A Multiple Biomedical Signals Synchronous Acquisition Circuit Based on Over-Sampling and Shaped Signal for the Application of the Ubiquitous Health Care

Abstract

Researches on the real-time health system have attracted much attention during recent years. The dynamic detection of health information requires the biomedical amplifier with small volume, low power consumption, multi-function, and high robustness to noise. In this paper, a novel biomedical signal acquisition circuit based on over-sampling and shaped signal technology is proposed, which is able to detect multiple biomedical signals. A novel algorithm based on the shaped signal has been investigated to extract multiple biomedical signals synchronously without complex signal processing scheme. Specifically, electrocardiograph, respiratory, and standard lead dropping signals are detected synchronously by the circuit. Experiments show that the circuit achieves dynamic monitoring of multiple biomedical signals, good manufacturability, low requirement to the device performance and high-noise suppression. The circuit is highly suitable for the biomedical signals as well as other weak signals monitoring.

1 Introduction

Biomedical signals of human body mainly include electrocardiograph (ECG), electroencephalogram (EEG), electromyography (EMG), electrogastrogram (EGG), electrooculogram (EOG), respiratory signal and so on, which are extremely weak with low frequency. Thus, the measurement of one bioelectrical signal will be significantly influenced by other signals [1, 13, 23]. The human body physiological parameters are expected to attract further attention by health care providers and high-performance services. With the development of dynamic body sensor networks, the sensors and actuators around the human body are required to display multi-function and high-performance behavior [9].

Most PC-based monitors are characterized by huge volume, comprehensive function, and high cost, which can only be afforded in hospital. Most portable dynamic monitors can be implemented in daily life but have limited function [4, 14, 24, 25]. Over the last decade, there have been many reports on related issues such as multiple biomedical signal synchronous acquisition circuit. Corbishley and Rodriguez-Villegas [5] chose acoustic sensing for the detector, and proposed an automated algorithm for detecting breathing and its cessation. Kim et al. [11] designed a simple electronic circuit which consists of respiratory motion and ECG detection. The respiratory motion was measured by constantan linear strain gages attached to elasticized belts which can be wrapped on chest or abdomen depending on patients’ breathing status. Felblinger et al. [6] developed an ECG amplifier, a 3D magnetic field sensor in microelectronic technology, and the associated motion sensors to acquire respiration displacement during an MRI exam. However, they are all able to record the ECG and respiratory signal, respectively, instead of in an integrated circuit.

A microcontroller with some vital signal detection circuits as well as the bulky and expensive monitor screen is the traditional architecture of the biomedical signal monitor. In this paper, according to the over-sampling and shaped signal technology [18], the bioelectric amplifier [16, 17], and the removal method for the interference [15], a novel acquisition system is investigated and subsequently implemented to detect the ECG and respiratory signals synchronously. The shaped signal is added to the preamplifier circuit to detect the respiratory signal and coupled with over-sampling technology to improve the system resolution. Firstly, the novel method based on the shaped signal to synchronously extract the bioelectrical and impedance signals is presented. Secondly, the synchronous detection circuit for the multiple biomedical signals is proposed. Lastly, the numerical calculation and human body test are implemented to detect the ECG, respiratory, and lead dropping signals.

2 Principle and Algorithm

2.1 Over-Sampling Technology

Over-sampling is to use $k$ times of the actual required sampling frequency, i.e., $F_{s} =kf_{s}.\,F_{s}$ is the over-sampling frequency, $f_{{s}}$ is the frequency according to the Nyquist sampling theorem, and $k$ is the over-sampling rate. After sampling, $k$ points should be accumulated to accomplish down-sampling. The signal-to-noise ratio (SNR) of the system can be improved after down-sampling. Based on theoretical derivation [18], the SNR can be improved by 6.02 dB with every four times increase of the sampling rate, which is equivalent to increase one bit resolution of the analog-to-digital-converter (ADC). Because over-sampling makes the bandwidth of the quantization noise increase $k$ times, and the amount of quantization noise is certain; the quantization noise after down-sampling in the effective frequency band reduces to $1/\sqrt{k}$ times of the origin, which shows that over-sampling can reduce the quantization noise of ADC and improve the system resolution. As a result, over-sampling technology can not only obtain a new ADC structure —$\sum -\Delta$ADC together with noise plastic and digital filtering technology, but also enhance the resolution of the existing N-bit ADC [2, 3, 7].

