Comparative Study on Contactless Breath Monitoring Techniques

DOI : 10.17577/IJERTV11IS070147

Download Full-Text PDF Cite this Publication

Text Only Version

Comparative Study on Contactless Breath Monitoring Techniques

K Shanmukha Vamshi

Department of Electronics and Communication RV College of Engineering

Bengaluru, India

R Sindhu Rajendran,

Assistant professor

Department of Electronics and Communication RV College of Engineering

Bengaluru, India

AbstractBreath monitoring techniques involving no physical contact provide very high flexibility in healthcare systems and can be used in wide variety of applications. This paper compares different technologies that exist for monitoring the breath along with their functioning and applications. The comparisons include Passive Radar, FMCW Radar, UWB Radar, Camera and Infrared Imaging. Environmental factors largely affect the performance of contactless breath monitors. The impact of the environmental conditions on the accuracy is also discussed. The mentioned methods have accuracy varying between 8596% and depends on environmental conditions.

Keywords FMCW (Frequency Modulated Continuous Wave); CW (Continuous Wave); UWB (Ultra-Wide Band); Radar; Doppler; FFT (Fast Fourier Transform).


    Biological signals extracted from the human body gives us critical information such as heart rate, body temperature, respiration rate, etc. The need to measure respiratory variables throughout a range of applications is constantly expanding. Because it offers essential data, respiratory rate seems to be one of the most promising and measurable variables. Applications for breath monitoring range widely including sleep research, newborn death early detection, and patient monitoring. Different medical disorders including anxiety, lung illness, asthma, and pneumonia may all be identified just by looking at breathing rate.

    There are various of traditional methods used to monitor breathing, most of them require a physical contact. Airtight coats, piezoelectric belts, and inductance are a few of them. Such techniques, though efficient, cause discomfort and cannot be used in all scenarios.

    Systems for contactless health monitoring employ sensor(s) that are put close to the person but are not in direct touch with them. The objective is to offer wearable technology with enhanced comfort while maintaining its functionality. The setting wherein health monitoring devices are employed has an impact on their efficacy. Environments like Smart Built Environments (SBE), which offer data gathering and analysis services embedded into a control architecture, might be beneficial for contactless devices.

    The most popular types of sensors for contactless measurement are thermal, optical, and radar sensors [1]. Contactless sensors can now monitor respiration in a range of circumstances, especially with the recent advancement of the mobile environment. In other words, a range of medical conditions that might arise in normal living, such as sleep

    apnea and respiratory failure, can now be detected and managed using contactless sensors [2].

  2. RADAR FOR BREATH MONITORING Millimeter-wave Radar operating at high frequency have

    high accuracy with respect to monitoring breath. Millimeter-wave Radar is power efficient and small, making it very suitable for this application [3]. The Radar system typically contains a transmitter to emit radio waves, one or more receivers to capture the reflected radio waves from the target, mixer to down convert the received wave into an intermediate frequency and an ADC.

    Different signal processing methods can be applied to the ADC signal from Radar to detect respiratory rate. General schematic diagram for breath rate detection using Radar is shown in figure 1.

    Different types of Radar can be used for this application such as passive Radar, Doppler Radar, FMCW and CW Radar and UWB radar. A pulse, together with breathing (inspiration) and exhalation (expiration) stages, comprise a whole respiratory cycle [4]. Radar uses the information about chest wall or abdominal displacement during inhalation and expiration to determine the type of breathing and to detect breathing patterns.

    Fig. 1. Schematic Diagram of breath rate estrimation using radar system

    1. Passive radar

      Passive Radars do not comprise of a transmitter, instead they rely on secondary sources such as TV, radio, cellular or Wi-Fi signals for their operation. A passive radar typically consists of two channels, Reference and Surveillance channel. The original signal from the transmitter is received on the reference channel and the signal reflected from the target is received on the surveillance channel.

      The received signal can be divided into three components, a strong signal from the transmitter (DPI), signal reflected off large obstacles and signal from the target object. Certain signal processing techniques such as matched filtering and DPI cancellation can be applied to the radar signal to extract only the signal from the target [5]. Figure 2 shows the components involved in a bistatic Passive radar.

