This repository contains the code, data, and notebooks used to analyze and diagnose faults in PIR sensors.
📓 src/ — Code used in the edge platform: *Arduino* and *Raspberry Pi*
| Folder |
Purpose |
edge_platform/arduino |
Data collection sketches used in the Arduino Microcontroller |
edge_platform/raspberrypi |
Data Analysis & Feature extraction python code used in the Raspberry Pi |
matlab |
MATLAB Visualization scripts for frequency visualization and K-S statistic computation |
📓 scripts/ — Helper scripts to plot data from different lens types and converting ADC values into voltages
| Folder |
Purpose |
plot_single_sensor_ADC_count.py |
Helper script to plot voltages from ADC values |
plot_single_sensor_lens_types.py |
Helper script to analyze nature of voltage waveforms from different lens types |
📊 dataset/ — Raw & Processed Data
| Path |
Description |
FaultDetection/faulty_vs_working.csv |
Time-domain of Aout signals from working and faulty sensors |
FaultDetectionFeatures/ |
Extracted FFT and FFT-based features for classifying PIR sensor health |
FineGrainedFaultAnalysis/ |
Time-domain of Aout signals from Class III faulty sensors |
FineGrainedFaultAnalysisFeatures/ |
FFT-based features for diagnosing specific Class III fault types |
deployments/ |
Data (raw and processed) from real-world deployments: Elevator, Lobby, Starbucks |
📓 notebooks/ — Jupyter Notebooks
| Folder |
Purpose |
FaultDetection/ |
Notebooks for detecting faulty vs. working sensors |
FineGrainedFaultAnalysis/ |
Notebooks for diagnosing Class III faults (paper, tape, dust) |
📓 docs/ — Back of the Envelope Calculations + Warm-up Analysis
| Folder |
Purpose |
back_of_envelope_analysis/ |
Data collected from sensors in the lab that were subjected to thermal stress |
warmup_analysis.md |
Notes containing observations from warm-up analysis in PIR sensors |
git clone https://github.com/synercys/PIRMedic.git- Flash the appropriate sketch on Arduino from
src/edge_platform/arduino depending on the number and type of PIR sensors
- Connect the Arduino to the Raspberry Pi using a serial cable and note the serial device
- e.g.,
/dev/cu.usbmodem1421
- Copy
src/edge_platform/arduino onto the Raspberry Pi using scp
- Run
main.py using python3 main.py.
main.py will read the data collected from the Arduino through the serial device and store the features in data/ directory on the Raspberry Pi
- For fault detection or fault diagnosis, when needed run the corresponding jupyter notebook from
notebooks/ on the Raspberry Pi
- The notebooks can also be run on a separate server, provided
data/ from the Raspberry Pi is copied to it
- ✅ Real-world PIR sensor fault data
- ⚙️ FFT and machine learning-based feature extraction
- 🔍 Fine-grained fault classification for Class III anomalies
- 📈 Includes deployment data from varied environments
Please read the paper from ACM TOSN and BuildSys for additional information