A method and device for rapid reconstruction of network topology based on link relationship discovery

Through deep learning network model and correlation analysis technology, the link relationship characteristics of network topology are extracted, and the problem of low topology reconstruction efficiency of traditional algorithms under dynamic network changes is solved, and accurate dynamic tracking and reconstruction of network topology is achieved. Technical support is provided for network surveying, mapping and management.

CN118337634BActive Publication Date: 2024-11-22NO 15 INST OF CHINA ELECTRONICS TECH GRP
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410562847.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-08
Publication Date
2024-11-22
Estimated Expiration
2044-05-08

AI Technical Summary

Technical Problem

Traditional machine learning algorithms are difficult to obtain random change information of network topology accurately in real time and in terms of rapid reconstruction of network topology, and cannot fully extract the multi-dimensional and distributed features of link relationships, resulting in low efficiency in tracking of dynamic change information.

Method used

Using deep learning network model, correlation analysis feedback mechanism and automatic classification aggregation algorithm, node weights, node strength and local path information are extracted through link weighted relationship learning indicators, and a hierarchical dynamic network logic topology is constructed to realize automatic link relationships. Learn and precise calibration.

Benefits of technology

It realizes automatic discovery and reconstruction of the link relationship of the accurate network topology node under dynamic network changes, providing technical support and data foundation for subsequent surveying and mapping, penetration testing, etc.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118337634B_ABST
    Figure CN118337634B_ABST
Patent Text Reader

Abstract

The present invention discloses a method and device for rapid reconstruction of network topology based on link relationship discovery, which is used to solve the problem that in the case of dynamic changes in the network, traditional machine learning algorithms are difficult to accurately obtain random change information of network topology in real time in terms of rapid reconstruction of network topology. The method includes: analyzing the target network to obtain link weighted relationship learning indicators; processing the link weighted relationship learning indicators to obtain link relationships within a network snapshot; using a link relationship feature extraction model to process the link weighted relationship learning indicators to obtain link relationships between network snapshots; processing the link relationships within a network snapshot and the link relationships between network snapshots to obtain network topology information; processing the network topology information to obtain a rapidly reconstructed network topology result. The present invention can accurately extract the distributed features of link relationships and realize automatic learning and precise correction of link relationships.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of network topology reconstruction, and in particular to a method and device for rapid network topology reconstruction based on link relationship discovery. Background Art

[0002] In a real network environment, network node links are in a state of constant dynamic change, which in turn leads to changes in the target network topology. In this dynamic network environment, it is difficult to obtain the comprehensive link status of the network topology in real time, and it is impossible to effectively support subsequent penetration testing of the target network. Therefore, it is an important research hotspot for target mapping analysis to flexibly and dynamically detect the potential structure of the network to synchronously update the target network topology information.

[0003] The key to the flexible reconstruction of the target network topology lies in the automatic learning of the communication relationship between different pairs of nodes in the target network, that is, analyzing the addition and deletion of edges in the network, determining the node-link relationship of the target network, and then combining information such as device attributes and node attributes to draw the target network topology in real time and update it as needed. At present, the link relationship learning methods based on machine learning mainly focus on supervised classification algorithms and related semi-supervised classification learning algorithms. These methods all use shallow learning algorithms, and the implementation process has shortcomings such as low real-time processing results and low efficiency in dynamically tracking the change information of the target network topology structure. Since the multi-dimensional and distributed characteristics of the link relationship cannot be fully extracted, it is impossible to construct a comprehensive and reasonable autonomous learning indicator that reflects the dynamic changes of the network. Under the condition of dynamic changes in the network, it is difficult for traditional machine learning algorithms to accurately obtain random changes in network topology in real time in terms of rapid reconstruction of network topology. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method and device for rapid reconstruction of network topology based on link relationship discovery, which is used to solve the problem that traditional machine learning algorithms are difficult to accurately obtain random change information of network topology in real time in terms of rapid reconstruction of network topology under dynamic network changes. The present invention comprehensively adopts deep learning network models, association analysis feedback mechanisms and automatic classification and convergence algorithms to accurately extract distributed features of link relationships, realize automatic learning and precise correction of link relationships, and then assist in the construction of hierarchical dynamic network logical topology, providing technical support and data basis for subsequent comprehensive mapping and penetration testing of target networks.

[0005] In order to solve the above technical problems, the first aspect of the embodiment of the present invention discloses a method for rapid reconstruction of network topology based on link relationship discovery, the method comprising:

[0006] S1, analyzing the target network to obtain link weighted relationship learning indicators; the link weighted relationship learning indicators include node weight information, node strength information and local path information;

[0007] S2, processing the link weighted relationship learning indicator to obtain the link relationship in the network snapshot;

[0008] S3, using a link relationship feature extraction model to process the link weighted relationship learning indicator to obtain the link relationship between network snapshots;

[0009] S4, processing the link relationship within the network snapshot and the link relationship between the network snapshots to obtain network topology information;

[0010] S5, processing the network topology information to obtain a rapidly reconstructed network topology.

[0011] As an optional implementation manner, in the first aspect of the embodiment of the present invention, the analyzing the target network to obtain the link weighted relationship learning index includes:

[0012] S11, dividing the target network into stages to obtain N target network stages, where N is a positive integer;

[0013] S12, for any node pair in the target network phase i<m,n> Processing to obtain node pairs<m,n> Node weight information, node pair<m,n> Node strength information and node pairs<m,n> Local path information, i = 1, 2, ..., N;

[0014] S13, for the node pair<m,n> The node weight information of<m,n> The node strength information and the node pair<m,n> The local path information is integrated to obtain the link weighted relationship learning index.

