CN109120465A - Target area network topology division methods based on die body - Google Patents
Target area network topology division methods based on die body Download PDFInfo
- Publication number
- CN109120465A CN109120465A CN201811236808.4A CN201811236808A CN109120465A CN 109120465 A CN109120465 A CN 109120465A CN 201811236808 A CN201811236808 A CN 201811236808A CN 109120465 A CN109120465 A CN 109120465A
- Authority
- CN
- China
- Prior art keywords
- corporations
- node
- network topology
- die body
- network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/12—Discovery or management of network topologies
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L2101/00—Indexing scheme associated with group H04L61/00
- H04L2101/60—Types of network addresses
- H04L2101/69—Types of network addresses using geographic information, e.g. room number
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses the target area network topology division methods based on die body, comprising: obtains target area network topology；Target area network topology is subjected to convergence arrangement, forms router level network topology；Data prediction is carried out to obtained router level network topology；Pretreated router level network topology is abstracted into one and haves no right undirected network topological diagram, the excavation of die body is carried out in network topological diagram；Candidate of the die body as initial seed is chosen, and merges beta pruning, determines final initial seed set；The community structure to be formed and not overlapped is extended to initial seed set based on fitness function F；Corporations are extended based on corporations' density, determine the ownership corporations of node on single-pathway；The method for being based respectively on terrestrial reference and the ballot of IP geographic information database carries out geolocation mapping building to the corporations of division.The present invention can get up most of entity in network with geolocation mapping, and accuracy rate is higher.
Description
Technical field
The present invention relates to network topologies fields, more particularly to the target area network topology division side based on die body
Method.
Background technique
With using Internet as the rapid development of the computer of representative communication and network technology, the development face of human society
Facing huge and deep change, internet is closely coupled with modern society, work, life and the amusement of people, or even
Operation, the expanding economy of national government affairs are all highly dependent on the safe operation of internet.Network security concerning national security, due to
Internet innately has the characteristics that the decentralized property of heterogeneous, dynamic and development, along with continuous development causes its scale to be got over
Come huger, in order to ensure the safe and stable operation of network, people's urgent need is preferably controlled and managing internet, and builds
Vertical network entity and the accurate mapping relations in its geographical location are to implement the basis of network management.
Operation is mainly responsible for by multiple Internet Service Provider ISP (Internet Service Provider) in internet
And maintenance, it can be by the analysis data message routing forwarding situation between each node in a network, and then obtain between node
Logic connecting relation constitutes network topology.
Interconnection net topology can be regarded as a kind of typical complex network, current grinding for internet topological connection relation
Study carefully, main is also to carry out with the relevant knowledge of complex network.With the physical significance and mathematical characteristic to network property
It researchs and analyses and deepens continuously, research finds that many complex networks including internet have a common property, i.e. society
Group's phenomenon or referred to as clustering phenomenon.The community structure of research network facilitates the deep institutional framework for understanding network and its generation
The system function of table, (Yuan Shaoqian, Zhao Hai, Zhang Xin wait the community structure of Internet topology to analyze [J] complication system to paper
With complexity science, 2007,4 (3): 17-27.) just disclose community structure in internet AS grades of network topology and AS is geographical
Relationship between position distribution.Although most of connection for interconnecting net topology interior joint on surface is unrestricted choice, on ground
There are great localized clusters phenomenon under the actual environment that area and national region divide, illustrate connection between internet node not
It establishes at random, node, which is more likely to connect with the node of local, (Zhang Guoqing internet topological structure Knowledge Discovery and its answers
Journal, 2010,31 (10): 18-25. are communicated with [J]).Mr. Qian Xuesen also indicates that the network between the inside of proximate region is opened up
Flutter the feature that height clustering is presented.
Tian et al. (Tian Ye, Dey R, Liu Y, et al.Topology mapping and geolocating
for China's Internet[J].IEEE Transactions on Parallel and Distributed
Systems, 2013,24 (9): 1908-1917) propose it is a kind of based on the heuristic cluster of network topology network topology mapping and
Geographic positioning (HC-Based), clusters network entity using the assemblage classification based on network structure, for the first time by base
Combine in data base querying and two IP like localization methods of network measure, is finally constructed and mapped according to the geographical location of cluster
Relationship.However, often number of nodes is very little for the cluster that clusters of this method, and will lead to using simple voting rule based on throwing
The positioning mistake of ticket result, therefore the mapping relations error finally established is larger.Liu et al. people (Liu S, Liu F, Zhao F,
et al.IP city-level geolocation based on the PoP-level network topology
analysis[C]//International Conference on Information Communication and
Management.IEEE, 2016:109-114.) foundation that divides the PoP structure in network as network topology, it is believed that PoP
Structure be include a series of set with strong ties relationship and the close node of location distribution, their mutual structures
Then will have at Bi-fan structure by finding the node set with close connection performance Bi-fan structure in a network
The Bi-fan of intersection, which gathers, to be merged, to realize the division to network topology.But they do not make a concrete analysis of to be ground
Whether Bi-fan structure largely there is in the network structure studied carefully, and whether there is also the node set of other strong ties,
Therefore algorithm has significant limitation.Li et al. people (Li M, Luo X, Shi W, et al.City-level IP
geolocation based on network topology community detection[C]//International
Conference on Information NETWORKING.IEEE, 2017:578-583.) propose one kind based on network node
NNC (Network Node Clustering) network topology division methods of cluster, by the cluster and complexity in network topology
Corporations in network theory are mapped, and are clustered, are adopted to the node in network topology using Louvain community discovery algorithm
The intensity that community structure is measured with modularity, is divided into different corporations for network topology and analyzes.But their method
The specific topological property of internet is not accounted for, such as measures community structure intensity using modularity, it is desirable that section inside community structure
There is intensive connection and the connection between external node is sparse between point, but is usually constituted in internet topological connection relation
The node of this harsh corporations' condition has focused largely on backbone network.
Internet is affected as maximum complex network in the world, development by factors such as politics, economy, different
The lack of uniformity of development is presented in region, therefore the network topology of different zones its structure is also different.Existing method exists
When dividing corporations in network topology, the specific structure characteristic of network topology is not fully taken into account, can only realize few portion
The mapping relations in subnetwork entity and geographical location construct, and still have the space of promotion in accuracy rate.