2.2 The Introduction of Shaped Signal

The shaped signal is a linear change, high frequency, and periodic signal. Triangle wave, sawtooth wave, and sine wave signal are often employed as shaped signal in actual applications. When the amplitude of the signal changes within a quantitative step height of ADC, the original signal information will be completely lost. Thus, the shaped signal is introduced to this system to ensure details of the signal that can be quantized when the pre-stage magnification is insufficient and resolution of ADC is low. In theory [19], the method of the superposition shaped signal and the over-sampling technology can improve the accuracy of ADC, and it can be employed to detect the direct current and weak alternating current signals. In this paper, by the method of adding shaped signal to the bioelectrical signal, we accomplish the accurate extraction of the ECG and respiratory signals.

2.3 The Theory of Respiratory Signal Detection and Novel Algorithm for Signal Separation

Human body chest is equivalent to a volume conductor, and the chest impedance constantly changes with respiratory [12]. When the current is injected into the chest, the respiratory signal can be recorded. The detection principle of respiratory signal can be equivalent to the circuit as shown in Fig. 1.

Fig. 1

Principle diagram for the detection of the respiratory signal

$U_{s}$ is the bioelectrical signal. Here, we take the ECG signal as one of the bioelectrical signal, it accesses to the input of the amplifier via chest impedance $R_{s}$ and contact impedance $R$ between skin and electrode, ${ U_{f}}$ is the shaped signal with the source resistance of ${ R_{f}}, and \,U_{o}$ is the output of amplifier A. According to linear superposition theorem, the input signal $U_{i}$ of A can be given by

$$\begin{aligned} U_i =\frac{R_f }{R_S +R+R_f }U_S +\frac{R+R_S }{R_S +R+R_f }U_f \end{aligned}$$

(1)

When designing this circuit, we maintain ${ R_{f}}$ is far greater than $R_{s }$ and $R$, (1) can be expressed by

$$\begin{aligned} U_i =\mathop U\nolimits _s +\frac{R+R_S }{ R_f }{ U_f} \end{aligned}$$

(2)

$R_{s}$ changes accordingly with breathing and is modulated by the shaped signal in (2). Generally, Wavelet transform and adaptive filtering are always employed to extract the ECG and respiratory signals [10, 20–23]. However, these algorithms exploit the respiratory-induced changes of the ECG to provide a surrogate respiratory signal, which contain the inherent uncertainty in measuring amplitude when ECG contains noise, and are complex when implemented in actual practice. In this paper, we present an easy to perform method to obtain the ECG and respiratory signal.

Triangle wave signal is used as the shaped signal with higher frequency than ECG and respiratory signal as shown in Fig. 2. According to (2), the ECG signal adds to the triangle wave signal, and the respiratory is modulated by the triangle wave signal. If we sample 2$m$ points in each cycle of the triangle wave signal, we define

$$\begin{aligned} { U_{fs}} (n)&= \sum \limits _{i=2mn}^{2m(n+1)-1} { U_f (i)} \quad n=1,2,3\ldots \end{aligned}$$

(3)

$$\begin{aligned} U_{is} (n)&= \sum \limits _{i=2mn}^{2m(n+1)-1} {U_i (i)} \quad n=0,1,2,3\ldots \end{aligned}$$

(4)

$$\begin{aligned} U_{os} (n)&= k*U_{is} (n)\quad n=0,1,2,3\ldots \end{aligned}$$

(5)

where $m$ and $n$ are the natural numbers, $k$ is the magnification of A, ${ U_{fs}}$ is the sum of the data in each period of the triangle signal, which is zero as shown in Fig. 2, $U_{is}$ is the sum of the data in each period of the mixed signal $U_i ,\,U_{os}$ is the output of A. Because the frequency of the ECG signal is far lower than the signal expressed by the second term in (2) carrying the respiratory signal, and the amplitude and frequency of the respiratory are far lower than the ECG signal, the ECG signal can be extracted from $U_{os}$ in (5). This method is equivalent to the low-pass filtering.