[0015] As an optional implementation manner, in the first aspect of the embodiment of the present invention, the pair of any node in the target network<m,n> Processing to obtain node pairs<m,n> Node weight information, node pair<m,n> Node strength information and node pairs<m,n> The local path information includes:

[0016] S121, the length of time that node m and node n remain connected in the Δt time period is processed to obtain the node pair<m,n> Node weight information;

[0017] The node pair<m,n> The calculation formula of node weight information is:

[0018]

[0019] In the formula, cmn For node pairs<m,n> The node weight information, t mn is the time length that nodes m and n remain connected within the Δt period, Δt 12 is the interval between the two network states corresponding to node m and node n in the Δt time period;

[0020] S122, traverse different nodes in the node set that has a communication relationship with node m in turn, and obtain node pairs<m,n> Node strength information;

[0021] The node pair<m,n> The node strength information of is the sum of the node weight information of each node in the node set that has a communication relationship with node m and node m;

[0022] S123, calculating and summing the node strength information of each node on the j-th path between node m and node n, except node m and node n, to obtain the associated path contribution information, where j is the path number;

[0023] S124, calculating the first associated path contribution information of the path with a length of l between node m and node n, and taking the inverse of the first associated path contribution information and summing them up to obtain the node pair<m,n> Local path information.

[0024] As an optional implementation, in the first aspect of the embodiment of the present invention, the link-weighted relationship learning indicator is processed by using a link-relationship-oriented feature extraction model to obtain a link relationship between network snapshots, including:

[0025] S31, using link weighted relationship learning indicators of different node pairs at different target network stages, training a preset feature extraction model to obtain a link relationship oriented feature extraction model;

[0026] S32, using the link relationship oriented feature extraction model, processing the link weighted relationship learning index to obtain the link relationship between network snapshots.

[0027] As an optional implementation, in the first aspect of the embodiment of the present invention, the link-relationship-oriented feature extraction model is used to process the link weighted relationship learning indicator to obtain the link relationship between network snapshots, including:

[0028] S321, using a first feature processing model, processing the link weighted relationship learning indicator to obtain first feature information;

[0029] S322, using a second feature extraction model to process the first feature information to obtain second feature information;

[0030] S323: Process the second feature information using the link relationship oriented feature extraction model to obtain link relationships between network snapshots.

[0031] As an optional implementation manner, in the first aspect of the embodiment of the present invention, the processing of the link relationship within the network snapshot and the link relationship between the network snapshots to obtain the network topology information includes:

[0032] S41, parsing the link relationship within the network snapshot and the link relationship between the network snapshots to obtain node topology information of the target network;

[0033] The node topology information of the target network includes node IP address, MAC address, subnet mask length, gateway address, gateway IP address, online status, parent node address, node name, node type and node system version;

[0034] S42, filtering the node topology information of the target network to obtain the node topology information of the first target network;

[0035] S43, normalizing the node topology information of the first target network to obtain the node topology information of the second target network;

[0036] S44, extracting key information from the node topology information of the second target network to obtain node topology information of the third target network;

[0037] S45, processing the node topology information of the third target network to obtain network topology information.

[0038] As an optional implementation manner, in the first aspect of the embodiment of the present invention, the processing of the network topology information to obtain a rapidly reconstructed network topology includes:

[0039] S51, analyzing the network topology information to obtain real-time node attribute information, network topology attribute information and target network link connection information;

[0040] S52, using a preset network aggregation model, classify and aggregate the real-time node attribute information, the network topology attribute information and the target network link connection information to obtain a rapidly reconstructed network topology.

[0041] A second aspect of an embodiment of the present invention discloses a network topology rapid reconstruction device based on link relationship discovery, the device comprising:

[0042] A target network analysis module is used to analyze the target network and obtain link weighted relationship learning indicators; the link weighted relationship learning indicators include node weight information, node strength information and local path information;

[0043] An indicator analysis module, used to process the link weighted relationship learning indicator to obtain the link relationship in the network snapshot;

[0044] A feature extraction module, used to process the link weighted relationship learning index using a link relationship-oriented feature extraction model to obtain the link relationship between network snapshots;

[0045] A network topology information calculation module, used to process the link relationship within the network snapshot and the link relationship between the network snapshots to obtain network topology information;

[0046] The network topology reconstruction module is used to process the network topology information to obtain a rapidly reconstructed network topology.

[0047] As an optional implementation manner, in the second aspect of the embodiment of the present invention, the analyzing the target network to obtain the link weighted relationship learning index includes:

[0048] S11, dividing the target network into stages to obtain N target network stages, where N is a positive integer;

[0049] S12, for any node pair in the target network phase i<m,n> Processing to obtain node pairs<m,n> Node weight information, node pair<m,n> Node strength information and node pairs<m,n> Local path information, i = 1, 2, ..., N;

[0050] S13, for the node pair<m,n> The node weight information of<m,n> The node strength information and the node pair<m,n> The local path information is integrated to obtain the link weighted relationship learning index.

[0051] As an optional implementation, in the second aspect of the embodiment of the present invention, the pair of any node in the target network<m,n> Processing to obtain node pairs<m,n> Node weight information, node pair<m,n> Node strength information and node pairs<m,n> The local path information includes:

[0052] S121, the length of time that node m and node n remain connected in the Δt time period is processed to obtain the node pair<m,n> Node weight information;

[0053] The node pair<m,n> The calculation formula of node weight information is:

[0054]

[0055] In the formula, c mn For node pairs<m,n> The node weight information, tmn is the time length that nodes m and n remain connected within the Δt period, Δt 12 is the interval between the two network states corresponding to node m and node n in the Δt time period;

[0056] S122, traverse different nodes in the node set that has a communication relationship with node m in turn, and obtain node pairs<m,n> Node strength information;

[0057] The node pair<m,n> The node strength information of is the sum of the node weight information of each node in the node set that has a communication relationship with node m and node m;

[0058] S123, calculating and summing the node strength information of each node on the j-th path between node m and node n, except node m and node n, to obtain the associated path contribution information, where j is the path number;

[0059] S124, calculating the first associated path contribution information of the path with a length of l between node m and node n, and taking the inverse of the first associated path contribution information and summing them up to obtain the node pair<m,n> Local path information.