As the scale of complex network in real world is increasing, the global information of network is difficult to hold, it is a kind of
Local Combo discovering method based on network local message is found to be more applicable for the bigger complex network of scale, because should
Method is usually to be diffused to the initial seed in complex network, it is only necessary to just can be carried out corporations according to the local message of network
It was found that.Internet is difficult to hold as maximum complex network, the overall permanence of network in the world, therefore is locally believed based on network
The community discovery algorithm of breath is more applicable on interconnection net topology.
By the analysis to existing complex network part Combo discovering method, find the selection of initial seed node to society
The final result of group's diffusion plays the role of vital.But the initial seed of local Combo discovering method chooses one at present
As be divided into following 3 class:
(1) random selection node is as initial seed；
(2) select central node as initial seed using some Measure Indexes；
(3) select very big subgraph as initial seed.
It, usually will be through it can be found that first kind initial seed choosing method can make algorithm stability poor by analysis
Relatively good division result can just be accessed by crossing multiple community division；Second class initial seed choosing method, if only selecting that list
A node not can guarantee the center that the node indeed is in corporations as initial seed；Third class initial seed choosing method
Algorithm complexity when finding very big subgraph is higher, and may meet very big subgraph due to the sparsity of network itself
Structure is not significant in a network to be occurred, therefore this method is more harsh, does not ensure that and excavates corporations all in network.
Summary of the invention
The present invention considers network topology specific structure characteristic deficiency and existing local community division side for existing method
In method initial seed choose there are the problem of, propose " die body " of network topology being used as initial seed, invent a kind of based on mould
The target area network topology division methods of body, the method for the present invention can be by the most of entities and geolocation mapping in network
Get up, and accuracy rate is higher.
To achieve the goals above, the invention adopts the following technical scheme:
Target area network topology division methods based on die body, comprising the following steps:
Step 1 obtains target area network topology；
Target area network topology is carried out convergence arrangement by step 2, and the IP grade network topology that detection is obtained is by alias
Parsing forms router level network topology；
Step 3 carries out data prediction to obtained router level network topology: institute in statistics router level network topology
There is the degree of node, the node that the degree being connected with the larger node of degree is 1 is all picked out, with the node phase for being 1 with the degree
The node that the biggish node substitution degree of degree even is 1；
Pretreated router level network topology is abstracted into one and haves no right undirected network topological diagram by step 4, in institute
State the excavation that die body is carried out in network topological diagram；
Step 5 chooses candidate of the die body as initial seed, and merges beta pruning, determines final initial seed collection
It closes；
Step 6 is extended initial seed set to form 1 or more corporations based on fitness function F, deletes duplicate
Corporations form the community structure not overlapped；
Step 7 is extended corporations based on corporations' density, determines the ownership corporations of node on single-pathway；
Step 8, the method for being based respectively on terrestrial reference and the ballot of IP geographic information database, by network entity and geographical location pair
It answers, geolocation mapping building is carried out to the corporations of division.
Further, the step 4 includes:
Pretreated router level network topology is abstracted into one and haves no right undirected network topological diagram by step 4.1, benefit
The method enumerated with subgraph carries out subgraph search in the network topological diagram, counts the number that each drawing of seeds occurs；
Step 4.2, the network topological diagram according to step 4.1 construct network topology described in multiple and step 4.1
Figure is in the identical random network topology figure of whole statistical property；
Step 4.3 carries out subgraph search on the random network topology figure of building, counts time that each drawing of seeds occurs
Number；
Step 4.4, comparison same sub-image network topological diagram described in step 4.1 and the Stochastic Networks described in step 4.3
The subgraph of Z-score>2 and P-value<0.01 is regarded as the die body of target area network by the frequency occurred in network topological diagram.
Further, the step 5 includes:
Candidate of the die body that step 5.1, selected size are 4 as initial seed；
The die body of shared 2 common nodes is merged into an initial seed by step 5.2, will only share 1 common node
Die body as the bridge node between two initial seeds.
Further, the step 6 includes:
Step 6.1 is extended initial seed set based on fitness function F, for each section adjacent with corporations S
Point v calculates separately the fitness that corporations S is added after node v, then by the fitness function value and addition after addition node v
Preceding fitness function value compares, and judges to add whether node v will increase the fitness of corporations S into corporations S, thus certainly
Determine, if will increase the fitness of corporations S, corporations S to be added in node v, if will not increase whether by node v addition corporations S
Add the fitness of corporations S, then corporations S is not added in node v；
Step 6.2 repeats step 6.1, until there is no can increase the suitable of corporations S in the neighbor node of corporations S
The node of response terminates and returns to final corporations S；
Step 6.3 deletes duplicate corporations, forms the community structure not overlapped.
Further, the step 7 includes:
The calculation formula of corporations' density are as follows:
Wherein M indicates the number of edges inside corporations S, the number of nodes after N expression addition node v inside corporations S；
According to the sequence being connect with corporations S, successively the node that the degree on the single-pathway being connected with corporations S is 1 is added
Into S, the f of calculate noded vValue, by fd vThe node that value is greater than threshold alpha is added in corporations S, obtains final Topology partition
As a result.
Further, the step 8 includes:
When step 8.1, execution network topology measurement, using terrestrial reference known to geographical location as the destination node of detection；
Step 8.2, the mapping construction method based on IP geographic information database determine corporations position using based on voting mechanism
Set: the location information by searching each corporations' internal node in known IP geographic information database, each node according to
The location information of itself votes to the corporations present position, using the consistent position of most nodes in corporations as the ground of corporations
Position is managed, the mapping relations of each node and geographical location in corporations are constructed.
Compared with prior art, the invention has the benefit that
The present invention is divided into research object with target area network topology, fully considers the structure of target area network topology
Characteristic excavates die body in network topology first, forms initial seed set by the merging beta pruning of die body, then utilizes adaptation
Degree function is extended, and considers the characteristic of node on single-pathway in network, proposes corporations' extension based on corporations' density,
More nodes are divided into corporations as far as possible, are finally based respectively on terrestrial reference and the ballot of known IP geographic information database in corporations
Method, network entity is mapped with geographical location, geolocation mapping building is carried out to the corporations of division.By a large amount of
Experiment obtain, for network topology division for, method of the invention is more reasonable, can not only by more nodes and ground
Reason position maps, and accuracy rate is higher.
Detailed description of the invention
Fig. 1 is the basic procedure signal of the target area network topology division methods based on die body of the embodiment of the present invention
Figure.
Fig. 2 be the embodiment of the present invention selected by experimental subjects, Hong Kong, Taiwan router level network topology Gephi
Visualization result figure.
Fig. 3 is experimental subjects selected by the embodiment of the present invention, " parachute " in the router level network topology of Taiwan
Phenomenon schematic diagram.
Fig. 4 is experimental subjects selected by the embodiment of the present invention, and Hong Kong of " parachute " structure is removed after data prediction
With the network topological diagram of Taiwan.
Fig. 5 is idealization expression schematic diagram of the embodiment of the present invention based on the fitness function F corporations S extended；Wherein, ellipse
Node in circle virtual coil is that initial seed die body passes through FsCorporations are saved on corporations S obtained from expansion, side shown in straight dotted line
Point is connected with its neighbor node.
Fig. 6 is experimental subjects selected by the embodiment of the present invention, single in the router level network topology structure of Taiwan
The Gephi visualization result figure in path.
Specific embodiment
With reference to the accompanying drawing with specific embodiment the present invention will be further explained explanation:
Embodiment one:
As shown in Figure 1, a kind of target area network topology division methods based on die body of the invention, including following step
It is rapid:
Step S101: target area network topology is obtained；
As an embodiment, the data that acquisition target area network topology uses were from 2 months 2017
The ITDK ipv4 data set of CAIDA, ITDK ipv4 data set are based on by being distributed in 42 national 120 sensing points
Traceroute principle, to the data that global Internet is detected, data reliability is higher.Extract ITDK ipv4
Nodal information in data set obtains target area network topology.
Step S102: target area network topology is subjected to convergence arrangement, router level network topology is formed: will detect
The IP grade network topology arrived passes through alias resolution, forms router level network topology；
Due to IP address dynamic and changeability the features such as, metastable open up can not be formed according to IP grades of network topologies
Flutter connection mapping, it is therefore desirable to research object be focused in router level network topology, the IP grade network that detection obtains is opened up
It flutters by alias resolution, edit forms router level network topology.
Step S103: data prediction is carried out to obtained router level network topology: statistics router level network topology
In all nodes degree, by with spend the node that degree that larger node is connected is 1 and all pick out, with the degree for 1 section
The node that the biggish node substitution degree of degree that point is connected is 1；
Target area router level network topology is abstracted as and haves no right undirected complex network topologies figure, node v expression is opened up
The router rushed the net in network, the side e between node and node indicate the two routers it is adjacent (be namely separated by a jump away from
From).It is utilized respectively Gephi software to be visualized, discovery has many " parachute " phenomenons in two networks, also
It is that the bigger node of degree connects the leaf node that many degree are 1.Such " parachute " phenomenon is general in network topology
Store-through exists, and is then visualized in the router level network topology including 11 other different zones such as the U.S., Japan, South Korea
When, also have been found that such phenomenon.
For the community division of target area network, more focused on to those, connection is complicated, is difficult to directly determine and return
The node for belonging to which area is analyzed, and " parachute " structure is a greatly interference, it is therefore desirable to be carried out to data
Preliminary pretreatment.The process of data prediction is exactly the mistake for rationally eliminating " parachute " phenomenon in original network topology
Journey.
Step S104: being abstracted into one for pretreated router level network topology and have no right undirected network topological diagram,
The excavation of die body is carried out in the network topological diagram；
The step S104 includes:
Step S104.1: pretreated router level network topology is abstracted into one and haves no right undirected network topology
Figure, the method enumerated using subgraph carry out subgraph search in the network topological diagram, count the number that each drawing of seeds occurs.
Step S104.2: it according to network topological diagram in step S104.1, constructs and multiple is opened up with network in step S104.1
Figure is flutterred in the identical random network topology figure of whole statistical property.
Step S104.3: carrying out subgraph search on the random network topology figure of building, counts what each drawing of seeds occurred
Number.
Step S104.4: comparison same sub-image network topological diagram and Stochastic Networks in step S104.3 in step S104.1
The subgraph of Z-score>2 and P-value<0.01 is regarded as the die body of target area network by the frequency occurred in network topological diagram.
Step S105: candidate of the die body as initial seed is chosen, and merges beta pruning, determines final initial seed
Set；
The step S105 includes:
Step S105.1: candidate of the die body that selected size is 4 as initial seed.
Step S105.2: merging into an initial seed for the die body of shared 2 common nodes, public by only shared 1
The die body of node is as the bridge node between two initial seeds；
Certain nodes are shared due to will appear some die bodys in live network data, or it is another for will appear some die bodys
The subset situation of a little die bodys.Therefore the die body of shared 2 common nodes is merged into an initial seed, greatly reduces and needs
The quantity of the initial seed of extension.The only die body of shared 1 common node, as the bridge node between two initial seeds
It treats.
Step S106: being extended initial seed set to form 1 or more corporations based on fitness function F, deletes weight
Multiple corporations form the community structure not overlapped, comprising:
Step S106.1: based on fitness function F to being extended in initial seed set, for adjacent with corporations S
Each node v, calculate separately corporations S be added node v after fitness, then by be added node v after fitness function value with
Fitness function value before addition compares, and judges to add whether node v will increase the fitness of corporations S into corporations S,
To decide whether node v corporations S is added, if will increase the fitness of corporations S, corporations S is added in node v, if
It not will increase the fitness of corporations S, then corporations S be not added in node v；
Step S106.2: repeating step S106.1, until there is no can increase corporations in the neighbor node of corporations S
The node of the fitness of S terminates and returns to final corporations S.
Step S107: being extended corporations based on corporations' density, determines the ownership corporations of node on single-pathway；
Based on the network topological diagram that Traceroute principle detects, its essence is the nets being made of a plurality of detective path
Network, display on the topology, just will appear the single-pathway of many not cross-connects, and such case can occur in backbone mostly
On router and network boundary router.In order to solve the partition problem of this kind of node, expanded again based on corporations' density
Exhibition, while guaranteeing that corporations are density stabilized, the ownership corporations of these nodes of the determination of maximum possible.Corporations' density of use
Formula are as follows:
Wherein M indicates the number of edges inside corporations S, the number of nodes after N expression addition node v inside corporations S.
According to the sequence being connect with corporations S, successively the node that the degree on the single-pathway being connected with corporations S is 1 is added
Into S, the f of calculate noded vValue, by fd vThe node that value is greater than threshold alpha is added in corporations S, obtains final Topology partition
As a result.
Step S108: the method for being based respectively on terrestrial reference and the ballot of IP geographic information database, by network entity and geographical position
Correspondence is set, geolocation mapping building is carried out to the corporations of division.
