Ultra large scale data intelligence analysis processing ai edge gateway of unlicensed own network

Abstract

To improve the reliability of a data transmission and reception applied to power consumption wide-band communication networks.SOLUTION: A ultra large scale data intelligence analysis processing AI edge gateway 300 of an unlicensed own network, includes, in a front step of an edge network controller, an AI edge computing part 330 that updates a sensor data analysis rule and a video image analysis model on the basis of a result obtained by analysing an IoT RF gateway that performs a radial transmission and reception of an IoT terminal and a data via a spontaneous polarization effect antenna (SOMPA), collect data, and transmits it to an AI edge computing part, a WiFi access point that performs the radial transmission and reception of an IoT terminal of a WiFi terminal and the data via the SOMPA, collect the data, and transmits it to the AI edge computing part, and an ultra large scale IoT sensor data and a WiFi video image / audio data IoT RF gateway that are input via WiFi access point on the basis of an AI algorithm.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present invention relates to an AI edge gateway for intelligent analysis and processing of ultra-large-scale data for unlicensed private networks, which is applied to a low-power wide-area communication network system and is located in front of an edge network controller so as to be operated close to the site, and which analyzes large-scale sensor data and video / audio data collected in real time based on an AI algorithm, and then performs a comprehensive intelligent analysis of the results to recognize the exact situation and make a judgment in real time, thereby enabling quick decision-making and measures to be taken. The present invention relates to an AI edge gateway for intelligent analysis and processing of ultra-large-scale data for unlicensed private networks, which improves the reliability of data transmission and reception by constructing a stable radio wave environment using a polarized effect antenna. [Background technology]

[0002] A low power wide area network (LPWAN) is a low power wireless wide area network that provides a very wide service area (10km or more) and communication speeds of several hundred kilobits per second (kbps). It is used in dedicated networks for the Internet of Things (IoT), which connects things to a network.

[0003] LPWANs include LoRa (registered trademark), SigFox (registered trademark), NB-IoT, etc., and most of them have technical limitations in terms of data stability and coverage. For example, some cities have installed LoRaWAN (Long Range WAN) and installed up to 45,000 meter reading devices for remote water meter reading, but problems have arisen on-site, such as a 4% non-reception rate due to the excessive number of connected devices.

[0004] Therefore, in the case of a hyper-connected smart city environment where various types of data are mixed, it is necessary to provide a stable and reliable connection environment for processing massive sensor data generated in applications such as transportation, environment, energy, and safety.

[0005] To solve this problem, the Korea Communications Technology Association (TTA) established the Differentiated Wireless Channel Access based Low Power Wide Area Network (dcaLPWAN) standard.

[0006] dcaLPWAN is an LPWAN technology that provides wireless channel access in a newly designed manner that is differentiated according to the characteristics of various IoT application services, while the physical layer is connected using the LoRa method. This technology is a low-power wide area network technology that can support priority-based competitive channel access and reservation-based non-competitive channel access in consideration of the characteristics of each service class.

[0007] As shown in FIG. 1, the dca LPWAN system includes an IoT device 10 that collects various types of data such as temperature, humidity, and PM2.5, a terminal 20 that transmits the data collected by the IoT device 10 to an RF gateway 30 through wireless communication, the RF gateway 30 that performs a packet transmission function between the terminal 20 and a network controller 40, the network controller 40 that manages the terminal 20 and the RF gateway 30, configures a network, processes user commands, etc., an edge gateway 50 that performs pre-processing and intelligent analysis of data collected in the dca LPWAN, and an application server 60 that performs computing functions for various applications and service processing.

[0008] In the configuration of such a dcaLPWAN system, the edge gateway 50 is a component that performs pre-processing and intelligent analysis of data collected in the dcaLPWAN, and is essential equipment for performing edge computing.

[0009] Edge computing means that computing resources are located close to where the data is generated, i.e., close to the "edge" of the data.

[0010] The goal is to process data on the device that generates it, or on an edge computing device, instead of transmitting the data to a central server such as a central data center or cloud.

[0011] This allows for real-time response and data processing, providing faster response times in bandwidth-limited situations.

[0012] In addition, some data can be processed at the edge instead of being transmitted to a central data center due to privacy issues.

[0013] Edge computing can be applied in a variety of industrial fields.

[0014] For example, smart grids can use edge computing to manage and optimize energy production and usage.

[0015] Additionally, in the smart healthcare field, edge computing can be used for patient monitoring and real-time data processing by medical devices.

[0016] In the manufacturing industry, edge computing can be used for monitoring and controlling production lines, and it also plays an important role in autonomous vehicles, smart cities, etc.

