A Method for Remotely Sensing Vital Signs of Human Subjects Outdoors

Chuantao Li, Fuming Chen, Jingxi Jin, Hao Lv, Sheng Li, Guohua Lü, Jianqi WangSensors 2015 · DOI

TL;DR

This paper presents a compact 24 GHz Doppler radar system and a novel signal processing approach to remotely detect human respiration signals, even in challenging outdoor environments with significant interference. This technology offers a promising solution for identifying survivors in disaster scenarios where traditional methods are insufficient.

What Problem Does This Paper Address?

In disaster situations like chemical or nuclear leaks, quickly identifying survivors is critical. While drones and smart vehicles can locate bodies, determining if someone is alive or dead is difficult. Existing methods like thermal cameras are expensive and prone to environmental interference. This paper tackles the challenge of reliably detecting vital signs, specifically respiration, from a distance in complex outdoor settings, where interference from moving objects and environmental factors can severely degrade signal quality.

Methods

The core of this work is a two-pronged approach:

  • Hardware: A compact and portable 24 GHz Doppler radar system is developed. This frequency was chosen for a balance between antenna size and directivity, allowing for a reasonable detection range and field of view. The system uses a single-channel design, with a local oscillator generating the transmit signal, a power amplifier, a receiving antenna, a low-noise amplifier, and a mixer to downconvert the received signal. An STM32F407 microcontroller handles analog-to-digital conversion and digital signal processing.
  • Signal Processing: To overcome the low Signal-to-Noise Ratio (SNR) and interference in outdoor environments, a multi-stage signal processing pipeline is employed:
  • Baseline and Clutter Reduction: A single-side slip averaging method is used to remove baseline drift and background noise.
  • Self-Correlation: This technique is applied to enhance the respiration signal and reduce colored Gaussian noise. It leverages the periodic nature of respiration to distinguish it from random noise.
  • Adaptive Line Enhancer (ALE): This method further reduces Gaussian noise by using a delayed version of the signal as a reference, effectively filtering out unwanted components while preserving the respiration signal. This is an adaptation of adaptive noise cancellation techniques, specifically designed to work without a separate reference sensor.

Datasets

Not discussed in the paper. The experiments were conducted with human volunteers in controlled indoor and simulated outdoor environments.

Key Insights and Ideas

  • Outdoor Respiration Detection is Challenging: The paper highlights that while indoor respiration detection with Doppler radar is relatively straightforward, outdoor environments introduce significant interference (wind, moving objects) that drastically reduces SNR and makes signal extraction difficult.
  • Multi-Stage Processing is Crucial: No single signal processing technique is sufficient for robust outdoor respiration detection. The combination of baseline removal, self-correlation, and ALE is shown to be effective in progressively improving the signal quality.
  • ALE Outperforms FIR for Noise Reduction: The study demonstrates that the Adaptive Line Enhancer (ALE) technique provides a significant SNR improvement (around 5 dB) compared to a standard Finite Impulse Response (FIR) filter when dealing with respiration signals in noisy environments.
  • Distance and Posture Matter: While the system can detect respiration at up to 7 meters, signal strength can vary with human body posture, with the radar facing the abdomen directly yielding a stronger signal.

Key Results

  • The developed 24 GHz Doppler radar system successfully detected human respiration signals at a distance of 7 meters outdoors, even in the presence of significant environmental interference.
  • The proposed signal processing method (baseline removal, self-correlation, ALE) achieved an average SNR improvement of 23.5 dB when a person was present in front of the radar.
  • Even when no person was present, the system showed an average SNR improvement of 17.9 dB, indicating its robustness against false positives.
  • When using ALE, the average SNR improvement was approximately 5 dB higher than using a FIR filter alone for respiration detection.
  • The authors conclude that when the SNR is greater than 10 dB after processing, the presence of vital signs can be reliably verified.

Limitations and Future Work

  • Limited Number of Volunteers: The experiments were conducted on only two male volunteers.
  • Simulated Outdoor Conditions: While outdoor experiments were performed, they were simulated, and real-world disaster scenarios might present even more complex interference.
  • Future Work: The authors plan to test the system in conjunction with smart vehicles equipped with cameras for large-area simulated target searching to further validate its reliability. They also aim to investigate its performance in more diverse and challenging environments.

Summary updated

Abstract

After chemical or nuclear leakage or explosions, finding survivors is a huge challenge. Although human bodies can be found by smart vehicles and drones equipped with cameras, it is difficult to verify if the person is alive or dead this way. This paper describes a continuous wave radar sensor for remotely sensing the vital signs of human subjects. Firstly, a compact and portable 24 GHz Doppler radar system is designed to conduct non-contact detection of respiration signal. Secondly, in order to improve the quality of the respiration signals, the self-correlation and adaptive line enhancer (ALE) methods are proposed to minimize the interferences of any moving objects around the human subject. Finally, the detection capabilities of the radar system and the signal processing method are verified through experiments which show that human respiration signals can be extracted when the subject is 7 m away outdoors. The method provided in this paper will be a promising way to search for human subjects outdoors.
BibTeX
@inproceedings{li2015method,
  author = {Chuantao Li and Fuming Chen and Jingxi Jin and Hao Lv and Sheng Li and Guohua Lü and Jianqi Wang},
  title = {{A Method for Remotely Sensing Vital Signs of Human Subjects Outdoors}},
  booktitle = {Sensors},
  year = {2015},
  doi = {10.3390/s150714830},
}
Figure