Monitoring device, monitoring method, program, and monitoring system

Abstract

Conventional monitoring systems can receive notifications when the monitored person deviates from the monitoring range, but they have difficulty in making quick situational judgments in emergencies, and there was room for improvement. One of the purposes of this disclosure is to provide a monitoring device, monitoring method, program, and monitoring system that can quickly and efficiently grasp the status of the monitored person. [Solution] The monitoring device comprises a location information acquisition unit that acquires location information of the person being monitored, a range setting unit that sets the monitoring range for the person being monitored, a deviation detection unit that detects when the person being monitored deviates from the monitoring range, a map information generation unit that generates map information including the current location and movement history of the person being monitored when the deviation detection unit detects a deviation, and a notification control unit that displays the map information in the notification area of ​​a push notification.

Description

Technical Field

[0001] The present disclosure relates to a monitoring device, a monitoring method, a program, and a monitoring system.

Background Art

[0002] Conventionally, in order to ensure the safety of children, a monitoring system using a GPS function has been provided. For example, Patent Document 1 discloses a monitoring method in which a wireless device is detected by a system mounted on a vehicle and a response is activated based on a preset status declaration.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In the conventional monitoring system, although it is possible to receive a notification when the monitored object deviates from the monitoring range, it is difficult to quickly judge the situation in an emergency, and there is room for improvement.

[0005] One object of the present disclosure is to provide a monitoring device, a monitoring method, a program, and a monitoring system capable of quickly and efficiently grasping the situation of the monitored object.

Means for Solving the Problems

[0006] A monitoring device according to one aspect of the present disclosure includes: a location information acquisition unit that acquires location information of a monitored object; a range setting unit that sets a monitoring range for the monitored object; a deviation detection unit that detects when the monitored object deviates from the monitoring range; a map information generation unit that generates map information including the current location and movement history of the monitored object when the deviation detection unit detects a deviation; and a notification control unit that displays the map information in the notification area of ​​a push notification. [Brief explanation of the drawing]

[0007] [Figure 1] Figure 1 is an overview diagram showing the overall configuration of the monitoring system according to this embodiment. [Figure 2] Figure 2 is a functional block diagram of the monitoring device (server) according to this embodiment. [Figure 3] Figure 3 is a detailed flowchart of the monitoring process according to this embodiment. [Figure 4] Figure 4 is a detailed hardware configuration diagram of the server. [Figure 5] Figure 5 is a software configuration diagram of the server. [Figure 6] Figure 6 is a UI diagram showing an example of a push notification screen. [Figure 7] Figure 7 is a UI diagram showing an example of the monitoring range setting screen. [Figure 8] Figure 8 is a UI diagram showing an example of an anomaly detection result display screen. [Figure 9] Figure 9 is a UI diagram showing an example of an AI learning results display screen. [Figure 10] Figure 10 is a UI screen diagram showing an example of a large-scale management screen. [Figure 11] Figure 11 is a diagram illustrating the IoT integration system configuration. [Figure 12] Figure 12 is a communication sequence diagram. [Figure 13] Figure 13 is a diagram of the database structure. [Figure 14] Figure 14 is a security configuration diagram. [Modes for carrying out the invention]

[0008] The embodiments of this disclosure will be described in detail below with reference to the drawings.

[0009] (First Embodiment) • Overall system configuration Figure 1 is an overview diagram showing the overall configuration of the monitoring system 1 according to this embodiment. The monitoring system 1 comprises a GPS terminal device 100 carried by the person to be monitored 10, a terminal device 200 used by the guardian 20, a server 300 that communicates with these devices, and an external service 400. The server 300 can communicate with the GPS terminal 100, the terminal device 200, and the external service 400 via the internet or the like.

[0010] The GPS terminal device 100 is a small location information acquisition device carried by the person being monitored 10 (for example, a child). The GPS terminal device 100 incorporates a GPS receiver, communication module, battery, acceleration sensor, temperature sensor, waterproof housing, etc., and periodically transmits location information to the server 300. More specifically, the GPS terminal device 100 receives signals from GPS satellites (including "Michibiki," the Japanese version of GPS) to calculate its own location information and transmits the location information to the server 300 using wireless communication means such as LTE-M, NB-IoT, Cat.1, Wi-Fi, and Bluetooth Low Energy (BLE).

[0011] The GPS terminal device 100 has a compact size, for example, approximately 50mm (height) x 30mm (width) x 15mm (thickness), and is designed to weigh less than 30g. The casing has IP67-rated water and dust resistance and is also shock-resistant. It uses a lithium-ion battery and can operate continuously for about 5 days to 1 week under normal use.

[0012] The GPS terminal device 100 has a function of dynamically adjusting the acquisition frequency of location information. For example, when within the monitoring range, it acquires and transmits location information at intervals of 10 minutes, and when outside the monitoring range, it does so at intervals of 1 minute. Also, adaptive control is performed to lower the acquisition frequency in a stationary state and raise it in a moving state based on the detection results of the acceleration sensor.

[0013] The terminal device 200 is a portable terminal device such as a smartphone, tablet, or smartwatch used by the protector 20. The terminal device 200 includes a CPU, memory, display, touch panel, communication module, camera, microphone, speaker, etc., and has a monitoring application installed. The terminal device 200 has a function of receiving a push notification from the server 300 and displaying map information within the notification area.

[0014] The operating system of the terminal device 200 is compatible with iOS (iPhone (registered trademark), iPad (registered trademark)), Android (Android smartphones, tablets), WatchOS (Apple Watch), Wear OS (Android Wear), etc. It displays push notifications in an optimal format according to the specifications of the notification APIs of each OS.

[0015] The server 300 is a server device that is the core of the monitoring system 1. It processes the location information received from the GPS terminal device 100 and sends appropriate notifications to the terminal device 200. The server 300 is configured as a cloud server and operates on a cloud platform such as Amazon Web Services (AWS), Microsoft Azure, or Google Cloud Platform (GCP). To ensure high availability, the server 300 is distributed and processed by multiple servers and is accessed via a load balancer.

[0016] Server 300 consists of a group of servers including a web server, application server, database server, and machine learning server. The web server handles HTTPS communication from the terminal device 200, and the application server executes monitoring logic. The database server stores location information, user information, configuration information, etc. (see, for example, Figure 13), and the machine learning server performs learning processing to realize AI functions.

[0017] External Services 400 is a group of external services that the monitoring system 1 integrates with. Specifically, these include map services (Google Maps API, Mapbox API, OpenStreetMap API, etc.), Points of Interest (POI) services (Google Places API, Foursquare API, etc.), push notification services (Apple Push Notification Service (APNs), Firebase Cloud Messaging (FCM), etc.), weather information services, traffic information services, emergency call services, etc.

[0018] Detailed configuration of the monitoring device Figure 2 is a functional block diagram of the monitoring device (server 300) according to this embodiment. The monitoring device includes a location information acquisition unit 310, a range setting unit 320, a deviation detection unit 330, a map information generation unit 340, a notification control unit 350, a learning unit 360, an automatic setting unit 370, an anomaly detection unit 380, a facility information presentation unit 390, a data management unit 391, a security management unit 392, a communication control unit 393, a log management unit 394, and a statistical analysis unit 395.

[0019] The location information acquisition unit 310 is a functional unit that acquires location information of the monitored object 10. The location information acquisition unit 310 receives location information transmitted from the GPS terminal device 100 and processes the location information, including coordinate data (latitude, longitude, altitude), acquisition time, accuracy information, movement speed, azimuth angle, number of satellites, HDOP (Horizontal Dilution of Precision), etc.

[0020] Regarding "location information" in the location information acquisition unit 310, as a higher-level concept, the location information acquisition unit 310 is a functional unit that acquires "location information". As an intermediate-level concept, "location information" is information indicating the geographical location of the monitored object 10, and includes GPS coordinates, time information, accuracy information, movement information, etc. More specifically, "location information" includes information such as latitude and longitude data transmitted from the GPS terminal device 100, acquisition time, GPS positioning accuracy, movement speed, azimuth angle, altitude, number of satellites, and received signal strength. As a lower-level concept, for example, "location information" would be specific data such as "35.6762 degrees North latitude, 139.6503 degrees East longitude, altitude 25.3m, acquisition time 2024 / 03 / 15 14:30:25, accuracy ±3.2m, movement speed 2.3km / h, azimuth angle North-Northeast (22.5 degrees), number of satellites 8, HDOP1.2".

[0021] The location information acquisition unit 310 also has a function to evaluate the quality of the received location information. Specifically, it quantifies the reliability of the location information based on accuracy information, number of satellites, HDOP value, etc., and filters or corrects low-quality location information. It also checks the consistency of consecutive location information, and if it detects physically impossible movement (e.g., instantaneous movement), it excludes it as abnormal data.