The step S108 includes:
Step S108.1: when executing network topology measurement, using terrestrial reference known to geographical location as the destination node of detection；
When executing network topology measurement, just selectively using terrestrial reference known to some geographical locations as detection
Destination node just has the presence of terrestrial reference in the corporations finally divided in this way, can determine community structure according to terrestrial reference
Possible position.
Step S108.2: the mapping construction method based on IP geographic information database determines society using based on voting mechanism
Cumularsharolith is set: the location information by searching each corporations' internal node in known multiple IP geographic information databases, each
Node votes to the corporations present position according to the location information of itself, using the consistent positions of node most in corporations as
The geographical location of corporations constructs the mapping relations of corporations' node and geographical location.
As an embodiment, multiple IP geographic information databases be MaxMind, IP2Location,
Ipligence, HostIP, Netacuity, Geobytes, purity IP database and Taobao's IP address library.
The present invention is divided into research object with target area network topology, fully considers the structure of target area network topology
Characteristic excavates die body in network topology first, forms initial seed set by the merging beta pruning of die body, then utilizes adaptation
Degree function is extended, and considers the characteristic of node on single-pathway in network, proposes corporations' extension based on corporations' density,
More nodes are divided into corporations as far as possible, are finally based respectively on terrestrial reference and the ballot of known IP geographic information database in corporations
Method, network entity is mapped with geographical location, geolocation mapping building is carried out to the corporations of division.By a large amount of
Experiment obtain, for network topology division for, method of the invention is more reasonable, can not only by more nodes and ground
Reason position maps, and accuracy rate is higher.
Embodiment two:
Another target area network topology division methods based on die body of the invention, comprising:
Step S201: the nodal information of Taiwan and Hongkong is obtained, target area network topology is obtained:
ITDK ipv4 data set of the data that the present embodiment uses from 2 months 2017 CAIDA, ITDK ipv4 data
Collection is to be based on Traceroute principle by being distributed in 42 national 120 sensing points, is detected to global Internet
The data arrived, data reliability are higher.Extract in ITDK ipv4 data set that clearly mark is located at Taiwan and Hongkong
Nodal information obtains target area network topology.
Step S202: target area network topology is subjected to convergence arrangement, forms the router level in Hong Kong and Taiwan
Network topology:
Due to IP address dynamic and changeability the features such as, the mapping of metastable Topology connection can not be formed, therefore
Using router level network topology as research object, the IP grade network topology that detection is obtained is by alias resolution, edit
Form router level network topology.
Specifically, the present invention extracts the section for clearly marking in ITDK ipv4 data set and being located at Taiwan and Hongkong
Point information arranges by convergence and obtains the router level topology connection of target area network topology.
Step S203: data prediction is carried out to the router level network topology in obtained Hong Kong and Taiwan:
Two zone controller grade network topologies are abstracted into two complex network topologies figures, node v indicates topological network
In router, side e of the node between node indicate that the two routers are adjacent (being namely separated by a distance from a jump).Respectively
It is visualized using Gephi software, discovery has many " parachute " phenomenons, that is, a degree in two networks
Bigger node connects the leaf node that many degree are 1.As shown in Fig. 2, a large amount of dark node group in two width figures, is exactly
Many degree largely assemble formation for 1 node.Such " parachute " phenomenon is generally existing in network topology, we are then
When the router level network topology for including 11 other different zones such as the U.S., Japan, South Korea is visualized, also have been found that
Such phenomenon.
Fig. 3 is the specific displaying of " parachute " phenomenon in the network topology of Taiwan, the deeper small group of black of color in figure
It is exactly such node group.The present invention first more focused on it is complicated to connection, be difficult to directly determine the section for belonging to which area
Point is analyzed, and " parachute " structure is a greatly interference, it is therefore desirable to which data are carried out with preliminary pretreatment.Data are pre-
The process of processing is exactly the process for rationally eliminating " parachute " phenomenon in original network topology.
The degree for counting all nodes in topological network all picks out the node that the degree being connected with the larger node of degree is 1
Come, by foldings merging, substitute them with the biggish node of the degree being connected with them.Such replacement method is taken, is to be based on
The understanding of network topology is determined, the node being largely only connected with a node, topological property is connected depending on this
Node.Fig. 4 is the network topological diagram in Hong Kong and Taiwan after data prediction, and the deeper small group of black of color is in figure
The node being completely embedded, it can be seen that network shows phenomenon of significantly uniting.
Step S204: die body digging is carried out on the router level network topological diagram in Hong Kong and Taiwan after the pre-treatment
Pick；
The step S204 includes:
Step S204.1: pretreated Hong Kong and Taiwan router level network topology are regarded as have no right it is undirected
Figure, the method enumerated first with subgraph carry out the excavation of die body, by the son of Z-score>2 and P-value<0.01 in figure
Figure regards as the die body of target network；
Step S204.2: for the local community discovery algorithm being extended based on initial seed, the choosing of initial seed
The influence selected to result is very big.And die body excavation is carried out for target area network, according to the difference of the die body size of setting, dig
It is also not identical to excavate the die body come, and even if setting same size, the subgraph constructor excavated in real network
Mold body selection criteria is more than one, therefore selects candidate of the die body of suitable size, Rational structure as initial seed
Just become the key point of this problem.
One die body needs to meet some basic items if the die body is accepted as the candidate of initial seed
Part:
On the one hand, die body size should be sufficiently large, is found out so that any bigger network area can be used as corporations
Come, when being otherwise extended based on initial seed, it is possible to extend the network area of not community structure into corporations, and obtain
To the result of mistake.For example, triangle can be received to become kind if die body size is selected as the subgraph of 3 nodes
Son, but within some network, not every triangle is all inside corporations.In this case, those will not had
Finally there is mistake as seed in triangle inside any corporations.
On the other hand, die body size also should be sufficiently small, all contains at least one to be hopeful the corporations detected
Die body.If the die body size of selection is too big, those small community structures for not including die body would not be detected, this
Kind situation is also not to be allowed to.
It is worth noting that die body size refers to the node number that die body includes.
The die body come out by analysis mining, in conjunction with network actual characteristic, the die body that size is 4 by the present invention is as initial
The candidate of seed.This also with the research achievement of scholar before (Lee C, Reid F, Mcdaid A, et al. Detecting
highly overlapping community structure by greedy clique expansion[J]. 2010.)
(still open-birth, Chen Duanbing, all great waves .Detecting Overlapping Communities Based on Community
Cores in Complex Networks [J] China Physics Letters: English edition, 2010,27 (5): 264-267.) match.
