Design and Development of a System for Biofeedback Experiments

Overview

This thesis describes the design and development of a low-cost, embedded system for biofeedback experiments. The system measures physiological parameters — heart rate and galvanic skin response (GSR) — using sensors connected to an Arduino Nano 33 BLE Sense board paired with a rechargeable battery module, making it fully stand-alone.

Collected data is transmitted in real time via Bluetooth Low Energy (BLE) to a custom mobile application built with the Flutter framework using the Dart programming language. The app displays live charts of the measured signals and periodically uploads data via HTTP to the Measurify IoT cloud framework, developed by the Elios Lab at the University of Genova, where it can be stored and further processed.

Hardware

The main components:

• Microcontroller — Arduino Nano 33 BLE Sense
• Battery module — LIR2450 Li-Ion rechargeable cell (3.6V), soldered onto the back of the board by Elios Lab
• Heart rate sensor — Pulse Sensor
• Skin conductance sensor — Grove GSR Sensor

The complete wiring diagram is shown below:

GSR SENSOR  ──► A2
PULSE SENSOR ──► A0
Both sensors ──► GND, +3.3V

The assembled device is housed in a repurposed pill box with cutouts for the board, the GSR chip, a reset button access hole, and cable pass-throughs. The lid allows battery replacement and system power control.

Software

Firmware (Arduino)

The firmware is developed in the Arduino IDE and follows a modular structure using header (.h) and source (.cpp) files. A central configuration file (config.h) controls which modules are compiled:

Pulse Sensor — reads the PPG signal from pin A0 using the PulseSensorPlayground library. The firmware detects beats, reads BPM estimates, and timestamps each beat. An adaptive threshold was implemented after testing to accommodate inter-individual variation.

GSR Sensor — reads 20 samples and averages them to reduce noise. Skin resistance is calculated using the formula from the sensor datasheet and converted to conductance (µS). A median filter removes outliers, and the signal is decomposed into:

• Tonic component (SCL) — slow baseline, computed via a first-order IIR low-pass filter
• Phasic component (SCR) — fast transient responses, obtained by subtracting the tonic from the raw signal

Heart Rate Variability (HRV) — computed from RR intervals using three time-domain metrics:

• SDRR — Standard Deviation of RR Intervals (ms)
• RMSSD — Root Mean Square of Successive RR Differences (ms)
• pNN50 — percentage of successive RR intervals differing by more than 50 ms

BLE — the Arduino acts as a BLE peripheral, advertising a Biofeedback-BLE Service with characteristics for each measured signal (Pulse Sensor, BPM, GSR, Tonic, Phasic, SDRR, RMSSD, pNN50) and a CTRL characteristic. Data sharing is triggered by the mobile app writing 0x01 to CTRL (streaming mode) and stopped with 0x02 (idle mode). Updates are rate-limited per characteristic using a macro-defined interval.

Mobile Application (Flutter)

The application is developed in Visual Studio Code using Flutter and Dart, and is built on top of the smart-collector app developed by Elios Lab, extended with the following new pages:

Page	Description
main.dart	Entry point; requests permissions and initiates BLE scanning
default.dart	Default parameters (auth tokens, service/characteristic UUIDs)
globals.dart	Global state and connection settings
configPage.dart	In-app settings editor
PeripheralDetailPage.dart	Connect/disconnect from selected device
startPage.dart	Browse BLE services and characteristics; start/stop data streaming; send data to Measurify
SelectedCharacteristicsPage.dart	Ordered list of characteristics selected for charting
chartCartesianPage.dart	Real-time scrolling chart (300-point window); landscape/portrait toggle; pause/resume

BLE connectivity uses the cross-platform Quick Blue plugin. Data is sent to Measurify in configurable batches via HTTP POST, with any remaining samples flushed on stop.

The Pulse Sensor chart displays the PPG waveform, current BPM (with animated heartbeat icon), and computed HRV metrics. The GSR chart overlays the raw conductance signal alongside its tonic and phasic components on a single graph.

Measurify

Measurify is an open-source RESTful cloud API developed by Elios Lab, University of Genova, used to store and visualize IoT measurements.

Data is uploaded as a JSON array of timestamped samples:

[
  { "timestamp": "1684833177652.00", "values": [0.1, -0.2] },
  { "timestamp": "1684833177752.00", "values": [0.7, -0.4] }
]

Quick Start

1. Set up the hardware

• Solder Dupont cables to the Arduino Nano 33 BLE Sense pins: A0, A2, GND, +3.3V
• Connect the Pulse Sensor cables (purple → A0, red → +3.3V, black → GND)
• Connect the GSR sensor cables (yellow → A2, red → +3.3V, black → GND)
• Insert the LIR2450 battery with the + pole facing outward
• House the assembled board in the enclosure and route cables through the provided openings

2. Flash the firmware

• Install the Arduino IDE
• Add the Arduino Mbed OS Nano Boards core via Tools → Board → Boards Manager
• Install the required libraries via Sketch → Include Library → Manage Libraries:
  • ArduinoBLE
  • PulseSensorPlayground
• Open biofeedback_main.ino, select the board (Arduino Nano 33 BLE Sense) and the correct port
• Click Upload and wait for Done uploading

3. Run the mobile application

• Install Visual Studio Code with the Flutter and Dart plugins
• Install the Flutter SDK
• Open the smart-collector project folder in VS Code
• Connect an Android device via USB and enable Developer Mode (Settings → About Phone → tap Build Number 7 times) and USB Debugging
• Select the device in the bottom-right of VS Code and run:

  flutter run

4. Collect data

1. Power on the Arduino (flip the switch on the battery module)
2. Open the app — ensure Bluetooth is enabled on the phone
3. Press startScan and select Arduino-BIOFEEDBACK from the list
4. On the PeripheralDetailPage, press connect, then Go to services page
5. On the startPage, select the characteristics of interest using the checkboxes
6. Press Obtain all data to begin streaming — the Arduino enters STREAMING mode
7. Press Selected to navigate to the chart list and tap any characteristic to view its live graph
8. Press Stop collecting to end the session — remaining data is flushed to Measurify

5. Visualize stored data

• Log in to the Measurify dashboard with the credentials provided by Elios Lab
• Navigate to Visualize Timeseries, select the Biofeedback measurement, and choose the number of samples to display
• The pulse signal appears in blue and the GSR signal in red; each can be toggled independently

Applying the sensors correctly

• Pulse Sensor — fix to the fingertip using the included Velcro strap; avoid overtightening (reduces blood flow) or too loose a fit (introduces noise)
• GSR fingertip pads — slide fully onto the fingers; positioning too close to the fingertip causes signal artefacts

Future Improvements

• Add Machine Learning models on the Measurify cloud to automatically detect emotional states or stress patterns from stored data
• Implement Fast Fourier Transform (FFT) for frequency-domain HRV analysis
• Redesign the enclosure for a more compact and functional usage
• Enrich the mobile app with additional UI features and guidance for clinical or self-monitoring use