CN103607292B - Fast distributed monitoring method for electric-power communication network services - Google Patents

Abstract

The invention relates to a fast distributed monitoring method for electric-power communication network services. The method includes the following steps: generating a Bernoulli matrix and selecting OD streams which need to be measured directly through the matrix and constructing a traffic matrix; obtaining a measurement matrix; constructing an optimization greedy adaptive dictionary; and reconstructing the traffic matrix through compression sensing. In the fast distributed monitoring method for the electric-power communication-network services, part of OD streams, which need to be measured directly, are selected through a random matrix according power-law distribution of electric-power communication-network traffic and a traffic acquisition function is operated in part of routers as needed so as to obtain traffic information of the OD streams. An end-to-end network traffic reconstruction model is constructed through the traffic information and the model is solved through use of a compression-sensing reconstruction algorithm so as to obtain monitoring values of the traffic. The method is capable of obtaining the monitoring values of the traffic accurately in a real-time manner and at the same time reducing network traffic acquisition loss effectively.

Description

Fast distributed monitoring method towards power communication network service

Technical field

The present invention relates to the measurement of large scale network end-to-end flux and analysis field, more particularly to towards power telecom network The fast distributed monitoring method of network business.

Background technology

With the fast development of powerline network, network carries for the production practices such as dispatching telephone, relay protection, automation The basic communication technology has been supplied to support.In order to realize the intellectuality of electrical network, the access terminal of power telecom network and the business that carries become Obtain diversified.This is to network traffic engineering pipes such as the network management of power telecom network, network monitoring, network design and the network plannings Reason is put forward higher requirement.End to end network flow (i.e. source-purpose stream or od stream) monitoring is network management, traffic engineering etc. The important |input paramete of operation and implementation basis, therefore end to end network flow monitoring has obtained extensive concern.

Traditional network monitor technology can realize the real-time reproduction and prediction to large scale backbone network flow, and passes through Traffic matrix describes state of network traffic.Network Traffic Monitoring technology can be divided into two big class, respectively direct measurement and flow to estimate Meter.With respect to flow estimation method, direct measurement can more accurately describe network traffics dynamic change situation.Therefore network Equipment supplier is each provided with network traffics acquisition function (for example, the netflow of cisco router) in the network device.As Fig. 1, traditional Network Traffic Monitoring method is passed through operating flux acquisition function on each router and is collected flow status information, And NMS is sent to by backbone network.This is the most straightforward approach obtaining flow status information.But run stream Amount acquisition function takies cpu and the memory source of router, therefore reduces the storage transfer capability of router, this external flux shape The transmission of state information adds additional the load of network again.In sum, the method substantially increases via net loss, therefore exists Apply less in practice.

By the flow collection function of network management stand control router, build distributed flow monitoring system collection portion Divide end to end network flow status information, and all of end to end network flow monitoring value can be obtained by flow reconstruction method, As shown in Figure 2.Reduce network operation energy consumption and offered load by reducing the number of end to end network flow measured directly. How to determine needs part end to end network flow measured directly, and how to reconstruct all of flow shape in NMS State information is to build the subject matter facing during distributed network flow monitoring system.Existing method is to optimize operation to adopt mostly The distribution of collection power router, on minimum router, operating flux acquisition function measures most end to end network flow Information.Although this method can significantly reduce via net loss, the flow information of meeting lost part.Therefore study distributed Flow monitoring system obtains whole network traffic informations and has important Research Significance.

Content of the invention

The shortcoming existing for prior art, solves the problems, such as end to end network flow real-time monitoring, present invention offer towards The fast distributed monitoring method of power communication network service.The method is selected according to Bernoulli Jacob's calculation matrix and power-law distribution characteristic Select partly od flow measured directly；Flowed according to these od, set up linear inference problem, then, using the greedy self adaptation of optimization Dictionary (ogad) learning algorithm, makes linear inference problem observe compressed sensing technical requirements.Finally, reconstructed using compressed sensing and calculate Method recovers network traffics end to end.

In order to realize the purpose of the present invention, the technical solution used in the present invention is:

Towards the fast distributed monitoring method of power communication network service, comprise the steps:

Step 1: generate bernoulli matrix, select to need od stream measured directly by this matrix, and build flow square Battle array；

Step 2: obtain observing matrix；

The construction of observing matrix depends on traffic matrix x_partIn non-zero row, the row of other neutral elements represents unknown need Od stream to be reconstructed.Calculating observation matrix y_mMethod is as follows:

y_m=b x_part(2)

Wherein, b is the bernoulli matrix of m × n, x_partIt is the traffic matrix of n × t.Then by b, x_partWith observing matrix y_mShape Become a linear system；

Step 3: build and optimize greedy self-adapting dictionary；

Step 4: traffic matrix is reconstructed by compressed sensing.

Described step 1 specifically includes following steps:

Step 1-1: generate bernoulli matrix；

Generate the bernoulli random matrix b of m × n (m ＜ n), n is the number of od stream in network, and it is equal to network node Quantity square.Bernoulli entry of a matrix element b (m, n) is independent identically distributed, and the probability that element is equal to 1 is pr, equal to 0 Probability be 1-pr.

Step 1-2: determine the number of the od stream needing measurement；

The each row of bernoulli matrix of m × n are carried out respectively with boolean ' or ' computing,Make s=[s (1),s(2),...,s(n)]^tFor a column vector, then the od flow amount measuring is needed to be l=| | s | |₁, | | | |₁Represent l₁ Norm.

Step 1-3: direct measurement od stream；

Historical traffic matrix x known to calculating₀Every od stream average:

$\mathrm{aver}\_x\left(n\right)=\underset{t=1}{\overset{t_0}{\mathrm{\σ}}}{x}_0(n,t)n=\mathrm{1,2},...,n---\left(1\right)$

Wherein, t₀For the length of historical traffic matrix, n is od flow amount.According to this average, network management stand control route The on off state of flow acquisition function on device, to reach l=| | s | | of measurement maximum₁The purpose of individual od stream.L by measurement =| | s | |₁Individual od stream is designated as gathering { x_mea(l) }, l=1,2 ..., l；

Step 1-4: { x is flowed according to known od_mea(l) }, l=1,2 ...., l builds traffic matrix；

Described step 1-4 specifically includes following steps:

Step 1-4-1: generate bernoulli random matrix b, and its each row is entered respectively according to step 1-1, step 1-2 Row boolean ' or ' computing；

Step 1-4-2: initialization flow moment matrix is empty matrix, that is,Make l=1, iterations j=1, maximum changes For frequency n；

Step 1-4-3: when s (j)=1, traffic matrix is changed intoOtherwise,

Step 1-4-4: iterations j adds 1, if j is ＜ n, return to step 1-4-3), till iteration n time.Obtain Traffic matrix x_part,

Step 1-4-5: traffic matrix builds and terminates.

Described step 3 specifically includes following steps:

Step 3-1: the digital dictionary d of initialization₀For sky, i.e. d₀=[]；

Step 3-2: singular value decomposition is carried out to historical traffic, and extracts k principal component；

Step 3-3: setting redundancyFor k principal component of historical traffic, and determine iterations iter= N, makes j=1,

Step 3-4: calculate r_jSparse index ξ_a, and corresponding column index a of sparse index_j；

$a_j=\arg \underset{a\&notelement;i_j}{\min }\{{\mathrm{\ξ}}_a={\left|\right|r_j\left(a\right)\left|\right|}_1/{\left|\right|r_j\left(a\right)\left|\right|}_2\}---\left(5\right)$

Wherein r_jThe a row of (a) redundancy for obtaining after iteration j, | | | |₂Represent l₂Norm；

Step 3-5: obtain the row r with minimum sparse index according to step 3-4_j(a_j), and will there is minimum in redundancy Sparse index ξ_aA_jRow are unitization, and it is set to the jth row of data dictionary, i.e. d_j=r_j(a_j)/||r_j(a_j)||₂；

Step 3-6: obtaining data dictionary according to step 3-5 is: d_j=[d_j-1|d_j], i_j=i_j-1∪{a_j}；Then to institute There is row a；

Step 3-7: update redundancy, each row: r_j+1(a)=r_j(a)-d_j＜ d_j,r_j(a) ＞, wherein ＜ ＞ table Show inner product；

Step 3-8: iterations j=j+1, if j is ＜ n, return to step 3-4, till carrying out n iteration.Obtain Optimize greedy self-adapting dictionary d, d=d_j.

Step 3-9: optimize greedy self-adapting dictionary structure and complete.

Described step 3-2 comprises the steps:

Step 3-2-1: obtain historical traffic, singular value decomposition is carried out to historical traffic；

The historical traffic matrix obtaining is designated as x_0his.To x_0hisCarry out singular value decomposition,

$x_{0\mathrm{his}}=\underset{k=1}{\overset{n}{\mathrm{\σ}}}{\mathrm{\σ}}_k{u}_k{v}_k^t---\left(3\right)$

Wherein σ_kFor singular value, u_kReferred to as feature stream, v_kReferred to as characteristic vector

Step 3-2-2: k big singular value before extraction, other little singular values are set to 0；

Then have according to formula (3),

$x_{0\mathrm{his}}^{\mathrm{pc}}=\underset{k=1}{\overset{k}{\mathrm{\&sigma;}}}{\mathrm{\&sigma;}}_k{u}_k{v}_k,k<n---\left(4\right)$

For being extracted the traffic matrix approximate matrix of k maximum singular value；

Described step 4 specifically includes following steps:

Step 4-1: the dictionary d being obtained according to step 3, by following l₁The row of norm minimum problem solving n × 1 to Amount

${\widehat{\mathrm{\&theta;}}}_t\arg \min {\left|\right|{\mathrm{\&theta;}}_t\left|\right|}_{1}s.t.\mathrm{bd}{\mathrm{\&theta;}}_t=y_t---\left(6\right)$

Wherein, y_tIt is the t row of observing matrix y；

Step 4-2: by t iteration, obtain n × t matrix

Step 4-3: calculated flow rate Matrix Estimation value

The invention has the advantages that

The present invention according to the power-law distribution of powerline network flow, using random matrix selected part od measured directly Stream, and operating flux acquisition function fetching portion od measured directly stream flow information in detail router according to demand.Logical Cross the flow information that obtains of measurement and construct end to end network flow reconstruction model, solve this mould using compressed sensing restructing algorithm Type is to obtain the monitor value of flow.Using the inventive method, can accurately obtain flow monitoring value in real time, simultaneously can be effectively Reduce network traffics collection loss, be conducive to that network research person carries out user's behaviors analysis and network operator is preferably carried out The network traffic engineerings such as network management, network monitoring, network design and the network planning manage.

Brief description

Fig. 1 is typical power communication network service flow monitoring system block diagram in the present invention；

Fig. 2 is distributed traffic monitoring system block diagram proposed by the present invention；

Fig. 3 is the structure new traffic matrix flow chart of the embodiment of the present invention；

Fig. 4 is that the structure of the embodiment of the present invention optimizes greedy self-adapting dictionary flow chart；

Fig. 5 is any one od stream csor method reconstruction value and the actual value schematic diagram of the embodiment of the present invention；

Fig. 6 is any one od stream srsvd method reconstruction value and the actual value schematic diagram of the embodiment of the present invention；

Fig. 7 is embodiment of the present invention csor reconstructing method root-mean-square error schematic diagram relative with srsvd method；

Fig. 8 is the estimated bias schematic diagram of embodiment of the present invention csor reconstructing method and srsvd method；

Fig. 9 is the standard deviation schematic diagram of embodiment of the present invention csor reconstructing method and the estimated bias of srsvd method；

Figure 10 is the cumulative distribution schematic diagram that embodiment of the present invention csor reconstructing method needs od flow amount measured directly.

Specific embodiment

The present invention is described in further detail with specific embodiment below in conjunction with the accompanying drawings.

The emulation experiment that the inventive method is carried out, using U.S. abilene backbone's network data.Abilene backbone host Education and scientific research to be used for, it has 12 nodes, 30 inner link, 24 peripheral links, totally 144 end-to-end fluxes, Take m=54, n=144 in the embodiment of the present invention.Emulation data is gathered by netflow, and time interval is 5min, altogether In 2016 moment, the data on flows in wherein front 500 moment, as the traffic matrix measuring in advance, takes in the embodiment of the present invention T=2016, t₀=500, to calculate the average of od stream and the optimum reconstruction model (csor) set up based on compressed sensing, such as Fig. 2 Shown.

, towards the fast distributed monitoring method of power communication network service, step is as follows for the present invention:

Step 1: generate bernoulli matrix, select to need od stream measured directly by this matrix, and build flow square Battle array；

Specifically include following steps:

Step 1-1: generate bernoulli matrix；

Generate 54 × 144 bernoulli random matrix b.Bernoulli entry of a matrix element b (m, n) is independent same distribution , take pr=0.01 in the present embodiment, the probability that is, element is equal to 1 is 0.01, and the probability equal to 0 is 0.99.

