CN112134738B - Network multidimensional data flow simulation device based on composite two-dimensional Sketch - Google Patents
Network multidimensional data flow simulation device based on composite two-dimensional Sketch Download PDFInfo
- Publication number
- CN112134738B CN112134738B CN202011012284.8A CN202011012284A CN112134738B CN 112134738 B CN112134738 B CN 112134738B CN 202011012284 A CN202011012284 A CN 202011012284A CN 112134738 B CN112134738 B CN 112134738B
- Authority
- CN
- China
- Prior art keywords
- data
- network
- simulation
- flow
- sketch
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
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
- H04L41/14—Network analysis or design
- H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/211—Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/216—Parsing using statistical methods
-
- 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
- H04L41/08—Configuration management of networks or network elements
- H04L41/0803—Configuration setting
- H04L41/0813—Configuration setting characterised by the conditions triggering a change of settings
- H04L41/082—Configuration setting characterised by the conditions triggering a change of settings the condition being updates or upgrades of network functionality
-
- 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
- H04L41/08—Configuration management of networks or network elements
- H04L41/0803—Configuration setting
- H04L41/0823—Configuration setting characterised by the purposes of a change of settings, e.g. optimising configuration for enhancing reliability
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a network multidimensional data flow simulation device based on composite two-dimensional Sketch, and belongs to the field of network protocol test data construction and simulation. The invention uses Sketch to store the network data flow change, greatly saves the storage space, and can ensure the characteristics of the data flow model compared with the prior scheme; the conversion from the discrete Sketch group to the continuous packet-sending matrix is realized by a curve fitting and interpolation calculation method, and the method can realize the compression and expansion of a flow model in a time domain through the interpolation number and position, and even support the variation and the flow shaping; the existing flow simulation can only be based on the existing probability distribution, repetition and deletion, and the scheme is superior to the existing scheme in reality; the learning function and the simulation function are deployed in the network together, so that the integration degree is high, and the management is easy; a network flow model storage method convenient for storage and indexing is designed, and the completeness and the practicability of the scheme are improved.
Description
Technical Field
The invention belongs to the field of network protocol test data construction and simulation, and particularly relates to a network multidimensional data flow simulation device based on composite two-dimensional Sketch.
Background
With the computer network becoming a key means for data communication in offices, businesses and communities, a great deal of network testing needs are also brought forward. Currently above the transport layer. In a communication interconnection network, an application layer data simulation test needs to simulate a large number of users and servers, and in order to test the performance of network equipment and services, complex business communication needs to be constructed, and the performance and the authenticity of a test data stream are considered. Especially, in order to have a real network traffic model, it is necessary to have multiple protocol service data streams at the same time, and the existing schemes have advantages and disadvantages in the aspects of reality, occupied storage space and flexibility.
Therefore, the scheme performs comparative analysis from three aspects of authenticity, occupied storage space and flexibility around the theory and the realization of the existing scheme:
(1) The data flow model defined by the RFC standard. In the RFC2544 document, a plurality of test data flow models are defined, the model is composed of connection topology, data structure, transmission rate, simulation address and the like, the traffic model defines that each test port transmits test data flow to all other ports, the transmission sequence is strictly controlled, and test data models such as full mesh, partial mesh unidirectional flow and the like are formed based on the requirements. The scheme occupies a lower storage space, but is only used for performance testing, so that the authenticity of the transmitted data is poor, and the flexibility is poor.
(2) A real communication data replay model. In many data playback products, real grab packets are used for transmission, and some products may also adjust the transmission rate. The scheme can faithfully send out recorded network communication historical data, but occupies a large amount of storage space after being saved into a file, and the currently known scheme does not change the content. The scheme is not suitable for long-time simulation and test of real network communication.
(3) A queuing theory theoretical model. A plurality of communication models are defined in the queuing theory, but the models are partial to the theory, do not describe the content of the data frame more finely, and can only be used for theoretical research.
In summary, the existing network communication traffic model and simulation test cannot achieve reality, lightweight and flexibility at the same time, and a scheme of abstracting mass data of network communication and recovering real data based on a recorded lightweight template is required. The two-dimensional Sketch is a data structure composed of a hash function group, a counter matrix and a data stream identification table, has the characteristics of low storage overhead and reliable counting precision, and can be specifically defined by referring to the paper "a improved data stream summary: the count-min Sketch and the bits applications". In the application, the collected object which runs for a long time is compressed by adopting a sampling technology, and key information and statistical values are stored into simulation. The Sketch groups are restored to be true test data through a specific data simulation strategy.
The application layer communication has a complex communication flow and flexible data load, so that the simulation test has high difficulty. The data flow model defined by the RFC standard is only used for performance testing, so that the authenticity of the sent data is poor, the flexibility is poor, and the occupied storage space is low. Real communication data playback models find application in many data playback products, some of which may also adjust the transmission rate using real captured data frames as test data. The scheme can faithfully send out recorded network communication historical data, but occupies a large amount of storage space after being saved into a file, the address and the communication content may need to be changed during actual test, and the scheme is not suitable for long-time simulation and test of real network communication. A plurality of communication models are defined in the queuing theory, but the models are partial to the theory, do not describe the content of the data frame more finely, and can only be used for theoretical research.
In view of the above disadvantages, the present invention aims to deploy multiple traffic model learning nodes in a whole network, record network real communication through two-dimensional Sketch, and form a continuous abstraction of a service through several continuous sketches. The advantage of this scheme is that Sketch can store a large amount of information with very little storage overhead, while preserving the data stream core data by clustering deduplication compression. When the network flow simulation test is executed, a curve fitting algorithm can be used for interpolation calculation to recover discrete Sketch into continuous flow, data frame filling is carried out according to core data, and data volume is increased through repetition, random modification and calculation functions under the necessary condition, so that the stored simulation data is compressed or prolonged, and the generation and simulation of the whole network complex flow are supported.
