Binary phase modulation array radar vital sign detection signal enhancement method

Abstract

The application discloses a kind of binary phase modulation array radar vital sign detection signal enhancement methods, comprising: control FMCW radar multiple transmitting channels simultaneously transmit mutually orthogonal binary phase encoding signals;Each receiving channel synchronously acquires complex echo and down-converts, obtains multiple-channel complex intermediate frequency signal;The intermediate frequency signal is subjected to fast time domain distance Fourier transform to extract target distance gate, and mutually orthogonal decoding matrix of transmitting end is subjected to slow time correlation operation, and each virtual channel signal is demodulated and separated;Online eliminate inter-channel static amplitude-phase mismatch using static calibration factor;After locking the space direction of multiple targets by angle measurement algorithm, carry out full-array digital beam forming;Finally, the slow-time phase history after beam focusing is extracted and high-order differentiation, and the fine thoracic movement information of human body is reconstructed with high fidelity;Under the premise of keeping the same equivalent slow-time sampling rate as time division multiplexing system, it has strong robustness to amplitude-phase non-ideal hardware damage of low-cost phase shifter.

Description

Technical Field

[0001] This invention belongs to the field of radar signal processing and non-contact medical detection technology, specifically relating to an array radar system and method for monitoring vital signs. More specifically, this invention discloses a signal enhancement method that utilizes binary phase modulation (BPM) technology to achieve high sampling rate and high signal-to-noise ratio non-contact monitoring of the fine mechanical activities of the human heart in a multi-target environment. This method is suitable for multi-target, micro-motion fine waveform reconstruction applications such as medical monitoring, sleep monitoring, and vehicle-mounted vital sign detection. Background Technology

[0002] Non-contact vital sign monitoring technology based on microwave radar has shown significant application value in routine health monitoring and clinical mobile telemetry because it eliminates the need for electrodes attached to the body surface or sensors worn on the skin. In existing technologies, frequency-modulated continuous wave radar systems have been used to detect physiological parameters such as respiration and heart rate. However, the microscopic vibrations on the body surface (such as seismogram signals) caused by cardiac ejection, myocardial contraction, and aortic pulsation in the human chest cavity are extremely weak (micrometer-level), and their energy spectrum extends widely from 0 to 60 Hz or even higher. To reconstruct this waveform with high fidelity and avoid spectral aliasing, radar systems impose extremely stringent requirements on the equivalent slow-time sampling rate and echo signal-to-noise ratio.

[0003] In application scenarios where multiple targets coexist, existing radar systems suffer from the following technical bottlenecks: (1) Sampling rate bottleneck of time-division multiplexed multichannel radar: This type of radar relies on each transmission channel to transmit signals alternately on the time axis. As the number of transmission channels increases, the equivalent cumulative time required to complete the acquisition of one frame of data increases exponentially, resulting in a sharp drop in the effective slow-time sampling rate (usually limited to tens of hertz), which cannot meet the Nyquist sampling requirements of high-frequency cardiac oscillation signals. Although theoretically the frame rate can be forcibly increased by shortening the duration of a single linear frequency modulated pulse, this is limited by the steep slope phase-locking capability of the voltage-controlled oscillator and phase-locked loop hardware, introducing a large amount of system phase noise, which seriously degrades the measurement sensitivity of micron-level micro-motions. In addition, the extraction of fine vital signs often requires second-order differential processing of displacement, which amplifies high-frequency noise exponentially, further placing higher demands on the signal-to-noise ratio of the original echo at the front end.

[0004] (2) Signal-to-noise ratio and degree of freedom attenuation of traditional phased array radar in multi-target scenarios: Although traditional phased array radar can obtain high signal-to-noise ratio gain through full-channel coherent integration in single-target scenarios, it must rely on mechanical or electronic beam scanning in multi-target space environments. This leads to a significant decrease in the coherent slow-time accumulation gain allocated to each specific target, resulting in a deterioration of the signal-to-noise ratio and severely limited multi-target resolution (system degrees of freedom).

[0005] In summary, existing technologies struggle to balance high slow-time sampling rates with high signal-to-noise ratios in the detailed detection of multiple target vital signs, presenting a significant technical challenge that urgently needs to be addressed. Summary of the Invention

[0006] To address the challenges of balancing slow sampling rate and signal-to-noise ratio in the fine detection of vital signs in multi-target time-division multiplexed multi-channel radars, and the severe attenuation of signal-to-noise ratio and degrees of freedom in phased array radars under multi-target conditions, this invention provides a method for enhancing vital sign detection signals in array radars using binary phase modulation.

