CN105245362B - Important node information collecting method in a kind of SDN environment - Google Patents
Important node information collecting method in a kind of SDN environment Download PDFInfo
- Publication number
- CN105245362B CN105245362B CN201510581282.3A CN201510581282A CN105245362B CN 105245362 B CN105245362 B CN 105245362B CN 201510581282 A CN201510581282 A CN 201510581282A CN 105245362 B CN105245362 B CN 105245362B
- Authority
- CN
- China
- Prior art keywords
- node
- parameter
- network
- represent
- forward node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses important node information collecting method in a kind of SDN environment, for choosing important node in network, so as to carry out the monitoring of the emphasis of network service and dispose with an emphasis.The relevant information of Forwarding plane and network server is collected first, quantitative analysis then is carried out to the significance level of forward node, and then the information collection for having emphasis is carried out to network service.Wherein, selected parameter includes four parts：1) traffic；2) network topology；3) forward node liveness；4) information of network service, wherein information of network service refer to the related index of urgent degree that the service is disposed for evaluation forward node, can be chosen in its index provided according to specific network service.The present invention optimizes the information collection work of network service, while information of network service acquisition node quantity is reduced, obtains preferable service effectiveness.
Description
Technical field
The present invention relates to network information processing fields, and in particular to important node information collecting method in a kind of SDN environment.
Background technology
Software defined network (Sof tware Defined Networks, SDN) is by the data forwarding in traditional network
A kind of new network architecture that level is decoupled with management layer and proposed.Since SDN for network management and developer provides system
The northbound interface of one programming can easily carry out network management and provide all kinds of new network services, which has obtained
It is widely applied.Such as, Google applies SDN frames to improve device resource utilization rate in its backbone network；Amazon AWS leads to
It crosses SDN frameworks and extensive elastic calculation cloud service is provided.
In SDN frameworks, controller Controller is not only that lower floor forward node Switch is provided as core component
Data forwarding flow table also provides disparate networks basic information for upper level applications App.Therefore, how Controller is reduced
Work load, improve its working efficiency, be the key that ensure the operation of SDN efficient stables.Although in the research work of existing SDN
In work, the existing correlative study for reducing Controller work loads, but they are all only for a certain specific works of Controller
It optimizes, there is certain limitation.Especially in large-scale network environment, when service request quantity all kinds of in SDN
When increasing, these researchs can not effectively reduce the work load of Controller.This will lead to the increase of network delay, drop
Its low service quality or even there is the situation of periods of network disruption.
In traditional network, the information collection work of network service is reduced, the main method for reducing network delay is to be directed to
Node in network carries out importance assessment.By way of only carrying out information collection to important node, net can be greatly promoted
Network performance.Traditional network node importance appraisal procedure can be roughly divided into following a few classes：(1) sequence based on node neighbour；
(2) sequence based on path；(3) sequence of feature based vector；(4) sequence for removing and shrinking based on node.It is however, above-mentioned
4 quasi-tradition network node importance ranking methods are not particularly suited for SDN environment, the main reason is that：
(1) basic thought of the sort method based on node neighbour and based on path is by analysis node surrounding neighbours section
The interstitial content that communication path is included between the number of point or certain two node, to determine the importance of the node.But
In SDN, forwarding flow table is formulated by Controller according to whole net state and routing policy.Therefore, in SDN frameworks, only in accordance with
The topology information of Forwarding plane is not sufficient to accurately evaluate the importance of forward node.
(2) sort method of feature based vector needs constantly to spread, and pass through in network-wide basis by parameter information
Final result can be just obtained after one section of convergence time.But in SDN, especially in large-scale SDN, due to
Controller has the whole network information, causes it that can constantly reformulate forwarding flow table.Therefore, such as according to feature based phase
Adjacent sort method assesses the importance of forward node in SDN, and the algorithm will be caused to be constantly in converged state, can not
It obtains effectively as a result, or even there is the situation that the algorithm enters endless loop.
(3) it is after being removed by assessing network node based on the basic thought of sort method that node removes and shrinks
To the destructiveness of whole net, so that it is determined that the importance of the node.But in SDN, when there is " paralysis " in certain forward node,
Controller can reformulate rapidly new forwarding flow table, it is ensured that whole Network Communication it is unimpeded.Therefore, in SDN, only work as shifting
During except certain special forward node, can just have larger destructiveness to whole net.So sequence for being removed and being shunk based on node
Method is also not particularly suited for SDN.
Traditional network services information collection optimization algorithm carries out network service generally using node importance index as foundation
Emphasis is acquired to reach the optimization to its efficiency of service, and the importance index can be applied to other network management activities, such as
Network structure safety analysis and network route management etc..Its main results is as follows：Chen et al. proposes a kind of passing
Node importance sort algorithm based on half local message in system network.Kitsak et al. proposes a kind of true with K- shell decomposition methods
Determine the algorithm of node importance in network.K- shell decomposition methods are combined by Hu et al. with community structure, propose a kind of improvement index,
And the experiment on SIR models shows that this method is slightly good compared with the algorithm of Kitsak et al..Martin et al. is to eigenvector centrality
Property algorithm is improved, and is proposed in the score value of calculate node, the score value of neighbours does not consider further that the influence of the node.Garas
Et al. propose a kind of algorithm for representing weighted networks interior joint importance jointly with neighbours' quantity of node and the side right of node.
Chen et al. proposes a kind of half local algorithm for directed networks, which not only allows for the quantity of neighbor node, also
Consider the influence that cluster coefficients propagate information.Due to the distributed nature of traditional network architecture and shortage uniform hardware programming
The property of interface so that traditional network services disposition optimization algorithm is usually there are convergence time, it is difficult to which reflection node exists in real time
Importance in network, and the limitation of the network equipment is limited by parameter selection, it is difficult to comprehensively weigh the important of node
Property.However problem above, the characteristic of SDN frameworks can be utilized to well solve in SDN environment.
