Digital signal processing method for real-time measurement of heart rate with cw doppler radar

Abstract

Recently, research about vital sign detection is being actively done with the emergence of health care. Among various measurement methods of the vital sign detection, a continuous-wave (CW) Doppler radar which uses Doppler-effect and measures cardiopulmonary activities, such as respiration, heartbeat, in non-contact and non-invasive manner has been studied for many decades. Developing methods which measure heart rate variability (HRV), physiological phenomenon about variation of the time interval between each heartbeat, is one of interesting topics in this research area. Because the HRV occurs in short time, measuring heart rate (HR) accurately in real-time is a key of the topic. In this thesis, to measure HR in real-time, a new digital signal processing technique, polyphase basis discrete cosine transform (PB-DCT), is explained. The fact that discrete cosine transform (DCT) gives frequency resolution which is twice as good as that of discrete Fourier transform (DFT) is proved mathematically and verified with simulations. Thanks to this fact, DCT can give more accurate HR than DFT can do, and it is also verified with simulations. When a signal which has random phase is transformed with general DCT, Type II DCT, value of magnitude of frequency spectrum at the frequency of the signal, is not always maximum, and the spectrum is distorted. The cause of this phenomenon is established in this thesis, and to solve the problem, PB-DCT is proposed. Finally, HR which is measured by existing method, Complex Signal Demodulation (CSD) and DFT, and proposed method, PB-DCT, is compared with the HR from an electrocardiogram (ECG) equipment. Over 95 % of all measurement cases showed that the HR values from the proposed method is more accurate than those from the existing method. Also, proposed method showed that it is stronger at undesired peak errors, and is more robust at shorter length of time window than the existing method, so it has been verified that the proposed method is suit for real-time measurement of the HR.