[0060] As an optional implementation, in the second aspect of the embodiment of the present invention, the link-weighted relationship learning indicator is processed by using a link-relationship-oriented feature extraction model to obtain a link relationship between network snapshots, including:

[0061] S31, using link weighted relationship learning indicators of different node pairs at different target network stages, training a preset feature extraction model to obtain a link relationship oriented feature extraction model;

[0062] S32, using the link relationship oriented feature extraction model, processing the link weighted relationship learning index to obtain the link relationship between network snapshots.

[0063] As an optional implementation, in the second aspect of the embodiment of the present invention, the link-weighted relationship learning indicator is processed by using the link-relationship-oriented feature extraction model to obtain the link relationship between network snapshots, including:

[0064] S321, using a first feature processing model, processing the link weighted relationship learning indicator to obtain first feature information;

[0065] S322, using a second feature extraction model to process the first feature information to obtain second feature information;

[0066] S323: Process the second feature information using the link relationship oriented feature extraction model to obtain link relationships between network snapshots.

[0067] As an optional implementation manner, in the second aspect of the embodiment of the present invention, the processing of the link relationship within the network snapshot and the link relationship between the network snapshots to obtain the network topology information includes:

[0068] S41, parsing the link relationship within the network snapshot and the link relationship between the network snapshots to obtain node topology information of the target network;

[0069] The node topology information of the target network includes node IP address, MAC address, subnet mask length, gateway address, gateway IP address, online status, parent node address, node name, node type and node system version;

[0070] S42, filtering the node topology information of the target network to obtain the node topology information of the first target network;

[0071] S43, normalizing the node topology information of the first target network to obtain the node topology information of the second target network;

[0072] S44, extracting key information from the node topology information of the second target network to obtain node topology information of the third target network;

[0073] S45, processing the node topology information of the third target network to obtain network topology information.

[0074] As an optional implementation manner, in the second aspect of the embodiment of the present invention, the processing of the network topology information to obtain a rapidly reconstructed network topology includes:

[0075] S51, analyzing the network topology information to obtain real-time node attribute information, network topology attribute information and target network link connection information;

[0076] S52, using a preset network aggregation model, classify and aggregate the real-time node attribute information, the network topology attribute information and the target network link connection information to obtain a rapidly reconstructed network topology.

[0077] The third aspect of the present invention discloses another network topology rapid reconstruction device based on link relationship discovery, the device comprising:

[0078] A memory storing executable program code;

[0079] a processor coupled to the memory;

[0080] The processor calls the executable program code stored in the memory to execute part or all of the steps in the network topology rapid reconstruction method based on link relationship discovery disclosed in the first aspect of the embodiment of the present invention.

[0081] The fourth aspect of the present invention discloses a computer-storable medium, which stores computer instructions. When the computer instructions are called, they are used to execute some or all of the steps in the network topology rapid reconstruction method based on link relationship discovery disclosed in the first aspect of an embodiment of the present invention.

[0082] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0083] The present invention takes the real target network environment with irregular dynamic changes as the background, focuses on real-time and accurate acquisition of the link connectivity status of the target network topology and the demand for rapid construction of the network topology, and designs a network topology rapid reconstruction method based on dynamic link relationship discovery. Mainly through the automatic discovery of node link relationships based on deep learning, hierarchical network topology dynamic clustering and other means, it solves the problems of low accuracy of topological relationship feature extraction and poor pertinence of learning indicators under dynamic network changes, realizes accurate automatic discovery of network topology node link relationships, automatic tracking and reconstruction of network topology, and provides theoretical and technical support for subsequent comprehensive mapping, penetration testing, management and maintenance of non-cooperative target networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0085] Figure 1 It is a flow chart of a method for rapid reconstruction of network topology based on link relationship discovery disclosed in an embodiment of the present invention;

[0086] Figure 2 It is a flow chart of another method for rapid network topology reconstruction based on link relationship discovery disclosed in an embodiment of the present invention;

[0087] Figure 3 It is a schematic diagram of a DBN model for chain relationship feature extraction disclosed in an embodiment of the present invention;

[0088] Figure 4 It is a structural schematic diagram of a network topology rapid reconstruction device based on link relationship discovery disclosed in an embodiment of the present invention;

[0089] Figure 5 It is a structural schematic diagram of another network topology rapid reconstruction device based on link relationship discovery disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0090] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0091] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, device, product or equipment that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or equipment.

[0092] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0093] The present invention discloses a method and device for rapid reconstruction of network topology based on link relationship discovery, which is used to solve the problem that it is difficult for traditional machine learning algorithms to accurately obtain random change information of network topology in real time in the case of dynamic changes in the network. The method includes: analyzing the target network to obtain link weighted relationship learning indicators; processing the link weighted relationship learning indicators to obtain link relationships within the network snapshot; using a link relationship feature extraction model to process the link weighted relationship learning indicators to obtain link relationships between network snapshots; processing the link relationships within the network snapshot and the link relationships between network snapshots to obtain network topology information; processing the network topology information to obtain a rapidly reconstructed network topology result. The present invention can comprehensively adopt a deep learning network model, an association analysis feedback mechanism, and an automatic classification and convergence algorithm to accurately extract the distributed features of link relationships, realize automatic learning and precise correction of link relationships, and then assist in the construction of a hierarchical dynamic network logical topology, providing technical support and data basis for subsequent comprehensive mapping, penetration testing, etc. for the target network. The following are detailed descriptions.

[0094] Embodiment 1

[0095] See also Figure 1 , Figure 1 1 is a flow chart of a method for rapid network topology reconstruction based on link relationship discovery disclosed in an embodiment of the present invention. Figure 1 The described method for rapid reconstruction of network topology based on link relationship discovery is applied to obtain the comprehensive link status of network topology, effectively supporting subsequent penetration testing of the target network, which is not limited in the embodiments of the present invention. Figure 1 As shown, the network topology rapid reconstruction method based on link relationship discovery may include the following operations:

[0096] S1, analyzing the target network to obtain link weighted relationship learning indicators; the link weighted relationship learning indicators include node weight information, node strength information and local path information;

[0097] S2, processing the link weighted relationship learning indicator to obtain the link relationship in the network snapshot;

[0098] Determine the node pair based on the link weighted relationship learning indicator<m,n> The matching relationship with the target network state at a certain stage, the matching relationship is the link relationship within the network snapshot. In this embodiment, within the network snapshot is within a target network stage, and between network snapshots is between different target network stages. The matching relationship calculation method is the existing technology in the field and is not limited in this embodiment.

