Wearable and Wireless Brain-Computer Interface and Its Applications

Chin‐Teng Lin, Li‐Wei Ko, Che‐Jui Chang, Yu-Te Wang, Chia-Hsin Chung, Fu-Shu Yang, Jeng‐Ren Duann, Tzyy-Ping Jung, Jin‐Chern ChiouLecture Notes in Computer Science 2009 · DOI · OpenAlex

Introduction

Electroencephalogram (EEG) technology has long been the gold standard for non-invasive brain imaging, yet its practical application has been largely confined to laboratory settings. Traditional EEG setups suffer from two primary "usability killers": the requirement for conductive gels, which are messy and uncomfortable for long-term use, and the "tether" problem, where bulky amplifiers and thick wire bundles restrict user movement.

Researchers have addressed these challenges by developing the NCTU BCI-headband, a truly wearable, wireless brain-computer interface. By integrating dry MEMS electrodes and miniature data acquisition nodes with Bluetooth telemetry, the system enables real-time monitoring of cognitive states—such as drowsiness—in everyday environments using standard mobile devices.

System Architecture

The system is designed as a full-stack hardware and software solution that bridges the gap between clinical-grade data and real-world portability. The architecture consists of a front-end signal acquisition unit and a back-end processing unit connected via wireless transmission.

High-level system architecture showing the flow from signal acquisition to real-time processing and application.

Hardware Innovation: Dry MEMS and Daisy-Chaining

The core innovation of the BCI-headband lies in its modular hardware design. To eliminate the need for skin preparation and conductive paste, the team utilized Micro-Electro-Mechanical System (MEMS) electrodes. These sensors feature 400 "ganged" silicon-based spiked contacts capable of acquiring high-quality signals from non-hairy sites like the forehead.

To solve the wiring complexity, the researchers designed miniature 20mm x 18mm PCB nodes. These nodes are daisy-chained, sharing power and clock signals while passing digital data along a single serial line. This approach moves the Analog-to-Digital Converter (ADC) and pre-amplification directly to the electrode site, significantly reducing weight and electromagnetic interference.

Hardware components of the BCI-headband, including (A) the circuit block diagram, (B) a single acquisition node, (C) the Bluetooth module, and (D) the full daisy-chained assembly.

Validation and Performance

The system samples brain activity at 512Hz with 12-bit precision. Data is transmitted via Bluetooth to either a custom Digital Signal Processing (DSP) unit or a mobile phone application. The mobile app performs real-time Power Spectrum Density (PSD) estimation to detect the "alpha rhythm"—a key indicator of the transition from alertness to drowsiness.

Engineering optimizations led to a massive leap in power efficiency compared to previous prototypes. The following table highlights the improvements achieved in the headband design:

Feature NCTU BCI-cap (Previous) NCTU BCI-headband (New)
Weight 185 g < 100 g
Precision 8 bit 12 bit
Sampling Rate 200 Hz 512 Hz
Battery Life 3–4 hours 33–34 hours
Output Current 480 mA 31.58 mA

Discussion and Future Outlook

While the BCI-headband represents a significant step forward, limitations remain. The current dry MEMS electrodes are optimized for non-hairy sites, and the impact of motion artifacts during vigorous activity requires further study. However, the ability to monitor brain health for over 30 hours on a single charge using a mobile phone opens new doors for clinical neurology, psychiatry, and rehabilitation medicine outside of traditional hospital settings.

Summary updated

BibTeX
@inproceedings{lin2009wearable,
  author = {Chin‐Teng Lin and Li‐Wei Ko and Che‐Jui Chang and Yu-Te Wang and Chia-Hsin Chung and Fu-Shu Yang and Jeng‐Ren Duann and Tzyy-Ping Jung and Jin‐Chern Chiou},
  title = {{Wearable and Wireless Brain-Computer Interface and Its Applications}},
  booktitle = {Lecture Notes in Computer Science},
  year = {2009},
  doi = {10.1007/978-3-642-02812-0\_84},
}
Figure