Bluetooth-Based Pulse Monitoring System

A two-node wireless heart rate monitoring system using Raspberry Pi and the MAX30102 pulse oximeter sensor. The sender node reads photoplethysmography (PPG) signals and streams real-time heart rate metrics over Bluetooth RFCOMM to a receiver node running a live GUI dashboard.

---

System Architecture

┌────────────────────────────┐        Bluetooth RFCOMM         ┌────────────────────────────┐
│     Raspberry Pi 1         │  ─────────────────────────────► │     Raspberry Pi 2         │
│     (Sender / Client)      │   MAC: 2C:CF:67:21:EA:19        │    (Receiver / Server)     │
│                            │         Port 3                  │                            │
│  MAX30102 ──► Signal       │                                 │   JSON Parse ──► GUI       │
│  Processing ──► JSON       │                                 │   Tkinter + Matplotlib     │
└────────────────────────────┘                                 └────────────────────────────┘

Pi 1 (Sender) reads 100 IR/Red samples per batch from the MAX30102 over I²C, checks signal quality, computes BPM/HRV metrics, and transmits a JSON packet over Bluetooth.

Pi 2 (Receiver) listens on an RFCOMM server socket, parses incoming packets, and renders real-time waveform plots and metric cards via Tkinter/Matplotlib.

---

Features

• Real-time BPM — peak detection with moving-average smoothing (window = 8) and rolling history buffer (window = 4 readings)
• HRV Metrics — RMSSD (Root Mean Square of Successive Differences) and HRSTD (Heart Rate Standard Deviation)
• Signal Quality Scoring — IR mean + variance checks gate processing; no-finger state resets display to zero
• Noise Rejection — range filter (40–180 BPM) + consistency filter (±20% median deviation)
• Auto-Reconnect — receiver automatically re-listens after a dropped connection
• Live GUI — animated IR, Red, and BPM plots; color-coded signal quality indicator

---

Hardware

Component	Quantity	Notes
Raspberry Pi (any model with I²C + Bluetooth)	2	Tested on Pi 5; requires gpiozero/libgpiod
MAX30102 Pulse Oximeter Sensor	1	I²C address 0x57
Jumper wires	—	SDA, SCL, 3.3V, GND
Power supply	2	Standard 5V USB-C

Wiring (MAX30102 → Pi 1)

MAX30102 Pin	Raspberry Pi Pin
VIN	3.3V (Pin 1)
GND	GND (Pin 6)
SDA	GPIO 2 / SDA (Pin 3)
SCL	GPIO 3 / SCL (Pin 5)
INT	GPIO 4 (Pin 7)

---

Software Dependencies

Install on both Pis:

pip install -r requirements.txt

requirements.txt:

numpy
smbus2
gpiozero
pybluez
matplotlib

Tkinter ships with Raspbian by default. If missing:

sudo apt install python3-tk

Enable I²C on Pi 1:

sudo raspi-config  # Interface Options → I2C → Enable

---

Setup & Usage

1. Pair the Pis over Bluetooth

On Pi 2 (receiver), find its MAC address:

hciconfig

On Pi 1 (sender), pair and trust Pi 2:

bluetoothctl
> scan on
> pair <PI2_MAC>
> trust <PI2_MAC>

Update RECEIVER_ADDRESS in src/sender.py with Pi 2's MAC address.

2. Start the Receiver (Pi 2 first)

python3 src/receiver.py

The receiver will start a Bluetooth RFCOMM server on port 3 and open the GUI window.

3. Start the Sender (Pi 1)

python3 src/sender.py

The sender initializes the MAX30102 and connects to Pi 2. Place a fingertip on the sensor — readings appear on the GUI within ~4 seconds (one 100-sample batch at 25 Hz).

---

Signal Processing Pipeline

Raw PPG (IR + Red)
       │
       ▼
DC Removal  ──►  subtract mean to center signal
       │
       ▼
Inversion   ──►  multiply by -1 (convert absorption valleys → peaks)
       │
       ▼
Moving Average  ──►  window = 8 samples to reduce jitter
       │
       ▼
Peak Detection  ──►  adaptive threshold (30–60 amplitude)
       │
       ▼
Interval Filtering:
  Range Filter       ──►  reject intervals outside 40–180 BPM
  Consistency Filter ──►  reject intervals >20% from median
       │
       ▼
BPM + RMSSD + HRSTD  ──►  transmitted as JSON over Bluetooth

---

Data Packet Format

Each transmitted packet is a newline-delimited JSON object:

{
  "timestamp": "2025-06-01T14:23:01.123456",
  "metrics": {
    "bpm": 72,
    "ipm": 72,
    "hrstd": 2.41,
    "rmssd": 18.53,
    "spo2": null
  },
  "signal_quality": 84,
  "raw_buffers": {
    "ir": [131072, 131456, ...],
    "red": [98304, 98560, ...]
  }
}

A bpm: 0 / signal_quality: 0 packet signals no finger detected and resets the GUI display.

---

Project Structure

pulse-monitor/
├── README.md
├── requirements.txt
├── .gitignore
└── src/
    ├── sender.py      # Pi 1: MAX30102 driver + signal processing + BT sender
    └── receiver.py    # Pi 2: BT server + JSON parser + Tkinter/Matplotlib GUI

---

Known Issues / Future Work

• SpO2 not yet computed — raw IR/Red buffers are transmitted and available; the SpO2 calculation using the R-ratio method is stubbed out
• Single-channel audio — a natural extension is driving a haptic or audio actuator on Pi 2 based on BPM thresholds
• BLE migration — switching from Classic Bluetooth RFCOMM to BLE GATT would improve range and reduce pairing friction
• Wrist-form factor — current setup requires fingertip placement; a wrist-worn reflective PPG sensor would enable continuous monitoring

---

Acknowledgements

• MAX30102 base driver adapted from Doug Burrell's MAX30102 library
• Peak detection algorithm derived from Maxim Integrated's reference implementation
• gpiozero used in place of RPi.GPIO for Pi 5 compatibility

---

👤 Author

Arya Sureshbhai Patel