[0099] S3, using a link relationship feature extraction model to process the link weighted relationship learning indicator to obtain the link relationship between network snapshots;

[0100] S4, processing the link relationship within the network snapshot and the link relationship between the network snapshots to obtain network topology information;

[0101] S5, processing the network topology information to obtain a rapidly reconstructed network topology.

[0102] Optionally, analyzing the target network to obtain a link weighted relationship learning indicator includes:

[0103] S11, dividing the target network into stages to obtain N target network stages, where N is a positive integer;

[0104] S12, for any node pair in the target network phase i<m,n> Processing to obtain node pairs<m,n> Node weight information, node pair<m,n> Node strength information and node pairs<m,n> Local path information, i = 1, 2, ..., N;

[0105] S13, for the node pair<m,n> The node weight information of<m,n> The node strength information and the node pair<m,n> The local path information is integrated to obtain the link weighted relationship learning index.

[0106] Optionally, feature fusion is used for integration, specifically:

[0107] The two feature matrices that need to be fused are X (such as node weight information) and Y (such as node strength information). X is an n′×m′ dimensional matrix, and Y is a p×m′ dimensional matrix. m′ represents the number of samples, and n′ and p represent the dimensions of the two features. The two matrices are projected to 1D for linear representation, corresponding to the projection vector a respectively. 1 and b 1 , then the feature matrix after projection becomes:

[0108]

[0109] The purpose is to maximize the correlation coefficient between X' and Y', so as to obtain the projection vector a when the correlation coefficient is maximum. 1 and b 1 ,Right now:

[0110]

[0111] Before projection, the data is standardized. The purpose of standardization is to make the mean of the data 0 and the variance 1. We can get:

[0112]

[0113]

[0114]

[0115] D(X)=cov(X,X)=E(X T X)

[0116] D(Y)=cov(Y,Y)=E(Y T Y)

[0117] cov(X,Y)=E(XY T ),cov(Y,X)=E(YX T )

[0118] S XX =cov(X,X), then the solution objective is transformed into:

[0119]

[0120] Step 1: Calculate the variance S of X and Y XX and S YY , the covariance of XY and YX

[0121] Step 2: Calculate the matrix

[0122] Step 3: Solve the singular values ​​of M′ to obtain the largest singular value and its preceding and following singular vectors u, v;

[0123] Step 4: Projection vector a of X and Y 1 and b 1 They are:

[0124]

[0125] Step 5: The result after fusion is

[0126] Optionally, any node in the target network<m,n> Processing to obtain node pairs<m,n> Node weight information, node pair<m,n> Node strength information and node pairs<m,n> The local path information includes:

[0127] S121, the length of time that node m and node n remain connected in the Δt time period is processed to obtain the node pair<m,n> Node weight information;

[0128] The node pair<m,n> The calculation formula of node weight information is:

[0129]

[0130] In the formula, cmn is the node pair<m,n> tmn is the length of time that nodes m and n remain connected in the Δt time period, Δt 12 is the interval between the two network states corresponding to node m and node n in the Δt time period; Δt is a preset time period and can be set according to the specific network;

[0131] S122, traverse different nodes in the node set that has a communication relationship with node m in turn, and obtain node pairs<m,n> Node strength information;

[0132] The node pair<m,n> The node strength information of is the sum of the node weight information of each node in the node set that has a communication relationship with node m and node m;

[0133] S123, calculating and summing the node strength information of each node on the j-th path between node m and node n, except node m and node n, to obtain the associated path contribution information, where j is the path number, according to the specific network settings;

[0134] S124, calculating the first associated path contribution information of the path with a length of l between node m and node n, and taking the inverse of the first associated path contribution information and summing them up to obtain the node pair<m,n> Local path information.

[0135] l is a preset path length, which can be set according to a specific network and is not limited in this embodiment.

[0136] The calculation method of the first associated path contribution information is to calculate the node strength information of each node except the node m and the node n on the path with a length of l between the node m and the node n and sum them up to obtain the first associated path contribution information;

[0137] Optionally, the link-relationship-oriented feature extraction model is used to process the link-weighted relationship learning indicator to obtain the link relationship between network snapshots, including:

[0138] S31, using link weighted relationship learning indicators of different node pairs at different target network stages, training a preset feature extraction model to obtain a link relationship oriented feature extraction model;

[0139] Optionally, the preset feature extraction model is a support vector machine classifier;

[0140] S32, using the link relationship oriented feature extraction model, processing the link weighted relationship learning index to obtain the link relationship between network snapshots.

[0141] Optionally, the link-relationship-oriented feature extraction model is used to process the link-weighted relationship learning indicator to obtain the link relationship between network snapshots, including:

[0142] S321, using a first feature processing model, processing the link weighted relationship learning indicator to obtain first feature information;

[0143] Optionally, the first feature processing model is a DBN model, and the distributed multi-dimensional features are extracted as the first feature information;

[0144] S322, using a second feature extraction model to process the first feature information to obtain second feature information;

[0145] Optionally, the second feature extraction model is a least squares support vector machine, and its output result is the second feature information;

[0146] S323: Process the second feature information using the link relationship oriented feature extraction model to obtain link relationships between network snapshots.

[0147] Optionally, the link relationship oriented feature extraction model is a support vector machine classifier.

