CN105490834A - Probe deployment method based on vertex cover and weak vertex cover - Google Patents

Abstract

The invention relates to a probe deployment method based on vertex cover and weak vertex cover. The probe deployment method comprises the following steps: determining the quantity of probes and deployment locations of test probes in a network. The probe deployment method provided by the invention has the advantages as follows: based on a greedy algorithm of minimal vertex-covering problems, the method is improved on the basis of the algorithm, and a probe deployment algorithm based on the vertex cover and weak vertex cover is provided. Firstly, minimum vertex cover is used, so that the quantity of flow monitors is minimum under the condition that the flow of each link is obtained, and then the minimum vertex cover is improved as minimum weak vertex cover. Simulation results show that compared with a deployment scheme based on the minimum vertex cover, the scheme provided by the invention has the advantages that fewer probes are used, in addition, the algorithm is simpler, the consumed time is short, and the scheme is a more excellent network flow monitoring probe deployment scheme.

Description

Probe deployment method based on vertex coverage and weak vertex coverage

Technical Field

The invention belongs to a probe-based network flow detection technology, and relates to a probe deployment method capable of realizing comprehensive detection of a network layer of a data network.

Background

For the service requirement of network flow monitoring, a NetFlow detection mode can be used. Setting up a probe at a node in the network can get traffic on all links associated with this node. Therefore, in order to obtain the network traffic of all links in the network, which can be generally realized by configuring the monitor at some switching nodes (routers), we consider the following problems: monitors are placed at which nodes of the network to minimize the number of traffic monitors required given the availability of traffic on each link.

And the network layer of the data network can monitor the service comprehensively through the probe. If the coverage is full, not only the cost is high, but also a large amount of measurement data is generated during measurement, which causes noise. In a network formed by n network elements, measuring traffic characteristics among k network elements requires picking k network elements from n, and in addition, considering the flow direction problem, also needs considering the permutation and combination of k nodes, and the measurement scheme requires C (n, k) P (k, k). If the time complexity resulting from such full coverage measurement methods and schemes is factorial, the spatial complexity is also enormous. Starting around the requirement of business monitoring, research is carried out on how to deploy a proper amount of probes to meet different business monitoring requirements so as to solve the problems of deployment positions and deployment capacity.

In order to solve the development situation of the prior art, the existing papers and patents are searched, compared and analyzed, and the following technical information with high relevance to the invention is screened out:

the technical scheme 1: a patent of 'network coding pairing relation management method in PON based on flow monitoring' with patent number CN103138833A relates to the technical field of communication, and is mainly completed by three steps: firstly, an OLT establishes a network coding connection relation of an ONU pair with bidirectional data transmission, and when the OLT detects bidirectional data stream interaction of the ONU pair, a network coding process can be started; secondly, the OLT counts all ONU pairs with bidirectional data transmission based on the ONU identifiers, and converts the bidirectional data flow into corresponding influence factors by adopting a certain algorithm; thirdly, the OLT monitors the change condition of the influence factor in real time, and if the influence factor IFn of the ONU pair not participating in the network coding is greater than the influence factor IFp of the ONU pair participating in the network coding, the ONU pair corresponding to IFp is removed, and meanwhile, a network coding connection is established for the ONU pair corresponding to IFn.

The technical scheme 2 is as follows: the patent of intelligent flow monitoring method based on wireless sensor network and 3G network with patent number CN102781014A relates to seamless connection technology of flow data collected by sensor and information of sensor-less network, and protocol conversion technology between 802.15.4 protocol and TCP/IP protocol. The data acquisition module acquires dynamically-changed parameters such as different fluid media, pipe diameters and the like, the acquired parameter values are transmitted to the ZigBee wireless communication module of the wireless sensor network in an asynchronous transmission mode, then transmitted to the microprocessor module to be processed and operated, and finally transmitted to the upper computer monitoring center through the 3G gateway, so that remote flow monitoring and management are realized.

Technical scheme 3: a patent of "method and device for determining a flow monitoring baseline", with a patent number of CN103036741A, relates to a method and device for determining a flow monitoring baseline, and determines a historical flow data graph in a preset period in a rectangular coordinate plane, wherein a first coordinate axis in the rectangular coordinate plane represents time in the preset period, and a second coordinate axis in the rectangular coordinate plane represents the historical flow data; determining a base line drawing area in the coordinate plane according to an area surrounded by the historical flow data graph and the first coordinate axis and the second coordinate axis; receiving a mouse event triggered by a user according to the historical flow data graph in the baseline drawing area; and determining a flow monitoring baseline value according to the coordinate position of the mouse event in the baseline drawing area. The device includes: the device comprises a first processing module, a second processing module, a receiving module and a third processing module.

The technical scheme 1 can determine whether the ONU is suitable for participating in network coding in real time according to the flow size characteristics of the ONU, so that the optimal network coding pair participating in the network coding is found, the utilization rate of network bandwidth is improved, and the network congestion is further reduced. Coding is only carried out on the basis of flow monitoring, and how to carry out flow monitoring is not intensively researched. The technical scheme 2 adopts a mode of combining wire and wireless, saves a large amount of manpower and material resources for inspection and meter reading, improves the management level of the automation degree, and realizes low power consumption, low cost, automation, networking and intellectualization of flow monitoring. However, the intelligent traffic monitoring method based on the scheme is based on the wireless sensor network and the 3G network, is high in cost, is easy to cause data congestion, and does not consider the optimization problem. Technical scheme 3 can accurately determine the flow monitoring baseline value and avoid false alarm and missed alarm. The emphasis is on studying the determination of flow sensing limits and there is no specific way to study flow monitoring.

Disclosure of Invention

The invention provides a probe deployment method based on vertex coverage and weak vertex coverage to overcome the defects. The invention aims to research a probe deployment method based on a greedy strategy, and the probe deployment problem is realized through a greedy strategy of a branch definition method by improving a greedy algorithm of a minimum vertex coverage problem. The method can ensure that the algorithm is simpler and more convenient, the time consumption is shorter, and the number of used probes is less.

The invention firstly provides a probe deployment scheme based on minimum vertex coverage, and experiments show that the algorithm can obtain the optimal solution of the minimum vertex coverage, but the algorithm has high computational complexity and long time consumption, and cannot be applied to large-scale problems. And the optimal solution covered by the minimum vertex is not necessarily the optimal solution for the deployment of the network traffic monitoring probe. Therefore, the probe deployment scheme based on the minimum vertex coverage is improved on the basis of the probe deployment scheme based on the minimum vertex coverage, compared with the probe deployment scheme based on the minimum vertex coverage, the probe deployment scheme based on the minimum vertex coverage has the advantages that the number of used probes is less, the algorithm is simpler, the time consumption is shorter, and the probe deployment scheme based on the minimum vertex coverage is more excellent.

The invention is realized by the following technical scheme:

the invention discloses a probe deployment method based on vertex coverage and weak vertex coverage, which is characterized by comprising the following steps: the number of test probes and the deployment positions of the test probes in the network are determined.

In the probe deployment method based on the fixed-point coverage, the steps of determining the number of the test probes and the deployment positions of the test probes in the network are as follows:

s1, sequencing degrees of all nodes in a network to obtain k which simultaneously meets the condition that the sum of the degrees of the first k-1 nodes is less than the sum of the degrees of the first k nodes and the sum of the degrees of the first k nodes is greater than the sum of the degrees of the first k nodes;

s2, constructing original image data into a node of a solution space tree, judging whether a solution exists or not by using a delimitation strategy, if no solution exists, adding 1 to k, entering S2 again, and if a solution exists, inserting the solution into a priority queue;

s3, if the priority queue is not empty, taking out the first feasible node from the priority queue, and entering S4, if the priority queue is empty, adding 1 to k, and entering S2 again;

s4, judging whether the current node meets the condition of solution, if so, outputting the solution to exit, and if not, entering S5;

s5, checking whether the current node can be expanded, if the current node cannot be expanded, entering S3 to continue circulation, if the current node can be expanded, expanding, then verifying whether solutions exist in the nodes expanded to the left and right, inserting the expanded nodes with the solutions into a priority queue, and then entering S3 to continue circulation.

In the probe deployment method based on the minimum weak vertex coverage, the steps of determining the number of test probes and the deployment positions of the test probes in a network are as follows:

s1, deleting all nodes with the degree of 1, namely deleting rows with the sum of all row elements of 1 in the incidence matrix;

s2, selecting a node with the maximum number of contained links and marking the node as v_i；

S3, deleting v in incidence matrix_iThe corresponding row and the column where the element with the value of 1 in the row is located; then, other rows with the sum of all row elements not exceeding 1 and columns corresponding to the elements with the median value of 1 in the rows are sequentially deleted in the rest incidence matrixes until new rows and columns can not be deleted;

s4, repeating the operations of S2 and S3 until all the links are included.

Before determining the number of test probes and the deployment positions of the probes in the network, the method further comprises: building a model using the optimal queue and the spatial tree; after the model is built, the number of test probes and the deployment positions of the probes in the network model are determined.

In the invention, the optimal queue and the space tree are used for establishing a model, comprising the following steps:

maintaining currently viable nodes using a priority queue, each node maintaining a number k of vertices that can be selected if the node is a member of the set of nodes₁The number of remaining edges e to be covered, the state of the vertex, and the number of edges D of the vertexegree and other information, the nodes are sorted from large to small according to the number of edges of the vertex, wherein for the state of the vertex, the vertex in the strategy has three states, namely a selected state s1, a non-selected state s2 and a selectable state s3, the selected state s1 corresponds to the left node in the solution space tree number, the node is selected, and then the node is set to be a selected state s 1; the unselected state s2 corresponds to the right node in the solution space tree, the node is not selected, and then the node is set as the unselected state s 2; the selectable state s3 corresponds to a parent node in the solution space tree, the node is selected, and then the node is set to the selectable state s 3; s1 and s2 are both determined states, and the node can only be selected from s3.

Before determining the number of test probes and the deployment positions of the probes in the network, the method further comprises: establishing a detection model based on the coverage of the minimum weak vertex; after the model is established, the number of test probes and the deployment positions of the probes in the network detection model are determined.

In the invention, the establishment of a detection model based on the minimum weak vertex coverage comprises the following steps:

and establishing an incidence matrix of the nodes and the links, wherein the matrix takes the links as columns and takes the nodes as rows, if the links comprise the nodes, the incidence value is 1, and if not, the incidence value is 0.

And the network layer of the data network can monitor the service comprehensively through the probe. If the coverage is full, not only the cost is high, but also a large amount of measurement data is generated during measurement, which causes noise. In a network formed by n network elements, measuring traffic characteristics among k network elements requires picking k network elements from n, and in addition, considering the flow direction problem, also needs considering the permutation and combination of k nodes, and the measurement scheme requires C (n, k) P (k, k). If the time complexity resulting from such full coverage measurement methods and schemes is factorial, the spatial complexity is also enormous.

Starting around the requirement of business monitoring, research is carried out on how to deploy a proper amount of probes to meet different business monitoring requirements so as to solve the problems of deployment positions and deployment capacity. An example of a deployment that requires resolution is "a certain data network topology".

1. Assuming that the link weights in the data network in the problem are all equal, and assuming that the link weights are 1;

2. assuming that a communication routing rule among network elements in a problem adopts a shortest-path algorithm;

3. in the minimum weak vertex coverage model, it is assumed that one network element can monitor the traffic of all links directly related to the network element and satisfy traffic conservation;

4. in the minimum set coverage model, it is assumed that any two network elements can monitor the states of all links on their paths through communication, and a network monitoring command occurs between probes;

5. since the number of degrees of network edge nodes is usually small, if the position of a probe station is chosen at the edge of the network, the number of probes sent from that probe station and the capability of the probes are correspondingly limited, thus assuming that the probe position is not at the edge of the network.

For the service requirement of network flow monitoring, a NetFlow detection mode can be used. A probe is arranged at a certain node in the network, so that the flow on all links connected with the node can be obtained; therefore, in order to obtain the network traffic of all links in the network, which can be generally realized by configuring the monitor at some switching nodes (routers), we consider the following problems: monitors are placed at which nodes of the network to minimize the number of traffic monitors required given the availability of traffic on each link.

This problem can be solved using minimum vertex coverage. The vertex coverage problem has proven to be NP-complete and can only be approximated by an approximation algorithm to compute a near-optimal solution. A greedy algorithm of a minimum vertex coverage problem is improved, and a probe deployment problem is achieved through a greedy strategy of a branch definition method.

When a network monitoring mode is considered, and the constraint of traffic conservation is added, fewer probes can be deployed by using the minimum weak vertex coverage to realize the network traffic monitoring of the whole network. Also the problem of minimal weak vertex coverage has been shown to be NP-complete, and only approximately optimal solutions can be found using approximation algorithms. We use a greedy strategy here to implement the probe deployment problem.

Evaluation indexes for flow monitoring test:

1. the number of deployed probes;

2. the link coverage.

We need to convert the network map that needs to deploy probes into an undirected graph G (V, E), where V ═ V (V)₁,v₂,…,v_n) For a set of network nodes (which may be considered routers in an IP network), E ═ E₁,e₂,…,e_m) Is a collection of links in the network. Where n ═ V |, m ═ E | respectively represent the number of nodes and links in G. With e_k＝(v_i,v_j) To represent_ekIs a connecting node_viAnd v_jA link of (2).

The number of links of the node v, i.e. the degree of the node v, is represented by Degree (v).

All the routers, switches and other devices of the original network graph are labeled according to the sequence of 1,2 and … …, all the link information is counted, and the graph is converted into an undirected graph. The network has 160 nodes and 185 links. Is a sparse graph (the number of edges, m, is much less than n)²The graph of (b) can be expressed as an undirected graph by an adjacency list, but in general, an undirected graph is expressed by an adjacency matrix so as to solve various graphs.

First, a probe deployment scheme based on minimum vertex coverage is introduced:

the problem model can be conveniently described according to the following definitions

Definition 1: given an undirected graph G (V, E), where V is a set of vertices, E is a set of edges, and S is a subset of V, S is a set of measurements of graph G on traffic if the traffic of any edge in E can be determined from the traffic of the edges associated with the vertices in S.

The goal of the efficient measurement problem is to find the minimum set of measurements for a given graph G with respect to flow. It can be translated into the minimum vertex coverage problem of graph G in definition 2.

Define 2 (minimum vertex coverage problem): given an undirected graph G (V, E), a minimum subset S of the set of vertices V is found, such that E ═ u, V ∈ E, and u ∈ S or V ∈ S, i.e. any edge in E contains at least one point in this subset as a vertex, i.e. the vertex in S covers the set of edges E.

The mathematical description is as follows:

mincx where x_i∈S

$\underset{i=1}{\overset{n}{\Σ}}\underset{j=1}{\overset{n}{\Σ}}{a}_{ij}(1-x_i)(1-x_j)=0---1$

Wherein c ═ 1,1, …,1]And x ═ x₁,x₂,…,x_n]Are all vectors of length n, (a)_ij)_n×nIs a contiguous matrix of the graph, an

$xi=\left\{\begin{array}{cc}1,& v_i\∈S\\ 0,& v_i\∉S\end{array}\right.---2$

The problem of minimum vertex coverage is proven to be NP-complete, a polynomial time algorithm is not available for solving so far, a greedy algorithm can be used for solving an approximate optimal solution, namely, a node with the maximum degree is selected as a probe in each iteration, then all nodes connected with the probe can be eliminated, the condition that the probe is selected each time is the optimal condition of the current situation is met, however, the optimal condition of each time cannot be obtained integrally, in order to find a smaller solution, the greedy is improved through the following limiting strategy, and the better solution is obtained through a branch boundary method.

The degrees of all nodes are sorted from large to small, the first k nodes are taken, and the conditions are met

$\left\{\begin{array}{c}\underset{i=1}{\overset{k-1}{\&Sigma;}}Degree\left(v_i\right)<\left|E\right|\\ \underset{i=1}{\overset{k}{\&Sigma;}}Degree\left(v_i\right)\&GreaterEqual;\left|E\right|\end{array}\right.---3$

It is possible to satisfy the bounds of vertex coverage.

By means of the constraint, a minimum possible optimal solution k is obtained before a greedy algorithm is used, and the problem is converted into the situation that whether k nodes in an undirected graph meet the minimum vertex coverage or not is judged. And omitting judgment through a greedy strategy and a limiting strategy, if the judgment is satisfied, k is an optimal solution, and if the judgment is not satisfied, adding 1 to k, and continuing the judgment.

Before introducing the specific decision process, the model is first built using an optimal queue and spatial tree, with a priority queue maintaining the currently feasible nodes, each node maintaining the number of vertices k that can be selected if the node is still in place₁And the number e of the residual edges to be covered, the state of the vertex, the number Degreee of the vertex and the like, and the sequencing of the nodes follows a greedy strategy and is sequenced according to the number of the edges of the vertex from large to small. For the state of the vertex. The top points in the strategy have three states, namely a selected state S1, an unselected state S2 and a selectable state S3. Wherein, the selected state S1 corresponds to the left node in the solution space tree number, selects the node, and then sets the node as the selected state S1; the unselected state S2 corresponds to the right node in the solution space tree, which is unselected and then set to the unselected state S2. The selectable state S3 corresponds to a parent node in the solution space tree, selects the node, and then sets the node to the selectable state S3. S1 and S2 are both determined states, and the node can only be selected from S3.

Taking out the first node in the optimal queue for expansion:

setting the node as S1 while modifying the information of the node according to equation 4, and then placing the node in the left node of the spatial tree; the node is set to S2 and is placed directly into the right node of the spatial tree without modifying other node information.

k₁＝k₁-1,e＝e-Degree,Degree＝04

By constraint conditionsJudging a right node, and if the right node meets the requirement, inserting the right node into an optimal queue; then selecting left node to judgeAnd if so, inserting the node into the optimal queue.

The iteration of the above steps is a specific decision process.

The corresponding flow chart is shown in fig. 1.

The probe deployment algorithm based on the minimum vertex coverage is improved, and the probe deployment algorithm based on the minimum weak vertex coverage is provided:

on the basis of definition 1, since the probe is disposed on a switching device such as a router or a switch, according to the conservation of traffic, the graph G further satisfies the following two constraints:

1. for any vertex V in the vertex set V of the graph G, the degree Degree (V) is more than or equal to 2;

2. for any vertex V in the set of vertices V of the graph G, the flow conservation equation is satisfied, i.e., flow in versus flow out.

Although the following causes will result in distortion of the flow conservation equations, such as: switching devices are sources or sinks of data, not just data forwarders; multicasting results in data replication at the output port; the switching device itself has data packets delayed or lost. Several studies have shown that the flow conservation equation has a relative error of less than 0.05%.

Definitions 3 (Weak vertex coverage) assume that the undirected graph G (V, E) mesh satisfies Degree (V) ≧ 2 for any V ∈ V, calledIs the weak vertex coverage set of graph G, all edges in E can be marked if and only if the following operations are performed:

(1) marking all edges associated with the vertices in S;

(2) if Degree (v) -1 associated edge of a certain vertex v has been marked, marking the rest associated edges;

(3) and (3) repeating the step (2) until a new edge can not be marked.

Obviously, the weak vertex coverage S can obtain the traffic of each link in the graph G. The concept of the incidence matrix is used herein to build a model for solving the weak vertex coverage problem, for which the following definitions are given first.

Definition 4 (association relationship) in undirected graph G (V, E), if V ∈ V is one of the vertices of E ∈ E, it is said that there is an association relationship between V and E, and it is denoted as vR_e

Define 5 (correlation matrix) the correlation matrix a ═ a of the graph G (V, E)_ij) Is that

Refers to an n × m matrix defined as:

from the definition of the correlation matrix, it is known that a subset of V constitutes an overlay of graph G if and only if there is at least one 1 per column in the row of the correlation matrix corresponding to the node it contains.

According to the analysis, the weak vertex coverage problem can be solved by adopting a greedy algorithm, and it is noted that the boundary point degree in the network graph is 1 and does not meet the condition constraint 1, so that the undirected graph needs to be preprocessed.

For vertex v with degree 1 in the initial case_iOnly one edge is associated with the point, the edge being denoted as e_j，e_jIs denoted as v_j。

When v is_jWith only one degree-1 adjacent vertex v_iIf v is obtained_jOther related side flow information can calculate e according to the flow_ijThe flow rate of (c). I.e. v_jSatisfying minimum weak vertex coverage, calculating the degree of the edge only requires e_ijAdding; or v_jIs also directly monitored when the probe is deployed_ijThe flow rate of (c).

When v is_jThere are more (k | k > 1) adjacent vertices with degree 1, if node v is not located_jIf probes are used, at least k probes need to be provided, so that only the node v is connected_jThe probe is set to ensure that the minimum number of probes can be deployed.

From the above analysis, it is known that the nodes whose initial degree is 1 may not be provided with any probe and may be discarded directly, but when one node whose degree is 1 is discarded by the pretreatment, the degrees of other nodes connected thereto are not reduced by 1.

In view of the above flow monitoring model based on the minimum weak vertex coverage, the algorithm flow is designed as follows:

step 1: deleting all nodes with the degree of 1, namely deleting the row with the sum of all row elements of 1 in the incidence matrix;

step 2: selecting a node with the maximum number of contained links, and marking the node as v_i；

Step 3: deleting v in the incidence matrix_iThe corresponding row and the column where the element with the value of 1 in the row is located; then, other rows with the sum of all row elements not exceeding 1 and columns corresponding to the elements with the median value of 1 in the rows are sequentially deleted in the rest incidence matrixes until new rows and columns can not be deleted;

step 4: the operations of Step2 and Step3 are repeated until all links are included.

The corresponding algorithm flow chart is shown in fig. 2.

The key points of the technology of the invention are as follows:

1. the algorithm improves a greedy algorithm of the minimum vertex coverage problem, and realizes the probe deployment problem through a greedy strategy of a branch definition method;

2. when a network monitoring mode is considered, and the constraint of traffic conservation is added, fewer probes can be deployed by using the minimum weak vertex coverage to realize the network traffic monitoring of the whole network.

The invention has the beneficial effects that: the probe deployment technology based on the vertex coverage and the weak vertex coverage has the advantages that the technology is based on a greedy algorithm for solving the problem of minimum vertex coverage, is improved on the basis of the algorithm, and is provided. Minimum vertex coverage is used first to minimize the number of traffic monitors required given the availability of traffic on each link. And then improved to minimum weak vertex coverage. Simulation shows that compared with a deployment scheme based on minimum vertex coverage, the deployment scheme based on the minimum vertex coverage has the advantages that the number of probes used is less, an algorithm is simpler, time consumption is shorter, and the deployment scheme based on the minimum vertex coverage is a more excellent network traffic monitoring probe deployment scheme.

Drawings

FIG. 1 is a flowchart of a probe deployment algorithm based on minimum vertex coverage;

fig. 2 is a flowchart of an algorithm based on minimum weak vertex coverage.

Detailed Description

The invention discloses a probe deployment method based on vertex coverage and weak vertex coverage, which is characterized by comprising the following steps: the number of test probes and the deployment positions of the test probes in the network are determined.

In the probe deployment method based on the fixed-point coverage, the steps of determining the number of the test probes and the deployment positions of the test probes in the network are as follows:

s1, sequencing degrees of all nodes in a network to obtain k which simultaneously meets the condition that the sum of the degrees of the first k-1 nodes is less than the sum of the degrees of the first k nodes and the sum of the degrees of the first k nodes is greater than the sum of the degrees of the first k nodes;

s2, constructing original image data into a node of a solution space tree, judging whether a solution exists or not by using a delimitation strategy, if no solution exists, adding 1 to k, entering S2 again, and if a solution exists, inserting the solution into a priority queue;

s3, if the priority queue is not empty, taking out the first feasible node from the priority queue, and entering S4, if the priority queue is empty, adding 1 to k, and entering S2 again;

s4, judging whether the current node meets the condition of solution, if so, outputting the solution to exit, and if not, entering S5;

s5, checking whether the current node can be expanded, if the current node cannot be expanded, entering S3 to continue circulation, if the current node can be expanded, expanding, then verifying whether solutions exist in the nodes expanded to the left and right, inserting the expanded nodes with the solutions into a priority queue, and then entering S3 to continue circulation.

In the probe deployment method based on the minimum weak vertex coverage, the steps of determining the number of test probes and the deployment positions of the test probes in a network are as follows:

s1, deleting all nodes with the degree of 1, namely deleting rows with the sum of all row elements of 1 in the incidence matrix;

s2, selecting a node with the maximum number of contained links and marking the node as v_i；

S3, deleting v in incidence matrix_iThe corresponding row and the column where the element with the value of 1 in the row is located; then, other rows with the sum of all row elements not exceeding 1 and columns corresponding to the elements with the median value of 1 in the rows are sequentially deleted in the rest incidence matrixes until new rows and columns can not be deleted;

s4, repeating the operations of S2 and S3 until all the links are included.

Before determining the number of test probes and the deployment positions of the probes in the network, the method further comprises: building a model using the optimal queue and the spatial tree; after the model is built, the number of test probes and the deployment positions of the probes in the network model are determined.

In the invention, the optimal queue and the space tree are used for establishing a model, comprising the following steps:

maintaining currently viable nodes using a priority queue, each node maintaining a number k of vertices that can be selected if the node is a member of the set of nodes₁The nodes are sorted from large to small according to the number of edges of the vertex, wherein for the state of the vertex, the vertex has three states, namely a selected state s1, an unselected state s2 and a selectable state s3, the selected state s1 corresponds to a left node in the solution space tree number, the node is selected, and then the node is set to be a selected state s 1; the unselected state s2 corresponds to the right node in the solution space tree, the node is not selected, and then the node is set as the unselected state s 2; the selectable state s3 corresponds to a parent node in the solution space tree, the node is selected, and then the node is set to the selectable state s 3; s1 and s2 are both determined states, and the node can only be selected from s3.

Before determining the number of test probes and the deployment positions of the probes in the network, the method further comprises: establishing a detection model based on the coverage of the minimum weak vertex; after the model is established, the number of test probes and the deployment positions of the probes in the network detection model are determined.

In the invention, the establishment of a detection model based on the minimum weak vertex coverage comprises the following steps:

and establishing an incidence matrix of the nodes and the links, wherein the matrix takes the links as columns and takes the nodes as rows, if the links comprise the nodes, the incidence value is 1, and if not, the incidence value is 0.

As shown in table 1, by deploying the network by using the minimum vertex coverage model of the branch definition method, the coverage rate of the traffic monitoring network can be 100% only by deploying 43 probes, that is, the traffic monitoring of each edge of the whole graph of the network graph can be realized by deploying 43 probes based on the probe deployment scheme of the minimum vertex coverage.

TABLE 1 evaluation index

Number of network nodes	Number of probes	Traffic monitoring network coverage
			160	43	100％

It should be noted that although the branch definition method can obtain the optimal solution of the minimum vertex coverage, it has high computational complexity and takes a long time, and cannot be applied to a large scale problem. Moreover, the optimal solution of the minimum vertex coverage is not necessarily the optimal solution for the deployment of the network traffic monitoring probe, so in order to find a deployment scheme with simpler algorithm, shorter time consumption and fewer probes, the probe is deployed through the minimum weak vertex coverage by combining the property of network traffic conservation.

As shown in table 2, by using the method of minimum weak fixed point coverage, only 32 probes need to be deployed to achieve 100% coverage of the traffic monitoring network. In this method, the network traffic of a partial path is calculated by traffic conservation.

TABLE 2 evaluation index

Number of network nodes	Number of probes	Traffic monitoring network coverage
			160	32	100％

Compared with a deployment scheme based on minimum vertex coverage, the deployment scheme has the advantages that the number of probes used is less, the algorithm is simpler, the time consumption is shorter, and the deployment scheme is a more excellent network traffic monitoring probe deployment scheme.

However, both the two probe deployment prevention cases are passive tests, and can collect data packets transmitted in the network by using a port mirror image and multi-path forwarding mode in a link series connection mode and the like, wherein the data packets include service data packets, signaling data packets, management information data packets and the like, analyze the network performance, and passively monitor the network performance.

Claims (5)

1. A probe deployment method based on vertex coverage and weak vertex coverage is characterized by comprising the following steps: determining the number of test probes and the deployment positions of the test probes in a network; wherein,

the method for determining the number of the test probes and the deployment positions of the test probes in the network comprises the following steps:

s1, sequencing degrees of all nodes in a network to obtain k which simultaneously meets the condition that the sum of the degrees of the first k-1 nodes is less than the sum of the degrees of the first k nodes and the sum of the degrees of the first k nodes is greater than the sum of the degrees of the first k nodes;

s2, constructing original image data into a node of a solution space tree, judging whether a solution exists or not by using a delimitation strategy, if no solution exists, adding 1 to k, entering S2 again, and if a solution exists, inserting the solution into a priority queue;

s3, if the priority queue is not empty, taking out the first feasible node from the priority queue, and entering S4, if the priority queue is empty, adding 1 to k, and entering S2 again;

s4, judging whether the current node meets the condition of solution, if so, outputting the solution to exit, and if not, entering S5;

s5, checking whether the current node can be expanded or not, if the current node cannot be expanded, entering S3 to continue circulation, if the current node can be expanded, expanding, then verifying whether solutions exist in the nodes expanded to the left and right, inserting the expanded nodes with the solutions into a priority queue, and then entering S3 to continue circulation;

the method for determining the number of the test probes and the deployment positions of the test probes in the network comprises the following steps:

s1, deleting all nodes with the degree of 1, namely deleting rows with the sum of all row elements of 1 in the incidence matrix;

s2, selecting a node with the maximum number of contained links and marking the node as v_i；

S3, deleting v in incidence matrix_iThe corresponding row and the column where the element with the value of 1 in the row is located; then, other rows with the sum of all row elements not exceeding 1 and columns corresponding to the elements with the median value of 1 in the rows are sequentially deleted in the rest incidence matrixes until new rows and columns can not be deleted;

s4, repeating the operations of S2 and S3 until all the links are included.

2. The method of claim 1, wherein before determining the number of test probes and the deployment positions of the probes in the network, the method further comprises:

building a model using the optimal queue and the spatial tree;

after the model is built, the number of test probes and the deployment positions of the probes in the network model are determined.

3. The method for probe deployment based on vertex coverage and weak vertex coverage according to claim 2, wherein the optimal queue and spatial tree are used for establishing the model, and the method comprises the following steps:

maintaining currently viable nodes using a priority queue, each node maintaining a number k of vertices that can be selected if the node is a member of the set of nodes₁The nodes are sorted from large to small according to the number of edges of the vertex, wherein for the state of the vertex, the vertex has three states, namely a selected state s1, an unselected state s2 and a selectable state s3, the selected state s1 corresponds to a left node in the solution space tree number, the node is selected, and then the node is set to be a selected state s 1; the unselected state s2 corresponds to the right node in the solution space tree, the node is not selected, and then the node is set as the unselected state s 2; the selectable state s3 corresponds to a parent node in the solution space tree, the node is selected, and then the node is set to the selectable state s 3; s1 and s2 are both determined states, and the node can only be selected from s3.

4. The method for deploying the probe based on the vertex coverage and the weak vertex coverage as claimed in claim 1,2 or 3, wherein before determining the number of the test probes and the deployment positions of the probes in the network, the method further comprises:

establishing a detection model based on the coverage of the minimum weak vertex;

after the model is established, the number of test probes and the deployment positions of the probes in the network detection model are determined.

5. The method for deploying the probe based on the vertex coverage and the weak vertex coverage as claimed in claim 4, wherein establishing the detection model based on the minimum weak vertex coverage comprises:

and establishing an incidence matrix of the nodes and the links, wherein the matrix takes the links as columns and takes the nodes as rows, if the links comprise the nodes, the incidence value is 1, and if not, the incidence value is 0.