Data Distribution Dynamics in Real-World WiFi-Based Patient Activity Monitoring for Home Healthcare

TL;DR

The paper investigates WiFi-based activity recognition in real-world home healthcare, highlighting substantial dataset shifts when moving from lab to living environments. It introduces AURA, a WiFi-sensing framework integrated with voice assistants, and presents a large, paired RGB/CSI dataset collected across eight environments with 16 participants. A compact DNN with GRL and few-shot adaptation is used to study cross-domain performance, revealing that environment-driven drift dominates and that domain adaptation partially mitigates but does not fully bridge the gap to unseen environments. The work demonstrates practical challenges and provides directions for robust, context-aware WiFi sensing to support elderly care, suggesting hypernetwork-based strategies for better adaptation in real-world settings.

Abstract

This paper examines the application of WiFi signals for real-world monitoring of daily activities in home healthcare scenarios. While the state-of-the-art of WiFi-based activity recognition is promising in lab environments, challenges arise in real-world settings due to environmental, subject, and system configuration variables, affecting accuracy and adaptability. The research involved deploying systems in various settings and analyzing data shifts. It aims to guide realistic development of robust, context-aware WiFi sensing systems for elderly care. The findings suggest a shift in WiFi-based activity sensing, bridging the gap between academic research and practical applications, enhancing life quality through technology.

Figures (12)

• Figure 1: User's activity-aware conversational assistant.
• Figure 2: Architecture of AURA.
• Figure 3: Processed WiFi CSI for (a) no activity, (b) fall, (c) jogging, and (d) squat.
• Figure 4: Data shift due to environment change.
• Figure 5: Data shift due to activity from different people.