[0148] Optionally, the processing of the link relationship within the network snapshot and the link relationship between the network snapshots to obtain network topology information includes:

[0149] S41, parsing the link relationship within the network snapshot and the link relationship between the network snapshots to obtain node topology information of the target network;

[0150] The node topology information of the target network includes node IP address, MAC address, subnet mask length, gateway address, gateway IP address, online status, parent node address, node name, node type and node system version;

[0151] The analysis method is a prior art in the art and is not limited in this embodiment;

[0152] S42, filtering the node topology information of the target network to obtain the node topology information of the first target network;

[0153] The purpose of filtering is to remove redundant information. The specific method is the existing technology in this field and is not limited in this embodiment.

[0154] S43, normalizing the node topology information of the first target network to obtain the node topology information of the second target network;

[0155] Normalization is done by dividing by the maximum value of the node topology information;

[0156] S44, extracting key information from the node topology information of the second target network to obtain node topology information of the third target network;

[0157] The method of extracting key information is:

[0158] The node topology information of the second target network is subjected to a lattice network construction, which is a prior art in the art. The set of nodes for constructing the lattice network is N={n 0 ,n 1 ,…}, the set of arcs is E={e 0 ,e 1 ,…}, each node n i ∈N contains a time stamp t(n i ), arc e u→v represents the arc connecting node u to node v. Let the nodes in the keyword network be Represents the node N in the keyword network i Connect to node N j A set of arcs.

[0159] 1. Any n i ∈N i , n j ∈N i , by t(n i )<t(n j ) we can get i≤j.

[0160] 2. Any n i ∈N i , n j ∈N i , by t(n i )=t(n j ) we can get i=j.

[0161] 3. Any e u→v ∈E, if for u∈N i and v∈N j , e u→v Equivalent to a group Then i≤m≤n≤j. (Here n=m+1, e u→v The two endpoints belong to N m and N n . )

[0162] The word grid network is converted into a keyword network, and the generation algorithm is:

[0163] 1. Convert keywords into corresponding syllable strings: K 1 …K M(M is the number of syllables of the keyword). Traverse all nodes N in the Lattice network and find the nodes that match K. 1 Matching node n k , set to

[0164] 2. As the end point, search for the nearest node n in reverse time order. k-1 If n k-1 and There is no arc between them, then n k-1 Also belongs to Continue Search k-1 Otherwise, the search stops.

[0165] 3. As the starting point, search for similar nodes n backward in chronological order k+1 If n k+1 and There is no arc between them, then n k+1 belong Continue Search k+1 Otherwise, the search stops.

[0166] 4. For e k→k+1 belong n k+1 belong

[0167] 5. All included in The arc e k→k+1 Form a keyword confusion network.

[0168] 6. For each arc If u belongs to N s , n i Belong to N t When t=s+1, belong otherwise, belong (s+1≤n≤t).

[0169]

[0170]

[0171] Where w(l) and w(e) are the words corresponding to arcs l and e, sim(·,·) refers to the similarity between two words. It refers to the arc and e overlap in normalized time. In addition, each node n in the Lattice i∈N contains the time stamp t(n i ) will be used as a constraint to determine the termination of the search.

[0172] The data normalization method is used to confirm the keywords. First, the linguistic probability likelihood score of each candidate is normalized:

[0173]

[0174] Among them, Value is the set of linguistic score values ​​in the keyword network; minValue and maxValue are the minimum and maximum values ​​in the set respectively; x is the score before normalization; y is the score after normalization. After that, the sum of the scores is calculated and marked as the final score of the arc in the keyword network.

[0175] In each keyword network, select the arc candidate with the highest score to determine whether it is a keyword. If it is, output the keyword and time tag; if not, skip and enter the next keyword network. In this way, the keyword detection is completed, and the detected keyword is the key information.

[0176] S45, processing the node topology information of the third target network to obtain network topology information. Specifically:

[0177] Comprehensively utilize active detection, traffic monitoring, control and guarding, node reconnaissance based on permission access and other means (all existing technologies) to parse and obtain the target network's node IP address, MAC address, subnet mask length, gateway address, gateway IP address, online status, parent node address, node name, node type, node system version and other information. Combined with redundant information filtering, normalization and other data processing, the topological relationship is analyzed through association analysis, and the node link connectivity in the auxiliary target network topology status information is used as the input data of the deep learning network model. Through relative superposition, the unified correction of the network topology link relationship is achieved, and the final topological data, i.e., network topology information, is formed by integration.

[0178] Optionally, the processing the network topology information to obtain a rapidly reconstructed network topology includes:

[0179] S51, analyzing the network topology information to obtain real-time node attribute information, network topology attribute information and target network link connection information;

[0180] S52, using a preset network aggregation model, classify and aggregate the real-time node attribute information, the network topology attribute information and the target network link connection information to obtain a rapidly reconstructed network topology.

[0181] Specifically: based on the acquired target network node reconnaissance information, node link connectivity and other data (network topology information), through deduplication and merging of node attribute and connection relationship data, the real-time node attributes, network topology attribute information, automatically learned target network link connectivity and other information acquired through detection are classified and aggregated through an automatic aggregation algorithm (such as single node aggregation, star node aggregation, full connectivity aggregation, etc., which are not limited in this embodiment) to automatically build a hierarchical dynamic network logical topology, accurately reflect the connection relationship, connection multiplicity, and logical relationship of each node in the target network topology, and obtain a rapidly reconstructed network topology.

[0182] It can be seen that the present invention takes the real target network environment with irregular dynamic changes as the background, focuses on real-time and accurate acquisition of the link connectivity status of the target network topology and the demand for rapid construction of the network topology, and designs a network topology rapid reconstruction method based on dynamic link relationship discovery. Mainly through the automatic discovery of node link relationships based on deep learning, hierarchical network topology dynamic clustering and other means, it solves the problems of low accuracy of topological relationship feature extraction and poor pertinence of learning indicators under dynamic network changes, realizes accurate automatic discovery of network topology node link relationships, automatic tracking and reconstruction of network topology, and provides theoretical and technical support for subsequent comprehensive mapping, penetration testing, management and maintenance of non-cooperative target networks.

