PIRMedic: A Physics-based Fault Diagnosis for Passive Infra-Red (PIR) Sensors

📌 Summary

PIRMedic is a physics-based method to detect and diagnose faults in Passive Infra-Red (PIR) sensors used in smart building occupancy systems. It improves reliability by examining the physical principles (like the pyroelectric effect) instead of relying purely on data-driven inference.

🧠 Motivation

PIR sensors are integral to smart building systems due to their cost-effectiveness and privacy-preserving nature. However, they are susceptible to various failures, such as environmental degradation, incorrect installation, and component aging, leading to unreliable occupancy data. Traditional fault detection methods often rely on data-driven approaches, which may not effectively capture the underlying physical anomalies.

🔬 Key Contributions

Physics-based Fault Diagnosis: PIRMedic introduces a method that utilizes the physical characteristics of PIR sensors, specifically the pyroelectric effect, to detect and diagnose faults.
Edge-level Implementation: The approach captures voltage and current behaviors at the sensor hardware level, enabling real-time fault detection at the edge, reducing the need for extensive data transmission.
Improved Data Quality: By ensuring the reliability of occupancy data, PIRMedic enhances the performance of building automation systems, leading to better energy efficiency and occupant comfort.

📊 Key Results

PIRMedic achieved high accuracy in detecting and diagnosing various classes of PIR sensor faults. The approach was evaluated using real-world deployments in three distinct locations, demonstrating its effectiveness in diverse environments.

PIRMedic System Architecture — **Figure 1:** Internals of the PIR Sensor showing both Analog and Discrete Outputs.

Classification Accuracy — **Figure 5:** Workflow of PIRMedic.

Figure 6a: Normal Operation

Figure 6b: Short-circuit Fault

📚 Citation

@inproceedings{pirmedic2021,
  title={PIRMedic: A Physics-based Fault Diagnosis for Passive Infra-Red (PIR) Sensors},
  author={Kashinath, Ashish and Mohan, Sibin and Nambi, Akshay and Marathe, Sumukh},
  booktitle={Proceedings of the 8th ACM International Conference on Systems for Energy-Efficient Built Environments},
  year={2021},
  organization={ACM}
}