Table 1 is the die body that the size in the network topology of Hongkong is 4, and wherein Frequency [Original] is die body
The frequency occurred in target network topological diagram；Mean-Freq [Random] is all random network topologies of the die body in generation
The frequency occurred in figure；Standard-Dev [Random] is die body frequency of occurrences in all random network topology figures of generation
Standard deviation；Z-Score and p-Value is respectively Z value and P value, is the index for measuring die body importance.
The die body that size in 1 Hongkong network topology of table is 4
Table 2 is the die body that size is 4 in the network topology of Taiwan.
The die body that size is 4 in 2 Taiwan network topology of table
Step S205: die body is merged into beta pruning, forms initial seed set:
It will appear some die bodys in live network data and share certain nodes, or will appear some die bodys is other moulds
The subset situation of body.If not merged beta pruning, directly using all die bodys as initial seed structure, then passing through these
It is also almost the same that die body finally extends obtained community structure.The die body of shared 2 common nodes is merged into one by the present invention
A initial seed greatly reduced the quantity for needing the initial seed extended.As for only share 1 common node die body,
It is treated as the bridge node between two initial seeds, two different zones network topologies of connection is equivalent in real network
Backbone node, extension when be temporarily not considered.By the way that shared duplicate node die body is merged beta pruning, thus not influencing
While experimental result, the calculating time of memory overhead and Reduction algorithm is saved significantly on.
Step S206: utilize fitness function F, on network topology figure after the pre-treatment, carry out based on die body
Community discovery deletes duplicate corporations, forms the community structure not overlapped:
After the structure determination of initial seed, need to carry out respective extension to initial seed, to obtain community structure.This
The fitness function F that invention is defined using Lancichinetti et al., according to the intra-node degree of corporations SWith external degree
It defines the fitness of corporations S, is shown below.Wherein(i.e. equal to twice of quantity of side of the beginning and end all in S
For the sum of the degree of internal node in S), andIt is only number of edges amount of the one end in corporations S.Fitness function FsIs defined as:
σ is an adjustable parameter, is a positive real number, for controlling the scale of corporations.σ value is taken in Hongkong
It is 1.1, taking σ that must be worth in Taiwan is 1.0, and the corporations' result finally divided is more satisfactory.
In addition, also needing one node of definition to estimate to determine that can a node v be added into corporationsSuch as following formula
It is shown:
Judge that can node v be added to the standard of corporations S and beWhenWhen, indicate that the addition of node v can make
The adaptive metrology of corporations S increases.
The step S206 concrete operation step includes:
Step S206.1: for each node v adjacent with corporations S, i.e. dotted-line ellipse circle is outer interior with straight empty with circle in Fig. 5
The node of line connection calculates separately the fitness that corporations S is added after v；
Step S206.2: judgementIt is whether true, i.e., add whether v will increase the fitness of S into corporations S；
Step S206.3: selection makes the maximum node V of fitness incrementmax, it is added into corporations S；
Step S206.4: circulation executes step S206.1 to step S206.3, until no full in the neighbor node of corporations S
The node of sufficient step S206.2 terminates and returns to final corporations S；
Step S206.5: because of the initial seed expanding policy of step S206.1 to step S206.4, as long as theoretically
The neighbor node v's of corporations S estimatesCorporations S will extend down always.So in a network two apart from close
Initial seed forms corporations S in extension1And S2During, it is possible to by the node of other side when the node being Myself is divided into certainly
Oneself just will form approximate duplicate corporations on one side.From the perspective of Network topology, this be it is worthless, so this hair
It is bright that these approximate duplicate corporations are merged into a corporations, because it is considered that its internal contiguity is also enough after merging
It closely (is all based on what Fs extended), reaches the standard as corporations.
Step S207: the phenomenon that being directed to single-pathway present in topology proposes corporations' extension based on corporations' density, most
The big possible node by single-pathway is divided into corporations.
The step S207 includes:
Based on the network topological diagram that Traceroute principle detects, its essence is the nets that a plurality of detective path is constituted
Network.For same detection target, the route result returned from different detection sources, something in common is exactly can be by detection target
The border router of place network or the backbone router of this area's network, therefore show on the topology, just it will appear
The single-pathway of many not cross-connects, such case can occur mostly on backbone router and network boundary router,
As shown in Figure 6.
For the node on such single-pathway, a part is the destination node of selection when carrying out network measure, such
Node in the network before pretreatment the degree of its node be also 1, can be direct and because be directly connected with the node in corporations S
It is incorporated to corporations S.But there are also part of nodes be then in process of data preprocessing through folding merge its leaf node after gained
To special joint, they are greater than 1 by the path length that single-pathway reaches corporations, are shown in real network topology
" parachute " structure often can individually represent a small-scale Local Area Network topological structure, that is, itself can
To represent a corporations.It finds, is detected in obtained router level network topology connection in goal in research network topology, which
Be afraid of only to be separated by the distance (path length 1) of a jump, in reality distance may also a good distance off, therefore can not simply by
The node on all single-pathways being connected with S is directly merged with S.
In order to solve the partition problem of this kind of node, the present invention proposes the extended method again based on corporations' density, maximum
The ownership corporations of possible these nodes of determination.The formula of corporations' density of use are as follows:
Wherein M indicates the number of edges inside corporations S, the number of nodes after N expression addition node v inside corporations S, fd vFor section is added
Corporations' density of corporations S after point v.
It, will be on the single-pathway with " parachute " structure being connected with corporations S successively according to the sequence being connect with corporations S
Node be added in S, the f of calculate noded vValue, by fd vThe node that value is greater than threshold alpha is added in corporations S, is obtained final
Topology partition result.The value of α influences the final accuracy for dividing corporations' result S, and the present invention is calculated to be obtained after initial extension
Corporations' density of all S arrived takes final value of the minimum value therein as α.It is based on the considerations of this, it is initial by front
After seed extends and the node that the path length (degree) being connected directly with corporations is 1 is incorporated into S, S, which has been obtained, to be filled
Point extension, the simple path not being attached thereto under limiting case, and α chooses minimum corporations' density in S
Value can unite the density of final resulting network topology division result S, be more advantageous to analysis comparison different zones net
The otherness of network topological structure.
No " parachute " structure node being divided into any S is individually finally classified as a corporations, as front
As analysis, it often can individually represent a small-scale Local Area Network topological structure, can determine for network entity
Position analysis and the network topology situation for analyzing its locating region provide corresponding support.
