Experimental demonstration of noncontact vital-sign measurement using pulse radar

2.1 Wall analysis and compensation

Considering wall effect which is the most serious issues of achieving noncontact sensing, the RF pulses propagated by the electromagnetic (EM) wave experiences different parts of medium including free air and wall which contributes unequal permittivity and permeability, respectively. Consequently, the velocity and degradation of RF probing pulse in each medium should be modeled in the radar system design. First of all, to further detect accurate physiological data and preliminary distance estimation for delay line design between the radar module and the investigated subject, the theoretically sensing distance ( d 𝑠𝑢𝑏𝑗𝑒𝑐𝑡 = C ⋅ t d / 2 ) should be corrected as follows:

d est. = t d 2 ⁢ ( v m ⁢ 1 + v m ⁢ 2 )(1)

where C and t d are the spreading EM speed and time interval, respectively. v m ⁢ 1 and v m ⁢ 2 are the factors of spreading velocity in each medium that need to establish into the system model estimation before radar module implementation. Additionally, the degradation factor with regard to wall parameters such as permittivity, permeability, and thickness is significant to affect vital-sign demodulation. Thus, these parameters are also incorporated into our model plan to simulate wall effects with different degradation for estimating radiated power because the transmitted energy of spectral emission mask should be limited to a signal level in compliance with effective isotropic radiated power (EIRP) density level of - 41.3 dBm, regulated by the United States Federal Communications Commission (FCC) for UWB applications [13]. The RF probing pulse degrade greatly while it penetrates obstacle. The attenuation loss can be given by [14, 15]

α = ω ⁢ ( μ ⁢ ε 2 ⁢ [ ω 2 ⁢ ε 2 + σ 2 ω ⁢ ε - 1 ] 1 2 )(2)

where ε and σ are the dielectric constant and conductivity of the medium. The attenuation loss is proportional to the modeled degradation factor. On the other hand, the free space loss can be calculated by the Friis transmission formula [16]

𝐹𝑆𝐿 = 20 ⁢ log ⁡ ( d ) + 20 ⁢ log ⁡ ( f ) + 20 ⁢ log ⁡ ( 4 ⁢ π C )(3)

where d and f are the distance of EM propagation and corresponding frequency, respectively. Therefore, considering the degradation of air and wall. For the firmware architecture, the main program comprises system initial, analog/digital transformation, and Bluetooth module. The detected vital-sign data is captured by the receiver of radar for preliminary analog processing with band-pass filtering of heartbeat and breathing bands, respectively. The analog signal is then fed into the MCU to execute analog to digital conversion, which is discussed in the following subsection. Section 3 discusses the effect of the proposed radar system with three test cases, where case I means that no obstacle is placed between the radar and subject. For case II, a cardboard with 3 centimeter thickness is placed between a transmitting/receiving antenna of the radar and subject. For case III, an artificial plywood with 3 centimeter thickness is placed between a transmitting/receiving antenna of the radar and subject.

OPEN IN VIEWER

Figure 2a illustrates the flowchart of data. If we select the signal monitoring, the vital-sign data could be divided into several channels for raw data, heartbeat, and breathing rate analyses. Moreover, the proposed algorithm is incorporated into the filter processing block to increase the quality of time-domain physiological data, thereby enabling Fast Fourier Transform (FFT) to observe the corresponding spectral responses on each channel. As a consequence, a short-time Fourier transform (STFT) with both the sinusoidal frequency and phase content can be also used to obtain a spectrogram with the variation of the physiological spectrum over time. The required data is monitored by the proposed pulse-based radar with digital signal processing mechanism for long-term monitoring. Specifically, all bandpass filters are constructed in direct form with difference equation, which can be implemented in a MCU unit. The filtered signals y ⁢ [ n ] can be expressed as

y ⁢ [ n ] = ∑ m = 0 M - 1 x ⁢ [ n - m ] ⋅ c ⁢ [ m ](4)

OPEN IN VIEWER

where c ⁢ [ m ] is the coefficient of the m th tap, M is the filter taps and x ⁢ [ m ] is the digital time domain samples. For the estimation of breathing rate, we use a bandpass filter with the Kaiser window and 60 taps ( M = 60). For the estimation of heart rate, we use a bandpass filter with the Hanning window and 62 taps ( M = 62). The corresponding frequency responses of these two bandpass filters are shown in Fig. 3. It can be found that the passband range for breathing rate is from 0.1 Hz to 0.8 Hz and the passband range for heart rate is from 1 Hz to 1.5 Hz. The out-of-band noise is reduced by more than - 20 dB for both filters. After filtering the data, the spectral response of the filtered signals are obtained by using FFT process, which is expressed as

Y ⁢ [ k ] = ∑ n = 0 N - 1 y ⁢ [ n ] ⋅ e - j ⁢ 2 ⁢ π ⁢ k ⁢ m N(5)

where Y ⁢ [ k ] denotes the k th frequency tones available in the signal and N is the FFT size. The frequencies where the maximum values of FFT output between 0.1 Hz to 0.8 Hz and 1 Hz to 1.5 Hz denote the desired breathing rate and heart rate, respectively.