Fig. 2

The triangle wave signal

Similarly, we can get the respiratory signal by the method of the subtotal of the first half cycle data minus the subtotal of the subsequent half cycle data. We define

$$\begin{aligned} { U_{fm}} (n)&= \sum \limits _{i=2mn}^{2mn+m-1} { U_f} (i)-\sum \limits _{i=2mn+m}^{2m(n+1)-1} { U_f} (i)\qquad n=0,1,2,3\ldots \end{aligned}$$

(6)

$$\begin{aligned} U_{im} (n)&= \sum \limits _{i=2mn}^{2mn+m-1} {U_i } (i)-\sum \limits _{i=2mn+m}^{2m(n+1)-1} {U_i } (i)\qquad n=0,1,2,3\ldots \end{aligned}$$

(7)

$$\begin{aligned} U_{om} (n)&= k*U_{im} (n)\qquad n=0,1,2,3\ldots \end{aligned}$$

(8)

where $U_{fm} (n)$ is a constant expressing the subtotal of the first half cycle data minus the subtotal of the subsequent half cycle data of the triangle wave signal, and $U_{im} (n)$ is the subtotal of the first half cycle data minus the subtotal of the subsequent half cycle data of $U_i$. The frequency of the ECG signal is far lower than the triangle wave signal, which leads to the ECG signal filtered in the process of the subtraction operation according to (7). Then the respiratory signal can be extracted from $U_{om} (n)$ in (8).

According to the above analysis, the method of adding shaped signal can detect and extract the bioelectrical and the respiratory signals. In addition, because the internal resistance $R_{f}$ of the constant current source $U_{f}$ and the input impedance of the amplifier are in parallel, this design does not reduce the input impedance of the amplifier. Therefore, the over-sampling and shaped signal technology can not only improve the collection accuracy but also reduce the requirement of the amplifier amplification and simplify the circuit structure design.

In order to provide a potential application of the proposed biomedical signals detection method, the acquisition of the ECG and respiratory signals is implemented by three electrodes with the differential amplifier. Moreover, the principle of signal detection and extraction is the same as the single electrode as shown in Fig. 1.

3 The Design of a Multiple Biomedical Signals Synchronous Acquisition Circuit

The biomedical signal collection circuit is required to have high input impedance, high common-mode rejection ratio (CMRR), and strong anti-interference capability. The acquisition system schematic for the biomedical signal is shown in Fig. 3, which is comprised of the shaped signal superposition circuit, the common- mode driving circuit, the high-pass filtering circuit, the parallel differential amplifier circuit, the low-pass filtering circuit, as well as the right leg driving circuit. This circuit design architectures are designed to reduce the noise and improve the CMRR.

Fig. 3

The schematic of the multiple biomedical signals synchronous acquisition system

In order to monitor multiple biomedical signals synchronously, the shaped signal is added to the bioelectrical signal and combined with the over-sampling technology, which can reduce the magnification, expand dynamic testing range, and improve the precision of measurement. The second outstanding merit is that the high-pass filtering circuit is used to decrease the polarization voltage; the common-mode driving circuit and the right leg driving circuit are designed to reduce the power frequency interference and high-frequency noise and to improve the CMRR.

3.1 The Shaped Signal Superposition Circuit

In Fig. 3, $I_{1}$ and $I_{2}$ are the current-mode triangle wave signals with opposite phase and equal amplitude, which are injected into the measurement object by the electrodes. $I_{1}$ and $I_{2}$ are produced by two digital to analog converters (DAC). The detection principle of the bioelectrical signal and respiratory signal is illustrated in Fig. 1.

3.2 The Parallel Differential Amplifier Circuit

The circuit can be divided into two stage circuits which are the high-pass filtering circuit and the parallel differential amplifier circuit with the corresponding CMRR of CMRR$_{\mathrm{F}}$ and CMRR$_{\mathrm{B}.}$ The total CMRR of the circuit can be expressed by

$$\begin{aligned} \mathrm{CMRR}_\mathrm{T} =\frac{\mathrm{CMRR}_\mathrm{F} *\mathrm{CMRR}_\mathrm{B} }{\mathrm{CMRR}_\mathrm{F} +\mathrm{CMRR}_\mathrm{B} } \end{aligned}$$

(9)

The parallel differential amplifier circuit is composed of amplifier $A_{2}$ and $A_{3}$ shown in Fig. 3. We in detail analyze the CMRR$_{\mathrm{B}}$ of the circuit shown in Fig. 4. The output of the parallel differential amplifier circuit can be expressed by

$$\begin{aligned}&\begin{array}{lll} U_o &{}=&{}U_{o2} -U_{o1} \\ &{}=&{}\left( 1+\frac{R_6 }{R_5 }+\frac{R_7 }{R_5 }\right) *(U_{i2} -U_{i1} ) \\ \end{array}\end{aligned}$$

(10)

$$\begin{aligned}&M=1+\frac{R_6 }{R_5 }+\frac{R_7 }{R_5 }, \end{aligned}$$

(11)

where $R_5$, $R_6$, and $R_7$ are the resistors, $U_{i1}$ and $U_{i2}$ are the difference input signals, and $U_{o1}$ and $U_{o2}$ are the output signals. From (10), $U_o$ only contains different-mode signal, and the common-mode signal is zero, which is shown that the CMRR of this stage is infinite when the parameters of A$_{2}$ and A$_{3}$ are infinitely near. However, the parameters of A$_{2}$ and A$_{3}$ actually exist difference, so we should increase the magnification of $M$ in (11) by reducing the polarization voltage using the high-pass filtering circuit to improve the CMRR$_\mathrm{B}$.