[0007] This invention utilizes independent digitally controlled phase shifters connected in series in each transmit channel of a Frequency Modulated Continuous Wave (FMCW) radar. This controls all antennas to synchronously transmit binary phase-coded signals modulated by orthogonal Hadamard codes at full power within the same linear frequency modulation (chirp) time interval. This eliminates the hard frame spread on the time axis inherent in time-division multiplexed multi-channel radar architectures while maintaining the same equivalent slow-time sampling rate, achieving synchronous full-power transmission across all channels. Specifically, it includes the following steps: Step 1: Multi-channel quasi-orthogonal binary phase-coded transmission: This method is applicable to multi-channel array radar systems containing M transmit channels and U receive channels, where M≥2 and U≥1.

[0008] Constructing an orthogonal coding matrix: Define the coding length (i.e., the number of linear frequency modulated pulses) within a single frame of the radar system as N. To ensure that the M transmission channels are mutually orthogonal on the slow time axis, first construct a Hadamard matrix of order N at the digital end, satisfying N ≥ M. Any two columns of this matrix are orthogonal to each other. Arbitrarily select M columns from the data and combine them to form a quasi-orthogonal emission coding matrix of size N×M. .

[0009] Each transmission channel uses an independent numerically controlled phase shifter connected in series to read the encoding matrix. The corresponding column elements. On the slow timeline of each frame, with the nth pulse... The ±1 elements of the encoding matrix are mapped to the binary phase values ​​that control the phase shifter. All transmission channels synchronously radiate frequency-modulated continuous wave radar signals encoded by BPM: , Where t is the fast time (the sampling time variable within a single chirp); The imaginary unit; The center carrier frequency of the radar; This is the frequency modulation slope; It is the binary phase modulation codeword of the m-th channel during the n-th pulse.

[0010] Step 2: Separation of composite echo reception and full-channel correlation decoding: The echo signals from each receiving channel are amplified with low noise, then multiplied, mixed, and low-pass filtered with the local oscillator signal to obtain a multi-channel composite intermediate frequency (IF) signal containing the coded features of all transmitting channels. ;in Here, c represents the target complex scattering coefficient corresponding to the m-th channel; c is the speed of light. For the goal in slow time Radial distance at time; The pulse repetition period; A fixed additional phase introduced for the mixer.

[0011] The digitized intermediate frequency signal is compared with the ideal orthogonal decoding factor corresponding to the k-th transmission channel. By performing slow-time one-dimensional cross-correlation summation, the equivalent single-transmitter single-receive intermediate frequency signal of the k-th virtual channel is separated: , .

[0012] It should be noted that the output of the above decoding operation is... Corresponding discrete slow time sampling time Where l=0,1,2,... are the frame numbers, and N is the single-frame coding length. Let be the pulse repetition period. Therefore, the equivalent slow-time sampling interval of the k-th virtual channel is . The equivalent sampling rate is 1 / ( To simplify the notation, the symbols used in the following description will still be represented by ). This represents a continuous, slow-time variable, but in actual implementations, the signal only occurs during... There is a definition for every moment.

[0013] The strict orthogonality of Hadamard coding is used to eliminate signals from the remaining M-1 uncorrelated channels. For the above... Performing a Fast Fourier Transform (FFT) on the fast time t in the signal yields the frequency domain expression corresponding to the target range gate. In the fast time frequency domain, the signal can be represented as: ; The sinc function is defined as follows: .

[0014] Addressing the quantization phase error present in numerically controlled phase shifters in actual chips and amplitude imbalance The actual transmission code becomes The corresponding gain is 1+ Through matrix derivation, the error term in the decoded signal is: ;in This error term is independent of the phase term, which contains information about the target's motion. Therefore, it only causes a fixed amplitude decay and phase shift, and does not introduce time-varying noise on the slow time axis.

[0015] Step 3, Static calibration of channel amplitude and phase errors: A strong scattering object (such as a corner reflector) is placed at a predetermined distance and known azimuth in front of the radar. The radar acquires the echo of this static target, and after completing the BPM orthogonal correlation decoding and separation in step two, obtains the static amplitude and phase response values ​​of M×U equivalent virtual channels. One virtual channel is selected as a reference, and the relative amplitude mismatch and relative phase offset of all other virtual channels relative to this reference are calculated to construct a one-dimensional complex static calibration vector. In real-time detection, the intermediate frequency data of each virtual channel separated by demodulation in each frame is compared with... The corresponding elements are multiplied by complex numbers to eliminate static amplitude-phase mismatch between channels.