The research of current SDN frameworks is still in the elementary step, and research direction is concentrated mainly on the advantage using SDN frameworks
Solve the problems, such as it is already present under the scene based on extensive dynamic network, for Controller optimizations in SDN frames
The research of problem is still in the starting stage.Existing main results are as follows：Perez et al. proposes a kind of attribute according to demand will
The processing data packets work of Controller is optimized in the algorithm of data packet Fast Classification.Thorat et al. is proposed
The self-regeneration work of Controller is optimized in a kind of quick self-regeneration algorithm of SDN frameworks.Malboubi etc.
People proposes a kind of optimization algorithm for fine granularity flow monitoring, and Controller is produced when alleviating fine granularity flow monitoring
Raw work load.The existing research about Controller optimization problems is unfolded both for particular problem so that grinds
The autgmentability for studying carefully achievement is extremely limited, lacks with the algorithm research compared with high scalability.
Invention content
In view of the deficiencies of the prior art, the present invention is intended to provide important node information collecting method in a kind of SDN environment,
It reduces Controller to postpone all kinds of service request responses, while improving network stabilization, further improve single
The network size that Controller can be managed so as to reduce the deployment number of Controller in network, reduces
The collaborative work amount of Controller.
To achieve these goals, the present invention adopts the following technical scheme that：
Important node information collecting method in a kind of SDN environment includes controller Controller and forward node, institute
The method of stating includes the following steps：
S1 controllers obtain evaluating each forwarding by collecting Forwarding plane and network server into row information
The parameter information of node significance level, the parameter include the traffic, network topology, forward node active degree and network service
Information；
S2 is polynary according to each forward node of the obtained each corresponding parameter information compositions of forward node of step S1
Group data carry out quantitative analysis to the multi-component system data of each forward node, show that the importance of each forward node in network refers to
Mark.
It should be noted that in the step S1, data volume of the traffic for forward node transmission in the unit time passes through
Following formula obtains：
CNi=cni(t)-cni(t-1), i=1,2 ..., m；
Wherein, CNiRepresent forward node viTraffic parameter, cni(t) v is representediNode t moment cumulative data forwards
Amount, cni(t-1) it is viNode t-1 moment cumulative data transfer amounts, m represent the interstitial content in network.
It should be noted that in the step S1, the network topology is without connected graph is mutually had no right, and is denoted as G=(V, E),
Wherein V={ v1, v2..., vmBe G node set, E={ e1, e2..., enBe G line set, m and n are the node of G respectively
Number and number of edges, m that is to say the forward node number in network；In addition, the adjacency matrix of G is denoted as Am×m=[aij], only as node vi
With vjBetween have even side when aij=1, otherwise aij=0；
The network topology parameters specifically include following subparameter：
2.1) centrality parameter is centrifuged：
That is the maximum distance of other forward node, is obtained by following formula in a forward node to network：
ECCi=maxj(dij), i, j=(1,2 ..., m)；
Wherein, ECCiRepresent forward node viCentrifugation centrality parameter, dijRepresent forward node viWith vjBetween shortest path
The length of diameter, m represent the interstitial content in network；
2.2) close to centrality parameter：
That is the average distance of other forward node, is obtained by following formula in a forward node to network：
Wherein, CCiRepresent node viClose to centrality parameter, dijRepresent node viWith vjBetween shortest path length,
M represents the interstitial content in network；
2.3) centrality parameter is spent：
That is neighbours' number of a forward node, is obtained by following formula：
DCi=degi, i=(1,2 ..., m)
Wherein, DCiRepresent node viDegree centrality parameter, degiRepresent node viNeighbours' number, m represented in network
Interstitial content.
It should be noted that in the step S1, forward node liveness parameter specifically includes following subparameter：
3.1) data packet forwarding frequency：
The quantity for the data packet that forward node is sent, is obtained by following formula i.e. in the unit interval：
Wherein, EFiRepresent node viData packet forwarding frequency, efi(t) forward node v is representediIn t moment cumulative data
Packet hop count, efi(t-1) forward node v is representediIn t-1 moment cumulative data packet hop counts, m represents the node in network
Number, Δ t represent t moment and the time difference between the t-1 moment；
3.2) node request frequency：
I.e. in the unit interval, forward node sends out the number of all kinds of requests to controller, is obtained by following formula：
Wherein, QFiRepresent forward node viNode request frequency, qfi(t) forward node v is representediAdd up to send out in t moment
Send the number of request, qfi(t-1) forward node v is representediAdd up to send the number asked at the t-1 moment, m represents the section in network
It counts out, time differences of the Δ t between t moment and t-1 moment；
3.3) flow table modification frequency：
It represents in the unit interval, the number of the forward node flow table of modification deployment thereon under controller instruction, i.e.,：
Wherein, CFiRepresent node viFlow table modification frequency, cfi(t) v is representediNode t moment adds up time of flow table modification
Number, cfi(t-1) v is representediThe node t-1 moment adds up the number of flow table modification, and m represents the interstitial content in network, when Δ t is t
Carve the time difference between the t-1 moment.
It should be noted that in step S1, the information of network service parameter includes the parameter of multinomial network service, each
The urgent degree that the parameter of network service disposes the service with specific reference to forward node is chosen, and represents as follows：
TSi={ Si1, Si2..., Sinoi, i=(1,2 ..., m)；
TSiRepresent viThe information of network service of node, wherein Si1, Si2..., SinoiRepresent selected every network service
Parameter, noi represent the sum of web service parameter included in information of network service parameter.