Step S208: the corporations finally divided through the above steps are as a result, be utilized respectively based on terrestrial reference and based on
The method for knowing IP geographic information database, corporations' progress geographical location ballot to division, building corporations and geographical location are reflected
Penetrate relationship.
The step S208 includes:
Step S208.1: the mapping relations building of corporations and geographical location based on terrestrial reference is fairly simple, is executing network
When topology measurement, selectively using terrestrial reference known to some geographical locations as the destination node of detection, in this way last
The presence of terrestrial reference is just had in the corporations of division, so that it may the position of community structure is determined according to terrestrial reference.
Step S208.2: the mapping construction method based on geographic information database, using the corporations position based on voting mechanism
Set positioning.By inside known multiple IP address information libraries, each corporations' internal router node of the lookup of maximum possible
Location information, then each node votes to the corporations present position according to the location information of itself, will be more in corporations
Geographical location of the several sections of consistent positions of point as corporations constructs the mapping relations of corporations' node and geographical location.
As an embodiment, multiple IP geographic information databases be MaxMind, IP2Location,
Ipligence, HostIP, Netacuity, Geobytes, purity IP database and Taobao's IP address library.
Network topology division result based on Hong Kong and Taiwan is described in detail the present invention:
Present invention introduces the concepts of " die body " in biological networks, propose the die body that will be excavated from network as expansion
The initial seed of exhibition.It is because die body is the feature mode for significantly occurring in complex network, interacting, compared to phase
With the random network of global statistics characteristic, die body is more common in primitive network, is defined as regular and statistically significant
Subgraph or mode.Specifically, die body be exactly in a network it is a large amount of it is recurrent there is mutually isostructural small-scale subgraph,
These subgraphs feature AD HOC interconnected inside complex network from localized micro level, and announcement is complex network
Basic information structure or basic building module, can regard as foundation structure specific to the network to a certain extent, can
Embody the local characteristics of network.
And actually existing numerous Combo discovering methods based on network local message, it is all divided into two steps mostly
It is rapid: the selection of first step progress initial seed；Second step is diffused to initial seed and is arranged the condition for terminating diffusion.Due to
Influence of the selection of initial seed to the final community division of network is very big, it is therefore necessary to scientific and reasonable selection initial seed,
So proposing the die body that will be excavated from network as extension present invention introduces the concept of " die body " in biological networks
Initial seed.
The content that the present invention studies is target area network topology division methods.Target area network why is confined to open up
It flutters, mainly there is two o'clock consideration:
Consider 1: obtaining accurate, comprehensive topological connection relation is the basis for carrying out Network topology, although currently having
Many individuals and organizations are devoted for years in the measurement and analysis of interconnection net topology, but due to technology and the limit of capital investment
System, their obtained network topologies be it is incomplete, that is, want the interconnection net topology in the entire whole world of Overall Acquisition, this is not existing
Real thing.
Consider 2: internet innately has the characteristics that the lack of uniformity of isomerism and regional development, and each area is due to warp
The building mode of the various reasons such as Ji, politics, network is different, and development scale and level be not also in a level, entirely
The overall permanence of network cannot represent the characteristic of local area network completely, therefore the general research that mixes is undesirable
's.
Therefore the present invention puts forth effort on the network topology of research specific target areas, not only has realistic feasibility, Er Qiegeng
What is added has specific aim, but does not represent the present invention and be not applied for other type complex networks.
In order to assess advance of the invention, reflected respectively from corporations' quantity of division, the number of nodes for being divided into corporations, corporations
Accuracy rate and node mapping 4 aspects of accuracy rate are penetrated, the present invention (is opened with existing classics HC-Based based on network topology
The IP location algorithm of hairdo cluster) method and NNC (Network Node Clustering) method be compared analysis.
Corporations' quantity of division refers to the corporations' number finally found in the network topology of target area, is divided into the node of corporations
Quantity refers to the number for the node for including in final all corporations.The two indexs combine analysis, can effectively measure net
The granularity of network topology community division.Because network topology is limited within the scope of target area, largely just reduce
The quantity of its internal community structure, after all in target area the adjacent domain of population collection be it is denumerable, ideally
Be divide corporations quantity it is suitable with the region quantity of population collection in target area.If corporations' quantity of algorithm partition is too
It is more, then the number of nodes inside corresponding each corporations will be seldom, it will be by the more fine granularity of the network topology in close region
Division, be heavily located at the corporations of the same area to generate, lead to repetitive work；If community division quantity is very little,
The network topology node belonged in different zones will be mistakenly considered to lead to the result of mistake in the same region.In division
Corporations' quantity and the number of nodes aspect for being divided into corporations, the results are shown in Table 3 for three kinds of methods.
Comparison of the 3 three kinds of methods of table to the number of nodes of community division quantity and division
As can be seen from Table 3, method of the invention is either still divided into the number of nodes of corporations in community division quantity
Aspect all has advantage.
In terms of corporations map accuracy rate, mainly in the finally result of building community structure and geolocation mapping relationship
It compares and analyzes.The mapping building based on terrestrial reference is respectively adopted and the mapping based on open IP address information library constructs two kinds of sides
Method carries out geolocation mapping building to the community structure that three kinds of methods divide.If the corporations that two kinds of mapping construction methods determine
Structure geographical location is identical, then it is assumed that mapping relations building is accurate；If the community structure geographical location that two methods determine is different,
Then think that mapping relations construct mistake；The community structure that not can determine that position for both of which is then regarded as not determining
The corporations of mapping relations also regard as mapping relations building mistake.It is correct that accuracy rate is finally set as mapping relations building
The ratio of corporations' quantity and total corporations' quantity, therefore accuracy rate is higher, shows that the effect of method is better.The result of three kinds of methods
As shown in table 4.
Corporations' mapping accuracy rate comparison of 4 three kinds of methods of table
As can be seen from Table 4, method of the invention can obtain higher in the experiment of two regional network topologys
Accuracy rate, especially on the corporations' quantity for not determining regional location, method of the invention is obviously than other two methods
Few very much, on corporations' quantity of mapping relations building mistake, method of the invention also has apparent advantage.
In terms of node maps accuracy rate, the total quantity that mapping constructs node inside correct corporations has also further been counted,
It can be used to illustrate that more network entities can accurately be mapped by which kind of method with geographical location, as a result such as 5 institute of table
Show.
The node mapping accuracy rate comparison of 5 three kinds of methods of table
Table 5 the result shows that, method of the invention can get up more network entities with geolocation mapping.