Step 1-2: determine the number of the od stream needing measurement；

The each row of bernoulli matrix to 54 × 144 carry out boolean ' or ' computing respectively,Make s =[s (1), s (2) ..., s (144)]^tFor a column vector, then the od flow amount measuring is needed to be l=| | s | |₁, | | | |₁ Represent l₁Norm.

Step 1-3: direct measurement od stream；

Historical traffic matrix x known to calculating₀Every od stream average:

$\mathrm{aver}\_x\left(n\right)=\underset{t=1}{\overset{t_0}{\mathrm{\&sigma;}}}{x}_0(n,t)n=\mathrm{1,2},...,144---\left(1\right)$

Wherein, t₀For the length of historical traffic matrix, in embodiment, we intercept the historical traffic in 500 moment, i.e. t₀= 500.According to this average, the on off state of flow acquisition function on network management stand control router, to reach the l of measurement maximum =| | s | |₁The purpose of individual od stream.L=by measurement | | s | |₁Individual od stream is designated as gathering { x_mea(l) }, l=1,2 ..., l.

Step 1-4: { x is flowed according to known od_mea(l) }, l=1,2 ...., l builds traffic matrix, as shown in Figure 3, Specifically comprise the following steps that

Step 1-4-1: generate bernoulli random matrix b, and its each row is entered respectively according to step 1-1, step 1-2 Row boolean ' or ' computing；

Step 1-4-2: initialization flow moment matrix is empty matrix, that is,Make l=1, iterations j=1, maximum changes Generation number 144；

Step 1-4-3: when s (j)=1, traffic matrix is changed intol=l+1；Otherwise,

Step 1-4-4: iterations j adds 1, if j ＜ 144, return to step 1-4-3, till iteration 144 times. Obtain traffic matrix x_part,

Step 1-4-5: traffic matrix builds and terminates.

Step 2: obtain observing matrix；

In the present invention, the construction of observing matrix depends only on traffic matrix x_partIn non-zero row, the row of other neutral elements Represent the unknown od stream needing reconstruct.Calculating observation matrix y_mMethod is as follows:

y_m=b x_part(2) wherein, b and x_partObtained by step one.Then by b, x_partAnd observation Matrix y_mDefine a linear system.

Step 3: build and optimize greedy self-adapting dictionary, as shown in Figure 4, specifically comprise the following steps that

Step 3-1: the digital dictionary d of initialization₀For sky, i.e. d₀=[]；

Step 3-2: singular value decomposition is carried out to historical traffic, and extracts k principal component, take k=6 in embodiment, then specifically Including:

Step 3-2-1: obtain historical traffic, singular value decomposition is carried out to historical traffic；

The historical traffic matrix obtaining is designated as x_0his.To x_0hisCarry out singular value decomposition,

$x_{0\mathrm{his}}=\underset{k=1}{\overset{144}{\mathrm{\&sigma;}}}{\mathrm{\&sigma;}}_k{u}_k{v}_k^t---\left(3\right)$

Wherein σ_kFor singular value, u_kReferred to as feature stream, v_kReferred to as characteristic vector

Step 3-2-2: extract front 6 big singular values, other little singular values are set to 0；

Then have according to formula (3),

$x_{0\mathrm{his}}^{\mathrm{pc}}=\underset{k=1}{\overset{6}{\mathrm{\&sigma;}}}{\mathrm{\&sigma;}}_k{u}_k{v}_k,k<n---\left(4\right)$

For being extracted the traffic matrix approximate matrix of 6 maximum singular values；

Step 3-3: setting redundancyFor 6 principal components of historical traffic, and determine iterations iter= 144, make j=1,

Step 3-4: calculate r_jSparse index ξ_a, and corresponding column index a of sparse index_j；

$a_j=\arg \underset{a\&notelement;i_j}{\min }\{{\mathrm{\&xi;}}_a={\left|\right|r_j\left(a\right)\left|\right|}_1/{\left|\right|r_j\left(a\right)\left|\right|}_2\}---\left(5\right)$

Wherein r_jThe a row of (a) redundancy for obtaining after iteration j, | | | |₂Represent l₂Norm；

Step 3-5: obtain the row r with minimum sparse index according to step 3-4_j(a_j), and will there is minimum in redundancy Sparse index ξ_aA_jRow are unitization, and it is set to the jth row of data dictionary, i.e. d_j=r_j(a_j)/||r_j(a_j)||₂；

Step 3-6: obtaining data dictionary according to step 3-5 is: d_j=[d_j-1|d_j], i_j=i_j-1∪{a_j}；Then to institute There is row a；

Step 3-7: update redundancy, each row: r_j+1(a)=r_j(a)-d_j＜ d_j,r_j(a) ＞, wherein ＜ ＞ table Show inner product；

Step 3-8: iterations j=j+1, if j ＜ 144, return to step 3-4, till carrying out 144 iteration. Obtain optimizing greedy self-adapting dictionary d, d=d_j.

Step 3-9: optimize greedy self-adapting dictionary structure and complete.

Step 4: traffic matrix is reconstructed by compressed sensing；

Specifically comprise the following steps that

Step 4-1: the dictionary d being obtained according to step 3, by following l₁The column vector of norm minimum problem solving nx1

${\widehat{\mathrm{\&theta;}}}_t\arg \min {\left|\right|{\mathrm{\&theta;}}_t\left|\right|}_{1}s.t.\mathrm{bd}{\mathrm{\&theta;}}_t=y_t---\left(6\right)$

Wherein, y_tIt is the t row of observing matrix y.

Step 4-2: by t=2016 iteration, obtain 144 × 2016 matrixes

Step 4-3: calculated flow rate Matrix Estimation value

In order to preferably assess the reconstruction property of the present invention, we calculate root-mean-square error, reconstruction bias and deviation relatively Standard deviation.

$\mathrm{rrmse}\left(t\right)=\frac{\sqrt{\underset{n=1}{\overset{n}{\mathrm{\&sigma;}}}{(x(n,t)-\hat{x}(n,t))}^2}}{\sqrt{\underset{n=1}{\overset{n}{\mathrm{\&sigma;}}}x{(n,t)}^2}}---\left(7\right)$

$\mathrm{bias}\left(n\right)=\frac{1}{t}\underset{t=1}{\overset{t}{\mathrm{\&sigma;}}}(\hat{x}(n,t)-x(n,t))---\left(8\right)$

$\mathrm{sd}\left(n\right)=\sqrt{\frac{1}{t-1}\underset{t=1}{\overset{t}{\mathrm{\&sigma;}}}{((\hat{x}(n,t)-x(n,t))-\mathrm{bisa}\left(n\right))}^2}---\left(9\right)$

It is right in emulation experiment to carry out carried for present invention algorithm (csor) and sparse canonical singular value decomposition (srsvd) algorithm Result display the carried algorithm of the present invention can accurately reconstruct end to end network flow to ratio very much.Fig. 5 is csor algorithm to the 20th The reconstruction result of bar od stream, the reconstruction result that Fig. 6 flows to the 20th article of od for srsvd algorithm.Wherein, solid line represents real stream Amount, dotted line represents its reconstruction value.Although it will be apparent from this figure that more or less there is estimation and owed to estimate in csor reconstructing method Meter phenomenon, but csor can follow the tracks of the dynamic change trend of od stream well, approaches very much the actual value of od stream.Fig. 7 is csor With srsvd algorithm relative to root-mean-square error.Wherein solid line is csor algorithm, and dotted line is srsvd algorithm.It can be seen that The relative root-mean-square error of the inventive method csor is less.In order to assess method proposed by the present invention further, we introduce weight Structure deviation is poor with deviation standard.As can be seen that the reconstruction bias of csor method are less from Fig. 8 and Fig. 9, standard deviation is also very in addition Little, this illustrates that institute of the present invention extracting method has stronger stability.Next assess the economic benefit of csor, Figure 10 depicts me Need the Cumulative Distribution Function of direct measurement od flow amount, it can be seen that we only need to survey in the case of about 96% 70 od streams of amount are that is to say, that the od stream that we only need to measure 49% just can obtain the monitoring result that all 144 od flow.

Claims (5)

1. towards power communication network service fast distributed monitoring method it is characterised in that comprising the steps:

Step 1: generate bernoulli matrix, select to need od stream measured directly by this matrix, and build traffic matrix；

Step 2: obtain observing matrix；

The construction of observing matrix depends on traffic matrix x_partIn non-zero row, the row of other neutral elements represents and unknown needs weight The od stream of structure, calculating observation matrix y_mMethod is as follows:

y_m=b x_part(2)

Wherein, b is the bernoulli matrix of m × n, x_partIt is the traffic matrix of n × t, then by b, x_partWith observing matrix y_mShape Become a linear system；

Step 3: build and optimize greedy self-adapting dictionary；

Step 4: traffic matrix is reconstructed by compressed sensing；

Described step 1 specifically includes following steps:

Step 1-1: generate bernoulli matrix；

Generate the bernoulli random matrix b of m × n (m ＜ n), n is the number of od stream in network, and it is equal to network node quantity Square；Bernoulli entry of a matrix element b (m, n) is independent identically distributed, and the probability that element is equal to 1 is pr, general equal to 0 Rate is 1-pr；

Step 1-2: determine the number of the od stream needing measurement；

The each row of bernoulli matrix of m × n are carried out respectively with boolean ' or ' computing,Make s=[s (1), s (2),...,s(n)]^tFor a column vector, then the od flow amount measuring is needed to be l=| | s | |₁, | | | |₁RepresentNorm；

Step 1-3: direct measurement od stream；

Historical traffic matrix x known to calculating₀Every od stream average:

Wherein, t₀For the length of historical traffic matrix, n is od flow amount, according to this average, on network management stand control router The on off state of flow collection function, to reach l=| | s | | of measurement maximum₁The purpose of individual od stream, by the l=of measurement | | s ||₁Individual od stream is designated as gathering { x_mea(l) }, l=1,2 ..., l；

Step 1-4: { x is flowed according to known od_mea(l) }, l=1,2 ...., l builds traffic matrix.

2. the fast distributed monitoring method towards power communication network service according to claim 1 it is characterised in that Described step 1-4 specifically includes following steps:

Step 1-4-1: generate bernoulli random matrix b, and cloth is carried out respectively to its each row according to step 1-1, step 1-2 You ' or ' computing；

Step 1-4-2: initialization flow moment matrix is empty matrix, that is,Make l=1, iterations j=1, greatest iteration time Number n；

Step 1-4-3: when s (j)=1, traffic matrix is changed intoL=l+1；Otherwise,

Step 1-4-4: iterations j adds 1, if j is ＜ n, return to step 1-4-3, till iteration n time, obtain flow Matrix x_part,

Step 1-4-5: traffic matrix builds and terminates.

3. the fast distributed monitoring method towards power communication network service according to claim 1 it is characterised in that Described step 3 specifically includes following steps:

Step 3-1: the digital dictionary d of initialization₀For sky, i.e. d₀=[]；

Step 3-2: singular value decomposition is carried out to historical traffic, and extracts k principal component；

Step 3-3: setting redundancyFor k principal component of historical traffic, and determine iterations iter=n, make j =1,

Step 3-4: calculate r_jSparse index ξ_a, and corresponding column index a of sparse index_j；

Wherein r_jThe a row of (a) redundancy for obtaining after iteration j, | | | |₂RepresentNorm；

Step 3-5: obtain the row r with minimum sparse index according to step 3-4_j(a_j), and minimum sparse by having in redundancy Index ξ_aA_jRow are unitization, and it is set to the jth row of data dictionary, i.e. d_j=r_j(a_j)/||r_j(a_j)||₂；

Step 3-6: obtaining data dictionary according to step 3-5 is: d_j=[d_j-1|d_j], i_j=i_j-1∪{a_j}；

Step 3-7: and then to all row a, update redundancy, each row: r_j+1(a)=r_j(a)-d_j＜ d_j,r_j(a) ＞, wherein ＜ ＞ represents inner product；

Step 3-8: iterations j=j+1, if j is ＜ n, return to step 3-4, till carrying out n iteration, optimized Greedy self-adapting dictionary d, d=d_j；

Step 3-9: optimize greedy self-adapting dictionary structure and complete.

4. the fast distributed monitoring method towards power communication network service according to claim 3 it is characterised in that Described step 3-2 comprises the steps:

Step 3-2-1: obtain historical traffic, singular value decomposition is carried out to historical traffic；

The historical traffic matrix obtaining is designated as x_0his, to x_0hisCarry out singular value decomposition,

Wherein σ_kFor singular value, u_kReferred to as feature stream, v_kReferred to as characteristic vector

Step 3-2-2: k big singular value before extraction, other little singular values are set to 0；

Then have according to formula (3),

For being extracted the traffic matrix approximate matrix of k maximum singular value.

5. the fast distributed monitoring method towards power communication network service according to claim 1 it is characterised in that Described step 4 comprises the steps:

Step 4-1: the dictionary d being obtained according to step 3, by followingThe column vector of norm minimum problem solving n × 1

Wherein, y_tIt is the t row of observing matrix y；

Step 4-2: by t iteration, obtain n × t matrix

Step 4-3: calculated flow rate Matrix Estimation value