It should be noted that step S2 is implemented as follows：
2.1) standard parameter；
2.2) historical parametric is handled；
2.3) parameter is finally fitted, and obtains the importance index of each forward node in network.
It needs further exist for illustrating, in step 2.1), standard parameter carries out according to the following formula：
Wherein aiFor node viThe initial data of a certain parameter, xiFor node viThis standard parameter after as a result, aj
For node vjThis parameter initial data.
Explanation is needed further exist for, step 2.2) is implemented as follows：
2.2.1 single treatment first) is carried out to it by weighted sum to the historical data of each parameter, i.e.,：
Wherein, Q represents one historical parametric for having already passed through processing of certain forward node, TjRepresent the forward node one
- 1 historical data of jth of historical parametric, αjRepresent corresponding TjWeight, K represent for fitting information number, that is, use
In the historical data of fitting and the total number of current data；
2.2.2) using Weight of Coefficient through Analytic Hierarchy Process αjAnd P, it is specific as follows：
2.2.2.1) according to timing information T={ T1, T2... Tj..., TKJudgement Matricies A：
Wherein, bijRepresent TiWith TjRelativeness, mathe-matical map are as follows：
2.2.2.2 it) will determine that each row normalization of matrix A：
2.2.2.5 feature root) is acquired：Calculate coincident indicator CI=
(tmax- K)/K-1, control Aver-age Random Consistency Index carries out consistency check, if not adjusting b ifijMathe-matical map
Judgment matrix A is rebuild, and jumps to step 2.2.2.1), wherein, (AW)iThe product of representing matrix A and characteristic vector W
I-th of component；
It should be noted that step 2.2.2.1) in TiWith TjRelativeness bijIt empirically sets, if step
Rapid 2.2.2.5) in consistency check do not pass through, then prove relativeness bijSetting it is unreasonable, so as to readjust bijNumber
Mapping and again development of judgment matrix A are learned, until consistency check until.
Explanation is needed further exist for,
2.2.2.7) Q obtained by weighted sum：
Explanation is needed further exist for, step 2.3) is implemented as follows：
2.3.1) by the following formula by all parameter fittings into node importance index：
Wherein, siRepresent node viImportance index, CNiAnd CNjNode v is represented respectivelyiAnd vjTraffic parameter,
TPiAnd TPjNode v is represented respectivelyiAnd vjNetwork topology parameters, VDiAnd VDjNode v is represented respectivelyiAnd vjNode liveness
Parameter, TSiAnd TSjNode v is represented respectivelyiAnd vjInformation of network service parameter, λ1、λ2、λ3、λ4Node v is represented respectivelyiIt is logical
Traffic parameter, network topology parameters, node liveness parameter and information of network service parameter weight；
The calculation formula of the network topology parameters, node liveness parameter and web service parameter is as follows：
Wherein, λ21、λ22And λ23Centrifugation centrality parameter in network topology parameters is represented respectively, close to centrality parameter
With the weight of degree centrality parameter, λ31、λ32And λ33Respectively represent forward node liveness parameter in data packet forwarding frequency,
The weight of node request frequency and flow table modification frequency, λikIt is then each web service parameter SikCorresponding weight；
2.3.2 the weight λ in each fitting formula) is determined using Principal Component AnalysisjIt is specific as follows with final fitting result：
2.3.2.1 input variable) is initialized, and generates the parameter matrix A (m × n) for recording all node input parameters,
In, the initializing variable is number of parameters n, the whole network interstitial content m and forward node viParameters vi={ pi1,
pi2..., pin}；
2.3.2.7 the whole network fit metric) is acquired：P=[P1, P2..., Pm]T。
The beneficial effects of the present invention are：
1st, the information collection of network service work is optimized, is reducing the same of information of network service acquisition node quantity
When, obtain preferable service effectiveness；
2nd, with the reduction of information on services acquisition node quantity, Controller can obtain following benefit in the activity：
1) speeds of response of the Controller to all kinds of work is improved；2) the network rule that single Controller can be managed are improved indirectly
Mould so as to reduce the deployment number of Controller in network, reduces collaborative work amount and the deployment of Controller
The capital consumption of Controller.
Description of the drawings
Fig. 1 is the implementing procedure figure of the present invention；
Fig. 2 is the implementing procedure figure of quantitative analysis step in Fig. 1；
Fig. 3 is attack detecting program network oversampling ratio in emulation experiment to the shadow of the Controller work speeds of response
It rings；
Fig. 4 is attack detecting program network oversampling ratio and the relationship of attack-response time delay；
Fig. 5 is attack detecting program network oversampling ratio and the relationship of attack detecting success rate；
Fig. 6 is relationship of the node selection percentage with selecting similarity.
Specific embodiment
Below with reference to attached drawing, the invention will be further described, it should be noted that the present embodiment is with this technology side
Premised on case, detailed embodiment and specific operating process are given, but protection scope of the present invention is not limited to this reality
Apply example.
As shown in Figure 1, important node information collecting method includes the following steps in a kind of SDN environment：
S1 is according to the property of SDN frames, and on the basis of scalability is ensured, selection can be with effecting reaction forward node weight
The parameter for the property wanted.Wherein parameter includes four major class：1) traffic；2) network topology 3) forward node active degree；4) network takes
Business information.Wherein, standard communication protocol Openflow of first three item data in SDN frames is provided interface and effectively anti-
The purpose for reflecting forward node importance is chosen, and information of network service is chosen according to specific network service.
(1) traffic
The data volume that i.e. forwarding node is sent in the unit interval.Since the most basic work of forwarding node is exactly to complete
The forwarding of data packet, therefore the parameter has the ability for weighing the forwarding node importance.