[0016] The static calibration vector In subsequent real-time processing, the data is directly multiplied by the virtual channel complex data separated in step two to obtain the calibrated channel data. .

[0017] Step 4, Digital Beamforming and Phase Extraction: After acquiring the calibration data for each channel, the azimuth angle of the human target is determined using an angle measurement algorithm (such as the MUSIC algorithm, FFT beamforming, or Capon beamformer). Regarding the desired orientation At the target location, perform digital beamforming and output a synthesized signal: ,in For DBF weight vectors, Its conjugate transpose; It is a diagonal matrix, where K = M × U is the total number of virtual channels. Let K be the complex amplitude envelope of the Kth channel; As the guide vector, It represents the spatial phase difference of the electromagnetic wave reaching the Kth array element.

[0018] To achieve maximum coherence gain, let Then to Phase extraction: ; The thoracic cavity displacement information can be separated and subjected to high-order differential processing to reconstruct the micro-motion signal waveform that reflects the delicate mechanical activity of the human heart.

[0019] Compared with the prior art, the present invention has the following significant advantages: This invention abandons the traditional time-division multiplexing and phased array mechanisms, achieving full-channel synchronous full-power transmission through BPM. It maintains the same equivalent slow-time sampling rate as the time-division multiplexing system (both are 1 / (M·m·K). Under the premise that all transmission channels radiate energy at full power simultaneously, the signal-to-noise ratio of each sampling point is improved by a factor of M compared to the time-division multiplexing system, while avoiding the additional phase noise introduced by hard frame spread. In addition, high signal-to-noise ratio motion signal extraction in multi-target scenarios is achieved by performing beamforming in the digital domain.

[0020] Theoretical analysis and simulations show (see details) Figure 2 Even in the most extreme 2-channel configuration, the signal-to-noise ratio degradation caused by phase shifter amplitude and phase errors still has a greater than 80% probability of remaining below 0.2dB, demonstrating extremely strong hardware robustness.

[0021] Other features and advantages of the invention will be set forth in the following description or may be learned by practicing the invention. Attached Figure Description

[0022] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts.

[0023] Figure 1 This is a flowchart illustrating the data flow and processing of multi-person fine vital sign detection in the overall scheme of this invention.

[0024] Figure 2 The graph shows the cumulative distribution function (CDF) of signal-to-noise ratio degradation as a function of the number of channels when the phase shifter has finite resolution quantization phase error and amplitude imbalance. The horizontal axis represents the signal amplitude loss (in dB) caused by amplitude and phase non-idealities, and the vertical axis represents the cumulative distribution probability (CDF, ​​ranging from 0 to 1).

[0025] Figure 3This is a comparison chart of the time alignment of the radar seismogram signal obtained by this invention with the physiological feature points of synchronous contact electrocardiogram (ECG) and acceleration seismogram (SCG) in a multi-person test scenario, as well as the results of time division multiplexing (TDM) radar. Detailed Implementation

[0026] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment takes the monitoring of chest cavity micro-movements (echocardiogram signals) of two people sitting side by side as an example, but it does not constitute any limitation on the present invention.

[0027] Example 1: Hardware configuration and parameters of an 8-transmit, 8-receive BPM radar: This embodiment uses Texas Instruments' (TI) AWR2243 cascaded radar chipset to construct a MIMO array with 8 transmit channels (M=8) and 8 receive channels (U=8). The radar center frequency... =77GHz, bandwidth B=4GHz, corresponding distance resolution is =3.75cm. Duration of a single linear frequency modulated pulse (chirp). FM slope = = Pulse repetition period = The corresponding maximum unambiguous distance is (The actual detection distance is only a few meters, with no blurring issues).

[0028] To achieve synchronous orthogonal coding of the eight transmission channels, the single-frame coding length is set to N=8. An 8th-order Hadamard matrix is ​​constructed. Extract all 8 columns as the transmission coding matrix. Each transmit channel is connected in series with a 6-digit digitally controlled phase shifter. At the start of each chirp, according to... The corresponding column element (+1 or -1) sets the phase shifter to... or Eight launch channels in Simultaneously transmits BPM-encoded FMCW signals at full power.

[0029] Figure 1 This illustrates the overall data flow and processing procedure of the present invention. The following steps correspond to this, and include: 1. Transmission and reception: Within each frame, N=8 chirps are transmitted sequentially. Within each chirp, all 8 transmission channels simultaneously radiate signals. (m=1,...,8; n=0,...,7). Eight receiving channels synchronously acquire echoes, which are then amplified, mixed, and low-pass filtered to obtain eight intermediate frequency (IF) signals. The analog intermediate frequency signal is digitized by simultaneously sampling using eight analog-to-digital converters (ADCs, sampling rate 20MHz, 12bit).