[0022] The range setting unit 320 is a functional unit that sets the monitoring range for the monitored subject 10. The range setting unit 320 receives manual setting instructions from the guardian 20 and dynamically sets the monitoring range in cooperation with the automatic setting unit 370, which will be described later. Figure 7 is an example of a setting screen.

[0023] Regarding the "monitoring range" in the range setting unit 320, as a higher-level concept, the range setting unit 320 is a functional unit that sets the "monitoring range". As an intermediate-level concept, the "monitoring range" is a geographical boundary used to determine whether the monitored object 10 is within its normal activity range, and is defined by shapes such as circles, ellipses, polygons, combinations of multiple sub-regions, and strip-shaped regions along roads. The monitoring range can also be dynamically changed according to conditions such as time of day, day of the week, and season. As a lower-level concept, for example, the "monitoring range" can be set as "a circle with a radius of 500m centered on the home," "the union of three circles including the school, cram school, and park," "a strip-shaped region with a width of 100m along the school route," or "the area around the school from 8:00 to 16:00 on weekdays, and the area around the home at other times."

[0024] The range setting unit 320 supports multiple setting methods to accommodate diverse monitoring scenarios. Firstly, the circular range setting function allows parents to intuitively set a circular monitoring area by tapping a center point on the map and adjusting the radius using a slider or pinch gesture. This method is suitable for setting a basic monitoring area centered on home or school. Secondly, the polygonal range setting function allows for setting a monitoring area with a more complex shape by tapping multiple vertices on the map. This makes it possible to accurately set an area that includes irregularly shaped plots of land or specific groups of buildings. Thirdly, the composite range setting function allows for the construction of a more complex monitoring area by combining multiple circular and polygonal ranges. For example, a comprehensive monitoring area can be achieved by combining a circular range around home, a circular range around school, and a polygonal range of the school route. Fourthly, the roadside range setting function works in conjunction with an external road data API to automatically generate a band-shaped monitoring area along school routes or commuting routes. This function allows for detection only when a child deviates significantly from the road they normally travel. Fifth, the building range setting function acquires shape data of buildings such as schools, cram schools, and libraries, and sets a precise monitoring range that matches the building's outline. This enables fine-grained control, such as ensuring safety inside the building and requiring caution outside the building.

[0025] The deviation detection unit 330 is a functional unit that detects when the monitored object 10 deviates from the monitoring range. The deviation detection unit 330 compares the location information acquired from the location information acquisition unit 310 with the monitoring range set by the range setting unit 320, and detects a deviation if the location information is outside the monitoring range.

[0026] The deviation detection unit 330 has advanced judgment functions in addition to simple point in / outside determination. Specifically, buffering processing sets buffer areas inside and outside the monitoring range to prevent false detections near the boundary. In addition, temporal filtering detects only deviations that continue for a certain period of time (e.g., 3 minutes), rather than instantaneous deviations. Furthermore, by considering the direction of movement, it distinguishes between movement away from the monitoring range and movement towards the monitoring range. In addition, by considering speed, it estimates the means of movement, such as walking, cycling, or driving, based on the speed of movement and applies appropriate judgment criteria.

[0027] The map information generation unit 340 is a functional unit that generates map information including the current location and movement history of the monitored object 10 when the deviation detection unit 330 detects a deviation. The map information generation unit 340 works in conjunction with external map service APIs (e.g., Google Maps API, Mapbox API, OpenStreetMap API, etc.) to generate dynamic map images.

[0028] Regarding "map information" in the map information generation unit 340, as a higher-level concept, the map information generation unit 340 is a functional unit that generates "map information". As an intermediate-level concept, "map information" is a map image that visually represents the current location and movement history of the monitored object 10, and includes map image data, location markers, movement route lines, timestamps, monitoring range display, surrounding facility information, etc. Map information is generated in still image format (PNG, JPEG, etc.), video format (GIF, MP4, etc.), and vector format (SVG, etc.). As a lower-level concept, for example, "map information" is generated as a "320x240 pixel PNG image that displays the current location with a red pin (with pulse effect), draws the movement route for the past 30 minutes with a blue line (transparency gradient), displays the monitoring range with a dashed line, and displays surrounding schools, parks, etc. with icons."

[0029] The map information generation unit 340 has detailed functions. Specifically, it allows selection from multiple map styles such as standard maps, satellite images, and topographic maps. Automatic scale adjustment automatically determines an appropriate scale according to the distance traveled. Layer management manages background maps, travel routes, position markers, monitoring areas, etc., in independent layers. Color design uses a color palette that is considerate of people with color vision deficiencies. Multilingual support displays text information on the map in multiple languages. Time display shows the time corresponding to each location.

[0030] The notification control unit 350 is a functional unit that displays map information within the notification area of ​​a push notification. The notification control unit 350 uses the iOS and Android notification APIs to send a push notification with embedded map information within the notification area to the terminal device 200. Figure 6 shows an example of a push notification. As shown in Figure 6, the map information is displayed, along with a message indicating that the device has deviated from the monitoring area, the time of the deviation, etc. In addition, at least a portion of the monitoring area is displayed as needed. Furthermore, a sub-detail display button, a confirmation button, and an emergency contact button, which will be described later, are also displayed.

[0031] Regarding the "notification area of ​​a push notification" in the notification control unit 350, as a higher-level concept, the notification control unit 350 is a functional unit that displays map information within the "notification area of ​​a push notification". As an intermediate-level concept, the "notification area of ​​a push notification" is the display area for notification messages displayed on the lock screen, notification center, status bar, widgets, etc., of the terminal device 200, and can include text, images, sound, vibration, action buttons, etc. As a lower-level concept, for example, in the case of iOS, a "Rich Notification" using "UNNotificationServiceExtension" is used, and in the case of Android, a map image is displayed within the notification area using "NotificationCompat.BigPictureStyle".

[0032] The notification control unit 350 implements advanced notification functions. Specifically, adaptive notifications optimize the notification layout according to the terminal's screen size, resolution, and OS version. Urgency classification sets the notification priority (high, medium, low) according to the deviation situation. Notification scheduling suppresses notifications during late night and early morning hours and resends them at appropriate times. Notification grouping organizes the display by grouping multiple notifications. Interactive notifications provide interaction functions such as button taps and swipes within the notification area. Notification history management saves and displays past notifications as a history. Detailed configuration of AI functions

[0033] The learning unit 360 is a functional unit that learns behavioral data including past location information, time spent, frequency of movement, day of the week, and time of day of the monitored subjects 10, and recognizes the behavioral patterns of the monitored subjects 10. The learning unit 360 extracts behavioral patterns using machine learning algorithms (e.g., clustering, pattern recognition, time series analysis, deep learning, etc.).

[0034] The 360 ​​learning unit strategically combines multiple machine learning algorithms to accurately recognize the behavioral patterns of the monitored subject. K-means clustering classifies location information into geographically similar groups and automatically identifies places where the monitored subject frequently stays (home, school, extracurricular activities, etc.). This algorithm is computationally efficient and can quickly extract major activity locations from large amounts of location data. DBSCAN, as a density-based clustering method, precisely extracts true locations while effectively removing GPS positioning errors and noise. This method is particularly effective in environments with radio wave shielding due to high-rise buildings in urban areas and in subway stations where positioning errors are common. The Hidden Markov Model (HMM) mathematically models the movement patterns of the monitored subject as state transitions and learns everyday movement flows such as "home → school → extracurricular activities → home". This allows for early detection of unusual movement patterns. Long Short-Term Memory (LSTM) learns movement patterns as time-series data using deep learning to capture long-term behavioral trends and seasonality. For example, it can learn behavioral patterns during summer vacation or differences in behavior depending on the day of the week. Random Forest combines multiple features such as location, time, weather, and events to analyze the behavioral patterns of monitored subjects from multiple angles. This method is excellent at predicting behavior under different conditions and contributes to understanding behavioral patterns involving complex factors. Support Vector Machine (SVM) learns a boundary that clearly separates normal and abnormal behavioral patterns, achieving highly accurate abnormal behavior detection.

[0035] The features processed by the learning unit 360 include: location features such as latitude, longitude, altitude, distance traveled, speed, and direction of travel; time features such as time of day, day of the week, month, season, and holiday flag; stay features such as duration of stay, frequency of stay, and type of stay location; movement features such as travel time and mode of transport (walking, cycling, vehicle, etc.); environmental features such as weather, temperature, precipitation, and sunshine duration; and event features such as school events, extracurricular activities, and family plans.