CNi=cni(t)-cni(t-1)
Wherein, CNiRepresent viThe traffic parameter of node, cni(t) v is representediNode t moment cumulative data transfer amount.
(2) network topology
Forward the physical connection situation of network.Since the physical connection situation of network largely determines network
Ability, therefore, network topology situation can be used to weigh significance level of the forward node in network.In the present invention
In, topological diagram is without mutually having no right connected graph, be denoted as G=(V, E), wherein V={ v1, v2..., vmBe G node set, E=
{e1, e2..., enBe G line set, m and n are the number of nodes and number of edges of G respectively.The adjacency matrix of G is denoted as Am×m=[aij],
aij=1 but and if only if node viWith vjBetween have even side, otherwise aij=0.
Due to the complexity of network topology situation, existing balancing method comprehensively can not weigh node by network topology
Importance, therefore the present invention weighs it by following three kinds of parameters：
A) centrality is centrifuged：
The maximum distance of other nodes in node to network is investigated, i.e.,：
Wherein, ECCiRepresent viThe centrifugation centrality parameter of node, dijRepresent node viWith vjBetween shortest path length
Degree, m represent the interstitial content in network.
B) close to centrality：
The average distance of other nodes in node to network is investigated, i.e.,：
Wherein, CCiRepresent node viClose to centrality parameter, dijRepresent node viWith vjBetween shortest path length,
M represents the interstitial content in network.
C) centrality is spent：
Neighbours' number of node is investigated, i.e.,：
DCi=degi, i=(1,2 ..., m)
Wherein, DCiRepresent node viDegree centrality parameter, degiRepresent node viNeighbours' number, m represented in network
Interstitial content.
(3) forward node liveness
I.e. the node participates in the degree of network activity.With reference to the concrete condition of SDN scenes, weighed with following three parameters
The liveness of forwarding node：
A) data packet forwarding frequency：
In i.e. between unit, the quantity for the data packet that forward node is sent.It is a small amount of due to only being carried there are some in network
The data packet of data, these data packets are usually used to the objective function for realizing various agreements, are of great significance, however to this
Partial data packet forwarding work meaning can not be merely weighed by data volume.Therefore, the parameter is chosen to weigh this portion
The meaning that the division of labor is made, i.e.,：
Wherein, EFiRepresent node viData packet forwarding frequency, efi(t) v is representediNode t moment cumulative data packet forwards
Number, m represent the interstitial content in network.
B) node request frequency：
I.e. in the unit interval, forward node sends out the number of all kinds of requests to Controller.The parameter reflects forwarding
Node, to the degree of dependence of Controller instructions, the node is shown in network activity from a side in network activity
In liveness, i.e.,：
Wherein, QFiRepresent node viNode request frequency, qfi(t) v is representediNode t moment adds up to send time of request
Number, m represent the interstitial content in network.
C) flow table modification frequency：
I.e. in the unit interval, the number of the forward node flow table of modification deployment thereon under Controller instructions.
In SDN frames, the number that the flow table on a usual forward node is changed by Controller is more, it was demonstrated that the node takes part in
More network activities.It further relates to when something goes wrong, the network activity of influence is more, therefore chooses the parameter on the node
Weigh the liveness of node, i.e.,：
Wherein, CFiRepresent node viFlow table modification frequency, cfi(t) v is representediNode t moment adds up time of flow table modification
Number, m represent the interstitial content in network.
(4) information of network service：
The selection of the parameter is specifically formulated, and number according to the specific network service to be optimized by network management personnel
Measure it is unlimited, such as：To Network Intrusion Detection System, situation of every abnormal alarm etc. is may be configured as, uses TSi={ Si1, Si2...,
SinRepresent viThe information of network service of node, wherein Si1, Si2..., SinRepresent selected every web service parameter.
S2 passes through to forward node viMulti-component system data quantitative analysis, show that the importance of each node in network refers to
Mark, process is as shown in Fig. 2, include the following steps：
2.1) standard parameter；
2.2) historical parametric is handled；
2.3) parameter is finally fitted, and obtains the importance index of each forward node in network.
By the flow shown by Fig. 2, the Parameter analysis being collected into from Forwarding plane and network server is handled, is obtained
Go out the quantized data of network node importance.
2.1) standard parameter
Due to otherness of the parameters in terms of numerical value performance, to make subsequent parameter processing work that can facilitate exhibition
It opens, crude sampling parameter is first standardized by the present invention：
Wherein aiFor node viThe initial data of a certain parameter, xiFor the result after the standard parameter.Pass through the mark of parameter
Quasi-ization processing so that the different parameter of script value range is mapped in the range of [0,1], is facilitated at the parameter of subsequent step
Science and engineering is made.
2.2) historical parametric is handled
Since network activity behavior is usually expressed as the duration variation of network parameter in a period of time, the parameter
Data can show the consecutive variations in sequential, therefore when evaluating the historic parameter of this kind of tool, be considered as its history
Its historical data is also included in investigation range by continuity.And the traffic, forward node are active in the parameter selected by the present invention
Degree and information of network service have apparent History Continuity, therefore using the historical data of this three classes parameter, are asked by weighting
Single treatment is first carried out with to it, i.e.,：
Wherein, P represents one historical parametric for having already passed through processing of certain node, such as traffic volume, data packet forwarding frequency
Deng, and P will be inputted as the data of this parameter and be used in the final fitting part of parameter, TjRepresent one history of the node
- 1 historical data of jth of parameter, αjRepresent corresponding TjWeight.