      Fig. 2. Passive bistatic radar

      The measurement of doppler shift caused by the movement of hand and body captured using a passive radar with Wi-Fi signal as the illuminator is discussed in [6]. The chest wall movement caused by respiration is very minute and hence the doppler shift produced is difficult to estimate using a passive Radar.

      One can extract doppler and range data from the radar from the reference and surveillance signal with the help of CAF mapping and batching process. A novel micro doppler-based method is presented in [7] based on the CAF mapping since the doppler resolution offered by the typical CAF mapping is insufficient to extract the entire breathing motion.

      Due to the little amount of chest movement, there are only slight doppler shifts that cannot alter the doppler bin. The pulse's shape changes somewhat with chest movement, and this variation may be exploited to acquire micro doppler data. [8]. Table 1 shows the comparative results of the Passive radar and chest belt signal with changing distance from [7].


      delay is proportional to beat frequency (fb) which can be used to estimate the distance of the target.

      Fig. 3. FMCW transmitter and receiver chirp

      The data acquired from the Radar will consist of noise pertaining to obstacles, respiration and heart signals. After removing all the beat frequencies pertaining to noise, the vital signals of respiration and heartbeat must be separated to extract the breathing signal.

      The vibration caused by breathing occurs between 0.1 and 0.6 Hz [9] and the vibrations from heartbeat around 0.8 and 4.0 Hz [10]. Both signals can be separated by applying bandpass centered at different frequencies. The process of acquiring breathing signal from the radar data can be done in few steps [11]

      • Performing range FFT on data acquired from the radar.

      • Extract phase information from range bin in the region of interest.

      • Apply bandpass filter on the phase signal to obtain only the breathing signal

        Table 2 shows the comparative results of the FMCW radar and respiration belt with breath rate in breaths per minute (bpm) from [11].

        Experiment no.

        FMCW radar [bpm]

        Respiratory belt, [bpm]

        Error [ %]












































        Correlation Coeff
















    2. FMCW radar

      Unlike passive Radar, Frequency Modulated Continuous Wave (FMCW) Radar consists of both transmitter and receiver. A linear chirp is produced by varying the frequency of the CW signal, either high to low or low to high, and then transmitted. The received signal which is reflected off the target is similar to the transmitted signal except for a time delay (td) as shown in Figure 3. Time

    3. UWB Radar

    High spatial resolution provided by UWB technology enables the detection of minute changes in the chest wall that occur while breathing. UWB radars may be utilized conveniently in complicated surroundings where interference by sensitive equipment can be an issue since the power density of the emitted pulses is dispersed over a very wide frequency range with values close to those of the ambient noise [12]. The time period of the pulses

    transmitted is very short in UWB radar. The radars can use two different gating techniques including sampling gate and range gate. Systems based on UWB include restrictions on the user's location and orientation as well as the coverage area [13].

    Fig. 4. Range gating UWB radar block diagram

    The radar signal transmitted from the antenna is reflected off the target and captured by the receiver antenna. The Repetition Rate generator's delayed output is utilized to power a pulse generator, whose output serves as the receiver's strobe signal. When the delay present in the strobe signal is equivalent to the round-trip time of the transmitted signal, the output signal is set on the movement of the chest wall. Hence displacement in the chest wall will produce equivalent change in the output signal [12]. For visual representation, the output signal can be further transformed from analogue to digital.

    Fig. 5. Output of UWB radar verses Piezoelectric respiratory belt

    The breath pattern generated by UWB radar in comparison with piezoelectric belt is show in figure 5 from [12]. The pattern was recorded when the subject was placed at 25 cm away from the radar. It is observed that the signal produced by the belt and radar have a phase difference of about 180 degrees. This shows us how the change in the radar output depends on the delay in the received output.


    In camera-based detection, respiratory signals are often retrieved from video by looking at variations in the chest movement. A general approach involves finding the difference image from two adjacent image frames and generate the breathing signal from the sum of all pixel values in the difference image [14]. This approach is most likely to capture any kind of changes and not only the chest movement. It also does not give any kind of distinction between inhaling and exhaling.

    Some other alternative solutions include using features and then tracking the features frame-by-frame to generate

    the breath pattern. Some challenges involved in this technique are that it is difficult to locate the features, and the chest movement does not exceed few millimeters. Hence it is difficult to track such minute movements and is not reliable [15].

