Federated Learning Framework for Real-Time Activity and Context Monitoring Using Edge Devices.

With the increasing need for effective elderly care solutions, this paper presents a novel federated learning-based system that uses smartphones as edge devices to monitor and enhance elderly care in real-time. In this system, elderly individuals carry smartphones equipped with Inertial Measurement Unit (IMU) sensors, including an accelerometer for activity recognition, a barometer for altitude detection, and a combination of the accelerometer, gyrometer, and magnetometer for location tracking. The smartphones continuously collect real-time data as the elderly individuals go about their daily routines. These data are processed locally on each device to train personalized models for activity recognition and contextual monitoring. The locally trained models are then sent to a federated server, where the FedAvg algorithm is used to aggregate model parameters, creating an improved global model. This aggregated model is subsequently distributed back to the smartphones, enhancing their activity recognition capabilities. In addition to model updates, information on the users' location, altitude, and context is sent to the server to enable the continuous monitoring and tracking of the elderly. By integrating activity recognition with location and altitude data, the system provides a comprehensive framework for tracking and supporting the well-being of elderly individuals across diverse environments. This approach offers a scalable and efficient solution for elderly care, contributing to enhanced safety and overall quality of life.