Due to TjWith sequential precedence relationship, they are to PiInfluence power have more apparent strong or weak relation, i.e. T1To PiInfluence power
Maximum and TnTo P influence powers minimum.Due to that can determine YjThe relativeness of (j=1,2 ..., n) between any two, and step analysis
Method (Analytic Hierarchy Process, AHP) can be by the relativeness two-by-two between parameter by all parameter fittings
Into a quantizating index, therefore the present invention selects AHP to determine weight αjAnd P, it calculates as follows：
In algorithm above, pass through T firstijTiming information initialization judgment matrix A, initialization procedure is as follows：
Wherein, bijRepresent TiWith TjRelativeness, mathe-matical map are as follows：
The characteristic vector W and characteristic root ω of judgment matrix A is calculated by matrix operation later1, ω2..., ωK, then pass through
Characteristic root further verifies whether constructed judgment matrix A has the contradiction of importance index between parameter, and weight is needed if contradictory
Newly the relationship two-by-two between parameter judges, the Evaluating Eigenvalues of this calculating can be applied to obtain weight α if no contradictionj, most
Q is obtained eventually.
2.3) parameter is finally fitted
After standard parameter and historical parametric handle two steps, obtain parameters and be used for the number being finally fitted
According in the final fit procedure of parameter, the present invention is by the following formula by all parameter fittings into node importance index：
Wherein, siRepresent node viImportance index, CNiRepresent node viTraffic parameter, TPiRepresent node vi's
Network topology parameters, VDiRepresent node viNode liveness parameter, TSiRepresent node viWeb service parameter, λ represent section
Point viEach parameter weight.
Further, since network topology parameters, node liveness parameter and web service parameter are acquired by multinomial subparameter,
Calculation formula is as follows：
Wherein, λ21、λ22And λ23Centrifugation centrality parameter in network topology parameters is represented respectively, close to centrality parameter
With the weight of degree centrality parameter, λ31、λ32And λ33Respectively represent forward node liveness parameter in data packet forwarding frequency,
The weight of node request frequency and flow table modification frequency, λikIt is then each web service parameter SikCorresponding weight；
Since the parameter meaning being fitted is different, the relationship two-by-two between parameter can not be subjectively judged.And it leads
Componential analysis (Principal Components Analysis, PCA) is in the situation that can not judge each parameter relativeness
Under, its weight is determined according to the dispersion degree of each parameter.Therefore the present invention selects PCA to determine the weight λ in each fitting formulaj
With final fitting result, algorithm is as follows：
In algorithm above, by the matrix operation to the whole network information matrix A, the eigenmatrix for obtaining matrix A leads to W, then
By the data of the middle each element of eigenmatrix W, the weight λ of parameters is obtainedj, then pass through the weighted sum to parameters
Obtain fitting result Pi, finally obtain the matrix P for recording the whole network node fitting result.To calculate TPiFor, initially set up the whole network
Information matrix A (m × 3), matrix include the ECC of the whole network nodei、CCiAnd DCiInformation is obtained by the dispersion degree of each column data
The weight λ of parameters1、λ2And λ3, the TP of node is obtained by weighted fittingiInformation.
Obtain P=[P1, P2..., Pm]TAfterwards, the method for obtaining important node can be determined on a case-by-case basis, such as by door
Several former nodes are important node after limit value determines or carry out descending arrangement.
The performance of the present invention will be further described by emulation experiment below.
It is to a specific network service in emulation experiment --- the monitoring of TCP flood attacks carries out important node information
Collecting work verifies effectiveness of the invention by the emulation experiment in different scales network, and process is as follows.
Experimental situation
The validity of this programme is tested in this experiment by simulation software, and environment is as follows：Test host (3.20GHz
CPU, 2G memory) on the networks of three kinds of scales is simulated by mininet2.0.0 and Pox.The network of these three scales possesses respectively
100th, 200 and 300 Switch nodes and with Switch interstitial content same hosts, wherein Switch nodes and a host be straight
Connect connected, the network between Switch is determined by the connected graph that program generates at random.This experiment is generated by Scapy on this basis
Network context flow and simulation TCP flood attacks.The experimental arrangement work period write based on the present invention is 1min, and select
The detection program for TCP flood attacks of increasing income of automatic network is fetched as specific network server, and will successfully be detected
Number of times of attack as information of network service.
Analysis of experimental data
(1) TCP flood attacks program network oversampling ratio is to the relationship of the Controller work speeds of response
In SDN frames, the speed of the Controller work speeds of response directly affects the operating condition of network, and most can
The information for embodying the Controller work speeds of response is exactly the speed of response that Controller communicates to network new life, and net
The speed of response of network new life communication can be weighed by the response delay of ping packets first in network.Additionally due to
Mininet simulated programs can not be practical display controller specific works situation, such as：CPU Expenditure Levels, memory usage feelings
Condition etc., therefore in emulation experiment, this experiment selects network head ping latencies to work to weigh oversampling ratio Controller
The influence that the speed of response generates, specific data are as shown in Figure 3.
By taking Fig. 3 as an example, in the network of three kinds of scales, as attack detecting program samples ratio increases, network head ping
Time delay also gradually increases, and the increase of this data represents the reduction of the Controller work speeds of response.It can be seen that Attack monitoring
The extensive sample activity of program can influence the speeds of response of the Controller to work.As can to attack detecting procedure service
In the case that quality influences less, reduce the oversampling ratio of attack detecting program, will preferably reduce it to Controller
The influence of operating rate.And when Attack monitoring program samples ratio reaches 100%, during first ping in 300 meshed networks
Prolong more than 300ms, and the time delay of daily access network is left in 20ms by taking domestic large-scale website (Baidu, Sina, Netease) as an example
It is right, it is seen that influenced caused by its speed of response that works Controller it is quite apparent, therefore, have it is very big it is necessary to its into
Row optimization.