[0183] Embodiment 2

[0184] See also Figure 2 , Figure 2 1 is a flow chart of another method for rapid network topology reconstruction based on link relationship discovery disclosed in an embodiment of the present invention. Figure 2 The described method for rapid reconstruction of network topology based on link relationship discovery is applied to obtain the comprehensive link status of network topology, effectively supporting subsequent penetration testing of the target network, which is not limited in the embodiments of the present invention. Figure 2 As shown in the figure, the network topology rapid reconstruction method based on link relationship discovery mainly includes four parts: autonomous learning indicator construction, link node relationship learning based on deep network, topology relationship correction and network topology aggregation classification, which finally forms a dynamic network topology.

[0185] Among them, the self-learning index construction mainly extracts and constructs node similarity impact indicators from aspects such as the weight between nodes, node strength and local paths, and determines the connectivity of each path relationship in the target network by calculating the evaluation value of the link relationship;

[0186] Based on the deep network, the link node relationship learning focuses on the rapid dimension reduction and denoising of the training samples, and constructs a multi-layer trapezoidal restricted Boltzmann machine (RBM); on the other hand, it focuses on fast convergence performance and adopts an adaptive learning rate; on the other hand, it focuses on distributed feature extraction and classifies and labels different training sample sets of the deep belief network model. Combined with the learning index values ​​of different network nodes, the link relationship corresponding to different network states is calculated autonomously;

[0187] Topology relationship correction: collect real-time network topology information and link relationship data as input, input the network link relationship learning and analysis model, and correct the node and link analysis results based on the feedback mechanism;

[0188] Network topology aggregation classification integrates node attributes, link relationships and other information, classifies and aggregates the node types and node connection relationships of the same type of equipment, automatically builds a hierarchical dynamic network logical topology, and finally realizes the dynamic construction of network topology. Figure 2 shown.

[0189] 1. Determine the network link relationship based on the autonomous learning indicator and obtain the link weighted learning relationship indicator

[0190] Specifically, the different structural states of the target network are characterized by stages and time periods through operations that change the node connection relationship, and the target feature objects corresponding to the dynamic network structure are extracted. Then, by calculating the node-link relationship value corresponding to the target network state at a certain stage, the node-link relationship value corresponding to the target network state at the next stage is quickly analyzed and discovered, as follows:

[0191] Step 1: Based on the historical network status and the node connectivity dimension, the specific target network structure status of each stage and time interval is divided;

[0192] Step 2: Determine the weight value between node m and node n, that is, the node weight information. That is, calculate the ratio of the time that node m and node n remain connected in the Δt time period to the interval between the two corresponding network states;

[0193] Step 3: Calculate the node strength of each node. That is, traverse different nodes n in turn in all the nodes that have a communication relationship with node m, calculate the sum of the weights between all nodes m and node n, and obtain the node strength information;

[0194] Step 4 Calculation<m,n> Link L a The contribution of the related nodes to the associated path. That is, take each node except m and n on the j-th path from node m to n, calculate the node strength of these nodes respectively and sum them up, La is the link name, which is determined according to the specific network, j is the path number, which is determined according to the specific network;

[0195] Step 5 calculates the sum of the contributions of the paths with a length of l between node m and node n. That is, calculate the number of paths with a length of l between node m and node n, calculate the contribution of each path with a length of l, take the inverse of it, and then calculate the sum to get the node pair<m,n> The local path information of , l is the path length, which is determined according to the specific network;

[0196] Step 6 calculates the node link connectivity index value corresponding to the target network in the next stage. That is, for paths with different path lengths between node m and node n, calculate the sum of the squares of the predefined parameters as the product of the sum of the contributions of the weights and sum them up.

[0197] Step 7 Finally, the node pair is determined based on the node link connection relationship index value, node weight information, node strength information and local path information.<m,n> The matching relationship with the target network status at a certain stage.

[0198] 2. Construction of DBN model for chain relationship feature extraction

[0199] The DBN-based automatic learning framework for target network link relations is as follows: Figure 3 shown.

[0200] The workflow of the network link relationship automatic learning framework includes: first, according to different nodes<m,n> The target network state N at different stages a The corresponding node link relationship learning indicators and the path connection status between the node pairs are trained to form a support vector classifier (SVC); then, the distributed multi-dimensional features are extracted through the DBN model as the input data of the least squares support vector machine (LS-SVR); finally, the output value of the LS-SVR regression model is input into the trained classifier SVC to learn and analyze the target network state N at a certain stage of the network. a+j Different node pairs<m,n> Inter-link connectivity status.

[0201] (1) Construct training samples of DBN and LS-SVR regression models based on the target network status information at a certain stage

[0202] Definition 1 (node ​​connection relationship learning index value) DBN basic sample, note:

[0203] A=value(x,y) a

[0204] Determine j as the initial dimension, and convert the original sample of link relationship learning index values ​​of j nodes {score(m,n) a ,…,score(m,n) a+j} as the input of the DBN network.

[0205] Definition 2 (Original training sample set) The original training sample set of LS-SVR is:

[0206] D={h(m,n) a ,…,value(m,n) a+j}

[0207] Among them, h(m,n) a is the independent variable, which is from the sample set {value(m,m) a ,…,value(m,m) a+j-1}, value(m,n) a+j is the dependent variable, which is the link relationship learning indicator value corresponding to the target network state in the next stage.

[0208] Definition 3 (label of training sample) value(m,n) a+j The value on (0,1] corresponds to the equal division label represented by {value(m,m) a ,…,value(m,m) a+j-1}'s classification label Y j+a .

[0209] Definition 4 (Accuracy): Link connectivity relationship analysis accuracy. Note:

[0210]

[0211] Among them, when<m,n> Link L a Corresponding instance test sample {value(m,n) a , L a}'s classification label Y j+a Represents the degree of connectivity and is used to calculate label values ​​through model analysis If they are consistent, it is recorded as T, otherwise it is recorded as F.

