Design and Implementation of an IoT-Based Low-Power Wearable EEG Sensing System for Home-Based Sleep Monitoring.

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Sleep Science

Background:

• Long-term home sleep monitoring necessitates wearable devices balancing signal quality and power efficiency.
• Existing systems often struggle with power constraints, limiting continuous, high-fidelity data collection for sleep staging.

Purpose of the Study:

• To design and implement a low-noise, low-power wearable single-channel electroencephalography (EEG) system for automatic sleep staging.
• To ensure the system meets requirements for overnight monitoring with extended battery life.
• To develop an energy-efficient edge-cloud architecture for accurate sleep staging.

Main Methods:

• Integrated a TI ADS1298 analog front-end with an STM32F4 microcontroller for differential sampling and hardware filtering.
• Implemented a lightweight deep learning model (SleePyCo) on a cloud backend for edge-cloud collaborative execution.
• Validated the system on the ISRUC dataset and conducted field trials with 10 healthy subjects.

Main Results:

• Achieved average power consumption of ~150.85 mW, enabling >24.6 hours of operation on a 1000 mAh battery.
• Demonstrated an overall sleep staging accuracy of 79.3% ± 3.0% on the ISRUC dataset, with an F1-score of 88.3% for Deep Sleep (N3).
• Field trials showed >97% valid data rate and only 0.8% Bluetooth packet loss, confirming engineering stability.

Conclusions:

• The hardware-software co-designed system offers a robust and energy-efficient solution for wearable sleep monitoring.
• The proposed IoMT (Internet of Medical Things) sensing system is suitable for daily sleep health management.
• The edge-cloud collaborative approach effectively balances computational demands and power constraints for wearable EEG devices.