(2) influence of the Attack monitoring program samples ratio to its service effectiveness
Since in this experiment, what network service was chosen is that TCP flood attacks detect program, therefore Web Service Deployment ratio
The relationship of example and its service effectiveness can be mapped as network samples ratio and attack detecting success rate and the attack detecting time delay that is averaged
Relationship.And it in this experiment, employs and result is chosen with this programme and randomly selects knot to prove the validity of this programme
The comparison of fruit, effect are as shown in Figure 4：
By taking Fig. 4 as an example, with the increase of attack detecting program oversampling ratio in a network, to the response efficiency of attack
It improves for it.And this programme is compared to random fashion, plays the role of that this is more significant, and certain to improving attack-response efficiency
After oversampling ratio (40%-50%), the oversampling ratio growth in this programme brings response efficiency increase rate to be greatly reduced.By
This ratio that can be seen that sampling should choose again more than 40%.In addition, the success rate of network samples ratio and attack detecting
Relationship is as shown in figure 5, wherein attack detecting success rate attacks the 2000 TCP floodings simulated in network by Attack monitoring program
The testing result hit obtains.
By taking Fig. 5 as an example, network node oversampling ratio has preferable attack when a certain range changes (40%-100%)
Detect success rate.The influence that variation of the oversampling ratio in the section successfully detects attack detecting program attack is little.Therefore, may be used
By selection of this programme to important node in network, to optimize mapping out the work for network service, to reduce network service pair
The influence of the Controller work speeds of response.
(3) this programme network node chooses the comparison of result and random fashion
This experiment passes through the average value of two minor node selection result similarities, to weigh the stabilization of this programme node selection
Property, i.e., the present invention selects the degree of stability of network-critical node.The results are shown in Figure 6 for it, wherein selecting similarity-rough set
Epicycle selection result and the average value of the ratio of last selection result same node point obtain in 1000 wheel node selections.
By taking Fig. 6 as an example, under three kinds of network sizes, selection similarity of the invention is consistently higher than stochastical sampling mode, says
The bright present invention has higher stability compared to stochastical sampling, and the stability illustrates that the present invention selects the stabilization of important node
Degree.Under three kinds of network sizes, the similarity that the present invention generates node selection result is basically identical, illustrates the change of network size
Change and the stability of the present invention is had not significant impact.And the present invention occurs a peak in node selection percentage 40% or so
Value, illustrates there are the important node of node total number mesh 40% or so in network, and the present invention can be very good to be selected
Come.When selection percentage is more than 40%, the situation of change of present invention selection similarity curve is mainly by the shadow of oversampling ratio growth
It rings.It is more than that decline after 40% be because with the growth of the ratio of selection in selection percentage, some importance index are relatively low and phase
The node that significance level changes greatly between mutually is added into the ranks of selected node.And present invention selection similarity curve is in the later stage
Growth be the generation directly when selection percentage improves.
Analysis and summary of the present invention to SDN scenes and historic survey achievement, utilize the Centralized Controller of SDN frames
Controller realizes the Assessment of Important to the whole network node, and chooses important node based on this and carry out network service
Information collection works.By experimental verification, the selection result can effectively reduce network service large scale deployment pair
The influence of the Controller work speeds of response, and the network service of carry out important node information collection can be made to have higher service
Quality.
For those skilled in the art, it can be made various corresponding according to above technical solution and design
Change and deform, and all these change and deformation should be construed as being included within the protection domain of the claims in the present invention.
Claims (5)
1. important node information collecting method in a kind of SDN environment, includes controller and forward node, which is characterized in that institute
The method of stating includes the following steps：
S1 controllers obtain evaluating each forwarding section by collecting Forwarding plane and network server into row information
The parameter information of point significance level, the parameter include the traffic, network topology, forward node active degree and network service letter
Breath；
S2 forms the multi-component system number of each forward node according to the corresponding parameter information of the obtained each forward node of step S1
According to carrying out quantitative analysis to the multi-component system data of each forward node, obtain the importance index of each forward node in network；
Step S2 is implemented as follows：
2.1) standard parameter；
2.2) historical parametric is handled；
2.3) parameter is finally fitted, and obtains the importance index of each forward node in network；
In step 2.1), standard parameter carries out according to the following formula：
Wherein aiFor node viThe initial data of a certain parameter, xiFor node viThis standard parameter after as a result, ajFor section
Point vjThis parameter initial data；
Step 2.2) is implemented as follows：
2.2.1 single treatment first) is carried out to it by weighted sum to the historical data of each parameter, i.e.,：
Wherein, Q represents one treated historical parametric of certain forward node, TjRepresent the historic ginseng of the forward node one
Several -1 historical datas of jth, αjRepresent corresponding TjWeight, K represent for fitting information number, i.e., for fitting
The total number of historical data and current data；
2.2.2) using Weight of Coefficient through Analytic Hierarchy Process αjAnd Q, it is specific as follows：
2.2.2.1) according to timing information T={ T1, T2... Tj..., TKJudgement Matricies A：