[0212] (2) Constructing a DBN multi-dimensional distributed feature extraction model by combining information entropy and adaptive learning rate

[0213] On the one hand, focusing on the requirements of distributed feature extraction for the validity of input information, the information entropy is mainly used to construct RBM, evaluate and analyze the effective information in the sample and determine the number of neurons in the hidden layer of the RBM, thereby improving the processing performance of the DBN model in dimensionality reduction and denoising of training samples; on the other hand, on the basis of ensuring that the DBN model has good modeling and feature abstraction capabilities, the optimal number of RBM layers is determined by comprehensively considering potential problems such as reduced generalization performance and overfitting caused by too many RBM layers; on the other hand, in response to the problems that the size of the learning rate may cause the calculation process to fail to converge and the training time to be extended, an automatically adjusted learning rate is adopted based on the feedback mechanism to ensure dynamic adjustment according to the actual training process, and the optimal learning rate value can be obtained at different stages of the calculation.

[0214] (3) Constructing LS-SVR regression model using radial basis function (RBF) kernel function and K-fold cross validation

[0215] On the one hand, focusing on computing efficiency, the least squares support vector machine is mainly used to reduce the computational complexity; on the other hand, focusing on the impact of the kernel function parameters on the performance of the regression model, the number of parameters, value range, convergence domain range, stability and other factors are comprehensively considered. Finally, the RBF kernel function is selected as the kernel function of the LS-SVR model, and the K-fold cross-validation algorithm is used to calculate the optimal value of the parameter to achieve the optimal configuration of the parameters. 3. Confirmation of topological relationships based on network topology collection information

[0216] Comprehensively utilize active detection, traffic monitoring, control and guarding, node reconnaissance based on permission access and other means to parse and obtain the target network's node IP address, MAC address, subnet mask length, gateway address, gateway IP address, online status, parent node address, node name, node type, node system version and other information. Combined with redundant information filtering, normalization and other data processing, the topological relationship is analyzed through correlation analysis, and the node link connectivity in the auxiliary target network topology status information is used as the input data of the deep learning network model. Through relative superposition, the unified correction of the network topology link relationship is achieved, and the final topological data, namely the network topology information, is formed by fusion.

[0217] 4. Aggregation grouping based on network topology

[0218] Based on the acquired target network node reconnaissance information, node link connectivity and other data (network topology information), through deduplication and merging of node attribute and connection relationship data, the real-time node attributes, network topology attribute information, automatically learned target network link connectivity and other information acquired through detection are classified and aggregated through automatic aggregation algorithm for the types of device nodes and node connection relationships of the same type, and a hierarchical dynamic network logical topology is automatically constructed, which accurately reflects the connection relationship, connection multiplicity and logical relationship of each node in the target network topology, and obtains a rapidly reconstructed network topology.

[0219] Embodiment 3

[0220] See also Figure 4 , Figure 4 Schematic diagram of a network topology rapid reconstruction device based on link relationship discovery disclosed in an embodiment of the present invention. Figure 4 The described network topology rapid reconstruction device based on link relationship discovery is applied to obtain the comprehensive link status of the network topology, effectively supporting the subsequent penetration test of the target network, which is not limited in the embodiments of the present invention. Figure 4 As shown, the network topology rapid reconstruction device based on link relationship discovery mainly includes:

[0221] S301, a target network analysis module, used to analyze the target network to obtain a link weighted relationship learning indicator; the link weighted relationship learning indicator includes node weight information, node strength information and local path information;

[0222] S302, an indicator analysis module, used to process the link weighted relationship learning indicator to obtain the link relationship in the network snapshot;

[0223] S303, a feature extraction module, used to process the link weighted relationship learning indicator using a link relationship-oriented feature extraction model to obtain the link relationship between network snapshots;

[0224] S304, a network topology information calculation module, configured to process the link relationship within the network snapshot and the link relationship between the network snapshots to obtain network topology information;

[0225] S305, a network topology reconstruction module is used to process the network topology information to obtain a rapidly reconstructed network topology.

[0226] Embodiment 4

[0227] See also Figure 5 , Figure 5 1 is a schematic diagram of another network topology fast reconstruction device based on link relationship discovery disclosed in an embodiment of the present invention. Figure 5The described network topology rapid reconstruction device based on link relationship discovery is applied to obtain the comprehensive link status of the network topology, effectively supporting the subsequent penetration test of the target network, which is not limited in the embodiments of the present invention. Figure 5 As shown, the network topology rapid reconstruction device based on link relationship discovery mainly includes:

[0228] A memory 401 storing executable program codes;

[0229] a processor 402 coupled to the memory 401;

[0230] The processor 402 calls the executable program code stored in the memory 401 to execute the steps of the network topology fast reconstruction method based on link relationship discovery described in the first and second embodiments.

[0231] Embodiment 5

[0232] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program enables a computer to execute the steps of the network topology rapid reconstruction method based on link relationship discovery described in Embodiments 1 and 2.

[0233] The device embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative work.