The present invention is divided into research object with target area network topology, fully considers the structure of target area network topology
Characteristic excavates die body in network topology first, forms initial seed set by the merging beta pruning of die body, then utilizes adaptation
Degree function is extended, and considers the characteristic of node on single-pathway in network, proposes corporations' extension based on corporations' density,
More nodes are divided into corporations as far as possible, are finally based respectively on terrestrial reference and the ballot of known IP geographic information database in corporations
Method, network entity is mapped with geographical location, geolocation mapping building is carried out to the corporations of division.By a large amount of
Experiment obtain, for network topology division for, method of the invention is more reasonable, can not only by more nodes and ground
Reason position maps, and accuracy rate is higher.
Illustrated above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered
It is considered as protection scope of the present invention.
Claims (6)
1. the target area network topology division methods based on die body, which comprises the following steps:
Step 1 obtains target area network topology；
Target area network topology is carried out convergence arrangement by step 2, and the IP grade network topology that detection is obtained is by alias solution
Analysis forms router level network topology；
Step 3 carries out data prediction to obtained router level network topology: all sections in statistics router level network topology
The degree of point all picks out the node that the degree being connected with the larger node of degree is 1, is connected with the node for being 1 with the degree
Spend the node that biggish node substitution degree is 1；
Pretreated router level network topology is abstracted into one and haves no right undirected network topological diagram by step 4, in the net
The excavation of die body is carried out in network topological diagram；
Step 5 chooses candidate of the die body as initial seed, and merges beta pruning, determines final initial seed set；
Step 6 is extended initial seed set to form 1 or more corporations based on fitness function F, deletes duplicate society
Group, forms the community structure not overlapped；
Step 7 is extended corporations based on corporations' density, determines the ownership corporations of node on single-pathway；
Step 8, the method for being based respectively on terrestrial reference and the ballot of IP geographic information database, network entity is corresponding with geographical location,
Geolocation mapping building is carried out to the corporations of division.
2. the target area network topology division methods according to claim 1 based on die body, which is characterized in that the step
Rapid 4 include:
Pretreated router level network topology is abstracted into one and haves no right undirected network topological diagram by step 4.1, utilizes son
The method of enumeration of graph carries out subgraph search in the network topological diagram, counts the number that each drawing of seeds occurs；
Step 4.2, the network topological diagram according to step 4.1 construct network topological diagram described in multiple and step 4.1 and exist
The identical random network topology figure of whole statistical property；
Step 4.3 carries out subgraph search on the random network topology figure of building, counts the number that each drawing of seeds occurs；
Step 4.4, comparison same sub-image network topological diagram described in step 4.1 are opened up with the random network described in step 4.3
The frequency occurred in figure is flutterred, the subgraph of Z-score>2 and P-value<0.01 is regarded as to the die body of target area network.
3. the target area network topology division methods according to claim 1 based on die body, which is characterized in that the step
Rapid 5 include:
Candidate of the die body that step 5.1, selected size are 4 as initial seed；
The die body of shared 2 common nodes is merged into an initial seed by step 5.2, will only share the mould of 1 common node
Body is as the bridge node between two initial seeds.
4. the target area network topology division methods according to claim 1 based on die body, which is characterized in that the step
Rapid 6 include:
Step 6.1 is extended initial seed set based on fitness function F, for each node v adjacent with corporations S,
The fitness that corporations S is added after node v is calculated separately, it then will be before the fitness function value and addition after addition node v
Fitness function value compares, and judges to add whether node v will increase the fitness of corporations S into corporations S, to decide whether
Corporations S is added in node v, if will increase the fitness of corporations S, corporations S is added in node v, if not will increase corporations S
Fitness, then not by node v be added corporations S；
Step 6.2 repeats step 6.1, until there is no the fitness that can increase corporations S in the neighbor node of corporations S
Node, terminate and return to final corporations S；
Step 6.3 deletes duplicate corporations, forms the community structure not overlapped.
5. the target area network topology division methods according to claim 4 based on die body, which is characterized in that the step
Rapid 7 include:
The calculation formula of corporations' density are as follows:
Wherein M indicates the number of edges inside corporations S, the number of nodes after N expression addition node v inside corporations S；
According to the sequence connecting with corporations S, the node that the degree on the single-pathway being connected with corporations S is 1 is successively added to S
In, the f of calculate noded vValue, by fd vThe node that value is greater than threshold alpha is added in corporations S, obtains final Topology partition result.
6. the target area network topology division methods according to claim 1 based on die body, which is characterized in that the step
Rapid 8 include:
When step 8.1, execution network topology measurement, using terrestrial reference known to geographical location as the destination node of detection；
Step 8.2, the mapping construction method based on IP geographic information database, determine corporations position using based on voting mechanism:
By searching the location information of each corporations' internal node in known IP geographic information database, each node is according to itself
Location information vote the corporations present position, using the consistent positions of node most in corporations as the geographical position of corporations
It sets, constructs the mapping relations of each node and geographical location in corporations.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811236808.4A CN109120465B (en) | 2018-10-23 | 2018-10-23 | Target area network topology division method based on motif |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811236808.4A CN109120465B (en) | 2018-10-23 | 2018-10-23 | Target area network topology division method based on motif |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109120465A true CN109120465A (en) | 2019-01-01 |
| CN109120465B CN109120465B (en) | 2021-07-02 |
Family
ID=64855399
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811236808.4A Active CN109120465B (en) | 2018-10-23 | 2018-10-23 | Target area network topology division method based on motif |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109120465B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110012120A (en) * | 2019-03-14 | 2019-07-12 | 罗向阳 | A kind of IP City-level location algorithm based on PoP network topology |