[0017] Although there are theoretical results that show that the dcaLPWAN system structure shown in Figure 1 can ensure performance in terms of protocols for ultra-large-scale data transmission and reception, in order to efficiently build a network environment for actual ultra-large-scale data transmission and reception in smart cities, it must be modified and applied to an ultra-large-scale edge computing network structure that suits the actual environment.

[0018] The network controller 40 in Fig. 1 plays a role in maintaining and optimizing the efficiency and security of the IoT network. Its main functions include communication management, sensor and terminal management, security management, data flow control, and network expansion.

[0019] The edge gateway 50 located behind the network controller 40 is a device that handles data pre-processing and AI-based data analysis.

[0020] However, in order to realize edge computing that analyzes, judges, and makes decisions closer to the site, it is better to position the edge gateway 50 in front of the network controller 40, which prevents the inefficiency of having to install the network controller 40 at the site where the data is generated.

[0021] However, the edge gateway 50 closest to the site must be integrated with the RF gateway 30 as much as possible to configure the network. Also, distributing some of the functions of the network controller 40, such as sharing sensor and terminal information and controlling data flow, is a more efficient way to build a super-large-scale IoT network.

[0022] On the other hand, in many smart city applications, there are environments where it is difficult to accurately grasp the situation on site simply by analyzing IoT data.

[0023] For example, in the case of flood detection applications in urban areas, accurate situation assessment can only be achieved by analyzing not only sensor data but also video data monitoring the site.

[0024] In addition, construction sites in urban areas may need to set up separate local networks that are separate from the smart city network, and collect and analyze various on-site sensor data and video data through these networks.

[0025] However, video data cannot be transmitted only through LPWAN, so video data must be transmitted through WiFi in most cases. Therefore, the edge gateway 50 must have a function to establish a WiFi network together with the LPWAN for more accurate situational awareness, collect sensor data and video data through this combined network, and perform combined analysis. The edge gateway 50 must have an edge intelligence function that can intelligently perform sensor data analysis, video data analysis, comprehensive situational judgment, and decision-making on-site.

[0026] The AI ​​edge gateway for processing WiFi data and dca LPWAN data in a combined manner must have a structure that integrates a WiFi access point module, a dca LPWAN RF gateway module, and an edge computing module. In order to place it close to the site, it must be located in front of the network controller 40.

[0027] On the other hand, one of the many challenges facing the spread of LPWANs is the stable and reliable acquisition of sensor data. Even if a differentiated wireless channel access-based low-power wide-area network (dCaLPWAN) is constructed, many problems can occur due to poor radio environment such as mutual radio interference and failures depending on the installation environment of large-scale sensors.

[0028] Solving these issues is essential for the AI ​​edge gateway to more accurately analyze situations and make decisions.

[0029] This problem can be solved by coupling with an antenna using the polarized current effect based on Korean Patent Registration No. 10-1017690 (polarized current effect and its applications).

[0030] The electro-polarization effect refers to a method disclosed by the applicant for controlling the direction of electric current flowing in a conductor.

[0031] If such a bias current effect is utilized in a communication circuit, it is possible to eliminate the need for passive elements such as a filter duplexer and an isolator in the output section, which are necessary when constructing a communication circuit and a communication device. This has the advantage that the loss generated by such passive elements can be reduced, thereby improving the performance of the communication device. Furthermore, by omitting the passive elements, it is possible to realize a communication device that is cost-effective and lightweight.

[0032] In addition to enabling long-distance communication, scattering phenomena caused by rain, fog, dust, etc. can be prevented, and scattering phenomena caused by obstacles such as signs and facilities in urban areas can also be prevented, resulting in the effect of creating a stable wireless communication environment.

[0033] In conclusion, to ensure an AI edge gateway that can process WiFi data and dcaLPWAN in an integrated manner, three core issues must be resolved: first, a system structure that can effectively integrate the WiFi access point module, dcaLPWAN RF gateway module, and AI edge computing module; second, changing the location of the network controller in the system network configuration and sharing some of the functions of the network controller to operate close to the site; and third, the inclusion of technology that can solve problems such as radio interference and failures, build a stable radio environment, and improve the reliability of data transmission and reception. [Prior art documents] [Patent documents]

[0034] [Patent Document 1] Korean Patent No. 10-2593932 [Patent Document 2] Korean Patent Publication No. 10-2023-0054097 [Patent Document 3] Korean Patent No. 10-1017690 [Patent Document 4] Korean Patent No. 10-1591920 [Patent Document 5] Korean Patent No. 10-2422163 Summary of the Invention [Problem to be solved by the invention]

[0035] The present invention has been made to solve the above problems, and an object of the present invention is to provide an AI edge gateway for intelligent analysis and processing of ultra-large-scale data for an unlicensed private network, which is located in front of an edge network controller so that it can be operated close to the site, analyzes large-scale sensor data and video / audio data collected in real time based on an AI algorithm, and then performs a comprehensive intelligent analysis of the results to recognize the accurate situation and make a judgment in real time, and can take quick decisions and measures based on the results. In addition, the gateway improves the reliability of data transmission and reception by building a stable radio wave environment using a polarized effect antenna.

[0036] Another object of the present invention is to provide an AI edge gateway for intelligent analysis and processing of ultra-large-scale data for an unlicensed private network, which is applicable to a dcaLPWAN system and effectively integrates a WiFI access point module, a dcaLPWAN RF gateway module, and an AI edge computing module, and is located in front of a network controller so as to be operated close to the site, shares some of the functions of the network controller, and uses a polarized effect antenna to create a stable radio wave environment and improve the reliability of data transmission and reception. [Means for solving the problem]

[0037] To achieve the above object, the present invention is applied to a low-power wide-area communication network including an IoT terminal, a WiFi terminal, an edge network controller, and an AI edge server, and is characterized by including: an IoT RF gateway located in front of the edge network controller, equipped with a SOMPA, wirelessly transmitting and receiving data to and from an IoT terminal via the SOMPA, collecting data, and transmitting the data to an AI edge computing unit; a WiFi access point equipped with a SOMPA, wirelessly transmitting and receiving data to and from a WiFi terminal via the SOMPA, collecting data, and transmitting the data to the AI ​​edge computing unit; and an AI edge computing unit that analyzes super-large-scale IoT sensor data and WiFi video / audio data input through the IoT RF gateway and the WiFi access point based on an AI algorithm, performs a composite intelligent analysis of the analysis results to make a judgment in real time, makes a quick decision and takes a measure based on the result, simultaneously transmits the edge analysis results and the composite data to an external AI edge server so that precise analysis data and situation data can be stored, and updates a sensor data analysis rule and a video analysis model based on the result of the precise analysis.

[0038] According to the present invention, the AI ​​edge computing unit includes a data input / output unit which performs signal processing on data input from the IoT RF gateway and the WiFi access point and transmits the processed data to a control unit; a control unit which transmits data input from the data input / output unit to a situation awareness intelligence processing unit and performs control of data flow between a system operation and each detailed function, security management, management of IoT terminals and WiFi terminals, and management of a gateway system; and a situation awareness intelligence processing unit which performs analysis, integration analysis, etc. of ultra-large scale IoT sensor data and WiFi video / audio data input through the control unit based on AI algorithms for each function, and performs real-time decision-making and measures.

[0039] Furthermore, according to the present invention, the data input / output unit is characterized by being composed of a sensor data processing unit that performs signal processing on sensor data input from the IoT RF gateway, and a video / audio data processing unit that performs signal processing on video or audio data input from the WiFi access point.

[0040] According to the present invention, the situational awareness intelligence processing unit includes: The sensor data intelligent analysis unit analyzes the sensor data input from the control unit using a rule-based AI algorithm to determine whether it is normal or abnormal, and if it is normal, stores the sensor data in an external AI edge server, and if it is abnormal, performs an AI situation analysis, stores the performed situation analysis data in an external AI edge server, and takes measures after decision-making; the video / audio data intelligent analysis unit analyzes the video / audio data input from the control unit using a lightweight model-based AI algorithm to determine whether it is normal or abnormal, and if it is normal, stores the video / audio data in an external AI edge server, and if it is abnormal, performs an AI situation analysis after a deep analysis by the external AI edge server, stores the performed situation analysis data in an external AI edge server, and takes measures after decision-making; and if abnormal, performs an AI situation analysis after deep analysis by the external AI edge server, stores the performed situation analysis data in the external AI edge server, and performs measures after decision-making. An analysis rule update unit connected to the sensor data analysis unit, the data analysis and event processing unit, and the external AI edge server to update and share analysis rules; and an analysis model update unit connected to the video / audio data analysis unit, the data analysis and event processing unit, and the external AI edge server to update and share analysis models.

[0041] In addition, according to the present invention, there is provided a process for intelligent analysis of large-scale data, comprising: The first process transmits and receives sensor data and video / audio data to and from IoT devices and WiFi devices using SOMPA, and processes the collected sensor data and video / audio data. The second process analyzes the sensor data collected by the first process using a rule-based AI algorithm to determine whether it is normal or abnormal, and if it is normal, stores the sensor data in an external AI edge server. If it is abnormal, an AI situation analysis is performed, and the performed situation analysis data is stored in an external AI edge server and measures are taken after decision-making. If the sensor data is normal in the second process, the video / audio data collected by the first process is analyzed using a lightweight model-based AI algorithm to determine whether it is normal or abnormal, and if it is normal, stores the video / audio data in an external AI edge server. The system is characterized by sequentially performing the following three steps: a third step of storing the data, and if abnormal, performing an AI situation analysis after a deep analysis by an external AI edge server, storing the performed situation analysis data in an external AI edge server, and performing post-decision measures; and a fourth step of analyzing the video / audio data using a lightweight model-based AI algorithm to determine whether it is normal or abnormal, and if it is normal, storing composite data in an external AI edge server, performing an AI situation analysis, storing the performed situation analysis result in an external AI edge server, and performing post-decision measures; and a fourth step of performing an AI situation analysis after a deep analysis by an external AI edge server, storing the performed situation analysis data in an external AI edge server, and performing post-decision measures. Effect of the Invention

