KR102537023B1 - Method for controlling network traffic based traffic analysis using AI(artificial intelligence) and apparatus for performing the method - Google Patents

Abstract

The present invention relates to a network traffic control method based on artificial intelligence-based traffic analysis and a device for performing the method. The network traffic control method based on artificial intelligence-based traffic analysis may comprise the steps of: by a network traffic control device, receiving traffic information from a network device; by the network traffic control device, determining packet transmission rules through an artificial intelligence engine learned based on the traffic information; and, by the network traffic control device, transmitting the packet transmission rules to the traffic control device.

Description

Method for controlling network traffic based traffic analysis using AI(artificial intelligence) and apparatus for performing the method}

The present invention relates to a method for controlling network traffic using artificial intelligence-based traffic analysis and an apparatus for performing the method. More specifically, it relates to a network traffic control method based on artificial intelligence-based traffic analysis for creating an optimal network environment by controlling traffic on a network based on artificial intelligence, and a device for performing the method.

SDN (Software Defined Network) is a next-generation networking technology that can conveniently handle network path setup and control and complex operation management through software programming.

OpenFlow is one of the technologies for realizing SDN. Research and development of an AI network integration system using OpenFlow technology is in progress, which enables monitoring and prompt action on failure information by remotely controlling without human intervention using network integration system technology, reducing communication network advancement costs and strengthening security. there is.

CISCO, which occupies more than 70% of the global network market, predicts that 5.3 billion people around the world will use the Internet by 2023. In the event of a failure of the network integration system, enormous damage occurs, and if immediate response to the failure is not possible, the scale of damage increases exponentially.

According to Internet data trends announced by CISCO, it is predicted that by 2023, 66% of the world's population will use the Internet. There is a strong demand for a change to a flexible and open network structure that can quickly introduce various business requirements beyond the transmission of large amounts of information.

Network The operational efficiency of the communication network must be strengthened to secure the basis for stable network service.

The object of the present invention is to solve all of the above problems.

In addition, an object of the present invention is to intelligently monitor and control traffic of a network communication network by applying an artificial intelligence engine/algorithm.

Another object of the present invention is to control traffic of a network communication network by applying a structured traffic transmission rule and an unstructured traffic transmission rule based on a clustering algorithm and a learning algorithm.

Representative configurations of the present invention for achieving the above object are as follows.

According to an embodiment of the present invention, a network traffic control method based on artificial intelligence-based traffic analysis comprises the steps of a network traffic control device receiving traffic information from a network device, the network traffic control device based on the traffic information The method may include determining a packet transmission rule through an artificial intelligence engine learned from and transmitting the packet transmission rule to the traffic control device by the network traffic control device.

Meanwhile, the artificial intelligence engine may operate based on a clustering algorithm and a learning algorithm.

In addition, the clustering algorithm may be an algorithm for learning and processing a standardized packet transmission rule, and the learning algorithm may be an algorithm for determining and learning an unstructured packet transmission rule variable.

According to another embodiment of the present invention, a network traffic control device for performing network traffic control based on artificial intelligence-based traffic analysis receives traffic information from a network device, and an artificial intelligence engine learned based on the traffic information. A packet transmission rule may be determined through and the packet transmission rule may be delivered to the traffic control device.

Meanwhile, the artificial intelligence engine may operate based on a clustering algorithm and a learning algorithm.

In addition, the clustering algorithm may be an algorithm for learning and processing a standardized packet transmission rule, and the learning algorithm may be an algorithm for determining and learning an unstructured packet transmission rule variable.

According to the present invention, traffic of a network communication network can be intelligently monitored and controlled based on an artificial intelligence engine/algorithm.

In addition, according to the present invention, traffic of a network communication network can be controlled by applying a standardized traffic transmission rule and an unstructured traffic transmission rule based on a clustering algorithm and a learning algorithm.

1 is a conceptual diagram illustrating a network control system according to an embodiment of the present invention.
2 is a conceptual diagram showing the operation of an artificial intelligence engine according to an embodiment of the present invention.
Figure 3 discloses the operation of the artificial intelligence engine according to an embodiment of the present invention.
4 is a conceptual diagram showing the operation of an artificial intelligence engine according to an embodiment of the present invention.
5 is a conceptual diagram showing the operation of an artificial intelligence engine according to an embodiment of the present invention.
6 is a conceptual diagram illustrating a method for determining an artificial intelligence-based packet transmission rule according to an embodiment of the present invention.
7 is a conceptual diagram illustrating a method of determining an artificial intelligence-based packet transmission rule according to an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The detailed description of the present invention which follows refers to the accompanying drawings which illustrate, by way of illustration, specific embodiments in which the present invention may be practiced. These embodiments are described in sufficient detail to enable any person skilled in the art to practice the present invention. It should be understood that the various embodiments of the present invention are different from each other but are not necessarily mutually exclusive. For example, specific shapes, structures, and characteristics described herein may be implemented from one embodiment to another without departing from the spirit and scope of the present invention. It should also be understood that the location or arrangement of individual components within each embodiment may be changed without departing from the spirit and scope of the present invention. Therefore, the detailed description to be described later is not performed in a limiting sense, and the scope of the present invention should be taken as encompassing the scope claimed by the claims and all scopes equivalent thereto. Like reference numbers in the drawings indicate the same or similar elements throughout the various aspects.

Hereinafter, various preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings in order to enable those skilled in the art to easily practice the present invention.

1 is a conceptual diagram illustrating a network control system according to an embodiment of the present invention.

1 discloses a method for controlling network traffic by learning data collected based on network monitoring.

Referring to FIG. 1 , the network control system may include a network device 100 and a network traffic control device 150 .

The network device 100 may be a device for forwarding packets on a network such as a network switch or an access point (AP). The network device 100 is included in the infrastructure layer and may form a path for forwarding packets generated on the network.

The network device 100 may transmit an initial flow table to the network traffic control device 150 . The flow table may include information about rules related to packet transmission. When a packet arrives, the network device 100 examines the flow table, and if there is a table matching a packet matched to the flow table, it can forward the packet. Conversely, if there is no table matching the flow table, the network device may request the flow table from the network traffic control device 150 . The network traffic control device 150 may transmit an optimized flow table determined based on artificial intelligence to the network device 100 to process packets. In this way, the network device 100 may process a packet based on an optimal flow table generated based on a packet transmission rule optimized based on an artificial intelligence engine.

The network traffic control device 150 may receive and monitor traffic information about traffic of the network device 100 and control a traffic path based on the traffic information of the network device 100 . Traffic information may include information related to packet forwarding on the network device 100 .

The network traffic control device 150 may learn traffic information based on the artificial intelligence engine 140 and set an optimal packet transmission rule for delivering packets based on the artificial intelligence engine 140. An optimal packet transmission rule may be generated as an optimal flow table and transmitted to the network device 100 .

More specifically, the network traffic control device 150 may include a network controller 110, a network manager 120, a traffic information manager 130, and an artificial intelligence engine 140.

The network controller 110 may be implemented to provide a flow table including optimized packet path rules based on network traffic information. The network controller 110 can control traffic on the network based on the flow table so that the network device can efficiently deliver packets on the network.

The network management unit 120 may be implemented for management of a wired network and a wireless network. The network management unit 120 may manage packets to be efficiently transmitted over the wired/wireless network through dynamic management of the wired network and the wireless network. A wired network is a network based on switches, and a wireless network is a network based on APs.

The traffic information management unit 130 may be implemented to manage traffic information such as flow information, traffic path information, traffic monitoring information, and traffic statistical information.

More specifically, the traffic information includes aggregation switch traffic information, edge switch traffic information, terminal traffic information, excessive traffic inflow (hacking), and hang-up collected in real time, minute, hour, day, week, and month. (when there is no inflow of traffic), information on power shutdown (shut down), etc. may be included.

Traffic information based on packets generated on a network switch can be collected and managed as traffic information.

The artificial intelligence engine 140 may be implemented to generate an optimal flow table for packet processing of the network device 100 by pre-processing traffic information. The artificial intelligence engine 140 may determine a packet processing rule based on traffic information generated on the network, and the packet processing rule may be transmitted to the network control unit 110 and applied to an optimal flow table.

A network application programming interface (API) may be implemented as an upper layer of the network traffic control device 150 .

2 is a conceptual diagram showing the operation of an artificial intelligence engine according to an embodiment of the present invention.

2 discloses a method for monitoring and controlling traffic on a network based on an artificial intelligence engine.

Referring to FIG. 2 , the artificial intelligence engine may include a data collection unit 210, a data processing unit 220, an execution unit 230, and an application unit 240.

The data collection unit 210 may be implemented to collect traffic information. Traffic information can be delivered by network devices and collected through big data databases and network controllers.

The data processing unit 220 may be implemented to preprocess traffic information and learn the preprocessed traffic information. Traffic information can be preprocessed through conversion, normalization, and encoding. According to an embodiment of the present invention, the learning method of the data processing unit 220 may include a clustering algorithm and a learning algorithm. The artificial intelligence engine can determine packet processing rules through learning results, and can insert, add, or delete flow entries in the flow table to apply packet processing rules for an optimal traffic environment.

The execution unit 230 may be implemented to simulate (or test, test) and tune the determined packet processing rule.

The application unit 240 may be implemented to apply the simulated and tuned packet processing rules to the network device.

For convenience of explanation, it is assumed that the learning algorithm and clustering algorithm determine packet processing rules, but the learning algorithm and clustering algorithm determine traffic bandwidth auto scaling and packet transmission rules based on traffic analysis and failure type analysis. Transmission Rule Decision), measures according to the type of failure may be determined, and such an embodiment may also be included in the scope of the present invention.

Figure 3 discloses the operation of the artificial intelligence engine according to an embodiment of the present invention.

3 discloses a learning method of an artificial intelligence engine based on traffic information.

Referring to FIG. 3 , an artificial intelligence engine may be learned based on a clustering algorithm and a learning algorithm.

First, traffic information may be received from a network device (switch and access point (AP)) for learning.

Traffic information may be sorted according to time, source/destination, and transmitted to the processing module.

Traffic information sorted by time and source/destination is transmitted to a processing module, and the sorted traffic information may be pre-processed through conversion, normalization, and/or encoding.

A clustering algorithm 310 and a learning algorithm 320 may be applied to the preprocessed traffic information.

(1) Clustering Algorithm (310)

Learning models defined through ‘learning’ (network integration system – network control unit – AI engine 1-cycle circulation) are accumulated in the database, and learning models can be stored in the learning model storage. The standardized packet transmission rule may be modeled into a learning model and processed in the clustering algorithm 310 .

(2) Learning Algorithm (320)

Traffic information collected in real-time and used as training data is stored in the learning model cache storage for fast processing network application, and training data collected in batches is stored through the network manager’s intervention. Policies can be applied and processed. Irregular traffic transmission rules (failure, traffic congestion, bottleneck, etc.) can be determined as variables and processed in the learning algorithm 320.

That is, according to an embodiment of the present invention, structured packet transmission rules (paths, etc.) are modeled into learning models and processed in the clustering algorithm 310, and unstructured packet transmission rules (failures, traffic congestion, bottlenecks, etc.) may be determined as a variable and processed in the learning algorithm 320.

For example, the clustering algorithm 310 may determine a packet transmission rule based on clustering for non-failure traffic information, and the learning algorithm 320 may determine a packet transmission rule based on clustering for failure traffic information. there is.

The clustering algorithm 310 and the learning algorithm 320 may establish packet transmission rules in synchronization with each other.

A simulation (or test) of a packet transmission rule to which the clustering algorithm 310 and the learning algorithm 320 are applied may be performed. Packet transmission rules can be tuned based on simulation results. A packet transmission rule tuned through simulation may be applied to a network device, and an application result of the packet transmission rule may be monitored by a monitoring system.

That is, according to an embodiment of the present invention, the clustering algorithm 310 and the learning algorithm 320 complement each other to derive an optimal analysis result, and the determined policy may go through an application monitoring process. An optimized path decision algorithm based on the data extracted through the pre-processing process (collecting, sorting, and processing wired and wireless data) is performed, and a merger artificial intelligence processing algorithm combining network line conditions and traffic bottleneck information is performed.

In addition, a transmission ranking algorithm that transmits traffic based on priority by setting importance for each user and application is performed, and an algorithm that can dynamically modify predefined constraints through traffic policies generated by artificial intelligence is performed. can

4 is a conceptual diagram showing the operation of an artificial intelligence engine according to an embodiment of the present invention.

4 discloses a processing algorithm for traffic information of an artificial intelligence engine.

Referring to FIG. 4 , wire traffic information may be collected (step S400).

The wired traffic information may include traffic information generated by a switch, information on path network congestion, and the like. Wired traffic information may be collected through a network control unit.

Collection of radio traffic information may be performed (step S410).

The wireless traffic information may include traffic information generated by an AP, information on path network congestion, and the like. Wireless traffic information may be collected through a network control unit.

Sorting of traffic information may be performed (step S420).

Sorting of traffic information can be based on time and source/destination. More specifically, recognition and sorting based on switch time frame, recognition and sorting based on switch source IP/destination IP, merge sorting algorithm, quick sorting algorithm, etc. may be used for sorting traffic information.

Pre-processing of traffic information may be performed (step S430).

Conversion, normalization, and encoding algorithms may be applied to the sorted traffic information.

An artificial intelligence engine for packet processing may be generated through learning about traffic information (step S440).

As described above, learning of traffic information can be performed based on a clustering algorithm and a learning algorithm. According to an embodiment of the present invention, a standardized packet transmission rule (path, etc.) is modeled into a learning model and processed in a clustering algorithm, and Customized packet transmission rules (failure, traffic congestion, bottleneck, etc.) can be determined as variables and processed in the learning algorithm.

A simulation of packet transmission rules may be performed (step S450).

The packet transmission rule determined by the artificial intelligence engine is transmitted to the network control unit and simulation can be performed. Application of the packet transmission rule transmitted to the network control unit to a network device (switch, AP) may be monitored and transmitted. The simulated packet transmission rules can be tuned through monitoring and applied to network devices.

A final packet transmission rule may be applied (step S460).

The final packet transmission rule determined through simulation completion and tuning may be applied to the network device.

Data collected by the network control unit is transmitted to the artificial intelligence engine, and a connection route API (Application Programming Interface) may be used to transmit the packet transmission rule determined by the artificial intelligence engine to the network control unit.

5 is a conceptual diagram showing the operation of an artificial intelligence engine according to an embodiment of the present invention.

In FIG. 5, a clustering algorithm and a learning algorithm performed by an artificial intelligence engine are disclosed.

Referring to FIG. 5 , traffic collection and failure type collection may be performed for a clustering algorithm and a learning algorithm.

Traffic is collected in real time, minute, hour, day, week, and month, but aggregate switch traffic information, edge switch traffic information, and terminal traffic information can be collected.

Failure type collection can collect information on excessive traffic inflow (hacking), hang-up (when there is no traffic inflow), and power shutdown (shut down).

After traffic collection and failure type collection, traffic analysis and failure type analysis may be performed.

Traffic analysis may include aggregation switch traffic analysis, edge switch traffic analysis, and terminal traffic analysis.

Failure type analysis may include excessive traffic flow analysis, hang-up analysis, and power shutdown analysis.

The learning algorithm and clustering algorithm can automatically adjust traffic bandwidth, determine packet transmission rules, and determine actions according to failure types based on traffic analysis and failure type analysis. Traffic bandwidth auto-adjustment is QoS setting auto-adjustment (bandwidth), and traffic bandwidth can be adjusted, for example, from 100M to 200M or from 200M to 100M for each port. In addition, measures according to the type of failure may include automatic blocking of ports suspected of hacking, automatic reset of ports suspected of hanging up, rebooting of shut-down devices, and the like.

After the above measures, monitoring can be performed. For example, aggregation switch monitoring, edge switch monitoring, and terminal traffic monitoring may be performed after traffic bandwidth is automatically adjusted. In addition, after collecting failure types, excessive traffic inflow monitoring, monitoring after automatic reset of a port suspected of hangup, and monitoring after automatic power rebooting may be performed.

6 is a conceptual diagram illustrating a method for determining an artificial intelligence-based packet transmission rule according to an embodiment of the present invention.

6 discloses a method for determining a packet transmission rule based on a clustering algorithm.

Referring to FIG. 6, in order to classify different learning models based on a clustering algorithm, learning models can be created by considering different packet characteristics occurring on different time units such as minutes, hours, days, weeks, and months. there is.

Through such learning in different time units, network conditions can be predicted in different time units on the time axis, and a more accurate packet transmission rule can be determined based on the prediction of network conditions in different time units.

In addition, according to an embodiment of the present invention, a network condition is predicted in consideration of network association between network devices such as a network switch that forwards packets, and based on this, a packet transmission rule for each of a plurality of network devices is determined on the entire network can be determined adaptively.

Hereinafter, for convenience of description, different learning models are generated through learning of traffic data for each first time unit, second time unit, and third time unit in order to determine and predict packet transmission conditions on the network in different time units. A method is disclosed.

The traffic transmission data (or traffic information) of the first time unit may include information on traffic conditions generated for each first time unit, and the first artificial intelligence engine 610 may transmit traffic data of the first time unit. Learning about the first time unit traffic transmission situation may be performed through learning about .

The traffic transmission data of the second time unit may include information on traffic conditions generated for each second time unit, and the second artificial intelligence engine 620 learns about the traffic transmission data of the second time unit. It is possible to perform learning on the traffic transmission situation in 2 hour units.

The traffic transmission data of the third time unit may include information on traffic conditions generated for each third time unit, and the third artificial intelligence engine 630 learns the traffic transmission data of the third time unit. It is possible to perform learning on the traffic transmission situation in units of 3 hours.

Each of the first artificial intelligence engine 610, the second artificial intelligence engine 620, and the third artificial intelligence engine 630 may be additionally differentiated and learned according to the property of the network device. For example, depending on whether it is an aggregation switch or an edge switch, each of the first artificial intelligence engine 610, the second artificial intelligence engine 620, and the third artificial intelligence engine 630 is a first artificial intelligence engine (aggregation) , 1st artificial intelligence engine (edge), 2nd artificial intelligence engine (aggregation), 2nd artificial intelligence engine (edge), 3rd artificial intelligence engine (aggregation) and 3rd artificial intelligence engine (edge) It could be.

The output values of the artificial intelligence engine generated based on the above learning results for each time unit may be used individually or collectively to determine a packet transmission rule.

First, in order to predict the network condition for a specific time point and determine the packet transmission rule, the first artificial intelligence engine 610 generates a prediction result of the first network state at the specific time point predicted by the first time unit, and second artificial intelligence The intelligence engine 620 generates a prediction result of the second network state at a specific time point predicted by the second time unit, and the third artificial intelligence engine 630 predicts the third network state at the specific time point predicted by the third time unit. can produce results.

According to an embodiment of the present invention, a network state may be determined and a packet transmission rule may be determined by statistically considering the first network state prediction result, the second network state prediction result, and the third network state prediction result. For example, the network prediction result may indicate a network state based on a number from 0 to 100, and network congestion may be higher as 100 increases. For example, if the first network state prediction result is 80, the second network state prediction result is 70, and the third network state prediction result is 60, the network congestion may be determined to be 70 by considering the average of 80, 70, and 60. there is.

In another method, different weights are adjusted for network state prediction results in consideration of the prediction timing and the prediction accuracy of the first artificial intelligence engine 610, the second artificial intelligence engine 620, and the third artificial intelligence engine 630. Thus, it is possible to predict the network state at a specific point in time.

For example, the first artificial intelligence engine 610 has a relatively high prediction reliability for the first time point from the current point in time compared to other artificial intelligence engines, and a prediction reliability after the first point in time compared to other artificial intelligence engines. can be relatively low. The second artificial intelligence engine 620 may have relatively high prediction reliability for the second time point from the current point in time compared to other artificial intelligence engines, and relatively low prediction reliability after the second point in time compared to other artificial intelligence engines. there is. The third artificial intelligence engine 630 may have a relatively high prediction reliability for the third time point from the current point in time compared to other artificial intelligence engines, and a relatively low prediction reliability after the third point in time compared to other artificial intelligence engines. there is.

In this case, the prediction result of the first network state of the first artificial intelligence engine 610 and the second network state of the second artificial intelligence engine 620 in consideration of the predicted time point and the difference between the first time point, the second time point, and the third time point As a result of the state prediction, the network state at the predicted time point may be determined by assigning a weight to each of the third network state prediction results of the third artificial intelligence engine 630 . More specifically, based on the difference between the prediction time point and the first time point, the second time point, and the third time point, a relatively higher weight is assigned as the difference is smaller, so that the network state for the prediction time point may be determined.

7 is a conceptual diagram illustrating a method of determining an artificial intelligence-based packet transmission rule according to an embodiment of the present invention.

7 discloses a method for determining a packet transmission rule by estimating a network state in consideration of a connection relationship between network devices.

Referring to FIG. 7 , a method for performing overall network state prediction through network state prediction at a specific time point of a plurality of network devices and determining a packet transmission rule based thereon is disclosed.

In order to predict the overall network state, the network device association degree may be considered, and when the network device association degree is greater than or equal to a threshold degree of association, the prediction of the network state may be performed by grouping the network devices.

The degree of association between network devices determines whether packets are transmitted or received directly or indirectly when packets are transmitted or received between network devices, when transmission and reception occurs indirectly, how many network devices exist in the middle, and whether network switches are It may be determined by considering the identity of serviced objects.

For example, it may be assumed that the network device association degree between the first network device and the second network device is 0.7, and the network device association degree between the second network device and the third network device is 0.3. When the threshold network device association degree for generating a network device group is 0.5, the network device group 700 may be created by grouping the first network device and the second network device.

The network device group 700 may generate a network state prediction result in units of network device groups. The network state prediction for each network device group 700 is based on the network state prediction result for each of a plurality of network devices included in the network device group, but the network state prediction result of a larger value predicts the network state of the network device group can be used as a result.

In addition, according to an embodiment of the present invention, network state prediction results of a network device or network device group may be set to affect each other. When a network device group 700 or network device specific network state prediction result determines a limit network state prediction value (for example, 90 or more), this result is determined by the network device group 700 or another network device group connected to the network device, or It can be configured to give network devices an additional +A value.

Embodiments according to the present invention described above may be implemented in the form of program instructions that can be executed through various computer components and recorded on a computer-readable recording medium. The computer readable recording medium may include program instructions, data files, data structures, etc. alone or in combination. Program instructions recorded on the computer-readable recording medium may be specially designed and configured for the present invention, or may be known and usable to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks and magnetic tapes, optical recording media such as CD-ROMs and DVDs, and magneto-optical media such as floptical disks. medium), and hardware devices specially configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include high-level language codes that can be executed by a computer using an interpreter or the like as well as machine language codes generated by a compiler. A hardware device may be modified with one or more software modules to perform processing according to the present invention and vice versa.

Although the present invention has been described above with specific details such as specific components and limited embodiments and drawings, these are only provided to help a more general understanding of the present invention, and the present invention is not limited to the above embodiments, and the present invention Those with ordinary knowledge in the technical field to which the invention belongs may seek various modifications and changes from these descriptions.

Therefore, the spirit of the present invention should not be limited to the above-described embodiments and should not be determined, and all scopes equivalent to or equivalently changed from the claims as well as the claims described below are within the scope of the spirit of the present invention. will be said to belong to

Claims (6)

A network traffic control method based on artificial intelligence-based traffic analysis,
receiving, by a network traffic control device, traffic information from a network device;
determining, by the network traffic control device, a packet transmission rule through an artificial intelligence engine learned based on the traffic information; and
Transmitting, by the network traffic control device, the packet transmission rule to the traffic control device;
The artificial intelligence engine operates based on a clustering algorithm and a learning algorithm,
The clustering algorithm is an algorithm for learning and modeling standardized packet transmission rules based on non-failure traffic information and processing them;
The learning algorithm is an algorithm for learning by determining failure traffic information as an unstructured packet transmission rule variable,
The non-failure traffic information includes aggregation switch traffic information, edge switch traffic information, or terminal traffic information,
The fault traffic information includes information on traffic inflow (hacking), hang-up, or power shutdown;
The clustering algorithm is determined based on a learning model accumulated in the database,
Wherein the learning algorithm is determined by storing traffic information collected in real time and used as learning data in a learning model cache storage. According to claim 1,
The artificial intelligence engine includes a first artificial intelligence engine, a second artificial intelligence engine, and a third artificial intelligence engine,
The first artificial intelligence engine may perform learning on a traffic transmission situation in a first time unit through learning on traffic transmission data in a first time unit,
The traffic transmission data of the first time unit includes information on traffic conditions generated for each of the first time units;
The second artificial intelligence engine may perform learning on a traffic transmission situation in a second time unit through learning on traffic transmission data in a second time unit,
The traffic transmission data of the second time unit includes information on traffic conditions generated for each second time unit,
The third artificial intelligence engine may perform learning of a traffic transmission situation in a third time unit through learning of traffic transmission data in a third time unit,
The traffic transmission data of the third time unit includes information on traffic conditions generated for each of the third time units;
Each of the first artificial intelligence engine, the second artificial intelligence engine, and the third artificial intelligence engine is additionally differentiated and learned according to the property of the network device,
Wherein the packet transmission role is determined based on output values of the first artificial intelligence engine, the second artificial intelligence engine, and the third artificial intelligence engine generated based on the learning result for each time unit. According to claim 2,
The network traffic control device predicts the overall network state based on the network device association,
When the network device association is greater than or equal to a threshold association, network devices are grouped to predict a network state;
The degree of association between the network devices determines whether packets are transmitted or received directly or indirectly when packets are transmitted or received between network devices, when transmission and reception occur indirectly, how many network devices exist in the middle, and network switches. A method characterized in that the determination is made in consideration of whether objects served by are identical. A network traffic control device that performs network traffic control based on artificial intelligence-based traffic analysis,
Receive traffic information from a network device;
Determine packet transmission rules through an artificial intelligence engine learned based on the traffic information;
Delivering the packet transmission rule to the traffic control device;
The artificial intelligence engine operates based on a clustering algorithm and a learning algorithm,
The clustering algorithm is an algorithm for learning and modeling standardized packet transmission rules based on non-failure traffic information and processing them;
The learning algorithm is an algorithm for learning by determining failure traffic information as an unstructured packet transmission rule variable,
The non-failure traffic information includes aggregation switch traffic information, edge switch traffic information, or terminal traffic information,
The fault traffic information includes information on traffic inflow (hacking), hang-up, or power shutdown;
The clustering algorithm is determined based on a learning model accumulated in the database,
The learning algorithm is determined by storing traffic information collected in real time and used as learning data in a learning model cache storage. According to claim 4,
The artificial intelligence engine includes a first artificial intelligence engine, a second artificial intelligence engine, and a third artificial intelligence engine,
The first artificial intelligence engine may perform learning on a traffic transmission situation in a first time unit through learning on traffic transmission data in a first time unit,
The traffic transmission data of the first time unit includes information on traffic conditions generated for each of the first time units;
The second artificial intelligence engine may perform learning on a traffic transmission situation in a second time unit through learning on traffic transmission data in a second time unit,
The traffic transmission data of the second time unit includes information on traffic conditions generated for each second time unit,
The third artificial intelligence engine may perform learning of a traffic transmission situation in a third time unit through learning of traffic transmission data in a third time unit,
The traffic transmission data of the third time unit includes information on traffic conditions generated for each of the third time units;
Each of the first artificial intelligence engine, the second artificial intelligence engine, and the third artificial intelligence engine is additionally differentiated and learned according to the property of the network device,
The packet transmission role is determined based on output values of the first artificial intelligence engine, the second artificial intelligence engine, and the third artificial intelligence engine generated based on the learning result for each time unit. According to claim 5,
The network traffic control device predicts the overall network state based on the network device association,
When the network device association is greater than or equal to a threshold association, network devices are grouped to predict a network state;
The degree of association between the network devices determines whether packets are transmitted or received directly or indirectly when packets are transmitted or received between network devices, when transmission and reception occur indirectly, how many network devices exist in the middle, and network switches. Network traffic control device, characterized in that the determination is made in consideration of the identity of the objects to be served.

Images