Fig. 4

The parallel differential amplifier circuit

3.3 The High-Pass Filtering Circuit

In Fig. 5, capacitor $C_1$, resistor $R_3$, capacitor $C_2$, and resistor $R_4$ constitute two high-pass filters, which can reduce the polarization voltage caused by the electrodes. The cutoff frequency of the high-pass filter is up to 0.5 Hz, which can effectively filter the polarization voltage, and create condition for improving the CMRR and dynamic range of the detection. However, we should match the parameters of these components intentionally to achieve a high CMRR$_{\mathrm{F}}$. In order to ensure the CMRR$_{\mathrm{T}}$ higher than 80 dB, we should ensure CMRR$_{\mathrm{F}}>$80 dB. It is easy to be proved that the matching errors of the components in the two high-pass filtering circuits should be less than 1/10000, which is impossible in normal capacitor and resistor components. In order to achieve high CMRR, we design the common-mode driving circuit, which can avoid common-mode signal converting into differential mode signal caused by the circuit components mismatch.

3.4 The Common-Mode Driving Circuit

The common-mode driving circuit is composed of resistors $R_{1}$, $R_{2}$, and operational amplifier $A_{1}$. In Fig. 5, when $U_{I1} =U_{I2} =U_C$, $U_{I1}$ and $U_{I2}$ are the input signals, and $U_C$ is common-mode signal, we can obtain that

$$\begin{aligned} \begin{array}{l} U_{AO} =U_{Ai} =U_C \\ U_1 =U_2 =U_C, \\ \end{array} \end{aligned}$$

(12)

where $U_1$ and $U_2$ are the output signals, $U_{Ai}$ and $U_{Ao}$ are the input and output signal of amplifier $A_{1}$. For the different-mode signal, $U_{Ao} =U_{Ai} =0$, so the output signal $U_o =U_2-U_1$ only contains different-mode signal.

From the above deduction, we can conclude that the design of the common-mode driving circuit enables the CMRR$_{B}$ infinite without any limitation and assumption for the components of $C_1$, $R_3$, $C_2$, and $R_4$.

Fig. 5

The common-mode driving circuit and two high-pass filtering circuits

In the actual measurement, there are other interferences such as the power frequency interference that limits the differential voltage gain, we design the right leg driving circuit as shown in Fig. 6a to further eliminate the common-mode signal and stable the working point of the whole circuit.

Fig. 6

a The right leg driving circuit. b The equivalent circuit of the common-mode driving circuit and the right leg driving circuit

3.5 The Right Leg Driving Circuit

The right leg driving circuit is an effective method of restraining the power frequency interference without losing the biomedical information. More importantly, it forms closed loop between human body and the front end circuit by an additional reference electrode to enable the stability of the circuit. Specially, the input signal fluctuation should be in the input range of ADC. To meet this demand, an external reference voltage $V_\mathrm{REF}$ is connected to the non-inverting input of the amplifier $A_{4}$ to ensure the baseline of signal locating in the input range of ADC. The output of A$_{4}$ interfaces with the right leg of human by the electrode. In order to evaluate the interference rejection capability of the right leg driving circuit, we make the common-mode driving circuit and the right leg driving circuit equivalent into the circuit shown in Fig. 6b.

In Fig. 6b, the common-mode signal from the human body can be equivalent into a voltage source $V_{IC}$ and an internal resistance $Z_{S}$. The output of the equivalent circuit can be expressed by

$$\begin{aligned} V_{OC} =\frac{R*V_{IC} }{R+Z_S +Z_S *K}, \end{aligned}$$

(13)

where $R$ is the contact impedance between skin and electrode, $K$ is the magnification of amplifier $A_{4}$. In general, the value of $R$ is about a few to tens of thousands of ohm, the value of $Z_{S}$ is about 10$^{7}~\Omega$, and $K$ is about 10$^{2}$ to 10$^{3}$, so Eq. (13) is approximately equal to

$$\begin{aligned} A_c =\frac{V_{OC} }{V_{IC} }\approx \frac{R}{Z_S *K}, \end{aligned}$$

(14)

where $A_{c }$ is the magnification of the common-mode signal, and $A_{d}$ is about 10$^{-5}$ to 10$^{-6}$. Therefore, the right leg driving circuit can effectively reduce the common-mode signal. On the whole, this biomedical signal detection circuit obtains a high CMRR with the design of the common-mode driving signal, the high-pass filtering circuit, the parallel differential amplifier circuit, and the right leg driving circuit.