[0030] 2. BPM decoding separation: For the digitized intermediate frequency signal of each receiving channel, a fast time-domain windowed FFT (1024 points) is first performed to obtain the range profile. For the frequency point located at the target's range gate (e.g., the chest cavity at 1.2m from the radar), a slow time sequence of 8 chirps is extracted. This sequence is then compared with the ideal decoding sequence of the k-th transmitting channel. By performing cross-correlation, the signals of the corresponding virtual channels (k-th transmit, u-th receive) can be separated. Repeating this operation yields a total of 8×8=64 baseband signals for the virtual channels.

[0031] 3. Static calibration: Place a corner reflector (RCS approximately 10 dBsm) 1 meter directly in front of the radar. Collect 100 frames of data, obtain the complex responses of 64 virtual channels following the steps described above, and calculate the average to obtain the vector. The virtual channel formed by the first transmit channel and the first receive channel is selected as the reference, and the calibration vector is calculated. = (Each element is the base value divided by the response of the current channel). In subsequent tests, the data of each virtual channel was multiplied by... The corresponding complex coefficients.

[0032] 4. Multi-target localization and beamforming: The calibrated 64 virtual channel data form a uniform linear array (equivalent element spacing) / 2=1.95mm). Angle estimation was performed using the Multiple Signal Classification (MUSIC) algorithm, resulting in the two targets being located at... and (Relative to the radar normal). For those located at The goal is to generate a guide vector. Set DBF weights The slow-time complex signal of the target is obtained by weighted summation of the 64-channel data. .

[0033] 5. Phase extraction and second-order differential reconstruction: extract phase Calculate chest cavity displacement To obtain the acceleration signal reflecting cardiac impact (corresponding to the seismogram SCG), the following steps are performed: Perform second-order time differentiation: In practice, finite difference approximation is used, and the breathing components and noise are removed by bandpass filtering (cutoff frequency 0.5-50Hz).

[0034] The 6-digit digital phase shifter used in this embodiment has a quantization phase error uniformly distributed between -5.625° and 5.625°, and an amplitude imbalance between -1dB and 1dB. According to... Figure 2 The Monte Carlo simulation results shown (100,000 random trials) are illustrated in the figure. The horizontal axis represents the amplitude loss (dB) due to signal-to-noise ratio (SNR) degradation, and the vertical axis represents the cumulative distribution probability (CDF). Different curves correspond to different numbers of transmit channels M (M=2, 4, 8, 16, 32). With an 8-transmit-channel configuration, the SNR degradation caused by the aforementioned hardware non-ideals is almost negligible (less than 0.05dB in over 99% of cases), verifying the excellent robustness of this invention to low-cost hardware impairments.

[0035] Comparison of experimental results: Figure 3 This paper presents a comparison between the radar-based electrocardiogram (RF-SCG) measured in this embodiment and the simultaneously acquired contact electrocardiogram (ECG) and acceleration-based electrocardiogram (SCG). As can be seen from the figures: The RF-SCG signal of this invention has excellent time alignment with the reference SCG at key physiological feature points (such as MC: mitral valve closure; AC: aortic valve opening), and the peak signal-to-noise ratio is significantly higher than that of traditional time-division multiplexing radar.

[0036] A traditional time-division multiplexing radar (8 time divisions) requires 8× time to complete one frame (containing 8 chirps). =800μs, its effective slow-time sampling rate is 1 / (8 While a sampling rate of 1250Hz can cover the Nyquist band of the cardiac oscillation signal (maximum 60Hz), the transmission energy is dispersed across eight time slots, with only one channel operating in each slot. This results in a severely insufficient echo signal-to-noise ratio, and the noise is drastically amplified after the second derivative, making it impossible to clearly distinguish fine features such as MC and AC. In contrast, the BPM radar of this invention transmits synchronously at full power across all channels. While maintaining the same equivalent slow-time sampling rate (1250Hz), it concentrates the transmission energy at each sampling moment, increasing the signal-to-noise ratio of the virtual channel by eight times, thereby enabling high-fidelity reconstruction of the cardiac oscillation waveform. Figure 3 The comparison results clearly demonstrate this advantage.

[0037] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for enhancing vital sign detection signals in a binary phase-modulated array radar, characterized in that, Includes the following steps: Step S1: Control the M transmission channels of the frequency modulated continuous wave radar to simultaneously transmit mutually orthogonal binary phase-coded signal sequences, where M≥2; Step S2: Receive the multi-channel composite echo signal using the receiving antenna array, and mix and low-pass filter it with the local oscillator signal to obtain the multi-channel composite intermediate frequency signal; Step S3: Perform a range-to-fast Fourier transform on the multi-channel composite intermediate frequency signal in the fast time domain to extract the target range gate, and perform slow time correlation operation through a binary phase modulation decoding matrix that is orthogonal to the transmitter to demodulate and separate the single-antenna intermediate frequency signal of M independent virtual channels. Step S4: Call the preset hardware error calibration coefficient to perform static amplitude imbalance and static phase offset calibration on the virtual channel signals separated by demodulation; Step S5: Perform multi-target spatial orientation estimation based on the calibrated virtual channel signal, and perform matched filtering on the orientation of a specific target after spatial digital beamforming to extract the phase history of the target as it evolves over slow time. Step S6: Extract the time-varying information of the slow-time phase to demodulate the chest cavity displacement, and perform high-order differential processing on the chest cavity displacement to reconstruct the micro-motion signal waveform that reflects the delicate mechanical activity of the human heart.

2. The method according to claim 1, characterized in that, The binary phase-coded signal sequence in step S1 is modulated using an orthogonal coding matrix constructed from a Hadamard matrix; the size of the orthogonal coding matrix is ​​N×M, where N is the single-frame coding length and satisfies N≥M; each transmission channel corresponds to a column of the orthogonal coding matrix, and each channel sets the digitally controlled phase shifter to 0 or π phase according to the ±1 elements of its corresponding column.

3. The method according to claim 1, characterized in that, The slow-time correlation operation in step S3 specifically involves: combining the digitized intermediate frequency signal with the ideal orthogonal decoding factor of the k-th transmission channel. The output signal of a slow-time one-dimensional cross-correlation summation operation is represented as: , ; The target complex scattering coefficients are those corresponding to the k-th channel. For frequency modulation slope, For the target at discrete slow-time sampling moments The radial distance, l is the frame number, and N is the single-frame coding length. Where c is the pulse repetition period and c is the speed of light. This is the center carrier frequency of the radar.

4. The method according to claim 3, characterized in that, The virtual channel signals demodulated and separated in step S3 are represented in the fast time-frequency domain as follows: ， The sinc function is defined as follows: , It is a continuous slow-time variable.

5. The method according to claim 1, characterized in that, The preset hardware error calibration coefficient mentioned in step S4 is obtained in the following way: a strong scatterer is placed at a preset distance and a known azimuth in front of the radar, the echo of the static target is collected, and after demodulation and separation in step S3 is completed, the static amplitude and phase response values ​​of M×U equivalent virtual channels are obtained, where U is the number of receiving channels and U≥1. Selected Using one virtual channel as a reference, the relative amplitude mismatch and relative phase offset of all other virtual channels relative to this reference are calculated to construct a one-dimensional complex static calibration vector. .

6. The method according to claim 5, characterized in that, The inter-channel static amplitude imbalance and static phase offset calibration mentioned in step S4 specifically involves: during real-time detection, the intermediate frequency data of each virtual channel separated by demodulation in each frame are compared with the static calibration vector. Multiply the corresponding elements by complex numbers to obtain the calibrated channel data matrix. .

7. The method according to claim 1, characterized in that, The multi-target spatial orientation estimation in step S5 uses any one of the following: MUSIC algorithm, FFT beamforming method, or Capon beamformer.

8. The method according to claim 1 or 7, characterized in that, The spatial digital beamforming described in step S5 employs a matched filtering method, where the weight vector is set as the conjugate transpose of the guide vector of the desired azimuth, i.e. ,in For the desired direction, For the guide vector, K = M × U is the total number of virtual channels. It represents the spatial phase difference of the electromagnetic wave reaching the Kth array element.

9. The method according to claim 1, characterized in that, The output composite signal after matched filtering in step S5 is: ， in This is the conjugate transpose of the weight vector. This is the calibrated channel data diagonal matrix. The guide vector is the actual direction of the target. The radial distance to the target. This is a slow time.

10. The method according to claim 1, characterized in that, The thoracic displacement mentioned in step S6 is obtained through The calculation yields the result; the higher-order differential processing includes at least a second-order time differential, the output signal of which is used to reconstruct the acceleration waveform corresponding to the angiograph.

Images