English

Design and Quantitative Evaluation of an Embedded EEG Instrumentation Platform for Real-Time SSVEP Decoding

Human-Computer Interaction 2026-03-12 v2 Systems and Control Systems and Control

Abstract

This paper presents an embedded EEG instrumentation platform for real-time steady-state visually evoked potential (SSVEP) decoding based on an ESP32-S3 microcontroller and an ADS1299 analog front end. The system performs 88-channel EEG acquisition, zero-phase bandpass filtering, and canonical correlation analysis entirely on-device, while supporting wireless communication and closed-loop operation without external computation. A central contribution is the quantitative characterization of the platform's measurement integrity. Reported results demonstrate a stable shorted-input noise floor (0.08 μVRMS\approx 0.08~\mu\text{V}_{\text{RMS}}), tightly bounded sampling jitter (0.56 μs0.56~\mu\text{s} standard deviation), and negligible long-term drift (<1 ppm< 1~\text{ppm}). Numerical fidelity analysis shows 100%100\% decision agreement between the mixed-precision embedded pipeline and a 6464-bit double-precision reference. Effective common-mode attenuation exceeded 112 dB112~\text{dB} under balanced conditions, with a localized 26.9 dB26.9~\text{dB} degradation observed under source-impedance mismatch. Closed-loop validation achieved 99.17%99.17\% online accuracy and an information transfer rate of 27.66 bits/min27.66~\text{bits/min}. These results position the proposed system as a quantitatively characterized embedded EEG measurement and processing platform for real-time SSVEP decoding.

Keywords

Cite

@article{arxiv.2601.01772,
  title  = {Design and Quantitative Evaluation of an Embedded EEG Instrumentation Platform for Real-Time SSVEP Decoding},
  author = {Manh-Dat Nguyen and Thomas Do and Nguyen Thanh Trung Le and Xuan-The Tran and Fred Chang and Chin-Teng Lin},
  journal= {arXiv preprint arXiv:2601.01772},
  year   = {2026}
}
R2 v1 2026-07-01T08:50:19.184Z