[0234] Through the specific description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution can be essentially or partly contributed to the prior art in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0235] Finally, it should be noted that the method and device for rapid reconstruction of network topology based on link relationship discovery disclosed in the embodiment of the present invention only discloses the preferred embodiment of the present invention, which is only used to illustrate the technical solution of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, ordinary technicians in this field should understand that it is still possible to modify the technical solutions recorded in the aforementioned embodiments, or to replace some of the technical features therein by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for rapid reconstruction of network topology based on link relationship discovery, characterized in that: The method comprises: S1, analyzing the target network to obtain a link weighted relationship learning indicator; the link weighted relationship learning indicator includes node weight information, node strength information and local path information, including: S11, dividing the target network into stages to obtain N target network stages, where N is a positive integer; S12, for any node pair in the target network phase i<m,n> Processing to obtain node pairs<m,n> Node weight information, node pair<m,n> Node strength information and node pairs<m,n> The local path information, i = 1, 2, ..., N, includes: S121, the length of time that node m and node n remain connected in the Δt time period is processed to obtain the node pair<m,n> Node weight information; The node pair<m,n> The calculation formula of node weight information is: In the formula, c mn For node pairs<m,n> The node weight information, t mn is the time length that nodes m and n remain connected within the Δt period, Δt 12 is the interval between the two network states corresponding to node m and node n in the Δt time period; S122, traverse different nodes in the node set that has a communication relationship with node m in turn, and obtain node pairs<m,n> Node strength information; The node pair<m,n> The node strength information of is the sum of the node weight information of each node in the node set that has a communication relationship with node m and node m; S123, calculating and summing the node strength information of each node on the j-th path between node m and node n, except node m and node n, to obtain the associated path contribution information, where j is the path number; S124, calculating the first associated path contribution information of the path with a length of l between node m and node n, and taking the inverse of the first associated path contribution information and summing them up to obtain the node pair<m,n> Local path information; S13, for the node pair<m,n> The node weight information of<m,n> The node strength information and the node pair<m,n> The local path information is integrated to obtain the link weighted relationship learning index; S2, processing the link weighted relationship learning indicator to obtain the link relationship in the network snapshot; S3, using a link relationship feature extraction model to process the link weighted relationship learning indicator to obtain the link relationship between network snapshots, including: S31, using link weighted relationship learning indicators of different node pairs at different target network stages, training a preset feature extraction model to obtain a link relationship oriented feature extraction model; S32, using the link relationship oriented feature extraction model, processing the link weighted relationship learning indicator to obtain the link relationship between network snapshots, including: S321, using a first feature processing model, processing the link weighted relationship learning indicator to obtain first feature information; S322, using a second feature extraction model to process the first feature information to obtain second feature information; S323, using the link relationship oriented feature extraction model, processing the second feature information to obtain the link relationship between network snapshots; S4, processing the link relationship within the network snapshot and the link relationship between the network snapshots to obtain network topology information; S5, processing the network topology information to obtain a rapidly reconstructed network topology.

2. The method for rapid network topology reconstruction based on link relationship discovery according to claim 1, characterized in that: The processing of the link relationship within the network snapshot and the link relationship between the network snapshots to obtain network topology information includes: S41, parsing the link relationship within the network snapshot and the link relationship between the network snapshots to obtain node topology information of the target network; S42, filtering the node topology information of the target network to obtain the node topology information of the first target network; S43, normalizing the node topology information of the first target network to obtain the node topology information of the second target network; S44, extracting key information from the node topology information of the second target network to obtain node topology information of the third target network; S45, processing the node topology information of the third target network to obtain network topology information.

3. The method for rapid network topology reconstruction based on link relationship discovery according to claim 1, characterized in that: The processing of the network topology information to obtain a rapidly reconstructed network topology includes: S51, analyzing the network topology information to obtain real-time node attribute information, network topology attribute information and target network link connection information; S52, using a preset network aggregation model, classify and aggregate the real-time node attribute information, the network topology attribute information and the target network link connection information to obtain a rapidly reconstructed network topology.

4. A network topology rapid reconstruction device based on link relationship discovery, characterized in that: The device comprises: The target network analysis module is used to analyze the target network and obtain the link weighted relationship learning index; the link weighted relationship learning index includes node weight information, node strength information and local path information, including: S11, dividing the target network into stages to obtain N target network stages, where N is a positive integer; S12, for any node pair in the target network phase i<m,n> Processing to obtain node pairs<m,n> Node weight information, node pair<m,n> Node strength information and node pairs<m,n> The local path information, i = 1, 2, ..., N, includes: S121, the length of time that node m and node n remain connected in the Δt time period is processed to obtain the node pair<m,n> Node weight information; The node pair<m,n> The calculation formula of node weight information is: In the formula, c mn For node pairs<m,n> The node weight information, t mn is the time length that nodes m and n remain connected within the Δt period, Δt 12 is the interval between the two network states corresponding to node m and node n in the Δt time period; S122, traverse different nodes in the node set that has a communication relationship with node m in turn, and obtain node pairs<m,n> Node strength information; The node pair<m,n> The node strength information of is the sum of the node weight information of each node in the node set that has a communication relationship with node m and node m; S123, calculating and summing the node strength information of each node on the j-th path between node m and node n, except node m and node n, to obtain the associated path contribution information, where j is the path number; S124, calculating the first associated path contribution information of the path with a length of l between node m and node n, and taking the inverse of the first associated path contribution information and summing them up to obtain the node pair<m,n> Local path information; S13, for the node pair<m,n> The node weight information of<m,n> The node strength information and the node pair<m,n> The local path information is integrated to obtain the link weighted relationship learning index; An indicator analysis module, used to process the link weighted relationship learning indicator to obtain the link relationship in the network snapshot; The feature extraction module is used to process the link weighted relationship learning index using a link relationship-oriented feature extraction model to obtain the link relationship between network snapshots, including: S31, using link weighted relationship learning indicators of different node pairs at different target network stages, training a preset feature extraction model to obtain a link relationship oriented feature extraction model; S32, using the link relationship oriented feature extraction model, processing the link weighted relationship learning indicator to obtain the link relationship between network snapshots, including: S321, using a first feature processing model, processing the link weighted relationship learning indicator to obtain first feature information; S322, using a second feature extraction model to process the first feature information to obtain second feature information; S323, using the link relationship oriented feature extraction model, processing the second feature information to obtain the link relationship between network snapshots; A network topology information calculation module, used to process the link relationship within the network snapshot and the link relationship between the network snapshots to obtain network topology information; The network topology reconstruction module is used to process the network topology information to obtain a rapidly reconstructed network topology.

5. A network topology rapid reconstruction device based on link relationship discovery, characterized in that: The device comprises: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the network topology rapid reconstruction method based on link relationship discovery as described in any one of claims 1-3.

6. A computer storable medium, characterized in that: The computer storable medium stores computer instructions, and when the computer instructions are called, they are used to execute the network topology rapid reconstruction method based on link relationship discovery as described in any one of claims 1-3.