    A monochrome camera can be used to construct a breath monitoring system. Firstly, a profile is created from the image as shown in figure 6 [16], which is projected onto the vertical axis. To remove the effect of any changes other that chest movement, a high pass filter is applied onto the profile (1D vector).

    Fig. 6. 1D vertical profilr from video frame

    The breathing pattern is obtained by correlating the profile of the current image frame with the previous one. The breathing signal should not be taken when the subject is under any bodily motion, hence an algorithm is used to make a binary decision, which indicates moving as true or stationary as false. Signal with both global motion and breathing movement is passed to the algorithm for it to make proper classification and take in only frames which are free of global motion for breath pattern generation [16].

    Varying light intensity with the chest movement can be used for breath monitoring, except it will have many dependency factors. The change in light intensity will greatly depend on the type of shirt the subject is wearing, like tightly worn or loosely worn. In the case of loosely worn clothes, the chest movement causes less movement of the clothes and therefore the change in intensity is minute and breathing signal is not accurate. The results obtained when the subject wears tight clothes are promising [17].

    The change in intensity for different scenarios was analyzed and a breath pattern is generated. The comparison of the breath pattern generated with a reference signal generated by a breath monitor placed at the nostril of the subject is shown in the figure 7 [17].

    Fig. 7. A) reference signal, B) reference signal after further processing, C) signal obtained from the proposed algorithm


    Emitted radiation energy from the objects can be measured using passive infrared detectors which can be used to measure different vital signals, including breathing signal. Moisture and expired air, while breathing, can be used for breathing measurements where the sensor placed facing the side view of the subjects face [18]. A time varying signal can be produced of the thermal intensity around the region of interest.

    The signal can be very noisy due to the variations in temperature around the region of interest. Hence signal received from the IR sensor is filtered using the moving- average filtering. Due to the low signal amplitude, an effective signal filtering technique is required to cancel out the constant noise and later a continuous wavelet transformation (CWT) can be applied. With a three-level decomposition, Debauchies function was used as it is stable and the vital signals are decomposed properly using it.

    The results obtained from the experiment conducted in [19] show that there is a varying thermal activity near the nasal region. It was observed that there was a variation of about 3 degrees Celsius between inhaling and exhaling in the region of interest which is depicted in figure 8.

    Fig. 8. Change in temperature while breathing captured by infrared detector

    The signal obtained can be further processed by applying continuous wavelet transformation. The results of the experiment conducted on several test subjects in summarized in table 3 [19].




    ROI mean

    temperature [°C]



    Pattern [%]












































This paper discussed about different techniques available for breath monitoring along with a basic approach for each of the technologies. Although each of the approaches have their own limitation, they can be used for breath monitoring under controlled conditions. Radar based approach seems to be more accurate and comparatively less prone to noise. The camera-based approach is very sensitive to lighting conditions.

The accuracy provided by the radar makes it suitable for applications where minute movements must be detected such as breathing of heartbeat. The accuracy of the all the devices compared depend on the distance of the target and is inversely proportional. Accuracy provided by passive radar ranges from 85.8 60.22 % for target distance varying from 20 cm to 100 cm respectively, accuracy obtained with FMCW radar is around 95 %. Camera based methods provide an accuracy of 90% which can be increased up to 95% post processing. Infrared imaging with a suitable ROI is accurate up to 96.43 %.


[1] Folke, M.; Cernerud, L.; Ekström, M.; Hök, B. Critical review of non-invasive respiratory monitoring in medical care. Med Biol.

Eng. Comput. 2003, 41, 377383

[2] Nam, Y.; Kim, Y.; Lee, J. Sleep monitoring based on a tri-axial accelerometer and a pressure sensor. Sensors 2016, 16, 750.

[3] Y. Almalioglu, M. Turan, C. X. Lu, N. Trigoni, and A. Markham, Milli-RIO: ego-motion estimation with low cost millimetre-wave radar, IEEE Sensors Journal, vol. 21, no. 3, pp. 33143323, 2021.

[4] Y. Lee, P. Pathirana, R. Evans, and C. Steinfort, Noncontact detection and analysis of respiratory function using microwave doppler radar, in Journal of Sensors, 2015, p. 13.

[5] Kuschel, Heiner, Diego Cristallini, and Karl Erik Olsen. "Tutorial: Passive radar tutorial." IEEE Aerospace and Electronic Systems Magazine 34.2 (2019): 2-19.

[6] M. C. Tang, F. K. Wang, and T. S. Horng, Human gesture sensor using ambient wireless signals based on passive radar technology, in Microwave Symposium (IMS), 2015 IEEE MTT-S International, May 2015, pp. 14.

[7] W. Li, B. Tan and R. J. Piechocki, "Non-contact breathing detection using passive radar," 2016 IEEE International Conference on Communications (ICC), 2016, pp. 1-6, doi: 10.1109/ICC.2016.7511389.

[8] Y. Yang, N. Tong, C. Feng, and S. He, Micro-doppler extraction of rotating targets based on doppler rate, in Signal Processing, Communications and Computing (ICSPCC), 2011 IEEE International Conference on, Sept 2011, pp. 14.

[9] I. Tomasic, N. Tomasic, R. Trobec, M. Krpan, and T. Kelava, Continuous remote monitoring of COPD patients-justification and explanation of the requirements and a survey of the available technologies, Medical, & Biological Engineering &Computing, vol. 56, no. 4, pp. 547569, 2018.

[10] U. Orhan and A. Aydin, Heart rate detection on single-arm ECG by using dual-median approach, Arabian Journal for Science and Engineering, vol. 45, no. 8, pp. 65736581, 2020.

[11] Wang, Qisong & Dong, Zhening & Liu, Dan & Cao, Tianao & Zhang, Meiyan & Liu, Runqiao & Zhong, Xiaocong & Sun, Jinwei. (2021). Frequency-Modulated Continuous Wave Radar Respiratory Pattern Detection Technology Based on Multifeature. Journal of Healthcare Engineering. 2021. 1-18. 10.1155/2021/9376662.

[12] P. Bernardi, R. Cicchetti, S. Pisa, E. Pittella, E. Piuzzi and O. Testa, "Design, Realization, and Test of a UWB Radar Sensor for Breath Activity Monitoring," in IEEE Sensors Journal, vol. 14, no. 2, pp. 584-596, Feb. 2014, doi: 10.1109/JSEN.2013.2285915.

[13] Buckingham, Nicole and Denis Graanin. Ultra-Wideband Radar Detection of Breathing Rate: A Comparative Evaluation. HEALTHINF (2021).

[14] K. Tan, R. Saatchi, H. Elphick, and D. Burke, Real-time vision based respiration monitoring system, in Communication Systems Networks and Digital Signal Processing (CSNDSP), 2010 7th International Symposium on, Newcastle upon Tyne, UK, July 2010, pp. 770 774.

[15] K. Nakajima, A. Osa, S. Kasaoka, K. Nakashima, T. Maekawa, T. Tamura, and H. Miike, Detection of physiological parameters without any physical constraints in bed using sequential image processing, Japanese Journal of Applied Physics, vol. 35, no. 2B,pp. L269L272, 1996.

[16] M. Bartula, T. Tigges and J. Muehlsteff, "Camera-based system for contactless monitoring of respiration," 2013 35th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 2013, pp. 2672-2675, doi: 10.1109/EMBC.2013.6610090.

[17] Massaroni, Carlo & Presti, Daniela & Formica, Domenico & Silvestri, Sergio & Schena, Emiliano. (2019). Non-Contact Monitoring of Breathing Pattern and Respiratory Rate via RGB Signal Measurement. Sensors. 19. 2758. 10.3390/s19122758.

[18] I. Pavlidis, J. Levine, and P. Baukol, Thermal imaging for anxiety detection, in Proc. 2000 IEEE Workshop in Computer Vision Beyond the Visible Spectrum: Methods and Applications, Hilton Head Island, South Carolina, June 16, 2000, pp. 104-109.

[19] Abbas, Abbas K. & Heiman, K. & Orlikowsky, Thorsten & Leonhardt, Steffen. (2009). Non-contact Respiration Monitoring based on Infrared Thermography Imaging.