Wherein, bijRepresent TiWith TjRelativeness, mathe-matical map are as follows：
2.2.2.2 it) will determine that each row normalization of matrix A：
2.2.2.3) matrix by rows after normalization is summed：
2.2.2.4) to vectorIt is normalized：
Then characteristic vector W=[ω1, ω2..., ωK]T；
2.2.2.5 feature root) is acquired：Calculate coincident indicator CI=(tmax-K)/
K-1, control Aver-age Random Consistency Index carries out consistency check, if not adjusting b ifijMathe-matical map rebuild
Judgment matrix A, and jump to step 2.2.2.1), wherein, (AW)iI-th point of the product of representing matrix A and characteristic vector W
Amount；
2.2.2.6) weight vector is normalized：Obtain final weights：α
=[α1, α2..., αK]T；
2.2.2.7) Q obtained by weighted sum：
Step 2.3) is implemented as follows：
2.3.1) by the following formula by all parameter fittings into node importance index：
Wherein, siRepresent node viImportance index, CNiAnd CNiNode v is represented respectivelyiAnd vjTraffic parameter, TPiWith
TPiNode v is represented respectivelyiAnd vjNetwork topology parameters, VDiAnd VDjNode v is represented respectivelyiAnd vjNode liveness parameter,
TSiAnd TSjNode v is represented respectivelyiAnd vjInformation of network service parameter, λ1、λ2、λ3、λ4Node v is represented respectivelyiThe traffic ginseng
Number, network topology parameters, node liveness parameter and information of network service parameter weight；
The calculating of the network topology parameters, node liveness parameter and web service parameter is as follows：
Wherein, λ21、λ22And λ23Centrifugation centrality parameter in network topology parameters is represented respectively, close to centrality parameter and degree
The weight of centrality parameter, λ31、λ32And λ33Represent that the data packet in forward node liveness parameter forwards frequency, node respectively
The weight of request frequency and flow table modification frequency, λikIt is then each web service parameter SikCorresponding weight；
2.3.2 the weight λ in each fitting formula) is determined using Principal Component AnalysisjIt is specific as follows with final fitting result：
2.3.2.1 input variable) is initialized, and generates the parameter matrix A (m × n) for recording all node input parameters, wherein,
The initializing variable is number of parameters n, the whole network interstitial content m and forward node viParameters vi={ pi1,
pi2..., pin}；
2.3.2.2) by each row normalization of parameter matrix A：
bijElement for parameter matrix A；
2.3.2.3) matrix by rows after normalization is summed：
2.3.2.4) to vectorIt is normalized：
And feature vector is calculated, it is denoted as W=[ω1, ω2..., ωn]T
2.3.2.5 parameters weight) is acquired：
2.3.2.6 node parameter fitting result) is acquired：
2.3.2.7 the whole network fit metric) is acquired：P=[P1, P2..., Pm]T。
2. important node information collecting method in a kind of SDN environment according to claim 1, which is characterized in that the step
In rapid S1, data volume of the traffic for forward node transmission in the unit time is obtained by following formula：
CNi=cni(t)-cni(t-1), i=1,2 ..., m；
Wherein, CNiRepresent forward node viTraffic parameter, cni(t) v is representediNode t moment cumulative data transfer amount, cni
(t-1) it is viNode t-1 moment cumulative data transfer amounts, m represent the interstitial content in network.
3. important node information collecting method in a kind of SDN environment according to claim 1, which is characterized in that step S1
In, the network topology is without connected graph is mutually had no right, and is denoted as G=(V, E), wherein V={ v1, v2..., vmBe G set of node
It closes, E={ e1, e2..., enBe G line set, m and n are the number of nodes and number of edges of G respectively, and m that is to say the forwarding section in network
It counts out；In addition, the adjacency matrix of G is denoted as Am×m=[aij], only as node viWith vjBetween have even side when aij=1, otherwise aij
=0；
The network topology parameters specifically include following subparameter：
2.1) centrality parameter is centrifuged：
That is the maximum distance of other forward node, is obtained by following formula in a forward node to network：
ECCi=maxj(dij), i, j=(1,2 ..., m)；
Wherein, ECCiRepresent forward node viCentrifugation centrality parameter, dijRepresent forward node viWith vjBetween shortest path
Length, m represent the interstitial content in network；
2.2) close to centrality parameter：
That is the average distance of other forward node, is obtained by following formula in a forward node to network：
Wherein, CCiRepresent node viClose to centrality parameter, dijRepresent node viWith vjBetween shortest path length, m generations
Interstitial content in table network；
2.3) centrality parameter is spent：
That is neighbours' number of a forward node, is obtained by following formula：
DCi=degi, i=(1,2 ..., m)；
Wherein, DCiRepresent node viDegree centrality parameter, degiRepresent node viNeighbours' number, m represents the node in network
Number.
4. important node information collecting method in a kind of SDN environment according to claim 1, which is characterized in that step S1
In, forward node liveness parameter specifically includes following subparameter：
3.1) data packet forwarding frequency：
The quantity for the data packet that forward node is sent, is obtained by following formula i.e. in the unit interval：
Wherein, EFiRepresent node viData packet forwarding frequency, efi(t) forward node v is representediTurn in t moment cumulative data packet
Send out number, efi(t-1) forward node v is representediIn t-1 moment cumulative data packet hop counts, m represents the number of nodes in network
Mesh, Δ t represent t moment and the time difference between the t-1 moment；
3.2) node request frequency：
I.e. in the unit interval, forward node sends out the number of all kinds of requests to controller, is obtained by following formula：
Wherein, QFiRepresent forward node viNode request frequency, qfi(t) forward node v is representediAdd up to send in t moment and ask
The number asked, qfi(t-1) forward node v is representediAdd up to send the number asked at the t-1 moment, m represents the number of nodes in network
Mesh, time differences of the Δ t between t moment and t-1 moment；
3.3) flow table modification frequency：
It represents in the unit interval, the number of the forward node flow table of modification deployment thereon under controller instruction, i.e.,：
Wherein, CFiRepresent node viFlow table modification frequency, cfi(t) v is representediNode t moment adds up the number of flow table modification,
cfi(t-1) v is representediThe node t-1 moment adds up the number of flow table modification, and m represents the interstitial content in network, Δ t for t moment with
Time difference between the t-1 moment.