[0036] The automatic setting unit 370 is a functional unit that automatically sets the monitoring range based on the behavioral patterns recognized by the learning unit 360. Based on statistical information of the behavioral patterns, the automatic setting unit 370 determines the optimal shape, size, and location of the monitoring range.

[0037] The automatic setting unit 370 performs the following steps to automatically set the monitoring range. First, it identifies major locations of stay (home, school, extracurricular activities, etc.) from the learning results by extracting locations of stay. Second, it analyzes the routes of stay between locations to understand the normal movement patterns. Third, it statistically calculates the activity range centered on each location of stay by calculating the activity range. Fourth, it sets the monitoring range according to the behavior patterns of weekdays / holidays and different times of day by setting by time of day. Fifth, it dynamically adjusts the monitoring range according to changes in seasons, semesters, events, etc. by dynamic adjustment. Sixth, it predicts future behavior patterns and adjusts the monitoring range in advance by predictive setting.

[0038] The anomaly detection unit 380 is a functional unit that analyzes the frequency with which the monitored object 10 deviates from a specific location and detects it as abnormal behavior when it exceeds a predetermined threshold. The anomaly detection unit 380 identifies abnormal patterns using statistical anomaly detection methods (e.g., outlier detection, frequency analysis, time-series anomaly detection, etc.).

[0039] The anomaly detection algorithms implemented in the anomaly detection unit 380 include: detecting deviations from the normal distribution using statistical anomaly detection; performing anomaly detection using Local Outlier Factor (LOF) with density-based anomaly detection; performing time-series anomaly detection using Seasonal Hybrid ESD (SH-ESD) with time-series anomaly detection; performing anomaly pattern detection using an AutoEncoder with deep learning anomaly detection; and performing detection by combining multiple anomaly detection methods with ensemble anomaly detection.

[0040] The facility information display unit 390 is a functional unit that, when the anomaly detection unit 380 detects abnormal behavior, acquires information about facilities in the vicinity of the location and displays it via push notification or on a map. An example of a notification screen is shown in Figure 8. The facility information display unit 390 acquires information about surrounding facilities by coordinating with external POI (Points of Interest) APIs (e.g., Google Places API, Foursquare API, OpenStreetMap Nominal API, etc.).

[0041] The facility information acquired and displayed by the facility information display unit 390 provides important clues for understanding the background of the monitored person's behavior. Educational facilities include not only schools such as elementary schools, junior high schools, high schools, and universities, but also extracurricular activity facilities such as cram schools, preparatory schools, English conversation schools, music schools, and art schools, as well as cultural and educational facilities such as libraries, museums, science museums, and art galleries. Deviations around these facilities are often likely to be normal behavior related to learning or cultural activities. Commercial facilities include retail stores such as convenience stores, supermarkets, department stores, shopping malls, and specialty stores. These facilities are closely related to daily life, and if the monitored person is near these facilities, it is possible that they are making a temporary stop for shopping or errands. Entertainment facilities include game centers, movie theaters, bowling alleys, karaoke boxes, amusement parks, theme parks, parks, and sports fields. Deviations around these facilities may be due to playing with friends or engaging in entertainment activities, and safety should be judged based on the time of day and duration of stay. Transportation facilities include train stations, bus stops, subway stations, taxi stands, airports, and highway service areas. These facilities serve as hubs for movement, making them crucial sources of information for understanding the movement patterns of those being monitored. Medical facilities include hospitals, clinics, dental offices, pharmacies, osteopathic clinics, acupuncture clinics, etc. Deviating from the vicinity of these facilities suggests the possibility of seeking medical attention for health reasons. Public facilities include city halls, ward offices, post offices, police stations, fire stations, community centers, gymnasiums, etc. These facilities serve as places for official business and emergency shelters. Food and beverage establishments include restaurants, cafes, fast food restaurants, family restaurants, izakayas, etc. These facilities are used for meals and rest. Other important sources of information include banks, credit unions, ATMs, coin-operated parking lots, gas stations, and accommodations. Integrating this diverse information from these facilities allows us to understand the context of the monitored person's behavior and determine appropriate responses.

[0042] The data management unit 391 is a functional unit that manages various types of data handled by the monitoring system 1. The data management unit 391 appropriately stores and manages location information, user information, configuration information, log information, etc., and ensures data integrity, availability, and security.

[0043] The database managed by the data management unit 391 stores location information received from the GPS terminal device 100 in chronological order as a location information database. It stores basic information of guardians and the person being monitored as a user information database. It stores various setting information such as monitoring range and notification settings as a settings information database. It stores behavior pattern information generated by the learning unit 360 as a behavior pattern database. It stores abnormal information detected by the abnormality detection unit 380 as an abnormality information database. It stores system operation logs, error logs, etc. as a log database.

[0044] The Security Management Department 392 is the functional department responsible for ensuring the security of the monitoring system 1. The Security Management Department 392 ensures privacy protection and security by performing tasks such as communication encryption, authentication and authorization, access control, and data anonymization.

[0045] The security functions implemented by Security Management Unit 392 include: encryption of communications using TLS 1.3; authentication and authorization using OAuth 2.0 and OpenID Connect; access control using Role-Based Access Control (RBAC); database encryption using AES-256; anonymization of location information using k-anonymity and l-diversity; audit logging functionality that records all access; and intrusion detection functionality that detects abnormal access (Figure 14).

[0046] • Processing flow details Figure 3 is a detailed flowchart of the monitoring process according to this embodiment.

[0047] In step S101, the location information acquisition process involves the location information acquisition unit 310 acquiring location information from the GPS terminal device 100. This process includes detailed operations such as receiving location information from the GPS terminal device 100, evaluating the quality of the location information (checking accuracy, number of satellites, HDOP value, etc.), filtering out abnormal data (removing physically impossible movements), normalizing the location information (unifying coordinate systems, standardizing accuracy), and storing it in a database.

[0048] In step S102, the deviation detection unit 330 compares the acquired location information with the monitoring range to determine whether a deviation has occurred. This process involves detailed operations such as acquiring the current monitoring range (applying a dynamic range according to the time of day and day of the week), geometric determination of the location information and the monitoring range, prevention of false detection through buffering, temporal filtering (determination of duration), and consideration of movement direction and speed. If no deviation has occurred, the process returns to step S101. If a deviation has occurred, the process proceeds to step S103.

[0049] In step S103, the map information generation process involves the map information generation unit 340 generating map information including the current location and movement history. This process includes detailed operations such as determining the display range (setting an appropriate scale according to the distance traveled), acquiring map tiles (acquiring map data from an external map service API), rendering the movement history (setting color and transparency according to the time axis), placing the current location marker (dynamic representation such as pulse effect), rendering the monitoring area (boundary lines, fill, etc.), rendering surrounding facility information (icons, labels, etc.), and generating the final map image (in PNG, JPEG, etc.).

[0050] In step S104, the push notification sending process involves the notification control unit 350 sending the generated map information to the terminal device 200 in a form displayed within the notification area of ​​the push notification. This process includes detailed operations such as generating the notification message (title, body, action, etc.), optimizing the map information (adjusting according to the terminal's screen size and resolution), setting the notification priority (setting the priority according to the urgency of the deviation), sending the push notification (using services such as APNs and FCM), confirming the transmission result (determining whether delivery was successful or unsuccessful), and resending in case of failure.

[0051] In step S105, the abnormal behavior determination process involves the abnormality detection unit 380 analyzing the deviation pattern and determining whether or not abnormal behavior is present. This process includes detailed steps such as acquiring past deviation history, statistical analysis of deviation frequency, execution of the abnormality detection algorithm, calculation of an abnormality score, and comparison with a threshold. If no abnormal behavior is detected, the process terminates. If abnormal behavior is detected, the process proceeds to step S106.

[0052] In step S106, the facility information display process involves the facility information display unit 390 acquiring facility information around the deviation point and displaying it as an additional notification or on the map information. This process includes detailed operations such as acquiring the coordinates of the deviation point, searching for surrounding facilities (searching for facilities within a 500m radius using the POI API), classifying and filtering the facility information, determining the display format of the facility information, generating and sending an additional notification, and adding and displaying the facility information on the map information.

[0053] Figure 12 is a communication sequence diagram. This diagram shows the main communication flow and processing steps between the GPS terminal device 100, the server 300, and the terminal device 200 in the monitoring system 1. In step 1, the GPS terminal device 100 transmits location information to the location information acquisition unit 310 of the server 300. At this time, the quality of the location information and filtering of abnormal data are also performed. In step 2, the server 300 sends an ACK (acknowledgment) in response to receiving location information from the GPS terminal device 100. In step 3, the deviation detection unit 330 of the server 300 detects that the monitored object 10 has deviated from the monitoring range. This detection includes geometric determination of the monitoring range, buffering processing, temporal filtering, and consideration of movement direction and speed. In step 4, the map information generation unit 340 of server 300 generates map information including the current location and movement history of the monitored object. This process includes determining the display range, acquiring map tiles, drawing the movement history, placing the current location marker, drawing the monitoring range, drawing surrounding facility information, and generating the final map image. The map data is obtained from an external service 400. Facility information is also obtained from the external service 400. In step 5, map information is generated based on the acquired map data and equipment information. In step 6, the notification control unit 350 of the server 300 sends the generated map information to the terminal device 200 in a form that is displayed within the notification area of ​​the push notification. This step includes generating the notification message, optimizing the map information, setting the notification priority, sending the push notification, and confirming the transmission result. In step 7, the terminal device 200 receives the push notification and sends an ACK or transmission response signal to the server 300. In step 8, an emergency contact request is received from terminal device 200. This allows for contact with ambulance services, fire departments, police, etc., in step 9.

[0054] • Modified form of the first embodiment In the first embodiment, a modified version is conceivable that provides various display patterns for the movement history. The map information generation unit 340 flexibly provides multiple display patterns for displaying movement history to meet diverse user needs. In time-based display, the entire movement route for the most recent 30 minutes, 1 hour, 3 hours, or the day is displayed chronologically, allowing for a detailed understanding of the monitored person's recent actions. The display for the most recent 30 minutes is suitable for quickly understanding the situation in emergencies, and the movement route is displayed in a dark color, gradually fading as time passes to visually highlight the most recent actions. The 1-hour display is suitable for checking daily activity patterns, and the 3-hour display is effective for understanding long outings or lessons. The display of the entire movement route for the day allows for an overview of all movements from morning to the present, enabling a comprehensive understanding of the day's activity patterns. In distance-based display, only routes within a radius of 500m, 1km, or routes deviating from the monitoring range from the current location are selectively displayed. The display within 500m is suitable for checking short-distance movement, and the display within 1km is effective for understanding medium-distance movement patterns. Displaying only routes that deviate from the monitoring area is suitable for analysis that focuses on problematic behavior. Location-based display shows routes from registered locations such as home and school, routes from the last confirmed safe location, or routes from the location where anomalies were detected. This enables behavioral analysis starting from specific locations, allowing for a deeper understanding of the monitored person's behavioral patterns. Activity-based display classifies and displays routes according to the mode of transportation and activity. Displaying routes that are only walked allows you to confirm the monitored person's spontaneous movements, while displaying routes that are only driven allows you to understand the use of pick-up / drop-off or public transportation. Displaying routes that include locations where the person stayed for a long time helps identify important activity locations and understand the monitored person's lifestyle patterns. Note that the concept of movement history may include the most recent 0 seconds or longer, and may also include displaying only the most recent 0 seconds of movement history, i.e., the current location, as the movement route.

[0055] In the first embodiment, a modified version is conceivable that provides a dynamic and interactive display of push notifications.

[0056] The notification control unit 350 provides advanced display functions within the notification area of ​​push notifications. For animated displays, it implements animations that draw the travel route in chronological order, pulse effects for the current location marker, and flashing effects for the monitoring range boundary. For interactive operations, it enables pinch-in / pinch-out (zoom in / out), map swiping (move the display range), and switching of display modes (standard map, satellite image, topographic map). For multi-layer displays, it displays the background map, travel route, and facility information on independent layers, and enables switching the display / hide of layers and adjusting transparency.

[0057] In the first embodiment, a modified version is conceivable that provides an extension of the emergency contact and response function.

[0058] The notification control unit 350 provides emergency response functions within the notification area of ​​push notifications. Emergency contact functions include automatic notification to the police (110), initiation of voice calls to guardians, mass notification to family and relatives, and notification to schools and extracurricular activities. On-site verification functions include requesting assistance from the nearest police station or station staff, notifying local neighborhood watch volunteers, and arranging taxis or ride-hailing services. Remote response functions include sending voice messages to the GPS terminal device 100, remote control of the GPS terminal device 100's buzzer and LED flashing, and requesting confirmation of surrounding security camera footage.

[0059] In the first embodiment, a modified version is possible that provides a more detailed visual representation.

[0060] The map information generation unit 340 provides detailed visual representations. In terms of color design, movement paths within the monitoring range are displayed in blue (safe), movement paths outside the monitoring range in red (warning), anomaly detection points in orange (caution), and points where the user has stayed for a long time in green (stable). For line representation, line thickness is determined according to movement speed, transparency gradients are applied according to the passage of time, and line types (solid, dashed, dotted) are applied according to the mode of transportation. For marker representation, the current location is displayed as a red pin with a pulse effect, past locations as small dots with time displays, and anomaly locations as markers with warning icons.

[0061] (Second embodiment: Large-scale monitoring system for businesses and schools) In the second embodiment, a system for large-scale monitoring of employees of a company, students of a school, etc., will be described.

[0062] In the second embodiment, the number of people to be monitored increases from several hundred to several thousand, requiring an expansion of the system configuration.

[0063] In the large-scale monitoring system of the second embodiment, the conventional system architecture has been fundamentally revised to accommodate hundreds to thousands of people, and an advanced technological foundation that balances scalability and performance has been adopted. In the horizontally scaling-enabled microservice architecture, the monitoring function is divided into independent service units, and a mechanism has been built in which each service automatically scales out as needed. This allows the overall system performance to be maintained even if the load is concentrated on a particular function, by strengthening only that function. Automatic scaling with a Kubernetes cluster monitors indicators such as CPU usage, memory usage, and network traffic, and dynamically adjusts the number of pods according to the load. By combining Horizontal Pod Autoscaler (HPA) and Vertical Pod Autoscaler (VPA), scaling in both horizontal and vertical directions is achieved, optimizing resource utilization efficiency. High-speed data access with a Redis cluster stores frequently accessed data such as location information, monitoring settings, and notification status in a distributed cache, reducing the load on the database. The automatic sharding function of Redis Cluster distributes data across multiple nodes, ensuring high availability and high performance. Apache Kafka's high-throughput data streaming enables high-speed processing of location information from a large number of GPS terminals, real-time monitoring and decision-making. Kafka's partitioning functionality allows for parallel data processing, achieving processing power capable of handling hundreds of thousands of location information updates per second. Furthermore, Kafka Connect streamlines integration with different data sources and automates data exchange with external systems. By integrating these technologies, it enables large-scale processing that is difficult to achieve with conventional single-server configurations, providing a comprehensive monitoring system at the organizational level for companies, schools, and other organizations.

[0064] In the second embodiment, a hierarchical management function is implemented, which includes a three-tiered management system for school administrators, class teachers, and parents, or a three-tiered management system for company administrators, departmental administrators, and individuals, and access control is implemented according to authority. Batch processing functions are implemented, which include batch processing of large amounts of location data, generation of daily and monthly statistical reports, and analysis of long-term behavioral patterns.

[0065] In the second embodiment, a batch notification function is provided. As group notifications, batch notifications are implemented by class or department, batch distribution of disaster and emergency information by region, and automatic grouping based on conditions. As hierarchical notifications, step-by-step escalation in the event of anomaly detection, automatic reporting to administrators, and information sharing with relevant parties are implemented. As customized notifications, notification settings are implemented according to the organization's operational rules, notification control outside of business hours, and selection of notification methods according to importance are implemented.

[0066] In the second embodiment, large-scale data processing technology is employed to efficiently perform large-scale data processing.

[0067] In the second embodiment, a NoSQL database such as MongoDB or Cassandra is adopted as the distributed database, and horizontal distribution through sharding and high availability through replication are ensured. Big data processing is implemented using Apache Spark as the distributed processing framework, real-time streaming processing using Apache Flink, and large-scale batch processing using Apache Hadoop. Distributed caching is implemented using Redis Cluster, high-speed data access using Memcached, and map delivery optimization is implemented using a CDN (Content Delivery Network).

[0068] In the second embodiment, a comprehensive dashboard function is provided for administrators.

[0069] In the second embodiment, the real-time monitoring function enables simultaneous display of location information for all monitored subjects, immediate visualization of anomaly detection, and monitoring of system load and performance. The statistical and analytical function enables daily, weekly, and monthly behavioral pattern analysis, trend analysis of anomaly detection, and statistical information by region and time. The report generation function enables periodic monitoring status reports, detailed reports of abnormal events, and monthly summaries for guardians. For example, the management screen can be shown in Figure 10.

[0070] (Third embodiment) In the third embodiment, the function for linking with IoT devices will be described.

[0071] In the third embodiment, a function for linking with smart home devices is provided. Figure 11 is a diagram illustrating the concept of the linking function.

[0072] In the third embodiment, the home arrival detection integration enables automatic door unlocking, automatic lighting activation and dimming, automatic control of the air conditioning system, and notification of the home arrival to the family being monitored upon the monitored person's return home. The security integration enables automatic recording start of security cameras, confirmation of presence through integration with motion sensors, intrusion detection using window and door sensors, and integration with fire and gas leak detection. The lifestyle support integration enables analysis of the monitored person's daily rhythm, monitoring of appliance usage, indirect monitoring of health status, and an automatic emergency notification system.

[0073] In the third embodiment, a function for coordinating with a vehicle system is provided.

[0074] In the third embodiment, passenger boarding and alighting detection includes matching with vehicle location information, automatic detection of boarding and alighting timing, support for optimizing pick-up and drop-off routes, and provision of traffic safety information. As a driving support system, it provides safe driving support when the monitored person is in the vehicle, suggests speed limits in dangerous areas, suggests slowing down around schools and hospitals, and provides optimal route guidance in emergencies. As an in-vehicle safety monitoring system, it monitors the temperature and humidity inside the vehicle, detects and warns of unattended passengers in the vehicle, detects abnormal behavior, and provides automatic emergency notifications.

[0075] In the third embodiment, a function for coordinating with wearable devices is provided.

[0076] In the third embodiment, health status monitoring includes continuous monitoring of heart rate and blood pressure, recording of activity levels (steps, calories burned), analysis of sleep patterns, and automatic notification when abnormal values ​​are detected. Emergency detection includes automatic detection of falls and impacts, detection of prolonged periods of inactivity, detection of abnormal heart rate patterns, and manual transmission of SOS signals. Behavioral pattern analysis includes learning of daily activity patterns, analysis of the relationship between health status and location information, suggestion of optimal lifestyle habits, and information sharing with medical institutions.

[0077] In the third embodiment, a function for coordinating with environmental sensors is provided.

[0078] In the third embodiment, weather information linkage will enable real-time monitoring of temperature, humidity, and wind speed, prediction of precipitation and snowfall, calculation of UV index and heatstroke risk, and automatic notification when warnings are issued. For atmospheric environment monitoring, it will enable monitoring of PM2.5 and PM10 concentrations, monitoring of pollen dispersion, calculation of air pollution index, and determination and suggestion of suitability for going outside. For disaster-related monitoring, it will enable immediate distribution of earthquake and tsunami information, advance notification of typhoon and heavy rain information, provision of evacuation shelter information, and automation of safety confirmation. Technical details and support for foreign patent applications. Improvement of computer capabilities

[0079] This disclosure will contribute to improving computer functionality.

[0080] This disclosure optimizes communication volume. In conventional technology, after receiving a deviation notification, it was necessary to launch a separate application to acquire map data, resulting in two communications and a total data communication of approximately 2-3 MB. This disclosure integrates push notifications and map information, reducing the number of communications to one and compressing the data volume to approximately 500 KB or less.

[0081] This disclosure employs a combination of advanced technologies to optimize communication volume. Adaptive image compression technology dynamically adjusts the compression ratio and quality of map images according to the screen resolution, processing power, and communication environment of the receiving terminal device. For example, it efficiently transmits necessary and sufficient information by sending detailed map images to high-resolution smartphones and simplified map images to low-resolution smartwatches. This technology can reduce the amount of communication data by up to 70% while maintaining visibility. Differential delivery technology extracts only the differences between the previously transmitted map information and the current map information, and transmits only the changed parts, thereby significantly reducing communication volume. This method is particularly effective when the monitored person is moving within the same area, and improves communication efficiency by avoiding duplicate transmission of background maps. Precaching technology analyzes the past behavior patterns of the monitored person and pre-downloads map tiles of frequently visited locations to the terminal device. This reduces the amount of communication when a notification is actually needed and improves display speed. Machine learning algorithms are used to predict the movement of the monitored person and predictively cache map data that is likely to be needed. For example, a screen like the one in Figure 9 will be displayed. Regarding compression algorithms, in addition to the conventional JPEG and PNG formats, the latest image compression formats such as WebP, AVIF, and JXL (JPEG XL) will be adopted. These next-generation compression formats achieve a 30-50% improvement in compression ratio compared to conventional formats while minimizing degradation of image quality. Furthermore, for geographical location information, algorithms such as Google Polyline encoding and simplify-js will be used to efficiently compress travel path data and reduce the amount of data transmitted.

[0082] This disclosure displays map information within the notification area of ​​push notifications, eliminating the need for application startup and significantly reducing the processing load on the terminal device 200. Specifically, the effects of reducing the processing load include: reducing memory usage by approximately 50-100MB during application startup; reducing CPU usage by approximately 30-40% due to reduced application startup and rendering processes; reducing battery consumption by approximately 20-25% due to reduced processing load; and shortening the response time, reducing the time until notification display from approximately 10-15 seconds to 2-3 seconds.

[0083] This disclosure employs advanced technologies to improve real-time performance.

[0084] This disclosure enables bidirectional real-time data updates via WebSocket communication. Server-Sent Events enable immediate notifications via server push. Edge Computing reduces latency by processing on geographically closer edge servers. CDN optimization minimizes delays in map tile delivery.

[0085] This disclosure aims to improve the efficiency of the data structure in order to efficiently manage location data.

[0086] This disclosure employs time-series databases such as InfluxDB and TimescaleDB as time-series databases to achieve high-speed search and aggregation of location information, as well as automatic data compression and retention period management. It also employs spatial indexes such as R-tree and Quadtree to achieve high-speed determination of the monitoring range (O(log n) computation time) and accelerate neighbor and range searches. For data compression, it implements path compression using the Douglas-Peucker algorithm, data size reduction through Polyline encoding, and removal and normalization of redundant data.

[0087] This disclosure describes the implementation of optimization techniques to improve the efficiency of AI processing.

[0088] This disclosure integrates the latest distributed processing and hardware optimization technologies to improve the efficiency of AI processing. In distributed machine learning, a distributed processing environment based on Apache Spark MLlib is built to process large location data sets in parallel across multiple nodes. This can reduce training time by up to 90% compared to conventional single-server processing. In distributed deep learning using TensorFlow Distributed, neural network training is distributed across multiple GPU servers to accelerate the training of complex models such as LSTM and CNN. In distributed reinforcement learning utilizing the Ray framework, dynamic adjustment of monitoring ranges and optimization of anomaly detection thresholds are performed in real time. For GPU acceleration technology, the NVIDIA CUDA architecture is made to its fullest extent to accelerate matrix and tensor operations through parallel processing. The cuDNN library is used to optimize the training and inference of convolutional neural networks (CNNs) and recurrent neural networks (RNNs). In the inference engine using TensorRT, the inference processing of trained models is optimized to minimize latency. Model optimization techniques involve converting 32-bit floating-point numbers to 16-bit (FP16) or 8-bit integers (INT8) through quantization, reducing memory usage and computational complexity. Pruning techniques remove less important parameters from trained models, reducing model size by 50-80%. Knowledge distillation techniques transfer knowledge from large training models to smaller student models, improving inference speed. Edge AI technology involves developing lightweight AI models that run on smartphones and GPS terminals, providing continuous monitoring even in environments with unstable network connections. Efficient AI inference on mobile devices is achieved by utilizing TensorFlow Lite, PyTorch Mobile, ONNX runtime, etc. User interface improvements

[0089] This disclosure contributes to improving the user interface.

[0090] This disclosure implements a function to prevent user errors.

[0091] This disclosure achieves a design that ensures a minimum touch area of ​​44x44 pixels through a large touch target, sufficient spacing between adjacent elements, and reliable operation even in emergencies. Visual feedback provides visual and haptic feedback during touch, clear progress display during processing, and clear display of completion and error status. Operation confirmation functions provide confirmation dialogs for important operations, a function to undo incorrect operations, and a function to display and restore operation history.

[0092] This invention implements a function to improve visibility.

[0093] This disclosure provides adaptive display that automatically adjusts brightness according to ambient light, adjusts color according to outdoor and indoor environments, and automatically adjusts font size according to age. High-contrast display ensures a contrast ratio compliant with WCAG 2.1 AA, optimizes background and text colors, and highlights important information. Multiple display modes are available, including normal mode, high-contrast mode, large font mode, simple mode, and mode for colorblind users.

[0094] This disclosure implements a function to reduce input load.

[0095] This disclosure enables voice control through voice recognition ("Show current location," "Emergency contact," etc.), multilingual support (Japanese, English, Chinese, Korean, etc.), and high-precision recognition in noisy environments. Gesture control enables intuitive gestures (pinch, swipe, double tap, etc.), customizable gesture settings, and a design optimized for one-handed operation. Predictive input enables predictions based on past operation history, presentation of operation candidates according to context, and accuracy improvement through learning functions.

[0096] This disclosure implements features to improve accessibility.

[0097] This disclosure enables VoiceOver (iOS) and TalkBack (Android) support through screen reader compatibility, voice explanations of map information, and voice guidance for operating procedures. Alternative input methods enable operation via external switches, eye-tracking, and breath / inhalation. Multisensory feedback enables tactile feedback (vibration patterns), auditory feedback (voices and sound effects), and visual feedback (colors and animations).

[0098] The following are specific implementation examples and numerical examples of this disclosure.

[0099] Let's explain the accuracy of location information.

[0100] The GPS terminal device 100 employs an advanced position measurement system that integrates multiple positioning technologies to provide highly accurate location information under various environmental conditions. Positioning accuracy in outdoor environments varies greatly depending on environmental conditions. In open spaces (coastlines, plains, large parks, etc.), it can receive signals directly from GPS satellites, achieving highly accurate positioning within ±2-3m with a 95% probability. This accuracy is sufficient for detailed location tracking of children and setting up monitoring in small areas. In urban areas, positioning accuracy decreases due to signal obstruction and reflection by tall buildings, but it still maintains an accuracy of within ±5-8m with a 95% probability. To mitigate multipath effects, multiple GPS satellite signals are integrated and processed to calculate highly reliable location information. In mountainous and forested areas, the accuracy is within ±10-15m due to signal obstruction by terrain and vegetation, but this is sufficient accuracy for general monitoring purposes. In indoor environments, GPS signal attenuation makes independent positioning difficult, so auxiliary positioning technology is utilized. By using Wi-Fi positioning in conjunction with GPS, accuracy within ±8-12m is achieved with 80% probability even indoors. A Wi-Fi access point location database is utilized, and position estimation is performed using triangulation based on signal strength from multiple access points. Bluetooth positioning utilizes Bluetooth Low Energy (BLE) beacons to achieve accuracy within ±5-10m with 70% probability. Combined positioning, which combines multiple positioning technologies, leverages the strengths of GPS, Wi-Fi, and Bluetooth to achieve accuracy within ±3-8m with 85% probability. In movement measurement, movement speed is measured with an accuracy of ±0.3km / h or less, allowing detection of even minute changes in walking speed. Movement direction is measured with an accuracy of ±3 degrees or less, accurately understanding the movement tendencies of the monitored subject. Movement distance is measured with an accuracy of ±5% or less for movements of 100m or more, providing sufficient reliability for analyzing everyday movement patterns.

[0101] Let me explain the communication performance.

[0102] For data transmission, the location information transmission interval is set to 10 minutes under normal circumstances, 1 minute in case of deviation, and 30 seconds in emergency situations. The amount of communication data is set to approximately 150-200 bytes per location information transmission, approximately 300-500KB (320x240 pixels) for map information, and approximately 1-2KB (including metadata) for push notifications. Regarding communication delay, the average time from location information acquisition to server reception is 1.5 seconds, the average server processing time is 0.8 seconds, the average push notification delivery time is 2.2 seconds, and the average overall response time is 4.5 seconds. Regarding communication success rates, the system achieves over 99.5% for LTE communication, over 98.8% for Wi-Fi communication, and over 99.8% for combined communication.

[0103] Let's discuss battery performance.

[0104] The GPS terminal device 100 has a battery performance that provides continuous operation time of 7-10 days in normal use mode, 14-21 days in power saving mode, and 3-5 days in emergency mode. Charging performance includes a charging time of approximately 1.5-2 hours (full charge), a battery capacity of 800-1200mAh, and approximately 500 charging cycles (maintaining 80% capacity). Power consumption is 25-35mA during GPS positioning, 50-80mA during communication, 3-8mA in standby mode, and 0.5-1.5mA during sleep mode.

[0105] This explains AI learning performance.

[0106] The AI ​​learning function employs a step-by-step learning approach to effectively learn the behavioral patterns of the monitored subjects, achieving highly accurate predictions and anomaly detection. The data collection period for learning is set considering the balance between system accuracy and practicality. The minimum learning period is one week, during which approximately 1,000 data points (location information, time, movement patterns, etc.) are collected. After one week of learning, basic daily behavioral patterns (weekday commutes, weekend outings, etc.) can be grasped, but the prediction accuracy at this stage is only about 75-80%. The recommended learning period is one month, and by collecting approximately 4,000 data points, it becomes possible to learn more stable behavioral patterns. After one month of learning, it becomes possible to capture behavioral patterns by day of the week, activity trends by time of day, and the initial stages of seasonal fluctuations, improving prediction accuracy to 85-90%. The maximum learning period is one year, and by collecting approximately 50,000 data points, it is possible to comprehensively learn the influence of long-term behavioral changes, seasonality, and annual events. After one year of training, the system grasps the overall behavioral patterns of the monitored subject, enabling highly accurate predictions of over 95%. Training time varies significantly depending on the hardware and training algorithm used. Initial training using a CPU takes approximately 5-10 minutes, during which clustering, pattern recognition, and statistical analysis are performed. Using a GPU reduces this to approximately 1-2 minutes, and the parallel processing of the deep learning model significantly improves training efficiency. Additional training (incremental training) is completed in approximately 30 seconds to 1 minute, and continuous accuracy improvement is achieved by integrating new data points into the existing model. Different performance targets are set for prediction accuracy depending on the application. For monitoring range optimization, an accuracy of 90-95% is achieved, enabling range settings that fit the actual behavioral range of the monitored subject. For behavioral pattern prediction, an accuracy of 85-90% is achieved, allowing the next action of the monitored subject to be predicted with a high probability. In anomaly detection, a practical anomaly detection system is realized by suppressing the false positive rate (the rate at which normal behavior is mistakenly identified as abnormal) to 3-5% or less and the false negative rate (the rate at which abnormal behavior is mistakenly identified as normal) to 1-2% or less.

[0107] Let's explain the system's performance.

[0108] The overall system performance achieves a maximum of 100,000 simultaneous connections, a processing throughput of 10,000 requests / second, and an average response time of less than 50ms. In terms of availability, it achieves a system uptime of over 99.9%, a service continuity of over 99.95%, and an average failure recovery time of less than 5 minutes. For scalability, it offers linear scaling up to 10x horizontally, performance improvement up to 5x vertically, and support for up to 5 geographically distributed regions. Hardware and software detailed configuration

[0109] The detailed specifications of the GPS terminal device 100 are described below.

[0110] Figure 4 shows the detailed specifications of the GPS terminal device 100. The processor is an ARM Cortex-M4 @120MHz CPU, with 512KB SRAM, 2MB of flash memory, and a power consumption of less than 50mW (operating). The communication module supports LTE Cat.M1 / Cat.NB1, Wi-Fi 802.11b / g / n (2.4GHz), Bluetooth BLE 5.0, and GPS L1 / L5 dual-band. The sensors include a 3-axis accelerometer with ±16G, a 3-axis geomagnetic sensor with ±4900μT, a barometric pressure sensor with a range of 300-1100hPa, and a temperature sensor with a range of -40℃ to +85℃.

[0111] This section describes the detailed specifications of Server 300.

[0112] Figure 5 shows the detailed specifications of Server 300. The production server features two Intel Xeon Gold 6248R (3.0GHz, 24 cores) CPUs, 256GB DDR4-3200 ECC RAM, 10TB NVMe SSD (RAID 1+0) storage, and two 10Gbps Ethernet ports (redundant). The machine learning server features two AMD EPYC 7763 (2.45GHz, 64 cores) CPUs, four NVIDIA A100 (80GB) GPUs, 512GB DDR4-3200 ECC RAM, and 20TB NVMe SSD (RAID 1+0) storage. The database server features two Intel Xeon Platinum 8380 CPUs (2.30GHz, 40 cores each), 1TB of DDR4-3200 ECC RAM, 50TB of NVMe SSD storage (RAID 1+0), and two 25Gbps Ethernet ports (for redundancy).

[0113] This section describes the software architecture.

[0114] The microservice architecture implements Kong and Zuul as API Gateways, Auth0 and Keycloak for authentication and authorization, HAProxy and Nginx for load balancing, gRPC for inter-service communication, and REST APIs. Databases include PostgreSQL 13 or later as a relational DB, MongoDB 5.0 or later as a NoSQL DB, InfluxDB 2.0 or later as a time-series DB, and Redis 6.0 or later and Memcached as caching. Messaging uses Apache Kafka 2.8 or later as a message queue, Apache Pulsar for streaming, and Celery and RQ for asynchronous processing.

[0115] This section describes the development and operation environment.

[0116] The development environment supports Python 3.9 or later, Node.js 16 or later, Go 1.18 or later as programming languages, Django, FastAPI, Express.js, Gin as frameworks, React 18 or later, Vue.js 3 or later as UI frameworks, and React Native and Flutter for mobile. The production environment supports Docker 20.10 or later and Podman as containers, Kubernetes 1.24 or later for orchestration, Prometheus, Grafana, ELK Stack for monitoring, and GitLab CI, GitHub Actions, and Jenkins for CI / CD. For security, it supports SonarQube and CodeQL for static analysis, OWASP ZAP and Nessus for vulnerability scanning, Let's Encrypt and HashiCorp Vault for encryption, and OAuth 2.0 and OpenID Connect for authentication. Quality Assurance & Testing Details

[0117] To ensure the quality of this disclosure, comprehensive testing will be conducted.

[0118] Let's discuss functional testing.

[0119] Location acquisition tests include GPS positioning accuracy tests (outdoor, indoor, and while moving), communication quality tests (LTE, Wi-Fi, and Bluetooth), combined positioning tests (GPS + Wi-Fi + Bluetooth), and battery consumption tests (for each mode). Deviance detection tests include deviation detection accuracy tests in various monitoring ranges, false detection rate measurements, detection delay time measurements, and operation tests near boundaries. Notification tests include push notification delivery success rate tests, notification display quality tests (for each OS and device), notification delay time measurements, and notification content accuracy tests.

[0120] Let's discuss performance testing.

[0121] Load testing includes simultaneous connection tests (10,000, 100,000, 1,000,000), processing throughput tests (10,000, 100,000, 1,000,000 requests / second), memory usage tests, and CPU utilization tests. Stress testing includes limit performance tests (confirmation of operation under maximum load), failure operation tests, recovery time measurement tests, and data consistency tests. Performance testing includes response time measurement tests (average, maximum, 99th percentile), throughput measurement tests, concurrent execution performance tests, and memory leak detection tests.

[0122] Let me explain security testing.

[0123] Vulnerability testing includes vulnerability assessment using OWASP Top 10, vulnerability detection through static analysis, vulnerability detection through dynamic analysis, and assessment by a third-party organization. Penetration testing includes external attack resistance testing, internal threat countermeasures testing, social engineering countermeasures testing, and physical security testing. Encryption testing includes communication encryption strength testing, data encryption testing, key management testing, and certificate management testing.

[0124] Let's discuss usability testing.

[0125] Usability testing includes usability tests for each age group, emergency operation tests, accessibility tests, and multilingual support tests. Device compatibility testing includes operational tests for each version of iOS and Android, operational tests for each device manufacturer, display tests for different screen sizes, and operational tests for different communication environments. Combination of each embodiment

[0126] The elements of each of the above embodiments may be combined as appropriate. For example, by combining the automatic setting by the AI ​​function of the first embodiment, the large-scale management function of the second embodiment, and the IoT integration function of the third embodiment, a more advanced and comprehensive monitoring system can be constructed.

[0127] Let's look at some specific combination examples.

[0128] The combination of AI and large-scale management enables integrated learning of student behavior patterns across the entire school, enabling the detection of abnormal behavior in groups (bullying, accidents, etc.), setting optimal monitoring ranges at the class level, and implementing a hierarchical notification system for teachers and parents. The combination of AI and IoT integration enables comprehensive learning of lifestyle patterns, including information from in-home IoT devices, analysis of the relationship between health status and location information, behavioral prediction using environmental sensor data, and comprehensive monitoring from multiple devices. The combination of large-scale management and IoT integration enables integrated management of employee location information and office IoT devices in a company, enabling evacuation guidance systems during disasters, safety management systems in factories and construction sites, and patient and staff management systems in medical institutions. In the integration of all implementations, an optimized large-scale IoT-linked monitoring system is realized through AI learning, creating a comprehensive monitoring network that spans regions, companies, and homes, enabling wide-area evacuation and safety confirmation systems during disasters, and an integrated monitoring system for the elderly, disabled, and children.

[0129] (General tasks) One of the purposes of this disclosure is to quickly and efficiently grasp the situation of the person being monitored.

[0130] (Note 1) The issue addressed in Appendix 1 is one of the objectives of this disclosure: to enable rapid situation assessment when a monitored individual deviates from the established pattern. Note 1 states that the monitoring device according to this embodiment includes: a location information acquisition unit that acquires location information of the object to be monitored; a range setting unit that sets the monitoring range of the object to be monitored; a deviation detection unit that detects when the object to be monitored deviates from the monitoring range; a map information generation unit that generates map information including the current location and movement history of the object to be monitored when the deviation detection unit detects a deviation; and a notification control unit that displays the map information in the notification area of ​​a push notification. According to one aspect of this disclosure, when a monitored individual deviates from the system, map information including the current location and movement history can be immediately displayed within the notification area of ​​a push notification. This eliminates the need to launch an application, enabling rapid situation assessment. This allows for rapid situation assessment when a monitored individual deviates from the system.

[0131] (Note 2) The issue addressed in Appendix 2 is one of the objectives of this disclosure: to enable a more intuitive understanding of the situation by dynamically displaying the movement route of the person being monitored. In Appendix 2, the notification control unit provides the monitoring device described in Appendix 1, which dynamically tracks and displays the movement path of the monitored object within the notification area of ​​the push notification. This allows for a more intuitive understanding of the situation by dynamically displaying the movement path of the person being monitored.

[0132] (Note 3) The issue addressed in Appendix 3 is to provide a flexible display of movement history according to the user's choice, which is one of the purposes of this disclosure. Appendix 3 provides the monitoring device described in Appendix 1, wherein the map information generation unit generates, based on the user's selection, one of the following as the movement history: a movement route within the most recent predetermined time, a movement route within a predetermined distance, or a route deviating from a registered point. This allows for flexible display of movement history based on user selection.

[0133] (Note 4) The challenge addressed in Appendix 4 is to automatically optimize the monitoring range through AI learning, which is one of the objectives of this disclosure. Appendix 4 provides the monitoring device described in Appendix 1, further comprising: a learning unit that learns behavioral data including past location information, time spent, frequency of movement, day of the week, and time of day of the monitored subject and recognizes the behavioral pattern of the monitored subject; and an automatic setting unit that automatically sets the monitoring range based on the behavioral pattern recognized by the learning unit. This allows the monitoring area to be automatically optimized through AI learning.

[0134] (Note 5) The challenge addressed in Appendix 5 is to achieve more advanced monitoring by detecting abnormal behavior and providing background information, which is one of the objectives of this disclosure. Appendix 5 provides the monitoring device described in Appendix 1, further comprising: an abnormality detection unit that analyzes the frequency with which the monitored subject deviates from a specific location and detects it as abnormal behavior when it exceeds a predetermined threshold; and a facility information display unit that, when the abnormality detection unit detects abnormal behavior, acquires facility information in the vicinity of the location and displays it in the push notification or the map information. This enables more sophisticated monitoring by detecting abnormal behavior and providing background information.

[0135] (Note 6) The issue addressed in Appendix 6 is to provide a means of rapid communication in emergencies, which is one of the purposes of this disclosure. Appendix 6 provides the monitoring device described in Appendix 1, further comprising an emergency contact unit that directly communicates with emergency contacts from the notification area of ​​the push notification. This provides a means of rapid communication in emergencies.

[0136] (Note 7) The issue addressed in Appendix 7 is one of the purposes of this disclosure: to facilitate the understanding of travel routes through visual distinctions. Appendix 7 provides the monitoring device described in Appendix 1, wherein the notification control unit displays movement paths within the monitoring range in a first color and movement paths outside the monitoring range in a second color. This makes it easier to understand movement paths through visual distinction.

[0137] (Note 8) The issue addressed in Appendix 8 is to provide seamless integration from push notifications to applications, which is one of the objectives of this disclosure. Appendix 8 provides the monitoring device described in Appendix 1, wherein the notification control unit seamlessly transitions to an application that displays detailed movement information when the map in the notification area of ​​the push notification is clicked. This allows for seamless integration from push notifications to applications.

[0138] (Note 9) The issue addressed in Appendix 9 is one of the objectives of this disclosure: to improve monitoring accuracy by acquiring highly accurate location information. Appendix 9 provides the monitoring device described in Appendix 1, wherein the location information acquisition unit acquires location information by using Wi-Fi positioning, Bluetooth positioning, and cellular base station positioning in addition to signals from GPS satellites. This allows for improved monitoring accuracy through the acquisition of highly precise location information.

[0139] (Note 10) The challenge addressed in Appendix 10 is to achieve optimal monitoring through dynamic adjustment of the monitoring range, which is one of the objectives of this disclosure. Note 10 provides the monitoring device described in Note 1, wherein the range setting unit dynamically adjusts the monitoring range based on the time of day, day of the week, season, and weather. This allows for optimal monitoring through dynamic adjustment of the monitoring range.

[0140] (Note 11) The issue addressed in Appendix 11 is one of the objectives of this disclosure: to prevent false positives and improve the reliability of monitoring. Appendix 11 provides the monitoring device described in Appendix 1, wherein the deviation detection unit sets inner and outer buffer areas at the boundary of the monitoring range and detects a deviation when the vehicle remains outside the buffer area for a predetermined period of time. This prevents false detections and improves the reliability of monitoring.

[0141] (Note 12) The challenge addressed in Appendix 12 is to provide diverse displays using multiple map styles, which is one of the objectives of this disclosure. Appendix 12 provides the monitoring device described in Appendix 1, wherein the map information generation unit generates the map information by selecting one of a standard map, satellite image, or topographic map. This allows for diverse displays using multiple map styles.

[0142] (Note 13) The issue addressed in Appendix 13 is to achieve appropriate information provision through notification priority control, which is one of the purposes of this disclosure. Appendix 13 provides the monitoring device described in Appendix 1, wherein the notification control unit sets the notification priority according to the status of the deviation, immediately sends a notification in the case of a high-priority deviation, and delays the notification in the case of a low-priority deviation. This allows for the provision of appropriate information through notification priority control.

[0143] (Note 14) The challenge addressed in Appendix 14 is to achieve continuous performance improvement through machine learning, which is one of the objectives of this disclosure. Appendix 14 provides the monitoring device described in Appendix 4, wherein the learning unit continuously learns new behavioral data of the monitored subject and improves the accuracy of recognizing the behavioral patterns. This enables continuous performance improvement through machine learning.

[0144] (Note 15) The challenge addressed in Appendix 15 is to achieve highly accurate anomaly detection using statistical methods, which is one of the objectives of this disclosure. Appendix 15 provides the monitoring device described in Appendix 5, wherein the anomaly detection unit applies a statistical anomaly detection algorithm to the past deviation data of the monitored object and detects deviations from normal behavior patterns. This enables highly accurate anomaly detection using statistical methods.

[0145] (Note 16) The issue addressed in Appendix 16 is one of the purposes of this disclosure: to support understanding the situation by providing diverse facility information. Appendix 16 provides the monitoring device described in Appendix 5, which acquires and displays information on educational facilities, commercial facilities, entertainment facilities, transportation facilities, medical facilities, and public facilities located around the deviation point. This allows us to support understanding the situation by providing diverse facility information.

[0146] (Note 17) The issue addressed in Appendix 17 is to ensure high security through data encryption, which is one of the purposes of this disclosure. Appendix 17 provides the monitoring device described in Appendix 1, which has a security function that encrypts and stores the location information and map information, and also encrypts and transmits it during communication. This allows for higher security through data encryption.

[0147] (Note 18) The issue addressed in Appendix 18 is to improve convenience through synchronization across multiple devices, which is one of the objectives of this disclosure. Appendix 18 provides the monitoring device described in Appendix 1, wherein the notification control unit synchronously sends the push notification to multiple terminal devices and reflects the confirmation operation on one terminal device to the other terminal devices. This allows for improved convenience through synchronization across multiple devices.

[0148] (Note 19) The issue addressed in Appendix 19 is to provide diverse notification methods using voice and vibration, which is one of the objectives of this disclosure. Appendix 19 provides the monitoring device described in Appendix 1, wherein the notification control unit provides notifications in conjunction with the push notification by one or a combination of voice alerts, vibration patterns, and LED flashing. This makes it possible to provide a variety of notification methods using voice and vibration.

[0149] (Note 20) The challenge addressed in Appendix 20 is to ensure scalability through integration with cloud services, which is one of the objectives of this disclosure. Appendix 20 provides the monitoring device described in Appendix 1, which collaborates with an external cloud service to acquire weather information, traffic information, and disaster information and reflect them in the map information. This ensures scalability through integration with cloud services. [Explanation of Symbols]

[0150] 1. Monitoring System 10. Persons to be monitored 20 Guardian 100 GPS terminal devices 200 terminal devices 300 servers 301 CPU 302 GPU 303 memory 304 Storage 305 Network Interface 306 Power Supply Unit 307 Cooling System 310 Location information acquisition unit 320 Range setting section 330 Deviation detection unit 340 Map Information Generation Unit 350 Notification Control Unit 360 Learning Department 370 Automatic setting unit 380 Anomaly detection unit 390 Facility Information Display Section 391 Data Management Department 392 Security Management Department 393 Communication Control Unit 394 Log Management Department 395 Statistical Analysis Department 400 External Services 501 Operating Systems 502 Middleware 503 Application

Claims

1. A location information acquisition unit that acquires location information of the person being monitored, A range setting unit for setting the monitoring range of the object to be monitored, A deviation detection unit that detects when the monitored object deviates from the monitoring range, When the deviation detection unit detects a deviation, a map information generation unit generates map information including the current location and movement history of the monitored object. A notification control unit that displays the aforementioned map information within the notification area of ​​a push notification, A monitoring device equipped with the following features.

2. The monitoring device according to claim 1, wherein the notification control unit dynamically tracks and displays the movement path of the monitored object within the notification area of ​​the push notification.

3. The monitoring device according to claim 1, wherein the map information generation unit generates, as the movement history, one of the following based on the user's selection: a movement route within the most recent predetermined time, a movement route within a predetermined distance, or a route deviating from a registered point.

4. A learning unit learns behavioral data including past location information, time spent, frequency of movement, day of the week, and time of day of the monitored subject, and recognizes the behavioral patterns of the monitored subject. An automatic setting unit automatically sets the monitoring range based on the behavioral patterns recognized by the learning unit, The monitoring device according to claim 1, further comprising:

5. An anomaly detection unit analyzes the frequency with which the monitored subject deviates from a specific location and detects it as abnormal behavior when it exceeds a predetermined threshold. When the anomaly detection unit detects abnormal behavior, the facility information display unit acquires facility information in the vicinity of the location and displays it in the push notification or the map information. The monitoring device according to claim 1, further comprising:

6. The monitoring device according to claim 1, further comprising an emergency contact unit that performs communication to an emergency contact directly from the notification area of ​​the push notification.

7. The monitoring device according to claim 1, wherein the notification control unit displays movement paths within the monitoring range in a first color and movement paths outside the monitoring range in a second color.

8. The monitoring device according to claim 1, wherein the notification control unit seamlessly transitions to an application that displays detailed movement information when the map in the notification area of ​​the push notification is clicked.

9. The processor acquires the location information of the person being monitored. The processor sets the monitoring range for the object being monitored, The processor detects that the monitored object has deviated from the monitoring range. When the processor detects the deviation, it generates map information including the current location and movement history of the monitored object. The processor causes the map information to be displayed in the notification area of ​​the push notification. Methods of monitoring.

10. The processor is instructed to acquire the location information of the person being monitored. The processor is instructed to set the monitoring range for the object being monitored. The processor is instructed to detect when the monitored object deviates from the monitoring range. When the processor detects the deviation, it generates map information including the current location and movement history of the monitored object. The processor is instructed to display the map information within the notification area of ​​the push notification. A program that executes a process.

11. A GPS terminal device carried by the person being monitored, The terminal device used by the guardian, The system comprises the GPS terminal device and a server that communicates with the terminal device, The aforementioned server, Based on the location information received from the GPS terminal device, the system detects when the monitored object deviates from the monitoring range. When the aforementioned deviation is detected, map information including the current location and movement history of the monitored object is generated. The aforementioned terminal device is The map information received from the server is displayed in the notification area of ​​the push notification. A monitoring system.

12. The monitoring system according to claim 11, wherein the server learns the behavioral data of the person being monitored, recognizes behavioral patterns, and automatically sets the monitoring range based on the behavioral patterns.

13. The monitoring system according to claim 11, wherein the server cooperates with an external POI information service to acquire facility information around the deviation point of the monitored object and includes it in the map information.

14. The monitoring system according to claim 11, wherein the server compresses the image data of the map information and optimizes the distribution format according to the communication environment of the terminal device.

15. The monitoring system according to claim 11, wherein the communication between the GPS terminal device and the server, and the communication between the server and the terminal device, uses an encrypted communication protocol.

Images