Disclosure of Invention
Aiming at the technical problems in the prior art, the invention provides a network multidimensional data flow simulation device based on composite two-dimensional Sketch, aiming at deploying a plurality of flow model learning nodes in the whole network, recording network real communication through the two-dimensional Sketch, and forming continuous abstraction of a service through a plurality of continuous Sketch; reasonable in design has overcome prior art's not enough, has good effect.
In order to achieve the purpose, the invention adopts the following technical scheme:
the network multi-dimensional data flow simulation device based on the composite two-dimensional Sketch comprises a full network data flow simulation control program module, a network communication model storage module and a network communication flow simulation node; the three components adopt a decoupling modular design;
the network communication traffic simulation control program module is configured to control a network communication traffic model learning module and a network communication traffic simulation module of the network communication traffic simulation node;
the network communication model storage module comprises a Sketch database, a load database and a recovery strategy database; configured to store, maintain and update network communication traffic model data required by the network-wide data flow simulation control program module;
the network communication flow simulation node comprises a network communication flow model learning module and a network communication flow simulation module; configured to be responsible for independently collecting data streams passing through the physical node and processing and abstracting the data streams into traffic model information; the network communication traffic simulation node supports an embedding mode, a direct connection mode and a serial connection mode, wherein the embedding mode is installed in equipment of a tested network in a software or hardware mode, the direct connection mode is connected with the tested network through specific network equipment, and the serial connection mode is accessed between two adjacent equipment of the tested network by using two ports.
Preferably, the full network data flow simulation control program module has a learning module management function and a simulation module management function;
the flow and the model of the full network data flow can not be comprehensively collected or simulated from one physical node, and the functions are realized by adopting a distributed mode of a full network data flow simulation control program module and a plurality of network communication flow simulation nodes;
the learning module management function comprises a control function and a return function;
control function
A full-network data flow simulation control program module needs to control a network communication flow model learning function of a network communication flow simulation node, and specifically comprises the on/off of the function, flow learning configuration information, an acquisition sampling timer and an acquisition cycle trigger; wherein, the flow learning configuration information comprises the following contents:
a) The shunt parameters give different matching domains, matching values and corresponding Sketch groups;
b) The Sketch group configures parameters, and gives the number of hash functions, the width of a counter and a data stream identification storage structure of each Sketch;
c) Selecting clustering parameters according to the type of a clustering algorithm by using basic configuration parameters of a clustering classifier, and extracting high-frequency short sentences with preferential lengths from the application layer data;
d) Basic configuration parameters of the high-frequency short sentence tree comprise the depth of the tree, a leaf node partition function, the minimum sample number of the leaf nodes and the maximum sample number parameters of the leaf nodes;
e) Basic configuration parameters of a data compression function comprise a compression method, a storage format and the like of model data;
f) Collecting the time length of a sampling timer;
g) Collecting the time length of a periodic trigger;
backhaul function
After learning of the network communication flow model is completed, the network communication flow simulation node compresses relevant information and uploads the information to the full-network data flow simulation control program module, and a data frame before compression comprises network simulation node basic information, network flow model basic information, a data flow identification set, sketch group data, time scale data and high-frequency short sentence data; the network traffic model data structure includes:
a) Network simulation node basic information;
b) Basic information of a network flow model;
c) A set of data stream identifications;
d) Sketch group data and time mark data;
e) High-frequency short sentence data;
the simulation module management function comprises a control function and a return function;
control function
The full network data flow simulation control program module needs to control the network communication flow simulation function of the network communication flow simulation node, and specifically comprises the steps of transmitting a restored network communication flow model to network communication flow, wherein the restored network communication flow model comprises a data flow discrete statistic value, a data flow identification list, a load template, a recovery strategy and a continuous packet-sending matrix sliding window, and the basic structure of a network flow simulation configuration file comprises the following steps:
a) Data flow dispersion statistic, namely real statistic data of network flow stored by Sketch;
b) A data stream identifier list, a set of data stream identifiers possessed by each Sketch;
c) A load template, a data template used by the load template, and a clustering abstraction from real data;
d) A restoration strategy, wherein the restoration strategy is a method for restoring network data flow based on a data flow dispersion statistic and a load template;
e) A continuous packet-sending matrix sliding window gives the number of the Sketch actually processed, and only the Sketch in the sliding window is simulated;
f) High-frequency short sentence data is used for filling the load part and increasing the simulation truth degree;
backhaul function
In the process of generating simulation flow, the simulation node transmits some basic operation state information back to the whole network data flow simulation control program module, wherein the operation state information comprises the sent data volume and the sliding window position information;
preferably, the network communication traffic model learning module is configured to store the data stream identifier, the traffic distribution condition, and the data of a large amount of real data streams when the data streams pass through the network communication traffic simulation node; different data streams are mapped to independent sketches by adopting a splitter, data frames are counted into discrete Sketch groups by the different sketches according to a splitting result, the interval of each group is controlled by an acquisition period trigger, the content of the data frames is processed by a clustering algorithm, and high-frequency short sentences are obtained and finally are compressed and uploaded to a full-network data stream simulation control program module; the network communication flow model learning module comprises the following basic components:
a) A flow divider; the flow divider comprises a Hash filter, a Bloom filter and a key value table, and the identifier of each independent data stream comprises a quintuple, a protocol type and a port number;
b) A two-dimensional Sketch counter; a two-dimensional counter structure including CountMin Sketch and Count Sketch;
c) A cluster classifier; because the data volume in the network flow is huge, in order to reduce the storage capacity burden of the simulation node, a large number of high-frequency short sentences are clustered according to different applications according to the characteristic that the data of the service layer has a fixed mode; the whole network data flow simulation control program module can configure clustering strategies comprising length priority, service priority and user priority;
d) A high-frequency short sentence tree; the high-frequency short sentence tree is responsible for storing the classified high-frequency short sentences;
e) A data compression module; the data compression module is responsible for compressing the network flow model before sending the network flow model, so that the network bandwidth occupation overhead is reduced;
f) Collecting a sampling timer; the acquisition sampling timer controls the period length of Sketch statistics to play a role in controlling collision probability;
g) Collecting a periodic trigger; the acquisition period trigger controls the network communication flow model learning module to use the pause time counted by Sketch, so that the effect similar to sampling is achieved;
the network communication flow simulation module is used for restoring a network flow data model issued by the full network data flow simulation control program module into test data again and supporting the shortening and prolonging of simulation data on a time domain; restoring the discrete Sketch statistical data into a simulation flow control matrix by adopting a curve fitting and interpolation combined mode, and then combining a restoration strategy and a load template to generate a restored session; the network communication flow simulation module comprises the following basic components:
a) The Sketch group comprises a data stream identification list and a data stream discrete statistic value, and is used for accurately indicating a simulated data stream set and giving a data sending rate during simulation, and the direct sending of the discrete list is equivalent to the compression of network flow;
b) The load template is an abstract application layer data template, and high-frequency short sentence data is inserted into the load template to form simulation data;
c) The recovery strategy defines information including a time domain stretch coefficient, a fitting function, an interpolation algorithm, a fitting insertion mode and a load parameter;
d) The high-frequency short sentence data are obtained by a network communication flow model learning module;
preferably, the network communication traffic model store comprises the following basic components:
1) Sketch database
The Sketch data frame comprises a data stream identification and a two-dimensional counter value, and the data stream identification is mapped into a specific count value through each Sketch independent hash function;
2) Load database
The load database stores communication templates and high-frequency data, including connection establishment, accounts, connection interruption, application data and Web data which are commonly used in simulation;
3) Restoration policy database
The restoration strategy database is used for storing restoration strategies for network communication flow simulation, and comprises a time domain stretch coefficient, a fitting function, a fitting insertion mode and a load parameter;
a) Time domain stretch coefficient
The time domain stretch coefficient is the ratio of the total sending time of the actual simulation data stream to the sending time of the Sketch group, and reflects the degree of compression and extension of the network flow model;
b) Fitting function
The fitting function obtains a flow distribution change curve through the discrete Sketch group, and then obtains a continuous flow sending matrix through interpolation;
c) Fitting interpolation patterns
The fitting interpolation mode can select different modes including an interpolation algorithm, interpolation frequency, uniform interpolation and non-uniform interpolation;
d) Load parameter
The load parameters define usage rules for the load data, including native load, splice load, and variant load.
Preferably, the work flow of the network communication traffic model learning module in the network communication traffic simulation node is as follows:
step 1: after a network communication flow learning task is started by a full network data flow simulation control program module, the acquisition time and period, and the kind and quantity of Sketch of a network communication flow model learning module are configured;
step 2: after the configuration is finished, acquiring a data stream to obtain a data frame, processing the data frame by using a classifier, and judging whether corresponding Sketch exists or not;
if: judging that the corresponding Sketch exists; executing the step 3;
or judging that no corresponding Sketch exists; executing the step 4;
and step 3: using special Sketch statistics;
and 4, step 4: using default Sketch statistics;
and 5: extracting high-frequency words by using a cluster classifier;
step 6: adding a timestamp label;
and 7: when the sampling timer is overtime, compressing a network communication flow model;
and 8: and uploading the data.
Preferably, the work flow of the network communication traffic model simulation module in the network communication traffic simulation node is as follows:
step S1: configuring a data flow identification list, a data flow dispersion statistic value, a load template, a recovery strategy and high-frequency short sentence data of a network communication flow model simulation module through a full network data flow simulation control program module;
step S2: performing fitting interpolation calculation on discrete Sketch groups in the data stream discrete statistic according to curve fitting parameters in a recovery strategy to obtain statistic interpolation meeting the requirement of playback time;
and step S3: combining the statistic interpolation with the original Sketch to form a data stream continuous statistic;
and step S4: acquiring a simulation rate matrix by using a data stream identification list through a query algorithm defined by Sketch;
step S5: the recovery strategy can specify a filling mode, application layer sessions are filled according to the load template and the high-frequency short sentence data, and a plurality of application layer sessions are merged to form simulation data;
step S6: and the network communication flow simulation module sends simulation data under the control of the simulation rate matrix to form test data flow which is the occurrence of the test data on the simulation node.
The invention has the following beneficial technical effects:
the method has the advantages that the Scut can store a large amount of information with extremely little storage overhead, and simultaneously, the core data of the data stream is subjected to clustering de-duplication compression storage. When a network flow simulation test is executed, a curve fitting algorithm can be used for interpolation calculation to recover discrete Sketch into continuous flow, data frame filling is carried out according to core data, and data volume is increased through repetition, random modification and calculation functions under necessary conditions, so that the stored simulation data is compressed or prolonged, and the generation and simulation of the whole network complex flow are supported; the method comprises the following specific steps:
the method has the advantages that (1) network data flow change is stored by using Sketch, so that storage space is greatly saved, and compared with the existing scheme, the characteristics of a data flow model can be guaranteed;
secondly, the conversion from the discrete Sketch group to the continuous packet sending matrix is realized through a curve fitting and interpolation calculation method, and the method can realize the compression and expansion of a flow model in a time domain through the interpolation number and position, and even support the variation and the flow shaping; the existing flow simulation can only be based on the existing probability distribution, repetition and deletion, and the scheme is superior to the existing scheme in reality;
thirdly, the learning function and the simulation function are deployed in the network together, so that the integration degree is high and the management is easy;
and (IV) a network flow model storage method convenient for storage and indexing is designed, and the completeness and the practicability of the scheme are improved.
Drawings
FIG. 1 is a diagram of a full network data flow simulation architecture.
Fig. 2 is a block diagram of a full network data flow simulation.
Fig. 3 is a block diagram of a data structure of a network traffic model.
Fig. 4 is a schematic diagram of a network traffic simulation configuration file structure.
FIG. 5 is a learning module workflow diagram.
FIG. 6 is a flow chart of simulation module data generation.
Detailed Description
The invention is described in further detail below with reference to the following figures and detailed description:
the technology comprises a full-network data flow simulation control program, a network communication model storage and a network communication simulation node, wherein the specific architecture is shown as 1, the three components adopt a decoupled modular design, the full-network data flow simulation control program is responsible for configuring a simulation node data flow model to learn and test data flow generation, and the network communication model storage is used for storing relevant data. The method mainly comprises three parts of architecture design, communication flow model learning and communication flow simulation.
Network multidimensional data flow simulation architecture of (one) composite two-dimensional Sketch
1.1 full network data flow simulation control program (hereinafter referred to as control program)
Because the flow and the model of the full network data flow cannot be comprehensively collected or simulated from one physical node, the scheme adopts a distributed mode of one control program and a plurality of network communication flow simulation nodes (hereinafter referred to as simulation nodes) to realize the functions. The control program mainly realizes the learning and simulation of the network communication flow model and the access of the related data of the network flow model.
1) Learning module management function
(1) Control function
The control program needs to control the network communication traffic model learning function of the simulation node, and specifically comprises the on/off of the function, traffic learning configuration information, an acquisition sampling timer and an acquisition period trigger. Traffic learning configuration information includes, but is not limited to, the following:
a) The shunt parameters give different matching domains, matching values and corresponding Sketch groups;
b) The Sketch group configures parameters, and gives the number of hash functions, the width of a counter, a data stream identification storage structure and the like of each Sketch;
c) Selecting clustering parameters according to the type of a clustering algorithm by using basic configuration parameters of a clustering classifier, and extracting high-frequency short sentences with preferential lengths from the application layer data;
d) The basic configuration parameters of the high-frequency short sentence tree, including but not limited to the tree depth, the leaf node partition function, the minimum sample number of the leaf nodes, the maximum sample number of the leaf nodes and other parameters, can be realized by adopting the existing method;
e) Basic configuration parameters of a data compression function, including but not limited to a compression method and a storage format of model data;
f) Collecting the time length of a sampling timer;
g) The length of time of the cycle trigger is collected.
(2) Backhaul function
After learning of the network communication flow model is completed, the network communication flow simulation node compresses relevant information and uploads the information to the control program, and a file structure before compression is a full network data flow simulation architecture diagram as shown in fig. 1.
Fig. 2 is a block diagram of a full network data flow simulation.
As shown in the figure, the data frame before compression comprises, but is not limited to, network simulation node basic information, network traffic model basic information, a data stream identification set, sketch group data, time mark data and high-frequency phrase data.
A) Network simulation node basic information;
b) Basic information of a network flow model;
c) A set of data stream identifiers;
d) Sketch group data and time mark data;
e) High frequency phrase data.
2) Simulation module management function
(1) Control function
The control program needs to control a network communication traffic simulation function of the network communication traffic simulation node, and specifically includes transmitting the restored network communication traffic model to the network communication traffic, including but not limited to a data flow discrete statistic value, a data flow identification list, a load template, a restoration policy, and a continuous packet-sending matrix sliding window, where a basic structure of a network traffic simulation configuration file is as shown in fig. 4.
A) Data flow dispersion statistic, namely real statistic data of network flow stored by Sketch;
b) A data stream identifier list, a set of data stream identifiers possessed by each Sketch;
c) The load template, a data template used by the simulation, is from the clustering abstraction of real data;
d) A restoration strategy, wherein the restoration strategy is a method for restoring network data flow based on a data flow dispersion statistic and a load template;
e) A continuous packet-sending matrix sliding window gives the number of the Sketch actually processed, and only the Sketch in the sliding window is simulated;
f) And high-frequency short sentence data is used for filling the load part and increasing the simulation truth.
(2) Backhaul function
In the process of generating the simulation flow, the simulation node will transmit some basic operation state information back to the control program, where the information includes, but is not limited to, the sent data amount, the sliding window position, and the like.
1.2 network communication traffic simulation node
The node can be deployed on network equipment or a test instrument in the modes of independent software, plug-ins and the like, and is composed of a network communication flow model learning module and a network communication flow simulation module. And the data flow processing system is responsible for independently collecting data flow passing through the physical node and processing and abstracting the data flow into flow model information. And in the simulation mode, the model issued by the control program is restored into simulation test data. The simulation node should support an embedding mode, a direct connection mode and a serial connection mode, wherein the embedding mode is installed in a device of the tested network in a software or hardware mode, the direct connection mode is connected with the tested network through a specific network device, and the serial connection mode is accessed between two adjacent devices of the tested network by using two ports.
1) Network communication flow model learning module
The network communication traffic model learning (hereinafter referred to as learning module) refers to storing data stream identification, traffic distribution and data of a large amount of real data streams when the real data streams pass through a network communication traffic simulation node. In the scheme, a splitter is adopted to map different data streams to independent Sketch, data frames are counted into discrete Sketch groups by different Sketch according to a splitting result, the interval of each group is controlled by an acquisition cycle trigger, the content of the data frames is processed by a clustering algorithm, and high-frequency short sentences are obtained and finally compressed and uploaded to a control program. The learning module includes, but is not limited to, the following basic components:
a) The stream splitter may be, but is not limited to, a Hash filter, a Bloom filter, a key value table, etc., and the identity of each individual data stream may be, but is not limited to, a common five tuple, protocol type, port number, etc.
B) The two-dimensional Sketch counter can be but is not limited to a two-dimensional counter structure such as CountMin Sketch, count Sketch and the like, and the two-dimensional Sketch counter uses the structures to Count and store the real change of the network flow;
c) The cluster classifier is used for clustering a large number of high-frequency short sentences according to different applications and the characteristic that the data of the service layer has a fixed mode in order to reduce the storage capacity burden of the simulation node due to the huge data volume in the network flow. The control program can configure clustering strategies such as length priority, service priority, user priority and the like;
d) The high-frequency short sentence tree is responsible for storing classified high-frequency short sentences, so that the efficiency of insertion, retrieval and deletion is improved;
e) The data compression module is responsible for compressing the network flow model before sending the network flow model, so that the network bandwidth occupation overhead is reduced;
f) The acquisition sampling timer controls the period length of Sketch statistics to play a role in controlling collision probability;
g) The acquisition period trigger controls the learning module to use the pause time counted by the Sketch to play a role similar to sampling, the counting can be continuously counted when the value of the counter is a special value, and the special value is defined in the design of a specific product.
2) Network communication flow simulation module
The network communication traffic simulation module (hereinafter referred to as simulation module) is used for restoring the network traffic data model issued by the control program into test data again, and supports shortening and prolonging of simulation data in a time domain. In the scheme, discrete Sketch statistical data are restored into a simulation flow control matrix by adopting a curve fitting and interpolation combined mode, and then a restored session is generated by combining a restoration strategy and a load template, and in order to further control the sending rate, a sliding window of the fitted Sketch matrix can be designed. The simulation module includes but is not limited to the following basic components:
a) The Sketch group comprises a data stream identification list and a data stream discrete statistic value, and is used for accurately indicating a simulated data stream set, giving a data sending rate and the like during simulation, and directly sending the discrete list is equivalent to compressing network flow;
b) The system comprises a load template, a high-frequency short sentence data processing module and a high-frequency short sentence data processing module, wherein the load template is an abstract application layer data template and can be inserted into the high-frequency short sentence data to form simulation data, and the load template comprises but is not limited to various load templates such as Web (Http, website, html, xml and the like), mail (POP 3 and the like), user accounts and the like;
c) The method comprises the steps of restoring a strategy, wherein the restoring strategy defines information such as a time domain stretch coefficient, a fitting function, an interpolation algorithm, a fitting insertion mode, a load parameter and the like;
d) And the high-frequency short sentence data is the data obtained by the learning module.
1.3 network traffic model storage
The network communication traffic model storage (hereinafter referred to as model storage) is composed of a Sketch database, a load database and a recovery strategy database, and is used for storing, maintaining and updating network communication traffic model data required by a control program, and the specific structure of the network communication traffic model storage is as shown in fig. 1.
As shown in the figure.
1) Sketch database
The Sketch data frame contains a data stream identifier and a two-dimensional counter value, and the data stream identifier is mapped to a specific count value by a hash function independent of each Sketch. The specific statistical and reading methods are mature in typical Sketch algorithms, and the scheme is not described again, and the supported sketches include but are not limited to Count Min, count, openSketch and the like. Unlike normal Sketch, to simulate network traffic, a timestamp for Sketch ordering and fitting calculations is added to each Sketch.
2) Load database
The load database stores communication templates and high-frequency data, specifically including but not limited to connection establishment, accounts, connection interruption, application data, web data and the like, which are often used in simulation. The combination of specific templates and phrases needs to be realized in cooperation with a recovery strategy.
A) The templates define a number of steps in the simulation process, and a number of templates constitute a communication process.
B) The data is the content of specific services, and the scheme supports simulation application data and Web data.
3) Restoration policy database
The restoration strategy database is used for storing restoration strategies for network communication traffic simulation, and specifically includes but is not limited to time domain stretch coefficients, fitting functions, fitting insertion modes and load parameters.
A) Coefficient of time domain stretch
The time domain stretch coefficient is the ratio of the total sending time of the actual simulation data stream to the sending time of the Sketch group, and reflects the degree of compression and extension of the network traffic model.
B) Fitting function
The fitting function obtains a flow distribution change curve through the discrete Sketch group, and then obtains a continuous flow sending matrix through interpolation;
c) Fitting interpolation patterns
The fitting interpolation mode can select different modes such as an interpolation algorithm, interpolation frequency, uniform interpolation, non-uniform interpolation and the like;
d) Load parameter
The load parameters define the usage rules of the load data, including the native load, the splice load, the variant load, and the like.
Network communication flow model learning method
The processing flow of the learning module for one data is as shown in fig. 5, and after the control program starts the network communication traffic learning task, the acquisition time and period, the kind and number of Sketch, and the like of the learning module need to be configured. After configuration is completed, the data stream is collected and then is shunted by the classifier according to the identifier, a Bloom filter and a Hash function can be used, and the obtained result points to a unique Sketch storage structure. Sketch typically performs statistics by mapping the count values to counters through a set of hash functions. On the other hand, in order to learn that the data is really data, the data except the mark is subjected to clustering analysis to obtain high-frequency short sentences with a first length, and the high-frequency short sentences are stored to obtain useful high-level data as much as possible under the condition that the storage space is saved. To restore the data stream for subsequent simulations, a timestamp is therefore added to Sketch. And finally, in order to prevent the distributed simulation nodes from bringing huge communication overhead to the network and a server where the control program is located, the learned network communication model is processed by using a compression algorithm and uploaded.
Network communication flow simulation method
The test data flow generation process of the simulation module is shown as a full network data flow simulation architecture diagram in fig. 1.
Fig. 2 is a block diagram of a full network data flow simulation.
Fig. 3 is a block diagram of a data structure of a network traffic model.
Fig. 4 is a schematic diagram of a network traffic simulation configuration file structure.
FIG. 5 is a learning module workflow diagram.
As shown in the figure. The figure only describes the test data generation process of one simulation node, each rectangular box in the figure is used or generated data in the simulation process, and a specific processing method is arranged on an arrow. The discrete Sketch groups carry out fitting interpolation calculation according to the selection of a curve fitting algorithm and parameters in the recovery strategy to obtain a large amount of statistical value interpolation meeting the requirement of the playback time, the statistical value interpolation is combined with the original Sketch groups to form continuous statistical values of the data stream, and the query algorithm defined by the Sketch can be used for acquiring a specific guideline rate by using the data stream identification. On the other hand, the recovery strategy can specify a filling mode, one sentence of the load template and the high-frequency short sentence restore application layer conversation, and a plurality of conversations are combined to form simulation data. And finally, the data sending module sends the simulation data under the control of the simulation rate matrix to form the generation of the test data on the simulation node.
The key points and the protection points of the invention are as follows:
(1) The invention utilizes the advantages of accurate two-dimensional Sketch statistics and small occupied storage space to distribute network communication simulation nodes in the computer network communication equipment, and each node learns the network communication model and stores the network communication model as continuous Sketch. And finally, uploading the full network flow model to a database where the network communication model is stored. The distributed architecture for learning and simulating network traffic through Sketch is a protection point of the patent;
(2) The invention has the learning function and the simulation playback function of the network communication flow model at the same time, and the self-learning network data simulation generation mode is a protection point of the patent;
(3) The method comprises the steps of respectively storing various data streams by using a plurality of discrete sketches with time sequence, extracting application layer loads into high-frequency short sentences through a clustering algorithm, and abstractively storing network flow into stream identifiers, the sketches and application data high-frequency short sentences. The control program can define configuration parameters including but not limited to acquisition duration, acquisition period, distribution rules, clustering parameters and the like, and finally, the learned real network traffic model is uploaded to the control program through data compression and stored in a database by the control program. The abstract method of the network communication data flow model is the key point and the protection point of the invention;
(4) Calculating the flow of each data stream according to the Sketch, restoring a discrete group of Sketch into a continuous network simulation data stream through curve fitting function and interpolation calculation, and realizing the compression or extension of the simulation flow by setting the number of calculation values inserted between two continuous Sketchs by a control program. And in the aspect of sending content, according to the data stream identification, the load template and the recovery strategy, the simulation communication data is formed after the conversation recovery. The simulation method of the communication flow is a key point and a protection point of the invention;
(5) Managing data stream tags through a data stream identification tree, wherein each tag points to a group of Sketch data, and each Sketch data frame contains a time stamp; meanwhile, templates, high-frequency short sentences and recovery strategies used by the simulation data streams are stored;
(6) And restoring a plurality of discrete network communication traffic matrixes into a simulation traffic model by using a curve fitting and interpolation method, and controlling a packet sending module to send test data according to the model.
It is to be understood that the above description is not intended to limit the present invention, and the present invention is not limited to the above examples, and those skilled in the art may make modifications, alterations, additions or substitutions within the spirit and scope of the present invention.
Claims (5)
1. Network multidimension degree dataflow simulation device based on compound two-dimentional Sketch, its characterized in that: the system comprises a full network data flow simulation control program module, a network communication model storage module and a network communication flow simulation node; the three components adopt a decoupling modular design;
the network communication traffic simulation control program module is configured to control a network communication traffic model learning module and a network communication traffic simulation module of the network communication traffic simulation node;
the network communication model storage module comprises a Sketch database, a load database and a recovery strategy database; configured to store, maintain and update network communication traffic model data required by the network-wide data flow simulation control program module;
the network communication flow simulation node comprises a network communication flow model learning module and a network communication flow simulation module; the system is configured to be responsible for independently collecting data streams passing through the physical nodes and processing and abstracting the data streams into flow model information; the network communication traffic simulation node supports an embedding mode, a direct connection mode and a serial connection mode, wherein the embedding mode is installed in equipment of a tested network in a software or hardware mode, the direct connection mode is connected with the tested network through specific network equipment, and the serial connection mode is accessed between two adjacent equipment of the tested network by using two ports;
the full network data flow simulation control program module has a learning module management function and a simulation module management function;
the flow and the model of the full network data flow can not be comprehensively collected or simulated from one physical node, and the functions are realized by adopting a distributed mode of a full network data flow simulation control program module and a plurality of network communication flow simulation nodes;
the learning module management function comprises a first control function and a first return function;
the first control function is specifically as follows:
the full network data flow simulation control program module needs to control a network communication flow model learning function of a network communication flow simulation node, and specifically comprises the on/off of the function, flow learning configuration information, an acquisition sampling timer and an acquisition period trigger; wherein, the flow learning configuration information comprises the following contents:
a) The shunt parameters give different matching domains, matching values and corresponding Sketch groups;
b) The Sketch group configures parameters, and gives the number of hash functions, the width of a counter and a data stream identification storage structure of each Sketch;
c) Selecting clustering parameters according to the type of a clustering algorithm by using basic configuration parameters of a clustering classifier, and extracting high-frequency short sentences with preferential length from data of the application layer;
d) Basic configuration parameters of the high-frequency short sentence tree comprise the depth of the tree, a leaf node partition function, the minimum sample number of the leaf nodes and the maximum sample number parameter of the leaf nodes;
e) Basic configuration parameters of a data compression function comprise a compression method and a storage format of model data;
f) Collecting the time length of a sampling timer;
g) Collecting the time length of a periodic trigger;
the first feedback function is specifically as follows:
after learning of the network communication flow model is completed, the network communication flow simulation node compresses relevant information and uploads the information to the whole network data flow simulation control program module, and a data frame before compression comprises network simulation node basic information, network flow model basic information, a data flow identification set, sketch group data, time scale data and high-frequency short sentence data; the network traffic model data structure includes:
a) Network simulation node basic information;
b) Basic information of a network flow model;
c) A set of data stream identifications;
d) Sketch group data and time mark data;
e) High-frequency short sentence data;
the simulation module management function comprises a second control function and a second return function;
the second control function is specifically as follows:
the full network data flow simulation control program module needs to control the network communication flow simulation function of the network communication flow simulation node, and specifically comprises the steps of transmitting a restored network communication flow model to network communication flow, wherein the restored network communication flow model comprises a data flow discrete statistic value, a data flow identification list, a load template, a recovery strategy and a continuous packet-sending matrix sliding window, and the basic structure of a network flow simulation configuration file comprises the following steps:
a) Data flow dispersion statistic, namely real statistic data of network flow stored by Sketch;
b) A data stream identifier list, a set of data stream identifiers possessed by each Sketch;
c) A load template, a data template used, a cluster abstraction from real data;
d) A restoration strategy, wherein the restoration strategy is a method for restoring network data flow based on a data flow dispersion statistic and a load template;
e) A continuous packet-sending matrix sliding window gives the number of the Sketch actually processed, and only the Sketch in the sliding window is simulated;
f) High-frequency short sentence data is used for filling the load part and increasing the simulation truth degree;
the second backhaul function is specifically as follows:
in the process of generating the simulation flow, the simulation node transmits some basic operation state information back to the whole network data flow simulation control program module, wherein the operation state information comprises the sent data volume and the sliding window position information.
2. The composite two-dimensional Sketch-based network multi-dimensional data flow simulation device according to claim 1, wherein:
the network communication flow model learning module is used for storing data flow identification, flow distribution conditions and data of a large number of real data flows when the real data flows through the network communication flow simulation node; different data streams are mapped to independent sketches by adopting a splitter, data frames are counted into discrete Sketch groups by the different sketches according to a splitting result, the interval of each group is controlled by an acquisition period trigger, the content of the data frames is processed by a clustering algorithm, and high-frequency short sentences are obtained and finally are compressed and uploaded to a full-network data stream simulation control program module; the network communication flow model learning module comprises the following basic components:
a) A flow divider; the flow divider comprises a Hash filter, a Bloom filter and a key value table, and the identifier of each independent data stream comprises a quintuple, a protocol type and a port number;
b) A two-dimensional Sketch counter; a two-dimensional counter structure including CountMin Sketch and Count Sketch;
c) A cluster classifier; because the data volume in the network flow is huge, in order to reduce the storage capacity burden of the simulation node, a large number of high-frequency short sentences are clustered according to different applications according to the characteristic that the data of the service layer has a fixed mode; the whole network data flow simulation control program module can configure clustering strategies comprising length priority, service priority and user priority;
d) A high-frequency short sentence tree; the high-frequency short sentence tree is responsible for storing the classified high-frequency short sentences;
e) A data compression module; the data compression module is responsible for compressing the network flow model before sending the network flow model, so that the network bandwidth occupation overhead is reduced;
f) Collecting a sampling timer; the acquisition sampling timer controls the period length of Sketch statistics to play a role in controlling collision probability;
g) Collecting a periodic trigger; the acquisition period trigger controls the network communication flow model learning module to use the pause time counted by Sketch, so that the effect similar to sampling is achieved;
the network communication flow simulation module is used for restoring a network flow data model issued by the whole network data flow simulation control program module into test data again and supporting the shortening and the lengthening of simulation data in a time domain; restoring the discrete Sketch statistical data into a simulation flow control matrix by adopting a curve fitting and interpolation combined mode, and generating a restored conversation by combining a restoration strategy and a load template; the network communication flow simulation module comprises the following basic components:
a) The Scut group comprises a data stream identification list and a data stream discrete statistic value, and is used for accurately indicating a simulated data stream set and giving a data sending rate during simulation, and the direct sending of the discrete list is equivalent to the compression of network flow;
b) The load template is an abstract application layer data template, and high-frequency short sentence data is inserted into the load template to form simulation data;
c) The recovery strategy defines information including a time domain stretch coefficient, a fitting function, an interpolation algorithm, a fitting insertion mode and a load parameter;
d) And the high-frequency short sentence data is the data obtained by the network communication flow model learning module.
3. The composite two-dimensional Sketch-based network multi-dimensional data flow simulation device according to claim 1, wherein: the network communication traffic model storage comprises the following basic components:
1) Sketch database
The Sketch data frame comprises a data stream identification and a two-dimensional counter value, and the data stream identification is mapped into a specific count value through each Sketch independent hash function;
2) Load database
The load database stores communication templates and high-frequency data, including connection establishment, accounts, connection interruption, application data and Web data which are commonly used in simulation;
3) Restoration policy database
The restoration strategy database is used for storing restoration strategies for network communication flow simulation, and comprises a time domain stretch coefficient, a fitting function, a fitting insertion mode and a load parameter;
a) Time domain stretch coefficient
The time domain stretch coefficient is the ratio of the total sending time of the actual simulation data stream to the sending time of the Sketch group, and reflects the degree of compression and extension of the network flow model;
b) Fitting function
The fitting function obtains a flow distribution change curve through the discrete Sketch group, and then obtains a continuous flow sending matrix through interpolation;
c) Fitting interpolation patterns
The fitting interpolation mode can select different modes including an interpolation algorithm, interpolation frequency, uniform interpolation and non-uniform interpolation;
d) Load parameter
The load parameters define usage rules for the load data, including native load, splice load, and variant load.
4. The composite two-dimensional Sketch-based network multi-dimensional data flow simulation device according to claim 1, wherein: the working process of the network communication flow model learning module in the network communication flow simulation node is as follows:
step 1: after a network communication flow learning task is started by a full network data flow simulation control program module, the acquisition time and period, and the types and the number of Sketch of a network communication flow model learning module are configured;
and 2, step: after the configuration is finished, acquiring a data stream to obtain a data frame, processing the data frame by using a classifier, and judging whether corresponding Sketch exists or not;
if: judging that the corresponding Sketch exists; executing the step 3;
or judging that no corresponding Sketch exists; executing the step 4;
and step 3: using special Sketch statistics;
and 4, step 4: using default Sketch statistics;
and 5: extracting high-frequency words by using a cluster classifier;
step 6: adding a timestamp label;
and 7: when the sampling timer is overtime, compressing a network communication flow model;
and 8: and uploading the data.
5. The composite two-dimensional Sketch-based network multi-dimensional data flow simulation device according to claim 1, wherein: the work flow of a network communication flow model simulation module in the network communication flow simulation node is as follows:
step S1: configuring a data flow identification list, a data flow dispersion statistic value, a load template, a recovery strategy and high-frequency short sentence data of a network communication flow model simulation module through a full network data flow simulation control program module;
step S2: performing fitting interpolation calculation on discrete Sketch groups in the data stream discrete statistic according to curve fitting parameters in a recovery strategy to obtain statistic interpolation meeting the requirement of playback time;
and step S3: combining the statistic interpolation with the original Sketch to form a data stream continuous statistic;
and step S4: acquiring a simulation rate matrix by using a data stream identification list through a query algorithm defined by Sketch;
step S5: the recovery strategy can specify a filling mode, application layer conversations are filled according to the load template and the high-frequency short sentence data, and a plurality of application layer conversations are combined to form simulation data;
step S6: and the network communication flow simulation module sends simulation data under the control of the simulation rate matrix to form a test data stream.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011012284.8A CN112134738B (en) | 2020-09-24 | 2020-09-24 | Network multidimensional data flow simulation device based on composite two-dimensional Sketch |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011012284.8A CN112134738B (en) | 2020-09-24 | 2020-09-24 | Network multidimensional data flow simulation device based on composite two-dimensional Sketch |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112134738A CN112134738A (en) | 2020-12-25 |
| CN112134738B true CN112134738B (en) | 2023-03-24 |
Family
ID=73839913
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011012284.8A Active CN112134738B (en) | 2020-09-24 | 2020-09-24 | Network multidimensional data flow simulation device based on composite two-dimensional Sketch |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112134738B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113377785B (en) * | 2021-08-16 | 2021-11-02 | 宜科(天津)电子有限公司 | Industrial data processing system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103647670A (en) * | 2013-12-20 | 2014-03-19 | 北京理工大学 | Sketch based data center network flow analysis method |
| CN108923975A (en) * | 2018-07-05 | 2018-11-30 | 中山大学 | A kind of traffic behavior analysis method of Based on Distributed network |
| CN110635943A (en) * | 2019-09-02 | 2019-12-31 | 北京航空航天大学 | Network traffic simulation system of network transmission process based on Spark computing framework |
| CN111159063A (en) * | 2019-12-25 | 2020-05-15 | 大连理工大学 | A Cache Allocation Method for Multilayer Sketch Network Measurement |
| CN111314166A (en) * | 2019-12-30 | 2020-06-19 | 中国人民解放军军事科学院国防科技创新研究院 | Multi-node comprehensive simulation system and method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7751325B2 (en) * | 2003-08-14 | 2010-07-06 | At&T Intellectual Property Ii, L.P. | Method and apparatus for sketch-based detection of changes in network traffic |
-
2020
- 2020-09-24 CN CN202011012284.8A patent/CN112134738B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103647670A (en) * | 2013-12-20 | 2014-03-19 | 北京理工大学 | Sketch based data center network flow analysis method |
| CN108923975A (en) * | 2018-07-05 | 2018-11-30 | 中山大学 | A kind of traffic behavior analysis method of Based on Distributed network |
| CN110635943A (en) * | 2019-09-02 | 2019-12-31 | 北京航空航天大学 | Network traffic simulation system of network transmission process based on Spark computing framework |
| CN111159063A (en) * | 2019-12-25 | 2020-05-15 | 大连理工大学 | A Cache Allocation Method for Multilayer Sketch Network Measurement |
| CN111314166A (en) * | 2019-12-30 | 2020-06-19 | 中国人民解放军军事科学院国防科技创新研究院 | Multi-node comprehensive simulation system and method |
Non-Patent Citations (1)
| Title |
|---|
| 网络流量测量中的Sketch方法;任高明;《电子世界》;20200630(第12期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112134738A (en) | 2020-12-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8179799B2 (en) | Method for partitioning network flows based on their time information | |
| CN118017564B (en) | Energy storage method based on open source hong Meng system | |
| CN111782620B (en) | Credit link automatic tracking platform and method thereof | |
| CN105284052B (en) | Systems and methods for dictionary-based compression | |
| CN107896160B (en) | A kind of data center network flowmeter factor method based on distributed system | |
| CN103209087B (en) | Distributed information log statistical processing methods and system | |
| CN106027406B (en) | Traffic import method of NS3 simulation system based on Netflow | |
| CN112134846B (en) | Method, system, device and medium for analyzing signaling data of communication network | |
| CN107622064A (en) | A data reading method and system | |
| CN104572885A (en) | Method and system for collecting and analyzing preference data of user function | |
| CN106951476A (en) | Race informix processing system based on the network communications technology | |
| CN106326280B (en) | Data processing method, device and system | |
| CN118317008B (en) | Multi-protocol multi-interface data distribution system and method based on industrial edge gateway | |
| CN108023895B (en) | Mass data directional classification transmission method and system | |
| CN112199154A (en) | Distributed collaborative sampling central optimization-based reinforcement learning training system and method | |
| CN112134738B (en) | Network multidimensional data flow simulation device based on composite two-dimensional Sketch | |
| CN117119535A (en) | Data distribution method and system for mobile terminal cluster hot spot sharing | |
| CN119996286A (en) | Dynamic optimization network routing method, system, electronic device and storage medium | |
| WO2018107373A1 (en) | Multi-level electronic map implementation method, and apparatus therefor | |
| De Aquino et al. | Data stream based algorithms for wireless sensor network applications | |
| CN103957119A (en) | Method for managing network equipment by using MIB file and browser | |
| Jang et al. | Using cnn to optimize traffic classification for smart homes in 5g era | |
| CN107832459B (en) | System and method for knowledge base content sharing learning based on distributed network environment | |
| Gentili et al. | Data summarization model for user action log files | |
| CN109802990A (en) | A kind of resource log reading/writing method and device reducing data redundancy |
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 | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 266555 No. 98 Xiangjiang Road, Huangdao District, Qingdao City, Shandong Province Applicant after: CLP kesiyi Technology Co.,Ltd. Address before: 266555 No. 98 Xiangjiang Road, Huangdao District, Qingdao City, Shandong Province Applicant before: CHINA ELECTRONICS TECHNOLOGY INSTRUMENTS Co.,Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |