Blood pressure is one of the important physiological parameters of the human body. Accurate measurement is beneficial for early detection and identification of hypertension types, and for providing reasonable treatment recommendations. Currently, non-invasive detection methods are mainly used for ordinary patients in clinical practice, which are broadly divided into two categories: the artificial Korotkoff sound method and the oscillometric method. While the artificial Korotkoff sound method is relatively accurate, it is difficult to operate and is greatly affected by subjective factors; the traditional oscillometric method, although simple to operate, has poor stability and individual adaptability, which is not conducive to its widespread clinical application. This paper improves the original measurement method based on the oscillometric method in terms of both hardware implementation and software design, and conducts comparative tests. 1. Hardware Design The main process of blood pressure detection using the oscillometric method is to acquire the pressure signal changing within the cuff, analyze the pulse signal separated from it, find the corresponding positions of systolic and diastolic blood pressure, and thus obtain the data. Traditional oscillometric measurement amplifies the signal from the sensor, performs low-pass filtering on the amplified signal to obtain the pressure signal, and sends it to the microcontroller via an A/D converter. Then, the pressure signal is band-pass filtered to obtain the pulse signal, which is sent to the microcontroller via another A/D converter. Its basic structure is shown in Figure 1. The use of the Σ-Δ microcontroller ADμC848 simplifies the circuit. Because it integrates a high-precision 16-bit Σ-Δ A/D converter, and its A/D reference voltage is programmable (down to a minimum of 10mV), it can directly perform A/D conversion while ensuring accuracy and dynamic range requirements, without the need for amplification. This eliminates a series of problems caused by the presence of an amplifier, such as changes in dynamic range, noise, and voltage offset, and reduces the number of components used, lowering implementation costs. Since this Σ-Δ A/D converter provides a differential input mode, the differential signal from the sensor can be directly fed into the A/D converter, theoretically achieving an infinite common-mode rejection ratio. Therefore, it can significantly reduce common-mode interference caused by mismatches in the pre-amplifier circuit. Because the Σ-Δ A/D converter process involves filtering through a low-pass filter, filtering is unnecessary before A/D conversion; the sensor can be directly connected to the A/D converter, and then digital filtering can be performed. Because the ADμC848 integrates a standard constant current source, the constant current value can be adjusted via software programming. Therefore, a standard pressure output can be sampled according to different application environments, then converted to digital (A/D) value, and the constant current source can be adjusted in real time based on the conversion result until the desired conversion value is output, thus achieving automatic product calibration. The improved hardware structure of the electronic blood pressure monitor is shown in Figure 2. 2. Software Design After the above hardware processing, the cuff pressure change curve is obtained. In the software processing, the pulse signal is first separated; then interference points are removed, an envelope curve is fitted, and the corresponding mean pressure is found; finally, the systolic pressure and mean pressure are calculated based on the coefficients. A morphological filtering algorithm is introduced in the process of separating the pulse signal. Since the frequency bands of the cuff pressure signal and the pulse signal are close, directly using bandpass filtering will reduce the signal amplitude and lower the signal-to-noise ratio, making subsequent processing difficult. However, applying a morphological filtering algorithm separates the signal from a morphological perspective, which can effectively extract the pulse signal. In order to achieve real-time signal separation, an opening operation will be used to process and flatten all the peaks in the original signal. The difference between the original signal and the processed signal will be used to obtain the separated pulse signal. Figure 3 shows the original signal and Figure 4 shows the separated pulse signal. In order to effectively suppress interference and repair the missing pulse wave, the reliability of each pulse wave will be determined according to the angle between the peak value of each pulse wave and the peak value of its adjacent pulse wave. Since the amplitude of the pulse wave is not monotonically changing, such judgment also needs to consider the amplitude factor. The specific method is described in reference [1]. The envelope fitting is performed using the weight information of each pulse wave obtained above. Since the obtained envelope is obviously asymmetrical (i.e., the second-order fitting cannot meet the requirements), the weighted third-order least squares fitting method will be used. After the fitting is completed, the pressure value corresponding to the position of the maximum value on the curve is the value of the mean pressure. Finally, referring to the method in reference [2], the amplitude coefficient is determined according to the magnitude of the mean pressure, and the corresponding position of the systolic pressure and diastolic pressure is calculated using the amplitude coefficient, so as to obtain the magnitude of the systolic pressure and diastolic pressure. To verify the accuracy of the obtained blood pressure monitor, some typical samples were selected, and their measurement results were compared with those obtained by manual auscultation using the Korotkoff sound method. First, blood pressure value a1 was measured using the manual Korotkoff sound method. After a 15-minute interval, the improved electronic blood pressure monitor was used to measure the blood pressure again, obtaining value b. After another 15 minutes, the measurement was repeated using the manual Korotkoff sound method, obtaining blood pressure value a2. The average value 'a' of a1 and a2 was used as the measurement value obtained by manual auscultation using the Korotkoff sound method. The obtained measurement data are shown in Tables 1 and 2. From the above typical measurement results, it can be seen that the electronic blood pressure monitor described in this paper can ensure an accuracy of blood pressure measurement within 5 mmHg, basically meeting the accuracy requirements for blood pressure measurement. This paper proposes a method for implementing a blood pressure monitoring instrument based on a SoC. This method has high hardware integration and is simple to design and implement; the software design incorporates various advanced algorithms such as morphological filtering, resulting in high accuracy and strong anti-interference capabilities. Experiments have shown that this blood pressure monitor has excellent accuracy and can meet the general requirements for blood pressure measurement. References [1] LIN CT, LIU Sh H, WANG JJ et al.Reduction of inteRFerence in oscillometric arterial blood pressure measurement using fuzzy logic[J].IEEE Trans. On Biomedical Engineering, 2003, 50(4). [2] MORAES J, CERILLI MA Strategy for determination of systolic, mean and diastolic blood pressures from oscillometric pulse profiles[J].IEEE Computers in Cardiology, 2000, 27: 211-214.