| CN110188954A (en) * | 2019-05-31 | 2019-08-30 | 中国人民解放军战略支援部队信息工程大学 | Terrestrial reference reliability estimation method and device based on POP network |
| CN112235134A (en) * | 2020-09-29 | 2021-01-15 | 中国人民解放军战略支援部队信息工程大学 | Router ownership detection method and system based on IP connection probability classification |
| CN113485385A (en) * | 2021-07-13 | 2021-10-08 | 中国人民解放军战略支援部队信息工程大学 | UUV cluster formation configuration design method based on error ellipse |
| CN115001971A (en) * | 2022-04-14 | 2022-09-02 | 西安交通大学 | Virtual network mapping method for improving community discovery under heaven-earth integrated information network |
| CN116248573A (en) * | 2022-12-01 | 2023-06-09 | 中国联合网络通信集团有限公司 | Link splicing method, device and storage medium |
| CN116708273A (en) * | 2023-07-28 | 2023-09-05 | 中国电子科技网络信息安全有限公司 | Method for generating stopping judgment classification model, network topology detection method and device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103457752A (en) * | 2012-05-30 | 2013-12-18 | 中国科学院声学研究所 | Virtual network mapping method |
| CN103500168A (en) * | 2013-09-02 | 2014-01-08 | 中国矿业大学 | Method and system for discovering communities in overlapped complex networks according to topology potential |
| US9749406B1 (en) * | 2013-03-13 | 2017-08-29 | Hrl Laboratories, Llc | System and methods for automated community discovery in networks with multiple relational types |
| CN107357858A (en) * | 2017-06-30 | 2017-11-17 | 中山大学 | A kind of network reconstruction method based on geographical position |
| CN107623594A (en) * | 2017-09-01 | 2018-01-23 | 电子科技大学 | A kind of three-dimensional level network topology method for visualizing of geographical location information constraint |
| CN107944705A (en) * | 2017-11-27 | 2018-04-20 | 国网河南省电力公司信息通信公司 | A kind of all terminal reliability computational methods that communication corporations are divided based on modularity |
-
2018
- 2018-10-23 CN CN201811236808.4A patent/CN109120465B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103457752A (en) * | 2012-05-30 | 2013-12-18 | 中国科学院声学研究所 | Virtual network mapping method |
| US9749406B1 (en) * | 2013-03-13 | 2017-08-29 | Hrl Laboratories, Llc | System and methods for automated community discovery in networks with multiple relational types |
| CN103500168A (en) * | 2013-09-02 | 2014-01-08 | 中国矿业大学 | Method and system for discovering communities in overlapped complex networks according to topology potential |
| CN107357858A (en) * | 2017-06-30 | 2017-11-17 | 中山大学 | A kind of network reconstruction method based on geographical position |
| CN107623594A (en) * | 2017-09-01 | 2018-01-23 | 电子科技大学 | A kind of three-dimensional level network topology method for visualizing of geographical location information constraint |
| CN107944705A (en) * | 2017-11-27 | 2018-04-20 | 国网河南省电力公司信息通信公司 | A kind of all terminal reliability computational methods that communication corporations are divided based on modularity |
Non-Patent Citations (1)
| Title |
|---|
| CLARA PIZZUTI等: "An evolutionary motifs-based algorithm for community detection", 《2017 8TH INTERNATIONAL CONFERENCE ON INFORMATION, INTELLIGENCE, SYSTEMS & APPLICATIONS (IISA)》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110012120A (en) * | 2019-03-14 | 2019-07-12 | 罗向阳 | A kind of IP City-level location algorithm based on PoP network topology |
| CN110188954A (en) * | 2019-05-31 | 2019-08-30 | 中国人民解放军战略支援部队信息工程大学 | Terrestrial reference reliability estimation method and device based on POP network |
| CN112235134A (en) * | 2020-09-29 | 2021-01-15 | 中国人民解放军战略支援部队信息工程大学 | Router ownership detection method and system based on IP connection probability classification |
| CN113485385A (en) * | 2021-07-13 | 2021-10-08 | 中国人民解放军战略支援部队信息工程大学 | UUV cluster formation configuration design method based on error ellipse |
| CN113485385B (en) * | 2021-07-13 | 2023-11-07 | 中国人民解放军战略支援部队信息工程大学 | UUV cluster formation configuration design method based on error ellipse |
| CN115001971A (en) * | 2022-04-14 | 2022-09-02 | 西安交通大学 | Virtual network mapping method for improving community discovery under heaven-earth integrated information network |
| CN115001971B (en) * | 2022-04-14 | 2023-06-20 | 西安交通大学 | Virtual network mapping method for improving community discovery under space-earth integrated information network |
| CN116248573A (en) * | 2022-12-01 | 2023-06-09 | 中国联合网络通信集团有限公司 | Link splicing method, device and storage medium |
| CN116708273A (en) * | 2023-07-28 | 2023-09-05 | 中国电子科技网络信息安全有限公司 | Method for generating stopping judgment classification model, network topology detection method and device |
| CN116708273B (en) * | 2023-07-28 | 2023-10-31 | 中国电子科技网络信息安全有限公司 | Method for generating stopping judgment classification model, network topology detection method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109120465B (en) | 2021-07-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109120465A (en) | Target area network topology division methods based on die body | |
| Wang et al. | Boundary recognition in sensor networks by topological methods | |
| Zhi-Xiao et al. | Overlapping community detection based on node location analysis | |
| Marcus et al. | Rage–a rapid graphlet enumerator for large networks | |
| CN1925437B (en) | System and method for detecting status changes in a network | |
| CN109784636A (en) | Fraudulent user recognition methods, device, computer equipment and storage medium | |
| Tian et al. | Topology mapping and geolocating for China's Internet | |
| CN108683448B (en) | Influence node identification method and system suitable for aviation network | |
| CN102880799A (en) | Method for comprehensively evaluating importance of complicated network node based on multi-attribute decision-making | |
| CN105227689B (en) | Target IP location algorithm based on local time delay distribution similarity measurement | |
| Ferrara et al. | The role of strong and weak ties in Facebook: a community structure perspective | |
| Wang et al. | A very fast algorithm for detecting community structures in complex networks | |
| Allard et al. | Percolation and the effective structure of complex networks | |
| CN107818534A (en) | A kind of mankind's activity network area division methods with space constraint | |
| CN103020267A (en) | Complex network community structure mining method based on triangular cluster multi-label transmission | |
| CN108809697A (en) | Social networks key node recognition methods based on maximizing influence and system | |
| Yu et al. | Characterizing the spatial-functional network of regional industrial agglomerations: A data-driven case study in China's greater bay area | |
| Hristova et al. | Mapping community engagement with urban crowd-sourcing | |
| CN111064817A (en) | City-level IP positioning method based on node sorting | |
| CN109639510A (en) | A kind of region PoP division methods based on subnets analysis | |
| CN110442800A (en) | A kind of semi-supervised community discovery method of aggregators attribute and graph structure | |
| Sharma et al. | Community mining in signed social networks-an automated approach | |
| CN106550387B (en) | A kind of wireless sensor network routing layer QoS evaluating method | |
| Haddadi et al. | Mixing biases: Structural changes in the AS topology evolution | |
| Wang | Using the relationship of shared neighbors to find hierarchical overlapping communities for effective connectivity in IoT |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |