019-11-08 | 2020-02-25 | 西北工业大学青岛研究院 | Method for determining wireless grid network flow |
| CN112994921A (en) * | 2019-12-17 | 2021-06-18 | 华为数字技术（苏州）有限公司 | Flow prediction method and related device |
| CN113141318A (en) * | 2021-03-29 | 2021-07-20 | 重庆邮电大学 | Improved TFRC congestion control system and method based on video traffic prediction model |
| WO2022268030A1 (en) * | 2021-06-24 | 2022-12-29 | 中兴通讯股份有限公司 | Traffic prediction processing method, device and system, storage medium, and electronic device |
| CN115102869A (en) * | 2022-05-31 | 2022-09-23 | 上海云从汇临人工智能科技有限公司 | Network flow prediction method, system and computer storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107483265A (en) | A kind of network traffics dynamic prediction method and system based on wavelet analysis | |
| Kisi | Wavelet regression model for short-term streamflow forecasting | |
| Olyaie et al. | A comparison of various artificial intelligence approaches performance for estimating suspended sediment load of river systems: a case study in United States | |
| Willems | A spatial rainfall generator for small spatial scales | |
| Adamowski | Development of a short-term river flood forecasting method for snowmelt driven floods based on wavelet and cross-wavelet analysis | |
| Lu et al. | Geographically weighted regression using a non-Euclidean distance metric with a study on London house price data | |
| Pechlivanidis et al. | Calibration of the semi-distributed PDM rainfall–runoff model in the Upper Lee catchment, UK | |
| CN110958639A (en) | Target state estimation method and system | |
| CN104168131A (en) | Flow generation method of power dispatching exchange network based on multicast communication | |
| CN103840988A (en) | Network traffic measurement method based on RBF neural network | |
| Sadek et al. | Multi-scale high-speed network traffic prediction using k-factor Gegenbauer ARMA model | |
| Nowak et al. | Wavelet Auto-Regressive Method (WARM) for multi-site streamflow simulation of data with non-stationary spectra | |
| Maheswaran et al. | Wavelets-based non-linear model for real-time daily flow forecasting in Krishna River | |
| Meshgi et al. | An empirical method for approximating stream baseflow time series using groundwater table fluctuations | |
| CN103268525A (en) | WD-RBF (wavelet denoising-radial basis function)-based analogue prediction method of hydrological time sequence | |
| CN109508788A (en) | A kind of SDN method for predicting based on arma modeling | |
| CN116029459B (en) | Extraction method of TMGCN traffic flow prediction model combined with graph Fourier transform | |
| Ahn | Modeling of groundwater heads based on second-order difference time series models | |
| CN106972949A (en) | A kind of fractional order network system situation method of estimation based on adaptive equalization technology | |
| Sahay et al. | Wavelet regression models for predicting flood stages in rivers: a case study in E astern I ndia | |
| Bulygina et al. | A comparison of rainfall-runoff modelling approaches for estimating impacts of rural land management on flood flows | |
| CN106874676A (en) | A kind of electric power meter state evaluating method | |
| CN103607292A (en) | Fast distributed monitoring method for electric-power communication network services | |
| Padmanabhan et al. | Maximum entropy spectral analysis of hydrologic data | |
| Rathinasamy et al. | Multiscale nonlinear model for monthly streamflow forecasting: a wavelet-based approach |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171215 |
|
| RJ01 | Rejection of invention patent application after publication |