Sleep Monitor – Real‑Time Sleep Activity Tracking

A complete sleep‑monitoring pipeline built around a Raspberry Pi Pico W, using DIY pressure sensors. It uses a Wi‑Fi web server for data transfer, a Python script to download the data in CSV format, and MATLAB for visualization and analysis. Future goals include interpreting the data and developing a smart alarm that wakes the user during light‑sleep phases

---

📌 Project Overview

The system collects pressure deltas in four different positions on a mattress during sleep and stores it for later analysis.
The figure below summarizes the full workflow:

✨ Features

• Real‑time movement detection and event logging
• Low‑cost DIY pressure‑sensor array using velostat
• Embedded Wi‑Fi web server running on the Pico W for data transfer
• One‑click data download via Python script
• MATLAB tools for visualization and analysis
• Calibration system to normalize sensor sensitivity
• End‑to‑end workflow: hardware → firmware → data → analysis

💻 Software Requirements

• Arduino IDE 2.x + Raspberry Pi Pico Board 4.5.2 or later
• Python 3 with the requests package (pip install requests)
• MATLAB 2019a or later

🧩 Components

Raspberry Pi Pico W + Firmware (Arduino IDE C++)

• Sensor calibration
• Reads the pressure‑sensor matrix
• Detects and stores movement events
• Serves a Wi‑Fi web server for data transfer in CSV format

DIY Pressure Sensors

• Based on velostat, a pressure‑dependent resistive sheet material
• Each sensor is a 3cm × 3cm velostat square between two 2.5cm × 2.5cm aluminum‑foil electrodes
• Resistance varies with applied pressure
• Four sensors are placed under the mattress at torso height
• Although physically in a line, they are electrically connected in matrix form to reduce wiring

🔌 Circuit Assembly

• Each sensor is modeled as a pressure‑dependent resistor
• To build a sensor, cut a 3cm × 3cm square from a velostat sheet, and place it between two 2.5cm × 2.5cm aluminum‑foil electrodes
• Two external 1.5 kΩ resistors connected between GP26/27 and GND are needed to provide a correct reference
• Other values in the range 0.5kΩ to 5kΩ can be used: edit #define R0 1.5 in Sensors.h
• Working principle: SENSOR_LINE_1/2 are alternately driven to Vcc and GND, while measuring SENSOR_COL_A/B voltages
• The system of four equations and four unknowns (the sensor resistances) is solved iteratively in Sensors.cpp
• Optional: A switch allows to turn ON/OFF the Wi-Fi. You can also use the serial-monitor command W to toggle Wi-Fi state, indicated by the built-in LED

a) Sensor detail        b) Breadboard assembly

Sensors connected in matrix form

---

🚀 How to Run

Wi-Fi setup

• To set up Wi-Fi, edit your credentials char ssid[] and char password[] in WiFiControl.cpp. Alternatively, create a different .cpp file and place them there (no header file needed)
• When you power the Pico W, it will try to connect to your Wi-Fi. Using a serial monitor, you can view its progress
• If Wi-Fi successfully connects, the IP running the web server will be printed; For example: IP: 192.168.1.21
• The built‑in LED in the Pico W shows the Wi-Fi status

User Commands for Serial Monitor

• W - Toggle Wi-Fi state ON/OFF. The built-in LED switches according to Wi-Fi State
• P - Prints the sensor readings in real-time. Useful to view in graph format with the Arduino IDE Serial Plotter
• CS - Start the calibration process (instructions below)
• CE - End the calibration process (instructions below)
• E - Prints recorded events in raw MATLAB format: [time (seconds) , sensor ID (0 to 3) , pressure]

Calibration

• The code comes with a default calibration based on prescribed dimensions, so it's not mandatory to calibrate it, although it's useful to compensate individual differences between sensors
• The sensor array should be in its final position under the mattress
• Use CS to start the calibration process. If Wi-Fi is OFF, the built-in LED will turn ON
• Press and release the mattress in each of the 4 sensor positions in sequence
• Apply approximately the same pressure to all four sensors
• Use CE to end the calibration process

Note

• The calibration algorithm adjusts the sensitivity of each sensor to map the pressure applied during calibration to the value of 1000
• The calibration is stored in permanent memory, so you don't need to recalibrate every time you power the Pico W

Real-Time Sensor Visualization

• Using P in the serial monitor activates real‑time printing of sensor data
• Open the Arduino IDE serial plotter to view in graphical format
• Below is an example of the serial plotter

Serial plotter during calibration

---

📈 Data Transfer and Analysis Tools

1 - Downloading the Events

• Make sure the Pico W web server is ON by checking the LED status
• Make sure your Python has requests installed
• Run analysis/get_events.py to download the data and create events.csv

Note

• get_events.py searches for local web servers that respond to 192.168.1.x/events, with x from 1 to 64, as it is very likely to be in that range
• Whenever you turn ON the Wi-Fi, its IP is printed on the serial monitor
• If you know the IP, you can view the data in the browser by accessing http://IP/events

2 - Creating a .mat dataset

• Use MATLAB script analysis/create_dataset.m to convert raw CSV logs in events.csv to the .mat format
• Select a name by editing output_name = 'example_name.mat'
• Optionally, you can add timestamps by editing labels and labels_time

3 - Visualizing the data

• Running analysis/sleep_analysis.m will load the .mat file and generate:
  • Plot of pressure over time
  • Annotated events (if present in the file)
  • Histogram of pressure distribution

Note:

• If you don't have a dataset, you can test the visualization with dataset/synthetic_sleep.mat

Example Outputs

Events over time and pressure distribution

---

🚧 Future Work

• Basic sleep‑stage estimation from pressure patterns
• Smart alarm triggered during light‑sleep phases
• Higher‑resolution sensor array (if needed for the goals above)
• Temperature tracking during sleep
• Battery‑powered portable version