BioGAP-Ultra is an open-source, ultra-low-power platform for synchronized multimodal biosignal acquisition, wireless streaming, and on-device inference. Its stackable architecture combines a compact mainboard with application-specific sensing shields, enabling the same processing platform to be integrated into headbands, armbands, chestbands, and custom wearable devices.
The mainboard pairs a GAP9 parallel ultra-low-power processor for digital signal processing and neural-network inference with an nRF5340 for sensor control and Bluetooth Low Energy (BLE) connectivity. Available sensing modules support EEG, EMG, ECG, PPG, 3-axis acceleration, QVAR, and audio acquisition.
- Modular and extensible hardware: stackable sensing shields and standardized interfaces for rapid adaptation to new biosignals and wearable form factors.
- High-channel-count biopotential acquisition: 16 simultaneous 24-bit ExG channels per shield, configurable for EEG, EMG, or ECG measurements.
- Energy-efficient edge AI: GAP9 with a programmable 9-core RISC-V cluster and NE16 neural accelerator, delivering up to 15.6 GOPS for DSP and 32.2 GMAC/s for machine learning at 370 MHz.
- High-throughput wireless streaming: nRF5340 with BLE 5.4 and a measured application throughput of up to 1.4 Mbit/s.
- Large on-device memory: 512 Mbit GAP9 PSRAM, 512 Mbit non-volatile flash, and 128 Mbit nRF5340 PSRAM for model storage, intermediate results, and temporary stream buffering.
- Multimodal synchronization: shared timing and firmware-level synchronization across concurrently active sensing modules.
- Flexible expansion: two I2C buses, two SPI buses, MIPI CSI-2, PDM audio, QSPI, and GPIO interfaces are exposed through the board-to-board connectors.
- Compact mainboard: 15 mm x 25 mm after removal of the breakaway development section.
- Open research platform: hardware, firmware, documentation, and expansion-board templates are provided for reproduction and further development.
The platform was experimentally validated in three fully wearable form factors. Power values include continuous acquisition and BLE streaming; battery-life estimates use a 150 mAh LiPo battery.
| Form factor | Acquired signals | Streaming power | Battery life |
|---|---|---|---|
| EEG-PPG headband | 16-channel EEG, PPG, acceleration | 32.8 mW | 16.9 h |
| EMG sleeve | 16-channel EMG, acceleration | 26.7 mW | 20.8 h |
| ECG-PPG chestband | Single-channel ECG, PPG, acceleration | 9.3 mW | 59.7 h |
Two representative real-time processing pipelines demonstrate the platform's on-device capabilities:
| Application | Processor | Result | System power |
|---|---|---|---|
| ECG-PPG pulse-arrival-time estimation | nRF5340 | Real-time Pan-Tompkins-based processing | 8.6 mW |
| EMG-ACC reach-and-grasp phase classification | GAP9 | 79.9% +/- 5.7% accuracy; less than 5 ms per inference | 23.6 mW |
More details are available in the BioGAP-Ultra paper.
| Board | Description |
|---|---|
| Mainboard | Core processing, power-management, memory, and wireless-connectivity board integrating the GAP9, nRF5340, MAX77654 PMIC, accelerometer/QVAR sensor, and PDM microphone. |
| ExG shield | Stackable 16-channel biopotential acquisition board based on two ADS1298 AFEs. It supports active or passive electrodes, monopolar or bipolar montages, configurable gain, and single- or dual-supply assembly options. |
| EMG shield | Stackable 16-channel ADS1298-based board with interchangeable electrode-routing boards for fully differential, partially shared, or common-reference configurations. |
| PPG shield | Compact red/infrared optical sensing board designed for flexible placement, including integration into an earlobe clip. |
| Debug board | Exposes power rails, communication buses, GPIOs, status LEDs, buttons, and programming/debug interfaces for the nRF5340 and GAP9. |
| Template shield | Reference design with pre-routed connectors for developing custom BioGAP-compatible sensing and expansion boards. |
The hardware projects are included as Git submodules. Clone the repository recursively to retrieve them.
| Path | Contents |
|---|---|
Hardware/ |
Mainboard, ExG, EMG, PPG, debug-board, and template-shield hardware projects, included as Git submodules. |
Firmware/ |
nRF5340 firmware, build configuration, sensor drivers, protocol documentation, and utilities. |
Firmware/src_NRF/ |
Zephyr application for acquisition, power management, synchronization, command handling, and BLE streaming. |
Firmware/src_NRF/sensors/ |
Sensor modules for EEG, EMG, IMU, microphone, PPG, mmWave radar, and WULPUS integration. |
Documentation/firmware/ |
Firmware architecture, configuration, protocol, data-format, and setup guides. |
Documentation/ |
System-level documentation and images used by this README. |
Changelog.md |
Hardware and repository revision history. |
git clone --recurse-submodules --branch feature/FW-refactoring https://github.com/pulp-bio/BioGAP.git
cd BioGAPIf the repository has already been cloned without its submodules, initialize them with:
git submodule update --init --recursiveThe nRF5340 firmware uses nRF Connect SDK v2.6.1, Zephyr RTOS, and the custom SENSEI SDK. Follow the firmware getting-started guide to install the toolchain, select a sensing configuration, build the application, and flash the board.
The command-line build uses the custom BioGAP board target:
west build --build-dir <build-dir> <path-to-BioGAP>/Firmware/src_NRF \
--board nrf5340_senseiv1_cpuapp --pristine
west flash --build-dir <build-dir>For desktop acquisition, visualization, recording, and live inference workflows, use BioGUI. BLE commands and binary packet formats are documented in the BLE protocol and data-format reference.
- Firmware documentation index
- Getting started
- Firmware architecture
- Firmware configuration
- BLE command protocol
- BLE data formats
- Sensor modules
- Hardware changelog
If BioGAP contributes to your research, please cite:
@ARTICLE{Frey_2026_BioGAP_Ultra,
author={Frey, Sebastian and Spacone, Giusy and Cossettini, Andrea and Guermandi, Marco and Schilk, Philipp and Benini, Luca and Kartsch, Victor},
journal={IEEE Transactions on Biomedical Circuits and Systems},
title={BioGAP-Ultra: A Modular Edge-AI Platform for Wearable Multimodal Biosignal Acquisition and Processing},
year={2026},
volume={20},
number={3},
pages={399--415},
keywords={Electrocardiography; Biomedical monitoring; Monitoring; Electromyography; Electroencephalography; Artificial intelligence; Heart rate; Estimation; Temperature measurement; Hardware; Biopotential; ExG; photoplethysmogram; human-machine interface; sensor fusion},
doi={10.1109/TBCAS.2026.3652501}
}@INPROCEEDINGS{Frey_2023_BioGAP,
author={Frey, Sebastian and Guermandi, Marco and Benatti, Simone and Kartsch, Victor and Cossettini, Andrea and Benini, Luca},
booktitle={2023 IEEE International Conference on Omni-layer Intelligent Systems (COINS)},
title={BioGAP: a 10-Core FP-capable Ultra-Low Power IoT Processor, with Medical-Grade AFE and BLE Connectivity for Wearable Biosignal Processing},
year={2023},
volume={},
number={},
pages={1-7},
keywords={Wireless communication;6G mobile communication;Wireless sensor networks;Ultrasonic imaging;Wearable computers;Machine learning;Electroencephalography;wearable EEG;wearable healthcare;ultra-low-power design;embedded system},
doi={10.1109/COINS57856.2023.10189286}}- Frey, Sebastian, et al. "GAPses: Versatile smart glasses for comfortable and fully-dry acquisition and parallel ultra-low-power processing of EEG and EOG." IEEE Transactions on Biomedical Circuits and Systems 19.3 (2024): 616-628.
- Santos, Carlos, et al. "Real-time, single-ear, wearable ECG reconstruction, R-peak detection, and HR/HRV monitoring." 2025 47th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). IEEE, 2025.
- Orlandi, Mattia, et al. "Real-time motor unit tracking from sEMG signals with adaptive ICA on a parallel ultra-low-power processor." IEEE Transactions on Biomedical Circuits and Systems 18.4 (2024): 771-782.
- Frey, Sebastian, et al. "A wearable ultra-low-power sEMG-triggered ultrasound system for long-term muscle activity monitoring." 2023 IEEE International Ultrasonics Symposium (IUS). IEEE, 2023.
- Ingolfsson, Thorir Mar, et al. "A wearable ultra-low-power system for EEG-based speech-imagery interfaces." IEEE Transactions on Biomedical Circuits and Systems 19.4 (2025): 743-755.
- Mei, Lan, et al. "An ultra-low-power wearable BMI system with continual learning capabilities." IEEE Transactions on Biomedical Circuits and Systems 19.3 (2024): 511-522.
- Meier, Fiona, et al. "A parallel ultra-low-power silent speech interface based on a wearable, fully-dry EMG neckband." 2025 IEEE SENSORS. IEEE, 2025.
- Frey, Sebastian, et al. "Live Demonstration: Wearable Edge-AI Meets Real-Time Saccadic Eye Movement Classification." 2025 IEEE Biomedical Circuits and Systems Conference (BioCAS). IEEE, 2025, p. 539. https://doi.org/10.1109/BioCAS67066.2025.00126.
- Ingolfsson, Thorir Mar, et al. "VowelNet: Enhancing communication with wearable EEG-based vowel imagery." 2024 IEEE Biomedical Circuits and Systems Conference (BioCAS). IEEE, 2024.
- Spacone, Giusy, et al. "Wearable and ultra-low-power fusion of EMG and A-mode US for hand-wrist kinematic tracking." 2025 IEEE Biomedical Circuits and Systems Conference (BioCAS). IEEE, 2025.
- Frey, Sebastian, et al. "Wearable, real-time drowsiness detection based on EEG-PPG sensor fusion at the edge." 2024 IEEE Biomedical Circuits and Systems Conference (BioCAS). IEEE, 2024.
- Orlandi, Mattia, et al. "An adaptive dynamic mixing model for sEMG real-time ICA on an ultra-low-power processor." 2023 IEEE Biomedical Circuits and Systems Conference (BioCAS). IEEE, 2023.
BioGAP-Ultra was developed at the Integrated Systems Laboratory (IIS) at ETH Zurich by:
- Sebastian Frey - Hardware, firmware, software, and documentation
- Victor Kartsch - Software and conceptualization
- Giusy Spacone - Firmware
- Giovanni Pollo - Firmware
- Philipp Schilk - Firmware
- Marco Guermandi - Conceptualization
- Luca Benini - Supervision and conceptualization
- Andrea Cossettini - Supervision and conceptualization
- Hardware design files are released under the Solderpad Hardware License v0.51.
- Firmware in
Firmware/is released under the Apache License 2.0. - Images are released under the Creative Commons Attribution 4.0 International License.
Hardware submodules and third-party source files may include their own license terms. Consult the corresponding repositories and source headers before redistribution.