EDGELESS Health Care Assistant Use Case sensors data (simulated)

This repository contains nine synthetic datasets generated by a senior person home simulator designed for the Health Care Assistant Use Case in the EDGELESS project. The data represents IoT sensor readings from a single-resident home of an elderly person, covering activities of daily living and various anomalous behaviors. The simulator models a realistic daily routine that includes activities such as sleeping, eating, cooking, personal hygiene, watching TV, taking medication, and leaving the house. Repository Content Three CSV files are included, each covering a simulation of different periods of time, differing by the daily probability of anomalies: sensors-data-30d-anomalies-0.csv: 30 days of routine data with 0.0 anomaly probability (Normal routine). sensors-data-60d-anomalies-0.csv: 60 days of routine data with 0.0 anomaly probability (Normal routine). sensors-data-180d-anomalies-0.csv: 180 days of routine data with 0.0 anomaly probability (Normal routine). sensors-data-30d-anomalies-02.csv: 30 days of data with a 0.2 (20%) daily anomaly probability. sensors-data-60d-anomalies-02.csv: 60 days of data with a 0.2 (20%) daily anomaly probability. sensors-data-180d-anomalies-02.csv: 180 days of data with a 0.2 (20%) daily anomaly probability. sensors-data-30d-anomalies-05.csv: 30 days of data with a 0.5 (50%) daily anomaly probability. sensors-data-60d-anomalies-05.csv: 60 days of data with a 0.5 (50%) daily anomaly probability. sensors-data-180d-anomalies-05.csv: 180 days of data with a 0.5 (50%) daily anomaly probability. Simulated Activities The datasets include the following ground truth labels for activities (found in the activityTag column): GET_UP: The routine of waking up and getting out of bed in the morning. GO_TO_THE_BATHROOM: Visits to the bathroom, including specific routines like taking a shower. PREPARE_FOOD: Preparation of meals (Breakfast, Lunch, and Dinner) in the kitchen. EAT: Consumption of meals in the living room. TAKE_MEDICINES: The routine of taking medication, typically scheduled after meals. HOUSEHOLD_ACTIVITIES: Performing general household chores involving movement between different rooms (Living Room, Kitchen, Bedroom). WATCH_TV: Sedentary activity watching television in the living room. ACTIVITIES_OUTSIDE_THE_HOME: Periods where the resident leaves the apartment (e.g., for shopping or walking). SLEEP: Nighttime sleep routine. INACTIVITY: Periods of unusual inactivity (typically associated with anomalies). Simulated Anomalies The datasets with anomalies include labels for the following irregular situations (found in the anomalyTag column): EXCESSIVE_INACTIVITY: Unusual periods without movement detected in the house. NOT_COMING_HOME: The resident leaves the house and does not return within the expected timeframe. EXCESSIVE_SLEEPING: The resident stays in bed significantly longer than usual. EXCESSIVE_BATHROOM_VISITS_AT_NIGHT: High frequency of bathroom visits during sleeping hours. NOT_TAKING_MEDICINES: Omission of the medication routine after meals. Simulated Sensors The virtual environment includes the following devices: Motion Sensors (PIR): Located in Corridor, Bedroom, Living Room, Bathroom, and Kitchen. Contact Sensors (Magnetic): Entrance Door, Medicine Drawer (Bathroom), Cabinet and Drawers (Kitchen). Pressure Sensors: Bed (Bedroom), Chair and Couch (Living Room). Power Sensors: TV, Stove/Ceramic Hob, and Microwave. Data Format The files are in CSV format with the following columns: datetime:Timestamp of the event (YYYY-MM-DD HH:mm:ss). gatewayEui: IoT gateway identifier. deviceIndexId: Unique sensor identifier. room: Room where the sensor is located (e.g., KITCHEN, BEDROOM). place: Specific location within the room (e.g., FRIDGE, WALL, BED). value: Sensor reading (e.g., boolean for motion/contact, or watts for power). activityTag: Label of the activity being performed (Ground Truth). anomalyTag: Label of the anomaly present, if any (Ground Truth).