This front end circuit contains two low-pass filters before the ADC to filter the high-frequency signal, which is constituted by $C_3,\,R_8$ and $C_4,\,R_9$. The component values used for the ECG monitor are given in Table 1. Figure 7 shows the snapshot of our hardware board with three electrodes. In this measurement, in order to improve the system integration and achieve small volume and low-power consumption, the biomedical signal acquisition circuit is integrated to CY8C55 from Cypress Semiconductor Corporation which is a high-performance Programmable System-on-Chip (PSoC) with a single power supply of 3.3 V, including four amplifiers, ADC, DAC, and so on [8]. The biomedical signal collection circuit coupled with the shaped signal allows us to detect the ECG and respiratory signal synchronously.

Table 1 Characteristics of the used components

Fig. 7

A snapshot of our hardware board using three electrodes

4 Experimental Verification

4.1 Simulation Experiment

4.1.1 Methods

In order to prove the validity and feasibility of the novel algorithm about superposition shaped signal, the numerical simulation is carried out to extract the ECG and respiratory signal with matrix laboratory (MATLAB) software. The triangle wave signal acts as the shaped signal, the sine wave signal represents the impedance variation signal caused by respiratory, and the sine signal acts as the ECG signal in this simulation experiment.

The frequency of triangle wave signal is assumed to be 50 kHz with the peak amplitude of 1 V; the frequency of human chest impedance is 0.2 Hz, the peak amplitude is 2 $\Omega$; the frequency of sine signal is 1 Hz with the peak amplitude of 1 mV; the contact resistance is 200 $\Omega$; and the source resistance is 10 k$\Omega$. The mixed signal is produced according to Eq. (2). The signal sample rate is 500 kHz, the total sampling time is 5 s, and 30 dB Gaussian noise is added to the mixed signal.

4.1.2 Results and Discussion

The mixed signal is shown in Fig. 8. As mentioned in Sect. 2, the simulative ECG and respiratory signal can be extracted from the mixed signal by the algorithm proposed in this paper. As expected, the simulative ECG signal is obtained from the sum of the data in each period of the triangle signal according to (4), which is shown in Fig. 9a. Similarly, Fig. 9b shows that the respiratory signal is extracted by the method of the subtotal of the first half cycle data minus the subtotal of the subsequent half cycle data according to (7). From Fig. 9, we can see that the simulative ECG and respiratory signal are seriously interfered by the noise. Thereby, the down-sampling method is performed in order to reduce the noise and obtain high-quality signals. The primitive and 250 points down-sampling ECG signal are shown in Fig. 10a, b. And the primitive and 10,000 points down-sampling respiratory signal are shown in Fig. 11a, b. Clearly, the simulative ECG and respiratory signal are demodulated well and truly. Therefore, the results illustrate that the proposed method is feasible and simple to detect the respiratory and ECG signals.

Fig. 8

The simulative mixed signal

Fig. 9

a The simulative ECG signal obtained by the method according to (4). b The simulative respiratory signal obtained by the method according to (7)

Fig. 10

a The primitive simulative ECG signal. b The ECG signal after 250 points down-sampling

Fig. 11

a The primitive simulation respiratory signal. b The respiratory signal after 10,000 points down-sampling

With the simple addition and subtraction operations of the proposed method we can synchronously detect and extract the ECG and respiratory signals. In addition, the results indicate that the accuracy of the extracted signals can be improved as the frequency of the shaped signal and the over-sampling frequency increasing. Therefore, the method of superposition shaped signal can improve the performance of the over-sampling technology, simplify the design of the complex filter circuit, decrease the amplifier magnification, and achieve the ability of software instead of hardware.

4.2 The Application of the Multiple Biomedical Signals Synchronous Acquisition Circuit

4.2.1 Methods

We perform the designed circuit illustrated in Sect. 3 for the biomedical signal detection as shown in Fig. 7. The frequency of the two triangle wave signals with opposite phase is 50 kHz with 500 kHz sample rate. To collect the ECG, respiratory signal, and the lead dropping signal, we put two Ag/AgCl electrodes on the chest of human and one Ag/AgCl electrode on the arm.

We divide the collected signal into two parts in the microprocessor according to (5) and (8), which has been illustrated in the simulation experiment. One part mainly contains the ECG signal, and the other part contains the respiratory signal, and finally the signal is uploaded to the computer. The sampling time is 10 s.

4.2.2 Results and Discussion

Figure 12a provides the collected signal mainly contains ECG signal, and it is corrupted by the interference including power frequency interference, muscle electricity, baseline drift, and the noise. The noise is generally caused by the electrode contact noise and the quantization noise of the ADC. However, down-sampling is an effective method to remove the power frequency interference and the noise, which can be easily performed. From Fig. 12b, the resulting ECG signal is acquired after removing the muscle electricity and the noise by the method of down-sampling. The respiratory signal after down-sampling is shown in Fig. 12c. Figure 12d provides the lead dropping signal with larger amplitude value, which can indicate the phenomenon of lead dropping. Figure 13a, b show the acquired ECG and respiratory signals in deep breathing of human, which illustrate that the respiratory signal is well collected with higher impedance change in deep breathing.

Fig. 12

a The collected signal mainly contains ECG signal. b The ECG signal after down-sampling. c The respiratory signal after down-sampling. d The lead dropping signal

Fig. 13

a The extracted ECG signal in deep breathing. b The extracted respiratory signal in deep breathing

Therefore, our experiments show that this biomedical signal acquisition system can be used to monitor the ECG and respiratory signal synchronously for the application of the ubiquitous health care and the ambulatory monitoring system.

Firstly, with the microcontroller of PSoC, the architecture of the circuit system is simple in comparison with other detection systems shown in [14, 24]. Also the circuit is easily performed only with four operational amplifiers and a few passive components and can be easily miniaturized. Secondly, for the ECG signal with 12-bit ADC and 250 times over-sampling frequency, theoretically, we can obtain high-sampling accuracy of about 20-bit ADC [18]. Thirdly, because of the use of common-mode driving circuit, the parallel differential amplifier circuit and the right leg driving circuit combining the over-sampling technology and the shaped signal, the detection system can reach a high CMRR and robustness to noise. In addition, with the use of over-sampling, we just need a low magnification about 20, we can obtain a high-quality ECG and respiratory, thus the proposed system is insensitivity to noise. Fourthly, the proposed system is portable and easy to be applied to the daily life with the wireless and bluetooth transmission. Lastly, the proposed system just needs several dollars and simple signal processing scheme without using the complex method such as in [20, 21]. And the design of a single power system can effectively result to low-power consumption and minimize the size dimensions. With the increasing use of ambulatory monitoring system, not only continuous signal collection and low-power consumption, but also smartness with robust operation under the presence of signal artifacts are required. Thus, the proposed system is highly suitable for wearable applications of long-term cardiac monitor.

5 Conclusion

A novel multi-function biomedical signal acquisition circuit is designed in this paper, which has great potential for the application of health care in the daily life. And we propose a novel method without complex calculation based on shaped signal and over-sampling to extract the ECG and respiratory signals from the mixed signal. We in detail analyze the design architectures of the circuit, which show that with the design of the common-mode driving circuit, the parallel differential amplifier circuit, and the right leg driving circuit combining the over-sampling technology that the circuit obtains high CMRR and wide dynamic measurement range without large magnification and the need of matching the resistance-capacitance components. Both numerical simulation and human experiment verify the feasibility of the proposed synchronous data acquisition circuit and the method for the detection of the ECG and respiratory signals.

The circuit is characterized by multi-function, high accuracy, strong robustness to noise, and portability according to the requirement of dynamic body domain nets and the internet of things. The main idea of this paper lies in the application of the shaped signal, which can detect the impedance variation signal and improve the performance of over-sampling. The method based on over-sampling and shaped signal can improve the measurement sensitivity or precision, and reduce the complexity of the analog circuit in the same hardware and gain. Using this method the ECG and respiratory signals are detected synchronously in this paper. So, with the proposed circuit and signal processing method, the impedance variation signal and the bioelectrical signal such as EMG, EOG, and EEG can also be detected synchronously in the condition that the frequencies of the shaped signal and the over-sampling are high enough, and the amplitude of the shaped signal is large enough.