KR102444229B1 - Management system and method for large-scale wireless lan based on big data - Google Patents

Abstract

The present invention relates to a big data-based large-scale wireless LAN management system and method to provide efficient wireless LAN service by detecting load imbalance of APs in a building and analyzing service usage patterns of users using big data in a large-scale wireless LAN environment. It relates to an AP data collection unit that collects AP data in a WLAN environment; a building characteristic classification unit that classifies building characteristics using a clustering algorithm based on the behavioral patterns of people in the building; each of the buildings with similar usage patterns An imbalance region search and analysis unit for evaluating and interpreting the degree of imbalance of resources; an optimal AP location search unit for adjusting the AP position by searching for an optimal AP position based on reinforcement learning using a deep neural network.

Description

Big data-based large-scale wireless LAN management system and method {Management system and method for large-scale wireless lan based on big data}

The present invention relates to large-scale wireless LAN management, and more specifically, to provide efficient wireless LAN service by detecting load imbalance of APs in a building using big data in a large-scale wireless LAN environment and analyzing user's service usage pattern. It relates to a data-based large-scale wireless LAN management system and method.

Due to the increase in mobile devices such as smart phones, laptops, and tablet PCs, the use of Internet services through a wireless LAN based on the IEEE 80211 WLAN standard is rapidly increasing.

Accordingly, buildings, offices, campuses, and public institutions provide wireless LAN services by installing APs (Access Points) for user convenience. In particular, the service area becomes narrower due to signal attenuation, etc. indoors.

In addition, when each AP is installed indiscriminately and operates independently, expected performance is significantly lowered due to mutual interference. In addition, in an independently installed AP environment, seamless handoff of mobile stations, which is one of the major issues in wireless LAN, is impossible.

In particular, as the number of devices supporting a wireless LAN (WLAN), such as a smartphone, increases, and consumption of high-capacity content increases, a user's demand for quality improvement of a wireless LAN is increasing.

In general, solutions such as increasing system bandwidth and improving peak transmission rates have been proposed to improve WLAN performance. In addition, in an environment in which a plurality of access points (APs) are dense and the coverage of the APs overlap, methods are proposed to support simultaneous access by many terminals while enhancing the sensible performance of users who require high capacity and high rate services. have.

AP is attenuated, reflected, refractioned, diffracted, dispersed (scattered) and blocked due to the straight-line distance during radio wave transmission depending on the location of the site where it is fixed, that is, due to environmental influences and the structure of buildings are affected by

Therefore, it is very important to set an access point (AP) at an appropriate location in consideration of various surrounding environments in a wireless LAN.

As described above, the optimal design of the location of the access point (AP) of the wireless LAN is an important issue directly related to service quality and cost reduction. In order to find the optimal AP location, not only the radio intensity of the service area according to the location of the AP, but also the regional characteristics of the environment and user demand traffic should be considered.

Nevertheless, the biggest problem of a large-scale wireless LAN environment is that it does not reflect the characteristics of service users in the space at the initial stage of installation and is deployed based only on the service area.

In order to solve this problem, the process of installing the AP 10 in an appropriate place is repeated in consideration of the uniform radio wave reception strength within the service area and the location where the demand for wide bandwidth is concentrated to provide the optimal wireless LAN service. Since it is necessary to set up an environment that does this, it takes a lot of time and money, and at the same time, there are problems such as non-uniform results.

Therefore, there is a need for development of a new technology that enables efficient wireless LAN service to be provided by detecting load imbalance of APs in a building and analyzing a user's service usage pattern.

Republic of Korea Patent Publication No. 10-2003-0086366 Republic of Korea Patent Publication No. 10-2011-0096458 Republic of Korea Patent Publication No. 10-2013-0116529

The present invention is to solve the problem of the large-scale wireless LAN management technology of the prior art, and periodically searches for a place where an imbalance of characteristics occurs within a specific space and optimizes the location of an access point (AP) of the wireless LAN to improve service quality An object of the present invention is to provide a big data-based large-scale wireless LAN management system and method that enables improvement.

The present invention provides a big data-based large-scale wireless LAN management system and method to provide efficient wireless LAN service by detecting load imbalance of APs in a building and analyzing service usage patterns of users using big data in a large-scale wireless LAN environment. Its purpose is to provide

The present invention is a big data-based large-scale wireless LAN that enables the optimization of the AP (Access Point) location of the wireless LAN by analyzing and utilizing the actual usage patterns that were not reflected in the initial installation by using the periodically measured data of the large-scale wireless LAN environment. An object of the present invention is to provide a LAN management system and method.

The present invention provides a big data-based large-scale wireless LAN management system and The purpose is to provide a method.

Other objects of the present invention are not limited to the above-mentioned objects, and other objects not mentioned will be clearly understood by those skilled in the art from the following description.

A big data-based large-scale wireless LAN management system according to the present invention for achieving the above object includes: an AP data collection unit that collects AP data in a wireless LAN environment; utilizes a clustering algorithm based on behavioral patterns of people existing in a building A building characteristic classification unit that classifies the building characteristics using and an optimal AP location search unit for adjusting the location.

A big data-based large-scale wireless LAN management method according to the present invention for achieving another object comprises: collecting AP data in a wireless LAN environment in an AP data collection unit; Classifying the building characteristics using a clustering algorithm with and adjusting the AP position by searching for an optimal AP position based on reinforcement learning.

The big data-based large-scale wireless LAN management system and method according to the present invention as described above has the following effects.

First, it is possible to improve service quality by optimizing the location of an access point (AP) of a WLAN by periodically searching for a place where a characteristic imbalance occurs within a specific space.

Second, it is possible to provide efficient wireless LAN service by detecting load imbalance of APs in buildings and analyzing service usage patterns of users using big data in a large-scale wireless LAN environment.

Third, it is possible to optimize the location of the AP (Access Point) of the WLAN by analyzing and utilizing the actual usage patterns that were not reflected in the initial installation by using the periodically measured data of the large-scale WLAN environment.

Fourth, by applying large-scale wireless LAN data analysis technology, it is possible to improve service quality and reduce management costs by adjusting the location of APs according to the characteristics of buildings and detection of dense areas of users.

1 is a block diagram showing a hierarchical structure of a big data-based large-scale wireless LAN management system and method according to the present invention;
2 is a configuration diagram showing the AP arrangement optimization characteristics reflecting the user demand to which the big data-based large-scale wireless LAN management system and method according to the present invention is applied.
3 is a block diagram of a big data-based large-scale wireless LAN management system according to the present invention;
4 is a block diagram illustrating a wireless LAN management concept through a big data-based large-scale wireless LAN management system according to the present invention.
5 is a block diagram showing a large-scale wireless LAN environment AP data collection process;
6 is a graph showing the characteristic classification characteristic of a building;
7 is a configuration diagram showing the process of searching and analyzing an imbalanced area using a comparison of imbalance indicators within a building
8 is a configuration diagram for finding an optimal AP location based on reinforcement learning.
9 is a configuration diagram of a current state simulation scenario
10 is a flowchart illustrating a big data-based large-scale wireless LAN management method according to the present invention;

Hereinafter, a preferred embodiment of the big data-based large-scale wireless LAN management system and method according to the present invention will be described in detail as follows.

Features and advantages of the big data-based large-scale wireless LAN management system and method according to the present invention will become apparent through detailed description of each embodiment below.

1 is a block diagram showing a hierarchical structure of a big data-based large-scale wireless LAN management system and method according to the present invention.

The big data-based large-scale wireless LAN management system and method according to the present invention are designed to periodically search for a place where an imbalance of characteristics occurs within a specific space to improve service quality through optimizing the location of an access point (AP) of a wireless LAN. did it

To this end, the present invention may include a configuration of classifying building characteristics using a clustering algorithm based on large-scale wireless LAN environment AP data collection and behavior patterns of people present in the building.

The present invention evaluates the degree of imbalance of resources for each of the buildings with similar usage patterns, selects a floor where each attribute overlaps and the imbalance occurs, and configures for interpretation and optimal AP location based on reinforcement learning using deep neural networks It may include a configuration for adjusting the location of the AP by searching.

As shown in FIG. 1, the hierarchical structure of the big data-based large-scale wireless LAN management system according to the present invention is an Analytics Platform for state diagnosis reflecting the spatiotemporal characteristics of wireless network equipment, an AP arrangement that optimizes load imbalance, and each building characteristic. A Northbound SDN Controller Application Program Interface (API) that provides an interface between an AP positioning means and an SDN controller based on deep learning that supports power management operation, a control layer having an SDN controller, and a control layer It consists of the SDN Control-Data Plane Interface that interfaces the (Control Layer) and the infrastructure layer, and the AP infrastructure layer (Infrastructure layer) and the client.

2 is a block diagram showing the characteristics of AP arrangement optimization reflecting user demand to which the big data-based large-scale wireless LAN management system and method according to the present invention is applied.

A comparison is made in the state of optimizing AP deployment by reflecting user demand according to the present invention in the initial AP deployment state (left) aimed at maximizing the service area (right).

The detailed configuration of the big data-based large-scale wireless LAN management system according to the present invention is as follows.

3 is a block diagram of a big data-based large-scale wireless LAN management system according to the present invention.

As shown in FIG. 3, the big data-based large-scale wireless LAN management system according to the present invention includes data used with personnel connected to each AP during operation, usage attributes including throughput and signal status, and flags or retransmissions generated by each AP during operation. An AP data collection unit 10 that collects AP data in a large-scale wireless LAN environment including the number of frame information counter properties recorded, and a building that classifies building characteristics using a clustering algorithm based on the behavioral patterns of people in the building The characteristic classification unit 20 and the imbalanced area search and analysis unit 30 that evaluates the degree of resource imbalance for each of the buildings with similar usage patterns, and selects and interprets the floor where each attribute overlaps and the imbalance occurs and an optimal AP position search unit 40 for adjusting the AP position by searching for an optimal AP position based on reinforcement learning using a deep neural network.

4 is a block diagram illustrating a wireless LAN management concept through a big data-based large-scale wireless LAN management system according to the present invention.

The big data-based large-scale wireless LAN management system according to the present invention (1) collects AP data in a large-scale wireless LAN environment, (2) classifies building characteristics using a clustering algorithm based on the behavioral patterns of people in the building, (3) Evaluate the degree of resource imbalance for each building with similar usage patterns, select and analyze the floor where each property overlaps and imbalance occurs, (4) search for the optimal AP location based on reinforcement learning using deep neural networks to locate the AP Perform an adjustment action.

5 is a block diagram illustrating a process of collecting AP data in a large-scale wireless LAN environment.

In order to efficiently manage large-scale wireless LAN services, an SDN environment is premised. This is because the interface located between each plane can facilitate data collection.

The collection period is one hour, and the properties collected by each AP are shown in the table below.

The attributes collected by each AP include a usage attribute and a frame information counter attribute. The usage attribute indicates the number of people connected to each AP during operation, data used, throughput and signal status, and the like.

The usage attribute includes the channel utilization rate, the number of connected devices, the number of packet retransmissions, RSSI, SNR, and throughput.

The frame information counter attribute indicates an attribute in which each AP records the number of flags or retransmissions generated during operation.

The frame information counter properties include the number of ACK failures, the number of FCS (Frame Check Sequence) errors, the number of retransmissions, the number of RTS (Request To Send) failures, the number of RTS successes, the number of TX failures, the number of TX fragments, and the number of TX frames. do.

It goes without saying that these collection properties are not limited thereto, and other items may be added or changed.

6 is a graph showing the characteristic classification characteristic of a building.

Attributes collected prior to selecting a building should ensure periodicity in time to detect periodic imbalances.

Buildings with similar patterns of collected properties are judged to have similar behavior patterns of people in the building.

To classify this, a clustering algorithm is applied, and density based clustering and distance based clustering are mixed and classified.

6 shows an example of classifying buildings into four types according to usage.

Group 1: Active spaces were classified only during lecture time.

Group 2: Active spaces were classified only during business hours, and the active usage was recorded in the afternoon and decreased at night.

Group 3: Libraries were classified and had a cycle showing active usage in the afternoon as well as at night.

Group 4: Spaces focused on meetings and meetings were classified.

7 is a configuration diagram illustrating the process of searching for and analyzing an imbalanced area using a comparison of imbalance indicators within a building.

Assess the degree of resource imbalance for each of the buildings with similar usage patterns.

Evaluate the degree of imbalance by using the indicators below for all attributes, and try to interpret by selecting the layer where each attribute overlaps and imbalance occurs.

To evaluate the degree of imbalance after assigning an index to the repeated data pattern, Jain's fairness index is used.

Fairness evaluates the degree of resource allocation and sharing, the closer to 1, the higher the fairness, and the closer to 0, the lower.

The reinforcement learning-based optimal AP location search configuration is as follows.

8 is a configuration diagram of an optimal AP location search based on reinforcement learning.

Reinforcement learning optimizes the learning target (Agent) so that it receives a reward for the action and undergoes trial-and-error to get the maximum score. Representative algorithms such as Q-Learning, neural networks, etc. There is this.

),에피소드의 요소를 갖는다.The basic elements of reinforcement learning are environment, state (S), agent, behavior (A), state transition probability (P), reward (R), profit (G), policy (

), which has elements of an episode.

To improve the disproportionate use of APs, the following strategies are established to adjust the positions of APs.

와 정책

는 다음과 같다.

and policy

is as follows

9 is a configuration diagram of a current state simulation scenario.

)는,situation(

)Is,

Deploy clients around the AP based on the average number of connected devices per week (per hour, 168 hours, 168 data). The client moves randomly and can roam according to the service quality of the existing AP.

Based on the weekly average data throughput of the AP, the throughput of clients placed around the AP is arbitrarily divided. Mbps)

The signal attenuation of the concrete wall is fixed at 15dB to consider the service coverage.

The time to obtain the imbalance indicator according to the building classification is as follows (considering the main use time of the building)

Group1 : 00:00 ~ 23:59, 24 hours a day, total 168 hours

Group2 : 09:00 ~ 18:00, 9 hours a day, total 56 hours

Group3 : 09:00 ~ 23:59, 15 hours per day, total 105 hours

Group4 : 09:00 ~ 20:00, 11 hours per day, total 77 hours

)은 AP의 위치를 조정하여 불균형 지표를 계산한다.action(

) adjusts the position of the AP to calculate the imbalance indicator.

The big data-based large-scale wireless LAN management method according to the present invention will be described in detail as follows.

10 is a flowchart illustrating a big data-based large-scale wireless LAN management method according to the present invention.

As shown in FIG. 10, the big data-based large-scale wireless LAN management method according to the present invention includes, first, the data used with the number of people connected to each AP in the AP data collection unit 10 during operation, the usage attribute including the processing rate and the signal state, Each AP collects large-scale wireless LAN environment AP data including the frame information counter attribute that records the number of flags or retransmissions generated during operation. (S1001)

Next, the building characteristic classification unit 20 classifies the building characteristics by using a clustering algorithm based on the behavioral patterns of people present in the building. (S1002)

Then, the imbalanced area search and analysis unit 30 evaluates the degree of imbalance of resources for each of the buildings with similar usage patterns, and selects and analyzes the floor where the imbalance occurs due to overlapping properties. (S1003)

Next, the optimal AP location search unit 40 searches for an optimal AP location based on reinforcement learning using a deep neural network and adjusts the AP location (S1004).

The big data-based large-scale wireless LAN management system and method according to the present invention described above utilizes periodically measured data of the large-scale wireless LAN environment to analyze and utilize actual usage patterns that were not reflected in the initial installation, Access Point) location optimization is possible.

It will be understood that the present invention is implemented in a modified form without departing from the essential characteristics of the present invention as described above.

Therefore, the specified embodiments are to be considered in an illustrative rather than a restrictive view, the scope of the present invention is indicated in the claims rather than the foregoing description, and all differences within the equivalent scope are included in the present invention. will have to be interpreted.

10. AP data collection unit
20. Building Characteristics Division
30. Unbalanced area search and analysis unit
40. Optimal AP location search unit

Claims (15)

AP data collection unit for collecting AP data in a wireless LAN environment;
a building property classification unit that classifies building properties using a clustering algorithm based on the behavioral patterns of people in the building;
an imbalance area search and analysis unit that evaluates and interprets the degree of resource imbalance for each of the buildings with similar usage patterns;
It includes; an optimal AP location search unit that adjusts the AP location by searching for the optimal AP location based on reinforcement learning using a deep neural network.
The imbalance area search and analysis unit uses Jain's fairness index to evaluate the degree of imbalance after assigning an index to the repeated data pattern, and Fairness evaluates the degree of resource allocation and sharing, and the closer to 1, the higher Fairness is 0 The closer to, the lower

Big data-based large-scale wireless LAN management system, characterized in that calculated as The method of claim 1, wherein the AP data collection unit,
Usage attributes, including data used, throughput and signal status, and the number of people connected by each AP during operation;
Big data-based large-scale wireless LAN management system, characterized in that each AP collects data of frame information counter properties that record the number of flags or retransmissions generated during operation. The method of claim 2, wherein the usage attribute includes channel usage rate, the number of connected devices, the number of packet retransmissions, RSSI, SNR, and throughput.
The frame information counter properties include the number of ACK failures, the number of FCS (Frame Check Sequence) errors, the number of retransmissions, the number of RTS (Request To Send) failures, the number of RTS successes, the number of TX failures, the number of TX fragments, and the number of TX frames. Big data-based large-scale wireless LAN management system, characterized in that The method of claim 1, wherein the imbalance area search and analysis unit,
A large-scale wireless LAN management system based on big data, characterized in that it selects and interprets a layer where each property overlaps and an imbalance occurs, and performs an imbalance area search and analysis. According to claim 1, The building characteristic classification unit,
A large-scale wireless LAN management system based on big data, characterized in that it is determined that buildings having a similar pattern of collected properties have similar behavior patterns of people in the building. The method of claim 5, wherein the building characteristic classification unit,
A big data-based large-scale wireless LAN management system characterized by mixing and classifying density-based clustering and distance-based clustering as a clustering algorithm for classifying building characteristics. delete The method of claim 1, wherein the optimal AP location search unit is configured to search for an optimal AP location based on reinforcement learning;

and policy

Is,

Big data-based large-scale wireless LAN management system, characterized in that it is defined as The method according to claim 8, wherein the state (

)Is,
Based on the average number of devices connected to the AP per week, clients are placed around the AP, and the clients move randomly and roam according to the AP's service quality.
A big data-based large-scale wireless LAN management system characterized by randomly dividing the throughput of clients placed around the AP based on the weekly average data throughput of the AP. collecting AP data in a wireless LAN environment by an AP data collection unit;
Classifying the building characteristics by using a clustering algorithm based on the behavior patterns of people present in the building in the building characteristic classification unit;
Evaluating and analyzing the degree of imbalance of resources for each of the buildings having a similar usage pattern in the imbalance area search and analysis unit;
Adjusting the AP position by searching for the optimal AP position based on reinforcement learning using a deep neural network in the optimal AP position search unit;
In the stage of evaluating and interpreting the degree of resource imbalance for each of the buildings with similar usage patterns in the area search and analysis unit, Jain's fairness index is used to evaluate the degree of imbalance after assigning an index to the repeated data pattern. and fairness evaluates the degree of resource allocation and sharing, the closer to 1, the higher the fairness, and the closer to 0, the lower,

Big data-based large-scale wireless LAN management method, characterized in that calculated as The method of claim 10, wherein in the step of collecting AP data,
Usage attributes, including data used, throughput and signal status, and the number of people connected by each AP during operation;
A big data-based large-scale wireless LAN management method, characterized in that each AP collects the data of the frame information counter property that records the number of flags or retransmissions that occurred during operation. 12. The method of claim 11, wherein the usage attribute includes channel usage rate, the number of connected devices, the number of packet retransmissions, RSSI, SNR, and throughput.
The frame information counter properties include the number of ACK failures, the number of FCS (Frame Check Sequence) errors, the number of retransmissions, the number of RTS (Request To Send) failures, the number of RTS successes, the number of TX failures, the number of TX fragments, and the number of TX frames. Big data-based large-scale wireless LAN management method, characterized in that The method of claim 10, wherein in the step of evaluating and interpreting the degree of imbalance of resources for each of the buildings,
Big data-based large-scale wireless LAN management method, characterized in that it selects and interprets a layer where each property overlaps and an imbalance occurs, and performs an imbalance area search and interpretation. The method of claim 10, wherein in the step of classifying building characteristics using a clustering algorithm,
A big data-based large-scale wireless LAN management method, characterized in that it is determined that buildings having a similar pattern of collected properties have similar behavioral patterns of people present in the building. 15. The method of claim 14, wherein the clustering algorithm for classifying building characteristics is classified by mixing density based clustering and distance based clustering.

Images