[0042] In this way, the present invention makes it possible to create a reliable radio wave environment for ultra-large-scale IoT data collection.

[0043] In other words, by using the polarized antenna to create a stable radio wave environment and improve the reliability of data transmission and reception, it can contribute greatly to the construction and spread of LPWAN when combined with the AI ​​edge gateway, which is the core equipment of LPWAN. This makes it possible to reduce the cost of network construction and build an unlicensed communication infrastructure that enables stable, high-quality, and high-performance communication.

[0044] In addition, the present invention enables the spread of edge intelligence technology that can process IoT data and WiFi data in a combined manner based on standard communication technology that can process ultra-large-scale IoT data.

[0045] In other words, if a data processing platform is located in a centralized cloud, the long delay between data generation and processing makes it difficult to apply to services that require ultra-low latency and real-time processing such as disaster response. Therefore, edge computing technology that enables real-time processing near sensors is essential to reduce the data load of cloud platforms and support distributed processing and ultra-low latency services.

[0046] The AI ​​edge gateway proposed by this invention is a platform that analyzes and processes data close to things, rather than via the cloud over the network, in order to accommodate services that require immediacy as numerous types of things are connected to the network.

[0047] Therefore, by processing the explosive data volumes of smart cities, including those for transportation, medical care, education, and the environment, in real time at the edge, it is possible to solve many of the problems that hinder the efficient use of limited network resources and the provision of ultra-low latency, highly reliable services.

[0048] In addition, sensitive and personal information among the data analyzed at the edge can be used in a limited manner only within the edge without being transmitted to the cloud, reducing security risks due to information leakage.

[0049] In addition, the present invention makes it possible to build an efficient AI analysis and processing system suitable for edge computing by adopting an AI configuration in which learning and inference are separated and a rule-based AI algorithm for sensor data.

[0050] The AI ​​edge gateway is a system configuration that completely separates learning and inference, and can perform frequent re-learning in response to rapid changes in learning models separately on a high-performance server. By performing only lightweight inference at the edge, it can adopt AI technology based on efficient data processing and intelligently respond to the exploding demand for data services.

[0051] In addition, by being positioned in front of a network controller, the present invention makes it possible to configure a more flexible network system in various application environments, such as mounting smart poles in smart cities and independent construction sites.

[0052] In addition, the present invention is a solution that can eliminate the bottleneck of traffic in network controllers when constructing a super-large-scale network, and can effectively respond to various application fields by constructing a distributed edge environment. Furthermore, it can provide real-time data processing, emergency response through instant decision-making, and reduction of data traffic. [Brief description of the drawings]

[0053] [Figure 1] FIG. 1 is a diagram illustrating a conventional TTA standard differentiated channel access based low power consumption wide area network (dcaLPWAN) system. [Diagram 2] FIG. 2 is a diagram showing a dca LPWAN system to which the present invention is applied. [Diagram 3] FIG. 3 is a control block diagram of the ultra-large-scale data intelligent analysis processing AI edge gateway for an unlicensed private network according to the present invention. [Figure 4]FIG. 4 is a diagram showing a manhole smart node system for explaining the SOMPA technology applied to the present invention. [Diagram 5] FIG. 5 is a diagram showing a configuration for testing the system of FIG. [Figure 6] FIG. 6 is a diagram showing a configuration in which the SOMPA of FIG. 4 is actually installed inside a manhole. [Figure 7] FIG. 7 is a table showing the results of technical comparison performance measurements between the SOMPA applied to the present invention and a conventional omni-antenna. [Figure 8] FIG. 8 is a flowchart showing the control of an AI edge gateway for intelligent analysis and processing of ultra-large-scale data in an unlicensed private network according to the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0054] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[0055] First, when assigning reference numerals to components in each drawing, it should be noted that the same reference numerals are used for the same components even if they are shown in different drawings. Furthermore, when describing the present invention, detailed descriptions of related publicly known functions or configurations are omitted if it is determined that such descriptions may unnecessarily obscure the gist of the present invention.

[0056] FIG. 2 is a diagram showing a dcaLPWAN system according to the present invention, which can be simultaneously connected to an ultra-large-scale IoT and WiFi network and is applied to an integrated edge terminal equipped with an ultra-compact AI.

[0057] As shown in the figure, the dca LPWAN system to which the integrated edge terminal 200 according to the present invention is applied is as follows: It includes an IoT device 110, a WiFi device 120, an IoT terminal 210, a WiFi terminal 220, an AI edge gateway 300, an edge network controller 400, and an AI edge server 500.

[0058] The IoT device 110 collects various types of data such as temperature, humidity, PM2.5, etc. and transmits it to the IoT terminal 210, and the WiFi device 120 includes devices such as large-capacity data sensors or cameras and transmits the collected data (e.g., video / audio signals) to the IoT terminal 210.

[0059] The IoT terminal 210 and the WiFi terminal 220 are connected to the IoT device 110 and the WiFi device 120, respectively, and are equipped with an antenna that uses the electro-polarization effect (hereinafter referred to as "SOMPA" or "electro-polarization effect antenna"), embodying ultra-compact AI functions and management functions, and transmitting data collected from the IoT device 110 and the WiFi device 120 to the AI ​​edge gateway 300 via the SOMPA.

[0060] The edge network controller 400 is connected to the AI ​​edge gateway 300 and serves to manage and control the integrated network.

[0061] The AI ​​edge server 500 is connected to the AI ​​edge gateway 300 and the edge network controller 400 to perform pre-processing and intelligent analysis of data collected in the dca LPWAN.

[0062] In the dca LPWAN system having such a configuration, the AI ​​edge gateway 300 according to the present invention is located in front of the edge network controller 400 and is equipped with SOMPA. It analyzes large-scale sensor data and video / audio data collected in real time at the edge close to the site based on an AI algorithm, and then performs a comprehensive intelligent analysis of the results to recognize the accurate situation and make a judgment in real time, and is configured to take quick decisions and measures based on the results.

[0063] In addition, the AI ​​edge gateway 300 is configured to transmit the edge analysis results and composite data to the external AI edge server 500 simultaneously to enable more precise analysis and storage of situation data, and to perform effective and efficient analysis and control based on an AI algorithm that can update the image analysis model and sensor data analysis rules according to the results of precise analysis.

[0064] Specifically, the AI ​​edge gateway 300 includes an IoT RF gateway 310, a WiFi access point 320, and an AI edge computing unit 330, as shown in FIG. 3.

[0065] The IoT RF gateway 310 is equipped with a SOMPA 311, wirelessly transmits and receives data to and from the IoT terminal 210 via the SOMPA, and collects and transmits data to the AI ​​edge computing unit 330.

[0066] The WiFi access point 320 is equipped with a SOMPA 321, and wirelessly transmits and receives data to and from the WiFi terminal 220 via the SOMPA to collect data and transmit it to the AI ​​edge computing unit 330.

[0067] Here, the SOMPA 311 and 321 are antennas that utilize the polarized current effect disclosed by the present applicant in order to solve the problem of poor radio wave environments.

[0068] Incidentally, the SOMPA technology utilizing the electric current polarization effect is based on the applicant's Korean Patent Publication No. 10-1017690 (Electric current polarization effect and its applications), and SOMPA is disclosed in Korean Patent Publication No. 10-1591920 (Directional MIMO antenna utilizing electric current polarization effect) and Korean Patent Publication No. 10-2422163 (Omni-directional antenna utilizing electric current polarization effect).

[0069] The MIMO (Multi-Input Multi-Output) antenna of Korean Patent No. 10-1591920 is a structure that radiates radio waves in a certain direction by aligning the phase of the array of antenna radiators to improve antenna gain, and provides the effect of directing radio waves only in a specific direction, thereby increasing the service distance in a specific direction. The omni antenna of Korean Patent No. 10-2422163 improves on the patch antenna that uses the polarized effect to radiate radio waves in a perfect circle in all directions 360°, thereby facilitating service in open areas and suppressing radio wave scattering to provide excellent long-distance and obstacle-overcoming characteristics.

[0070] Such SOMPA has demonstrated outstanding performance when combined with WiFi terminals, WiFi access points, IoT terminals, IoT RF gateways, etc.

[0071] In other words, by utilizing the bias current effect in a communication circuit, there is an advantage in that it is not necessary to use passive elements such as a filter duplexer and an isolator in the output section, which are necessary when constructing a communication circuit and a communication device. This reduces the loss generated by such passive elements, thereby improving the performance of the communication device. Furthermore, by omitting the passive elements, it is possible to realize a communication device that is cost-effective and lightweight.

[0072] In addition to enabling long-distance communication, scattering phenomena caused by rain, fog, dust, etc. can be prevented, and scattering phenomena caused by obstacles such as signs and facilities in urban areas can also be prevented, resulting in the effect of creating a stable wireless communication environment.

[0073] To verify its effectiveness, a terminal combining LoRa technology and SOMPA technology was installed underground, and monitoring data on facilities such as internal cables was transmitted through a manhole to an IoT gateway located 100m outside, and performance was measured.

[0074] FIG. 4 is a diagram showing a manhole smart node system for explaining the SOMPA technology applied to the present invention, FIG. 5 is a diagram showing a test configuration of the system of FIG. 4, and FIG. 6 is a diagram showing a configuration in which the SOMPA of FIG. 4 is actually installed inside a manhole.

[0075] The manhole smart node system is a disaster prevention IoT system that installs a sensor node inside a manhole to transmit and receive data on the internal status of the manhole to the ground in order to grasp the safety status inside the manhole of an underground power distribution line.

[0076] Up until now, various types of antennas have been applied and tested, but radio communication was not possible through the thick double manhole, so a wired method was used in which an optical cable was connected by drilling a manhole. However, by applying SOMPA technology, there is no need to drill a manhole at all, and it was confirmed that its performance far surpasses that of conventional technology.

[0077] In other words, as shown in the technical comparison performance measurement result table of the SOMPA applied to the present invention and the conventional omni-antenna in FIG. 7, the SOMPA technology method installed inside the manhole without drilling a hole can achieve a performance improvement of 20 dB compared to the method of using an existing antenna to transmit through a manhole.

[0078] According to the present invention, the SOMPA 311 applied to the IoT RF gateway 310 is preferably an omni-antenna that utilizes the polarized current effect to radiate radio waves in a perfect circle in all directions of 360°, and the SOMPA 223a applied to the WiFi access point 320 is preferably a directional MIMO antenna that utilizes the polarized current effect to radiate radio waves in a fixed direction.

[0079] The AI ​​edge computing unit 330 includes a data input / output unit 340, a control unit 350, and a situational awareness intelligence processing unit 360.

[0080] The data input / output unit 340 is a component that processes data input from the IoT RF gateway 310 and the WiFi access point 320 and transmits the processed data to the control unit 350, and is composed of a sensor data processing unit 341 that processes sensor data input from the IoT RF gateway 310, and a video / audio data processing unit 342 that processes video or audio data input from the WiFi access point 320.

[0081] The control unit 350 transmits data input from the data input / output unit 340 to the context awareness intelligence processing unit 360, and performs control of data flow between system operations and each detailed function, security management, management of the IoT terminal 210 and the WiFi terminal 220, management of the gateway system, etc.

[0082] The situation awareness intelligent processing unit 360 is a component that analyzes the super large-scale IoT sensor data and WiFi video / audio data input through the control unit 350 based on AI algorithms for each function, such as analysis and integrated analysis, makes real-time decisions, and commands and executes measures, and includes a sensor data intelligent analysis unit 361, a video / audio data intelligent analysis unit 362, a data complex intelligent analysis and event processing unit 363, an analysis rule update unit 364, and an analysis model update unit 365.

[0083] The sensor data intelligent analysis unit 361 analyzes the sensor data input from the control unit 350 using a rule-based AI algorithm to determine whether the data is normal or abnormal, and if the data is normal, stores the sensor data in an external AI edge server 500, and if the data is abnormal, performs an AI situation analysis, stores the performed situation analysis data in the external AI edge server 500, and performs measures after decision-making.

[0084] The video / audio data intelligent analysis unit 362 analyzes the video / audio data input from the control unit 350 using a lightweight model-based AI algorithm to determine whether the data is normal or abnormal, and if the data is normal, stores the video / audio data in the external AI edge server 500, and if the data is abnormal, performs an AI situation analysis after a deep analysis by the external AI edge server 500, stores the performed situation analysis data in the external AI edge server 500, and performs measures after decision-making.

[0085] The data complex intelligent analysis and event processing unit 363 analyzes the sensor data input from the control unit 350 using a rule-based AI algorithm, and if it is determined to be abnormal, analyzes the video / audio data using a lightweight model-based AI algorithm to determine whether it is normal or abnormal, and if it is normal, stores the complex data in the external AI edge server 500, and if it is abnormal, performs an AI situation analysis after a deep analysis by the external AI edge server 500, stores the performed situation analysis data in the external AI edge server 500, and performs measures after decision-making.

[0086] The analysis rule update unit 364 is connected to the sensor data intelligent analysis unit 361, the data complex intelligent analysis and event processing unit 363, and the external AI edge server 500 to update and share analysis rules.

[0087] The analysis model update unit 365 is connected to the video / audio data intelligent analysis unit 362, the data complex intelligent analysis and event processing unit 363, and the external AI edge server 500 to update and share analysis models.

[0088] FIG. 8 is a flowchart showing the control of an AI edge gateway for intelligent analysis and processing of ultra-large-scale data in an unlicensed private network according to the present invention.

[0089] As shown in the figure, the control process of the AI ​​edge gateway for large-scale intelligent analysis processing of unlicensed private networks of the present invention is as follows: The present invention includes a first process S10 for transmitting and receiving sensor data and video / audio data to and from an IoT terminal 210 and a WiFi terminal 220 using SOMPA and processing the collected sensor data and video / audio data, a second process S20 for analyzing the sensor data collected by the first process S10 using a rule-based AI algorithm to determine whether it is normal or abnormal, and if it is normal, storing the sensor data in an external AI edge server 500, and if it is abnormal, performing an AI situation analysis, storing the performed situation analysis data in an external AI edge server 500, and performing measures after decision-making, and if the sensor data is normal in the second process S20, analyzing the video / audio data collected by the first process S10 using a lightweight model-based AI algorithm to determine whether it is normal or abnormal, and if it is normal, storing the video / audio data in an external AI edge server 500. and a third process S30 of storing the sensor data in the external AI edge server 500, performing an AI situation analysis after a deep analysis by the external AI edge server 500 if the sensor data is abnormal, storing the performed situation analysis data in the external AI edge server 500, and performing a post-decision measure. If the sensor data is abnormal in the second process S20, the video / audio data is analyzed using a lightweight model-based AI algorithm to determine whether the data is normal or abnormal, and if the data is normal, storing composite data in the external AI edge server 500, performing an AI situation analysis, storing the performed situation analysis result in the external AI edge server 500, and performing a post-decision measure. If the sensor data is abnormal, the fourth process S40 of performing an AI situation analysis after a deep analysis by the external AI edge server 500, storing the performed situation analysis data in the external AI edge server 500, and performing a post-decision measure.

[0090] The second process S20 includes a step S21 of analyzing the collected sensor data using a rule-based AI algorithm to determine whether the data is normal or abnormal, a step S22 of storing the sensor data in an external AI edge server 500 if the data is determined to be normal in the step S21, and a step S23 of performing an AI situation analysis, storing the performed situation analysis data in the external AI edge server 500, and performing post-decision measures if the data is determined to be abnormal in the step S21.

[0091] The third process S30 includes a step S31 of analyzing the video / audio data collected in the first process S10 using a lightweight model-based AI algorithm to determine whether the data is normal or abnormal if the sensor data is normal in the second process S20, a step S32 of storing the video / audio data in an external AI edge server 500 if the data is determined to be normal in the step S31, and a step S33 of performing an AI situation analysis after a deep analysis by the external AI edge server 500, storing the performed situation analysis data in the external AI edge server 500, and performing measures after decision-making if the data is determined to be abnormal in the step S31.

[0092] The fourth process S40 includes a step S41 of analyzing video / audio data using a lightweight model-based AI algorithm to determine whether the data is normal or abnormal if the sensor data is abnormal in the second process S20; a step S41 of storing composite data in an external AI edge server 500, performing an AI situation analysis, storing the performed situation analysis result in the external AI edge server 500, and performing post-decision measures if the sensor data is determined to be abnormal in the step S41; and a step S43 of performing an AI situation analysis after a deep analysis by the external AI edge server 500, storing the performed situation analysis data in the external AI edge server 500, and performing post-decision measures if the sensor data is determined to be abnormal in the step S41.

[0093] Although the limited embodiments of the present invention have been described above, the present invention is not limited to these, and it should be noted by those skilled in the art that various embodiments are anticipated. [Explanation of symbols]

[0094] 110 IoT devices 120 WiFi devices 210 IoT devices 220 WiFi terminals 300 AI Edge Gateway 310 IoT RF Gateway 320 WiFi Access Points 211, 221, 311, 321 SOMPA 330 AI Edge Computing Department 340 Data Input / Output Unit 341 Sensor data processing unit 342 Video / audio data processing section 350 Control section 360 Situational Awareness Intelligence Processing Unit 361 Sensor Data Intelligence Analysis Department 362 Video / Audio Data Intelligence Analysis Department 363 Data Complex Intelligence Analysis and Event Processing Unit 364 Analysis Rule Update Section 365 Analysis Model Update Division 400 Edge Network Controller 500 AI Edge Servers

Claims

1. This is applicable to a low-power wide-area communication network including an IoT terminal, a WiFi terminal, an edge network controller, and an AI edge server, An IoT RF gateway located in front of the edge network controller, equipped with a SOMPA, wirelessly transmits and receives data to and from an IoT terminal via the SOMPA, collects data, and transmits the data to an AI edge computing unit; A WiFi access point equipped with SOMPA and wirelessly transmitting and receiving data to and from a WiFi terminal via the SOMPA to collect data and transmit the data to an AI edge computing unit; an AI edge computing unit that analyzes the super-large-scale IoT sensor data and the WiFi video / audio data input through the IoT RF gateway and the WiFi access point based on an AI algorithm, performs a composite intelligent analysis of the analysis results to make a judgment in real time, makes a quick decision and takes a measure based on the analysis results, transmits the edge analysis results and the composite data simultaneously to an external AI edge server so that precise analysis data and situation data can be stored, and updates the sensor data analysis rules and the video analysis model based on the precise analysis results; An unlicensed private network ultra-large scale data intelligent analysis processing AI edge gateway comprising:

2. The AI ​​edge computing unit includes a data input / output unit that processes data input from the IoT RF gateway and the WiFi access point and transmits the processed data to a control unit; a control unit that transmits data input from the data input / output unit to a situational awareness intelligence processing unit, and controls data flow between a system operation and each detailed function, manages security, manages IoT terminals and WiFi terminals, manages a gateway system, etc.; The ultra-large scale IoT sensor data and WiFi video / audio data input through the control unit are analyzed based on AI algorithms for each function, such as analysis and integrated analysis, to perform real-time decision-making and measures.

3. The data input / output unit includes a sensor data processing unit that processes sensor data input from the IoT RF gateway; and a video / audio data processing unit for signal processing video or audio data input from the WiFi access point.

4. The situational awareness intelligence processing unit includes: A sensor data intelligent analysis unit that analyzes the sensor data input from the control unit using a rule-based AI algorithm to determine whether the sensor data is normal or abnormal, stores the sensor data in an external AI edge server if the sensor data is normal, and performs an AI situation analysis if the sensor data is abnormal, stores the performed situation analysis data in an external AI edge server, and performs measures after decision-making; a video / audio data intelligent analysis unit that analyzes the video / audio data input from the control unit using a lightweight model-based AI algorithm to determine whether the data is normal or abnormal, stores the video / audio data in an external AI edge server if the data is normal, and performs an AI situation analysis after a deep analysis by the external AI edge server if the data is abnormal, stores the performed situation analysis data in the external AI edge server, and performs measures after decision-making; a data composite intelligence analysis and event processing unit that analyzes sensor data input from the control unit using a rule-based AI algorithm, and if it is determined to be abnormal, analyzes video / audio data using a lightweight model-based AI algorithm to determine whether it is normal or abnormal, stores composite data in an external AI edge server if it is normal, and performs AI situation analysis after deep analysis by the external AI edge server if it is abnormal, stores the performed situation analysis data in the external AI edge server, and performs measures after decision-making; An analysis rule update unit connected to the sensor data intelligent analysis unit, the data complex intelligent analysis and event processing unit, and the external AI edge server to update and share analysis rules; The AI ​​edge gateway for ultra-large scale intelligent analysis processing of an unlicensed private network according to claim 2, characterized in that it is configured to include the video / audio data intelligent analysis unit, the data complex intelligent analysis and event processing unit, and an analytical model update unit connected to the external AI edge server to update and share analytical models.

5. The ultra-large scale data intelligent analysis process according to claim 1, A first step of transmitting and receiving sensor data and video / audio data to and from an IoT terminal and a WiFi terminal using SOMPA, and processing the collected sensor data and video / audio data; A second process of analyzing the sensor data collected by the first process using a rule-based AI algorithm to determine whether the data is normal or abnormal, storing the sensor data in an external AI edge server if the data is normal, and performing an AI situation analysis if the data is abnormal, storing the performed situation analysis data in an external AI edge server, and performing measures after decision-making; If the sensor data is normal in the second step, the video / audio data collected in the first step is analyzed by a lightweight model-based AI algorithm to determine whether the data is normal or abnormal, and if the data is normal, the video / audio data is stored in an external AI edge server, and if the data is abnormal, an AI situation analysis is performed after a deep analysis by the external AI edge server, the performed situation analysis data is stored in the external AI edge server, and a third step is performed after the decision-making process; In the second step, if the sensor data is abnormal, the video / audio data is analyzed by a lightweight model-based AI algorithm to determine whether it is normal or abnormal, and if it is normal, the composite data is stored in an external AI edge server, an AI situation analysis is performed, the performed situation analysis result is stored in an external AI edge server, and a post-decision measure is performed; and if it is abnormal, an AI situation analysis is performed after a deep analysis by the external AI edge server, the performed situation analysis data is stored in an external AI edge server, and a post-decision measure is performed; and An unlicensed private network ultra-large scale data intelligent analysis processing AI edge gateway, characterized by sequentially performing the above.

Images