5. important node information collecting method in a kind of SDN environment according to claim 1, which is characterized in that step S1
In, the information of network service parameter includes the parameter of multinomial network service, and the parameter of each network service is with specific reference to forwarding
The urgent degree of the node deployment service is chosen, and represents as follows：
TSi={ Si1, Si2..., Sinoi, i=(1,2 ..., m)；
TSiRepresent viThe information of network service of node, wherein Si1, Si2..., SinoiRepresent selected every web service parameter,
Noi represents the sum of web service parameter included in information of network service parameter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510581282.3A CN105245362B (en) | 2015-09-14 | 2015-09-14 | Important node information collecting method in a kind of SDN environment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510581282.3A CN105245362B (en) | 2015-09-14 | 2015-09-14 | Important node information collecting method in a kind of SDN environment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105245362A CN105245362A (en) | 2016-01-13 |
| CN105245362B true CN105245362B (en) | 2018-07-03 |
Family
ID=55042873
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510581282.3A Active CN105245362B (en) | 2015-09-14 | 2015-09-14 | Important node information collecting method in a kind of SDN environment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105245362B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106529562A (en) * | 2016-09-09 | 2017-03-22 | 浙江工业大学 | OSS (Open Source software) project developer prediction method based on Email networks |
| CN110213279B (en) * | 2019-06-10 | 2021-11-30 | 安徽理工大学 | Privacy protection-based dynamic network average consensus method |
| CN110768906B (en) * | 2019-11-05 | 2022-08-30 | 重庆邮电大学 | SDN-oriented energy-saving routing method based on Q learning |
| CN112087488B (en) * | 2020-08-03 | 2023-08-25 | 山东浪潮科学研究院有限公司 | Method, device, equipment and medium for determining important cloud robot nodes |
| CN112367692B (en) * | 2020-10-29 | 2022-10-04 | 西北工业大学 | Air-ground integrated vehicle networking relay selection method based on link service quality |
| CN112910720B (en) * | 2021-05-06 | 2021-08-03 | 成都云智天下科技股份有限公司 | Intelligent network scheduling method and system based on user experience quantitative index |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101853261A (en) * | 2009-11-23 | 2010-10-06 | 电子科技大学 | Network public-opinion behavior analysis method based on social network |
| CN103944748A (en) * | 2014-02-17 | 2014-07-23 | 北京航空航天大学 | Network-key-node self-similar-traffic generation simplification method based on genetic algorithm |
| CN104394202A (en) * | 2014-11-13 | 2015-03-04 | 西安交通大学 | A node vitality quantifying method in a mobile social network |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8504504B2 (en) * | 2008-09-26 | 2013-08-06 | Oracle America, Inc. | System and method for distributed denial of service identification and prevention |
| US8452871B2 (en) * | 2011-08-27 | 2013-05-28 | At&T Intellectual Property I, L.P. | Passive and comprehensive hierarchical anomaly detection system and method |
-
2015
- 2015-09-14 CN CN201510581282.3A patent/CN105245362B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101853261A (en) * | 2009-11-23 | 2010-10-06 | 电子科技大学 | Network public-opinion behavior analysis method based on social network |
| CN103944748A (en) * | 2014-02-17 | 2014-07-23 | 北京航空航天大学 | Network-key-node self-similar-traffic generation simplification method based on genetic algorithm |
| CN104394202A (en) * | 2014-11-13 | 2015-03-04 | 西安交通大学 | A node vitality quantifying method in a mobile social network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105245362A (en) | 2016-01-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105245362B (en) | Important node information collecting method in a kind of SDN environment | |
| CN107483487B (en) | TOPSIS-based multi-dimensional network security measurement method | |
| CN102075352A (en) | Method and device for predicting network user behavior | |
| CN104967629A (en) | Network attack detection method and apparatus | |
| CN101572623A (en) | Method for comprehensively evaluating network performance based on subjective and objective combination evaluation | |
| CN109218304B (en) | Network risk blocking method based on attack graph and co-evolution | |
| CN114513470B (en) | Network flow control method, device, equipment and computer readable storage medium | |
| CN108900513A (en) | A kind of DDOS effect evaluation method based on BP neural network | |
| CN116846565A (en) | SAA-SSA-BPNN-based network security situation assessment method | |
| Lou et al. | Cyber intrusion detection through association rule mining on multi-source logs | |
| CN106487906A (en) | A kind of mobile Web application protocol changing method of context aware | |
| CN107402851A (en) | A kind of data recovery control method and device | |
| CN104811336A (en) | Intelligent optimization-based distributed network flow predicting method | |
| CN103944748B (en) | Network-key-node self-similar-traffic generation simplification method based on genetic algorithm | |
| CN101986608B (en) | Method for evaluating heterogeneous overlay network load balance degree | |
| CN109324906A (en) | The method of selection processing node in cloud computing platform | |
| CN103825963A (en) | Virtual service transition method | |
| Chen et al. | A traffic identification based on PSO-RBF neural network in peer-to-peer network | |
| Lei et al. | Optimizing traffic classification using hybrid feature selection | |
| CN109492677A (en) | Time-varying network link prediction method based on bayesian theory | |
| Liu et al. | Identifying user clicks based on dependency graph | |
| CN111917589B (en) | Power communication network resource backup method and related device | |
| Chu et al. | A new P2P traffic identification methodology based on flow statistics | |
| Gong et al. | MSTP Network Data Traffic Anomaly Optimization Detection Algorithm | |
| CN115314500B (en) | Dynamic load balancing method based on improved TOPSIS model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |