Biological signal processing device, biological signal processing program, and biological signal processing method

Abstract

An output signal y capable of acquiring vital information with high accuracy is efficiently obtained from the reflected waves obtained by transmitting radio waves to a living body. [Solution] In an FMCW radio wave sensor, vital signals are acquired by applying CMA to IQ data, which are phase signals derived from transmitted and received waves. When the number of samples in one chirp set is n and the number of antennas is m, a radar cube is created in which n x m pieces of IQ data are arranged in the time direction. Of these, n' x m' pieces of IQ data are used and CMA is applied to the time information (time direction) to determine the beamforming weight w, and vital signals are integrated by multiplying the used IQ data by the weight. Here, n' and m' are selected so that m' x n' is less than m x n, and so that both m' and n' are not 1.

Description

[Technical Field]

[0001] The present invention relates to a biological signal processing device, a biological signal processing program, and a biological signal processing method, and relates to a technique for analyzing vital information using phase signals of a transmitted wave and a reflected wave transmitted to a living body by an FMCW method, for example. [Background technology]

[0002] Measuring vital information (biological signals) from living organisms is becoming increasingly important for understanding the health and psychological state of vehicle drivers, hospital patients, and everyday people, as well as for detecting signs of illness. In response to such demands, various technologies have been proposed for measuring vital information without contacting the subject, such as transmitting radio waves to the subject and using the Doppler shift (frequency transition) of the reflected waves to detect changes in distance to the subject. For example, Non-Patent Document 1 describes a technology for measuring the vital signs of a subject at a fixed distance using a radar device with a CW (Continuous Wave) system and a MIMO (Multiple Input Multiple Output) configuration. In this document, the received signals are integrated by beamforming using CMA, which controls the weight w so as to minimize the distortion component of the envelope of the received signals received by multiple virtual antennas, to obtain vital information.

[0003] However, the technology described in Non-Patent Document 1 is a technology for a CW method that transmits a transmission wave with a constant frequency, and there is no disclosure or suggestion as to how to apply CMA to a radar cube consisting of IQ data obtained by FMCW, in which the frequency changes continuously with each chirp of the transmission wave, and therefore it is unclear. In beamforming using CMA, the weight w to be multiplied by the input signal x is determined and updated to synthesize the output signal y, so it is necessary to appropriately determine the input signal x to which CMA is applied from the radar cube. [Prior art documents] [Patent documents]

[0004] [Non-Patent Document 1] Yuta Ogawa, Naoki Homma, and three others, "Vital Sign Tracking Method Using an Improved CMA Adaptive Array," IEICE Technical Report A·P2022-162 (2022-11) Summary of the Invention [Problem to be solved by the invention]

[0005] The present invention aims to more appropriately apply beamforming by CMA to a radar cube obtained by FMCW. [Means for solving the problem]

[0006] In the present invention, a transmitting means for transmitting a frequency-continuously modulated transmission wave toward a subject; an acquisition means for acquiring n phase signals per s chirps of the transmission wave from the transmission wave and its reflected wave for each of the m array antennas; a selection means for selecting a predetermined number (m' x n') of target phase signals as selected phase signals from the radar cube in which the (m x n) phase signals acquired by the acquisition means are arranged; a weight update means for determining and updating weights w that minimize distortion components of an envelope in an output signal y obtained by multiplying each of the selected phase signals x by a corresponding weight w by applying CMA to the input signal x; an output means for outputting an output signal y obtained by combining the input signals x using the updated weights w; The present invention provides a biological signal processing device comprising: [Effects of the Invention]

[0007] According to the present invention, by using a predetermined number m' x n' of selected phase signals selected from m x n phase signals per chirp number s as the input signal x, beamforming using CMA can be more appropriately applied to the radar cube obtained by frequency continuous modulation. [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 is a diagram illustrating a configuration of a biological signal processing device. [Figure 2] FIG. 2 is a diagram illustrating a hardware configuration of a signal processing device. [Figure 3] 10 is a flowchart illustrating a procedure for processing a biological signal. [Figure 4] This is an explanatory diagram conceptually showing the created radar cube and the application of CMA. [Figure 5] 10 is a flowchart illustrating a procedure for calculating weights by CMA. [Figure 6] FIG. 1 is an explanatory diagram showing mathematical formulas used in biological signal processing including CMA. [Figure 7] FIG. 10 is an explanatory diagram conceptually illustrating application of CMA in the second embodiment. [Figure 8] FIG. 11 is an explanatory diagram conceptually illustrating application of CMA in the third embodiment. [Figure 9] FIG. 10 is an explanatory diagram for evaluation of biosignals acquired according to the first to third embodiments. [Figure 10] FIG. 10 is an explanatory diagram showing a measurement model in the fourth embodiment. [Figure 11] FIG. 10 is an explanatory diagram showing mathematical expressions used in the fourth embodiment. [Figure 12] FIG. 13 is an explanatory diagram showing the results of verifying the effect of an evaluation function Q to which a limiting term A is added in the fourth embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, preferred embodiments of the biological signal processing device 1, the biological signal processing program, and the biological signal processing method of the present invention will be described in detail with reference to FIGS. (1) Overview of the embodiment In the biological signal processing device 1 of this embodiment, a frequency-continuous modulation wave (FMCW) type radio wave sensor applies a constant modulus algorithm (CMA) to IQ data, which is a phase signal of transmitted waves and received waves, to acquire a vital signal. That is, as shown in FIG. 1, the biological signal processing device 1 uses an array antenna (23, 25) to create a radar cube from IQ data generated by transmitting and receiving waves using the FMCW method with millimeter waves, applies CMA to this to determine the weight w for beamforming processing, and extracts vital signals by integrating the reflected waves hv(t) from the subject (living body) 7. Here, the IQ data is data that indicates the change over time in the phase angle between the transmitted wave and the received wave, and functions as a phase signal.

[0010] In this embodiment, instead of applying CMA to the entire created radar cube, the optimal IQ data is selected as selected IQ data 433, and CMA is applied to this data. The selected IQ data functions as selected phase data. In an FMCW array antenna, if the number of samples (distance) in one chirp set is n and the number of virtual array antennas is m, a radar cube is created in which n x m pieces of IQ data are arranged in the time direction. Of this radar cube, n' x m' pieces of IQ data are used and CMA is applied to the time information (time direction) to determine (n' x m') weights w for beamforming, and vital signals are integrated by multiplying the used IQ data by the weights. Here, n' and m' are selected in the range where m'×n' is less than m×n and both m' and n' are not 1.

[0011] Specifically, the first embodiment describes the case where n'=5, m'=m(=8), the second embodiment describes the case where n'=1, m'=m(=8), and the third embodiment describes the case where n'=5, m'=1, but other values ​​can also be used as long as they are within the range that satisfies the above conditions. According to this embodiment, by limiting the IQ data to be applied to selected IQ data 433, the burden of applying CMA and the processing of acquiring vital signals is reduced, and a larger vital signal can be obtained by integrating vital signals distributed across multiple distance bins. Furthermore, in the fourth embodiment, the directivity is controlled by providing an arrival direction restriction term in the evaluation function of the CMA. By using the arrival direction restriction term to restrict the direction from which the biological signal is thought to be present, it is possible to restore and acquire the signal from the direction from which the biological signal is thought to be present preferentially, even when a stronger reflected wave is received from another direction.

[0012] In CMA, the weight w is updated so that the distortion components of the envelope of the output signal y, which is obtained by multiplying multiple input signals x by the weight w, are minimized (the amplitude is constant). Therefore, by applying CMA to the selected IQ data 433, it is possible to obtain an output signal y that reduces noise while amplifying vital signals such as heart rate and breathing, which have an almost constant amplitude, thereby enabling highly accurate vital measurement.

[0013] (2) Details of the embodiment 1 is a diagram illustrating the configuration of a biological signal processing device 1 that measures biological signals (vital signs, vital signals) in this embodiment. In this embodiment, breathing, heart rate, pulse wave, etc. are measured as biological signals. In the embodiment described below, the biosignals of the subject 7 seated in a chair were measured, but the device can be widely used in a variety of situations, for example, by installing it in a vehicle to measure the biosignals of the driver, by detecting the biosignals of users in a hospital or nursing home, by installing it on a bathroom sink at home to detect the biosignals of a person washing their face, or by detecting the biosignals of a person sleeping in bed at home.

[0014] The biological signal processing device 1 is composed of a millimeter wave circuit 2, a control device 3, a signal processing device 4, and the like. The millimeter wave circuit 2 functions as an FMCW radar, irradiating millimeter waves hf(t) toward the subject 7 and receiving reflected waves hv(t) from the subject 7, mixing the transmitted waves (millimeter waves) with the reflected waves, and outputting the result to the signal processing device 4. The millimeter-wave circuit 2 includes an oscillator 21, a phase-shifting division unit 24, mixers 26 and 27, a transmitting antenna 23, and a receiving antenna 25. The millimeter wave circuit 2 of this embodiment functions as a transmitting means for transmitting a transmission wave that is frequency-continuously modulated and directed toward the subject.

[0015] The control device 3 is a control device that controls the driving of the transmitter 21, and supplies power to the transmitter 21 and drives it to generate millimeter waves. The control device 3 of this embodiment controls the transmitter 21 to output a millimeter wave transmission wave (chirp) modulated by the FMCW (frequency continuous modulation) method so that the frequency increases linearly over time. It should be noted that control device 3 controls transmitter 21 so that it outputs millimeter waves with no directionality, rather than millimeter waves with directionality in the direction of a predetermined area of ​​person 7.

[0016] The transmitter 21 includes a millimeter wave transmission device and generates and transmits millimeter waves of a predetermined frequency. In the present embodiment, the transmitter 21 transmits millimeter waves in the 79 GHz band among millimeter waves in the 30 GHz to 300 GHz band, but it may also transmit millimeter waves in other frequency bands (for example, the 60 GHz band). Alternatively, a transmitter 21 that transmits microwaves in the 3 GHz to 30 GHz band (for example, the 24 GHz band) instead of millimeter waves may be used. In the biological signal processing device 1 of this embodiment, by using radio waves (millimeter waves) in the high frequency band, the reflected waves from the skin surface (body surface) of the subject 7 become dominant, and therefore phase fluctuations can be effectively observed as fluctuations on the body surface. The transmission wave (millimeter wave) generated by the oscillator 21 is distributed to the transmission antenna 23 and the distribution phase shift unit 24 via a transmission path.

[0017] The distribution phase shift unit 24 has a distribution function and a phase shift function (for example, it is configured by combining a distributor and a phase shifter), and divides the transmission wave into two waves using the reference wave as a reference wave, inputs one of the waves to the mixer 26 in phase with the reference wave, and inputs the other wave to the mixer 27 after shifting the phase by 90°. In this way, the distributing and phase shifting unit 24 generates a reference wave that is in phase with the millimeter wave output from the transmitting antenna 23 and a reference wave that is orthogonal in phase to the millimeter wave output from the transmitting antenna 23, and inputs these to the mixers 26 and 27, respectively.

[0018] Transmitting antenna 23 is configured as a two-element patch array antenna, and irradiates the millimeter waves generated by transmitter 21 toward subject 7 . Receiving antenna 25 is configured as a four-element patch array antenna, and receives the millimeter waves transmitted by transmitting antenna 23 and reflected from subject 7, distributes them, and sends them to mixers 26 and 27. In this way, the biological signal processing device 1 uses the MIMO (Multiple Input Multiple Output) method to transmit and receive using multiple antennas, thereby forming a virtual array antenna for a total of eight channels, but Figure 1 shows an antenna set for one of these channels. Although the number of channels m in the virtual array antenna of this embodiment is m=2×4=8 channels (8ch), a virtual array antenna with a smaller or larger number of channels may be used. For example, a single input multiple output (SIMO) antenna or a multi-input single output (MISO) antenna may be used as the multiple antennas.

[0019] The body surface of the subject 7 moves slightly due to vital movements such as heartbeat and breathing, causing the distance to the receiving antenna 25 to fluctuate, and the phase of the reflected wave fluctuates in response to this change in distance. In this way, the reflected wave received by the receiving antenna 25 contains vital information. On the other hand, the reflected waves may contain small signals from vital signs or may contain reflected waves from sources other than the subject 7.

[0020] The mixer 26 mixes the reflected wave received by the receiving antenna 25 with the transmitted wave from the transmitter 21 to output an in-phase I(t) signal to the signal processing device 4. The mixer 27 mixes the reflected wave received by the receiving antenna 25 with a reference wave (transmitted wave) phase-shifted by 90°, and outputs an orthogonal Q(t) signal to the signal processing device 4. The millimeter wave circuit 2 is equipped with an A / D converter (not shown), and outputs to the signal processing device 4 an I(t) signal and a Q(t) signal (hereinafter referred to as IQ data) after analog-to-digital conversion.

[0021] The signal processing device 4 functions as a biosignal processing device as defined in claim 1, and the biosignal processing device 1 including the signal processing device 4 shown in FIG. 1 functions as a biosignal processing device as defined in claim 2. FIG. 2 is an explanatory diagram mainly showing the hardware configuration of the signal processing device 4. As shown in FIG. 2, the signal processing device 4 includes a CPU 41, a ROM 42, a RAM 43, an input / output unit 44, a storage device 45, and the like.

[0022] The CPU 41 executes various programs such as the biological signal processing program 451 and the CMA program 452 of this embodiment, and thereby performs various processes such as creating a radar cube, applying CMA, and acquiring vital information using the I(t) signal and the Q(t) signal acquired from the millimeter wave circuit 2, thereby realizing various functions. That is, the CPU 41 that executes the biological signal processing program 451 and the CMA program 452 functions as an acquisition means that acquires a phase signal by acquiring the IQ data output from the millimeter wave circuit 2, functions as a selection means that selects a selected phase signal by selecting selected IQ data to which the CMA is to be applied from the created radar cube, functions as a weight update means by applying the CMA to the selected IQ data to determine and update the weight w, and functions as an output means by outputting an output signal y that is the sum of the values ​​obtained by multiplying the selected IQ data by the updated weight. The ROM 42 is a read-only memory that stores basic programs and parameters that the CPU 41 uses to operate the signal processing device 4. RAM 43 is used as a working area when the CPU 41 performs biosignal processing, etc., and in this embodiment, IF data (IF raw data) 431, radar cube 432, selected IQ data 433, weight information 434, vital information 435, and various other data are temporarily stored.

[0023] In the IF data 431, the I(t) signal and Q(t) signal output from the millimeter wave circuit 2 are repeatedly acquired and stored. The transmitting antenna 23 successively transmits frequency-modulated millimeter waves, and n I(t) signals and Q(t) signals are sampled for each transmission unit (chirp set) and stored for each mch (channel). Therefore, for each chirp set, m×n I(t) signals and Q(t) signals are stored in the IF data 431. In this embodiment, the number of samples n per chirp set stored in the IF data 431 is 256, but it can be less (for example, 128) or more (for example, 512). Here, a chirp set is made up of two chirps, which matches the number of transmitting antennas 23 (two). On the other hand, the number of channels m is 8 (2 transmissions x 4 receptions), but it is also possible to set it to (f transmissions x v receptions) channels, for example, (3 transmissions x 4 receptions) = 12 channels.

[0024] In the radar cube 432, IQ data consisting of I(t) and Q(t) stored in the IF data 431 is generated for each channel, sampling, and chirp set and stored. The IQ data is a complex signal generated from both signals I(t) and Q(t) according to equation 22 (h(t) = I(t) + jQ(t)) shown in Figure 1, where j is the imaginary unit and t represents time. The radar cube 432, details of which will be described later, forms a virtual cube by arranging n pieces of IQ data in the sampling direction (hereinafter referred to as the distance direction) and m pieces in the virtual array antenna direction (hereinafter referred to as the channel direction) per chirp set, i.e., n x m pieces of IQ data per chirp set (for example, p pieces) in the direction of the number of chirp sets (hereinafter referred to as the time direction) (see Figure 4(a)). Hereinafter, among the created radar cubes 432, n×m pieces of IQ data per chirp set will be referred to as a unit radar cube (unit RC).

[0025] In the selected IQ data 433, the IQ data to which the CMA in this embodiment is applied is selected from the radar cube 432 and stored. As the IQ data to which CMA is applied, n' x m' units of IQ data arranged in the time direction are selected from the radar cube 432 in which units RC (n x m) are arranged in the time direction, and p' units of IQ data arranged in the time direction are stored in the selected IQ data 433. In this embodiment, as shown in FIG. 4(a), n′ (=5)×m′ (=m=8) pieces of IQ data in the time direction are stored in the selected IQ data 433 out of the units RC. Details of the n' data to be selected will be described later, but by utilizing the properties of the radar cube created using the FMCW method, a predetermined number of IQ data items before and after the distance bin determined corresponding to the distance from the reference point of the antenna 23 (25) to the subject 7 are targeted. The reference point of the antenna is predetermined at a position of the antenna. For example, the reference point may be the position of any one of the multiple antennas 23 (25), the point between the transmitting antenna 23 and the receiving antenna 25 arranged side by side on the same plane, the point between two transmitting antennas 23, or the point between two receiving antennas 25.

[0026] The vital signals of the subject 7 are distributed across multiple IQ data. Therefore, in the biological signal processing device 1 of this embodiment, the phase signals of the IQ data of multiple distance bins (=n') x multiple channels (m'), where not all distance bins and channels, and not both are 1, are integrated by CMA, thereby integrating the vital signals distributed across multiple IQ data to obtain a large vital signal. This makes it possible to acquire more accurate vital information. This is because CMA works by using a large weight w for phase signals with large vital signals, and a small weight w for phase signals with small vital signals and a lot of noise. The magnitude of the weights used by CMA depends on the relative magnitude of the vital signal and noise. Even if the vital signal is small, if the noise is even smaller, it will be synthesized with a large weight. Conversely, if the noise is so large that the cycle of the vital signal cannot be seen, the waveform will be synthesized with a small weight. is synthesized.

[0027] The weight information 434 stores n' x m' pieces of weight information w obtained by CMA for the IQ data stored in the selected IQ data 433 while updating them. The n' x m' pieces of weight information w stored in this weight information 434 are used for beamforming processing. That is, by summing values ​​multiplied by the weights w corresponding to the n' x m' pieces of selected IQ data 433, an output signal y is output in which signal components from the direction of the subject 7, including vital information, are combined.

[0028] Vital information 435 stores vital information such as heart rate and respiratory rate acquired using output signal y integrated (combined after multiplication) using selected IQ data 433 and weight information 434. Other data stored in the RAM 43 include an average radar cube formed from average IQ data, distance bin data, etc., which will be described in detail later.

[0029] The input / output unit 44 includes an input unit and an output unit. The input unit includes input devices such as a touch panel, a keyboard, and a mouse, and is a device for accepting operations from the user (operator) of the signal processing device 4. The output unit includes output devices such as a display, a speaker, and a printer, and displays the operation screen of the signal processing device 4 on the display and outputs analyzed biosignals (vital information) to these output devices.

[0030] The storage device 45 is composed of a readable / writable storage medium and a drive device for reading and writing various information such as programs and data from and to the storage medium. The storage media used in this storage device 45 are mainly semiconductor storage devices such as hard disks and SSDs (Solid State Drives), but external storage media such as semiconductor storage media such as memory chips and IC cards, and storage media from which information can be optically read, such as CD-ROMs, MOs, and PDs (phase change rewritable optical disks), may also be used. The storage device 45 stores, for example, a biological signal processing program 451, a CMA program 452, vital information 453, and other programs and data.

[0031] The biological signal processing program 451 is a program for finally extracting a biological signal such as a pulse wave from the I(t) signal and Q(t) signal output from the millimeter wave circuit 2. The biological signal processing program 451 may be a program that performs processing up to obtaining and outputting an original signal (output signal y) from which a biological signal can be appropriately extracted, without performing processing up to obtaining a specific biological signal such as a pulse wave. In this case, the specific biological signal is obtained from the output original signal at a different timing by the biological signal processing device 1 or another device.

[0032] The CMA program 452 is a program that calculates the weight w used in the beamforming process and stores it in the weight information 434 in the RAM 43 . The CMA program 452 in this embodiment functions as a subroutine program of the biological signal processing program 451, but may be saved as an independent processing program. The CMA performed by the CMA program 452 is a process for calculating a weight w so as to minimize the distortion component of the envelope of the selected IQ data 433. In other words, it is a process for calculating a weight w so as to restore the entire signal to that amplitude by regarding the part of the selected IQ data 433 with the largest amplitude as the correct answer. The CMA program 452 determines the weight w for each selected IQ data 433 according to the formula of the evaluation function Q and the update formula of the weight w, which will be described later.

[0033] The vital information 453 is data that is obtained by reading the vital information 435 obtained by the biological signal processing program 451 from the RAM 43 and saving it in the storage device 45 together with information such as the processing date and time and the subject. Other data stored in storage device 45 include, for example, channel ch and distance bins that are the subject of CMA. Regarding channel ch, all channels are stored in the first and second embodiments, and channel ch1 is stored in the third embodiment. Regarding distance bins, they are stored when a predetermined number of distance bins are determined in advance corresponding to the distance to the subject 7 who is located at a predetermined position.

[0034] Next, the biological signal processing performed by the biological signal processing device 1 to obtain a biological signal (vital information) from the subject 7 will be described. FIG. 3 is a flowchart showing the procedure of the biological signal processing performed by the signal processing device 4. The biological signal processing is performed by the CPU 41 executing a biological signal processing program 451 and a CMA program 452 stored in the storage device 45. As a premise for performing biosignal processing, it is assumed that millimeter waves are output from the millimeter wave circuit 2 to the subject 7, and complex signals, such as IQ data (=h(t)=I(t)+jQ(t)), are sequentially supplied, as shown in Figure 1.

[0035] The CPU 41 acquires the data necessary for analysis (step 10). That is, for each chirp set, the CPU 41 acquires n (=256) I(t) signals and Q(t) signals output from the mixers 26 and 27 of the millimeter-wave circuit 2 for each channel (ch1 to ch8) (acquisition means) and stores them sequentially in the IF data 431 of the RAM 43. Here, 256 x 8 I(t) signals and Q(t) signals are stored per chirp set.

[0036] Next, the CPU 41 creates a radar cube (step 12). Figure 4 conceptually shows the created radar cube 432 and the application of CMA. As shown in Figure 4(a), the radar cube 432 forms a virtual cube in which unit RCs (unit radar cubes), each consisting of n (= 256 samplings) IQ data (= h(t)) in the distance direction and m (= 8 channels) IQ data in the channel direction within each chirp set, are arranged in the time direction.

[0037] Specifically, the CPU 41 creates one IQ data from a pair of I(t) and Q(t) signals from one sample in one channel stored in the IF data 431 using h(t) = I(t) + jQ(t) in equation 22 shown in Figure 1. Then, for one chirp set, the CPU 41 creates a unit RC consisting of IQ data of all channels m (=8) x the number of samples in the chirp set n (=256). The CPU 41 sequentially creates IQ data for each chirp set that is sequentially output from the millimeter wave circuit 2 and stored in the IF data 431 , and stores the data in the radar cube 432 . For convenience of explanation, FIG. 4(a) shows three n×m unit RCs arranged in the time direction, but the unit RCs are stored in the radar cube 432 in order until the measurement is completed.

[0038] Next, the CPU 41 performs frame averaging of the radar cube 432 created in step 12 (step 14). The CPU 41 defines a group of m (number of channels) x n (number of samples) x s (number of chirp sets) in the radar cube 432 as one frame. Then, the CPU 41 averages the s number of IQ data aligned in the time direction, and stores the average radar cube, which is a single average IQ data, in the RAM 43. In one frame, s pieces of IQ data arranged in the time direction are averaged to obtain average IQ data, thereby making it possible to reduce noise in each average IQ data (IQ data).

[0039] Here, the number s of chirp sets (unit RC) that make up one frame is set to 3 in this embodiment as shown in Figure 4(a), but it can also be set to more or less than 3. For example, s may be set to 1, in which case the average IQ data is equal to the IQ data, thereby shortening the processing time.

[0040] In the following explanation, for the sake of convenience of the drawings used for the explanation, we will explain the case where s=3, and will explain the IQ data as the average IQ data of signals arranged in the chirp set number direction, three (s) pieces of IQ data for the same channel and same distance bin. Even when s is a number other than 3, the IQ data explained in each process in the following explanation can be replaced with average IQ data.

[0041] In addition, the process of step 14, in which s pieces of IQ data arranged in the time direction are averaged to create one average IQ data, can be omitted as appropriate if the measurement environment is good and noise is relatively small, such as when the movement of the subject 7 is relatively small.

[0042] Next, the CPU 41 acquires selected IQ data for performing CMA from the radar cube 432 (steps 16 and 18). First, the CPU 41 determines the channel ch and distance bin to be subjected to CMA (step 16). The target channel is determined by determining a specific channel from which IQ data is to be selected from among the array antennas of all channels. Furthermore, the determination of the distance bin is the determination of the distance bin from which IQ data is selected, and the distance bin corresponding to the distance from the antenna 23 (25) to the subject 7 is determined. When determining the distance bin, it is preferable to select IQ data that includes more biometric information of the subject 7.

[0043] In this embodiment, the CPU 41 makes the determination for all channels (m=8) and for five distance bins out of a total of n distance bins. In this embodiment, the five distance bins to be used are determined in advance as follows. By applying a fast Fourier transform (FFT) to the sampling direction in one chirp set acquired by the FMCW method, the distance from the antenna 23 (25) to the subject 7 can be obtained. Therefore, when the subject 7 is in a predetermined position, such as when he or she is driving a vehicle or sitting in a designated chair, the distance L to the subject 7 is actually measured, and five distance bins sampling at distances corresponding to the actual measured value L are calculated in advance and stored in the memory device 45 as described above. In this way, when the position where the person being measured is present is known in advance, the corresponding distance bin may be determined so as to include the area around that position.

[0044] In this embodiment, the CPU 41 determines (reads out) the target channel and distance bin that are previously stored in the storage device 45 (step 16), and obtains the IQ data, i.e., the IQ data of ch1-bin1, IQ data of ch1-bin2, ..., IQ data of ch1-bin5, IQ data of ch2-bin1, ..., ch8-bin5, which are displayed in color in Figure 4(a), from the radar cube 432 (selection means), and stores this as selected IQ data 433 in the RAM 43 (step 18). Then, the CPU 41 applies CMA to the phase signals of the 40 IQ data from ch1-bin1 to ch8-bin5 of the saved selected IQ data 433, thereby calculating and updating the weight w for each IQ data (step 20, weight update means).

[0045] That is, the CPU 41 reads out the CMA program 452 , which is a subroutine, from the storage device 45 and executes it, thereby applying CMA to the selected IQ data 433 . This CMA is an algorithm that controls the weight w so that the distortion component of the envelope in the selected IQ data 433 is minimized, taking advantage of the fact that the amplitude of the measurement signal fluctuates due to noise, while the amplitude (envelope) of the biological signal can be assumed to be constant.

[0046] FIG. 5 is a flowchart for explaining the procedure for calculating the weights by the CMA program 452. FIG. 6 is an explanatory diagram showing mathematical expressions used in biological signal processing including CMA. The CPU 41 uses the selected IQ data 433 as the input signal x and calculates the signal strength in accordance with the evaluation function Q shown in FIG. 6(a) (step 201). In this evaluation function Q, the expression in [ ] represents the power of the output signal, and the value of the evaluation function Q decreases as the power approaches the desired power δ×δ. Here, w, x, δ, E, and H in the formula of the evaluation function Q are as follows: w: weight (vector) x: Input signal (vector: Select IQ data 433) δ: desired amplitude of the output signal E: Ensemble average H: Hermitian transpose

[0047] Next, the CPU 41 uses the update formula W based on the steepest descent method shown in Figure 6(b) for the signal strength calculated according to the evaluation function Q to calculate a weight w that reduces the value of Q (step 202). The weight w calculated here is a vector consisting of (n' x m') weights w corresponding to each IQ data that makes up the selected IQ data 433. In addition, μ in the equation of FIG. 6(b) is a coefficient (μ>0) that determines the update width. The CPU 41 updates the value of the weight information 434 in the RAM 43 to the load w calculated using the update formula W (step 203), and returns to the main routine shown in FIG.

[0048] Returning to FIG. 3, the CPU 41 integrates (combines) and outputs the selected IQ data 433 (step 22). That is, as shown in the equation of Figure 6(c), the CPU 41 calculates an output signal y by combining an input signal x having each IQ data of the selected IQ data 433 as an element and a weight w of the updated weight information 434 (vector), and outputs the output signal y as an output signal from which vital information can be obtained (output means) and also stores it in the RAM 45. In addition, N in FIG. 6(c) is N=n′×m′=5×8=40.

[0049] Furthermore, the CPU 41 acquires vital information from the output signal y synthesized based on the equation of FIG. 6(c) using various well-known methods (step 24), stores the information in the vital information 435 of the RAM 43, and ends the process. A well-known processing method for acquiring vital information is, for example, to apply a band-pass filter of 0.7 to 2 Hz, detect peaks, and calculate the instantaneous heart rate from the peak interval.

[0050] In the biosignal processing explained with reference to FIG. 3, the processing flow from data acquisition (step 10) to vital information acquisition (step 24) has been mainly explained. In actual biological signal processing, data acquisition in step 10 and radar cube creation in step 12 are performed continuously from the start of measurement to the end of processing. On the other hand, the processes from step 14 to step 24 are repeatedly and continuously performed on the radar cube 432 created in step 12 for each frame unit (a group of number of channels m × number of samplings n × number of chirp sets s).

[0051] As described above, according to the biological signal processing device 1 of this embodiment, from the radar cube 432 created using signals from the subject 7 obtained using the FMCW method, IQ data of a predetermined number n' (bin1 to bin5) of distance bins corresponding to the distance to the subject 7 in all channels (m' = m), i.e., m × n' = 8 × 5 = 40 pieces of IQ data, are selected as IQ data 433. These 40 IQ data (selected IQ data 433) are arranged as shown in Figure 4(b), and the 40 weights w for each IQ data are updated by applying CMA assuming that they are input signals x from 40 virtual array antennas.The output (synthesized) signal y is calculated from the updated weights w and the selected IQ data 433 using the synthesis formula in Figure 6(c).

[0052] In CMA, the weight w is updated so that the distortion components of the envelope of the output signal y, which is obtained by multiplying multiple input signals x by the weight w and synthesizing them, are minimized (the amplitude is constant). Therefore, by applying CMA to the selected IQ data 433, it is possible to obtain an output signal y that reduces noise while amplifying vital signals such as heart rate and breathing, which have an almost constant amplitude, thereby enabling highly accurate vital measurement. In particular, since the distance bin corresponding to the distance to the subject 7 and the distance bins before and after it are targeted for the selected IQ data 433, it is possible to more reliably include distance bins with large vital signals, and by applying CMA, it is possible to obtain an output signal y with a larger vital signal. Therefore, highly accurate vital information can be obtained from the output signal y.

[0053] Next, a second embodiment and a third embodiment of the biological signal processing device 1 will be described. In the first embodiment described above, as shown in FIG. 4(a), a case was described in which CMA is applied to input signals x per frame with a chirp set number s=3, where m′=m=8 pieces of IQ data for all channels in the virtual array direction and n′=5 pieces in the sampling direction, totaling 8×5=40 pieces of IQ data (selected IQ data 433). In contrast to this, in the second and third embodiments, for an m×n radar cube 432 per frame, selected IQ data 433 are IQ data with different numbers of channels m′ and distance bins n′. In the second and third embodiments of the biosignal processing device 1, although the target is different, vital information is acquired using selected IQ data (selected IQ data 433) in the same manner as the biosignal processing and CMA described in Figures 3 and 5 in the first embodiment.

[0054] FIG. 7 is an explanatory diagram conceptually showing application of CMA in the second embodiment. In the biological signal processing device 1 of the second embodiment, as shown in Figure 7(a), the number of channels m' selected per frame is all channels, i.e., m' = m = 8, as in the first embodiment, and the number of distance bins n' is 1 (n' = 1). Then, as shown in Figure 7(b), m' x n' = 8 x 1 = 8 pieces of IQ data (selected IQ data 433) per frame are used as the input signal x, and by applying CMA to this, eight weights w are updated, and an output signal y is obtained by combining the selected IQ data 433 and the weights w. According to the second embodiment, the number of selected IQ data 432 per frame is small, so it is possible to reduce the processing load on the CMA more than in the first embodiment. However, in the second embodiment, since there is one target distance bin, it is preferable that the position of the person being measured 7 is fixed. Furthermore, it is preferable to target distance bins that have a higher degree of agreement with the actually measured distance L to the subject 7.

[0055] FIG. 8 is an explanatory diagram conceptually showing application of CMA in the third embodiment. In the biological signal processing device 1 of the third embodiment, as shown in FIG. 8(a), unlike the first and second embodiments which target all channels, the number of channels m′ per frame is set to only one channel, and the number of distance bins n′ is set to 1. <n′<nとしている。 Specifically, there is one channel (ch1), and the number of distance bins n' is set to 5, the same as in the first embodiment. Then, as shown in Figure 8(b), m' x n' = 1 x 5 = 5 pieces of IQ data (selected IQ data 433) per frame are used as input signal x, CMA is applied to update the five weights w, and an output signal y is obtained by integrating the selected IQ data 433 and the weights w. According to the third embodiment, as in the second embodiment, the number of selected IQ data 433 per frame is small, so that it is possible to reduce the processing load on the CMA.

[0056] FIG. 9 is an explanatory diagram for evaluation of the biological signals acquired according to the first to third embodiments. FIG. 9 shows the error (heart rate error) between the heart rate calculated from an ECG (electrocardiogram) and the heart rate obtained by the first to third embodiments for the same subject 7. 9, the heart rate error is 0.86 (first embodiment), 1.14 (second embodiment), and 1 (third embodiment), and in all cases it falls within the allowable error range (for example, ±5%). For example, assuming a human heart rate is 60 bpm, the allowable error range is preferably ±3 bpm, which is ±5% of that.

[0057] Next, a fourth embodiment will be described. In the biological signal processing apparatus 1 of the first to third embodiments described above, CMA is applied using an evaluation function Q (see FIG. 6(a)) that has no restrictions on the direction from which the reflected signal arrives. In contrast to this, in the biological signal processing device 1 of the fourth embodiment, CMA is applied using an evaluation function Q to which a restriction term A regarding the signal arrival direction is added. The radar cube 432 created in the fourth embodiment and the selected IQ data 433 to be selected are the same as those in the first to third embodiments and their modifications.

[0058] FIG. 10 shows a measurement model in the fourth embodiment. FIG. 11 is an explanatory diagram showing the mathematical expressions used in the fourth embodiment. 10 shows a measurement model in which eight antenna elements and five distance bins are targets of selected IQ data 433, as in the first embodiment. Then, under the assumption that the distance between the object (subject 7) and the antenna is sufficiently long compared to the wavelength, and therefore the signals from distances bin1 to 5 can be considered as plane waves, the signals from distances bin1 to 5 are represented by a single arrow. In the biological signal processing device 1 of the fourth embodiment, an evaluation function Q shown in FIG. 11(a) is used.

[0059] The evaluation function Q used in the fourth embodiment is composed of a first term and a second term as shown in FIG. 11(a), and the first term is the same as the evaluation function Q in the first embodiment (see FIG. 6(a)) that does not restrict the arrival angle, to which a restriction term A (second term) regarding the direction in which the signal arrives is added. In the limiting term A, α is a contribution coefficient of 0<α≦1 that represents the contribution of the signal arrival angle limiting term. If the contribution coefficient α is small, the value of the entire term becomes small regardless of the value of the arrival angle limiting term, and the contribution becomes small.

[0060] In the constraint term A, a is a steering vector indicating the position of the subject 7, and is expressed in FIG. 11(b). In this steering vector a, λ is the wavelength of the millimeter wave transmitted from the transmitting antenna 23. As shown in FIG. 10, d1 to d8 are the propagation path length differences between the propagation path length of the millimeter wave to each antenna element (2 transmission × 4 reception = 8 elements) and the propagation path length to the reference element. g1 to g5 are the propagation path length differences between the propagation path length of the millimeter wave to each distance bin and the propagation path length to the reference element.

[0061] The denominator of the first term in the parentheses in the constraint term A (the part excluding α of constraint term A) represents the power obtained when a plane wave arriving from the direction of steering vector a is received with weight w, and is maximized when the direction of steering vector a and the direction of weight w (vector) match. The numerator of the first term in the parentheses is a coefficient for normalization. When the direction of the steering vector a and the direction of the weight w (vector) match, the maximum value of the first term in the parentheses in the constraint term A is 1, and the minimum value of the terms in the parentheses as a whole is 0.

[0062] The CPU 41 calculates the signal strength of the selected IQ data 433 according to the evaluation function Q to which the limiting term A has been added, and uses the update formula W shown in Fig. 11(c) to calculate and update the weight w that reduces the value of Q. Note that the update formula W in Fig. 11(c) is the same as the update formula W in Fig. 6(b) described in the first embodiment. The CPU 41 integrates (combines) the selected IQ data 433 using the weight w of the updated weight information 434 and outputs it.

[0063] FIG. 12 shows the results of verifying the effect of the evaluation function Q to which the constraint term A is added. In this verification, as shown in Figure 12(a), two speakers S1 and S2 were placed at equal distances (2.7 m) from the millimeter-wave circuit 2 of the biological signal processing device 1, and speaker S1 was vibrated at a frequency of 1 Hz and speaker S2 was vibrated at a frequency of 2 Hz. Figure 12(b) shows the measurement results when CMA is applied to an evaluation function Q without the constraint term A. As shown in Figure 12(b), since there are no constraints on the direction from which the signal arrives, peaks are detected at frequencies of 1 Hz and 2 Hz output from both speakers S1 and S2. On the other hand, Fig. 12(c) shows the measurement results when CMA is applied to an evaluation function Q to which a restriction term A is added, which restricts the signal arrival direction to the direction of speaker S1. As shown in Fig. 12(c), by adding the restriction term A, the peak value at a frequency of 1 Hz output from speaker S1 is maintained, and the peak at a frequency of 2 Hz output from speaker S2 is suppressed. In this way, by adding the restriction term A to the evaluation function Q, it becomes possible to receive (synthesize) signals mainly from the restricted direction.

[0064] According to the fourth embodiment, CMA is applied to the selected IQ data 433 using an evaluation function Q to which a restriction term A regarding the direction from which the signal arrives is added, so that the signal from the direction of the subject 7 can be effectively extracted and an output signal y including a biological signal can be output.

[0065] According to each of the embodiments and variants described above, CMA is applied to m' pieces of IQ data in the channel direction x n' pieces of IQ data (selected IQ data 433) in the sampling direction per frame (s chirp sets) from the radar cube 432 created using signals from the subject 7 obtained using the FMCW method, thereby reducing the processing load and enabling biological signals to be obtained quickly. Further, as the target of the selected IQ data 433, n' distance bins corresponding to the distance to the subject 7 are used as the target of the selected IQ data 433. Therefore, distance bins with large vital signals can be more surely included, an output signal y with a larger vital signal can be obtained by applying CMA, and furthermore, highly accurate vital information can be obtained from the output signal y.

[0066] As described above, one embodiment of the biological signal processing apparatus 1 of the present invention has been described. However, the present invention is not limited to the described embodiment, and it is possible to make modifications to the scope described in each claim and the scope described in the embodiment, and further make further modifications to other modification examples. For example, in the described first to fourth embodiments, the case where a predetermined number n' of distance bins are obtained in advance by actually measuring the distance L to the subject 7 has been described. However, n' distance bins may be determined based on the created radar cube 432. That is, any one channel (for example, ch1, the central channel ch3 or ch4, etc.) is selected, and a distance FFT is applied to the n (= 256) IQ data in the distance direction to determine a peak distance bin that takes the peak value of the distance. When the number of distance bins is n' = 1, it is determined as the peak distance bin. On the other hand, when 2 ≤ n' < n and n' is odd, it is determined as the peak distance bin and (n' - 1) / 2 distance bins before and after it. When n' is even, it is determined as n' / 2 distance bins on one side and (n' / 2) - 1 distance bins on the other side before and after the peak distance bin. Further, in parallel with the creation of the radar cube 432, n' distance bins may be determined every time the number p of chirp sets reaches a predetermined number.

[0067] When the number of channels m' is m' = 1, in the determined one channel, the above distance FFT is applied to determine n' distance bins by the peak distance bin and the distance bins before and after it. When the number of channels m′ satisfies 2 ≤ m′ < m, for any one channel described above, the peak distance bin determined by applying the distance FFT (when m′ = 1), or the determined peak distance bin and the distance bins before and after it (when m′ ≠ 1) are determined. That is, all m′ channels use the same distance bin.

[0068] Also, in the third embodiment, m′ = 1 and ch1 is the target channel, but it may also be 2 ≤ m′ ≤ m (however, when m′ = m, n′ < n). For this target channel, an arbitrary channel determined in advance may be used, or a channel that satisfies a predetermined criterion may be selected. As the predetermined criterion, for example, a channel with an SN ratio of a predetermined threshold or higher, when m′ = 1, the channel with the largest SN ratio, the m′ channels in descending order of SN ratio, etc. can be selected. The method of selecting the n′ distance bins for these m′ channels is as described in the embodiment and the above modification.

[0069] In each of the described embodiments and their modifications, when selecting one distance bin or five distance bins, that is, when n′ = 1, n′ = 5 is described, but it is also possible to select other values such as n′ = 3, 7, 8, 25, 50, etc. It is also possible to set n′ = n - 1, but when all channels are selected and m′ = m, the number of weights w determined and updated by applying CMA becomes extremely large, increasing the processing burden and time. Therefore, it is desirable not to make the number of n′ too large. For example, it is preferable that n′ ≤ n / 10. As described in each embodiment, when the number of samplings n = 256, it is preferably 25 or less. Also, when m′ < m, it is possible to set n / 10 < n′ < n, but it is preferable to select within the range of m′ × n′ < m × (n / 10).

[0070] In the embodiment and modified examples described above, the case where a phase signal is acquired using the IQ data acquired from the millimeter wave circuit 2 has been described. That is, a case has been described in which a complex signal consisting of an in-phase signal I and a quadrature signal Q, which are made up of a radio wave transmitted by a transmitting antenna and a reflected wave received by a receiving antenna, is acquired. On the other hand, the phase signal may be a complex signal consisting of only an in-phase signal I or a quadrature signal Q. Even in this case, by applying a range FFT to a radar cube formed from an in-phase signal I or a quadrature signal Q, a complex number containing phase information can be obtained. Therefore, even when the CPU 41 acquires only the in-phase signal I or the quadrature signal Q, it can obtain a vital signal using the same processing procedure as the biological signal processing described with reference to FIG.

[0071] Furthermore, in the described embodiments and variants, millimeter waves are used for biological signal processing, but other radio waves, such as the above-mentioned microwaves or submillimeter waves (decimeter waves) in the 300 GHz to 3 THz band, may also be used. In this case, the transmitter 21 is made to emit microwaves or submillimeter waves.

[0072] In the embodiment and modified examples described above, the I(t) signal and Q(t) signal output from the millimeter wave circuit 2 are directly acquired by the signal processing device 4. On the other hand, the signal processing device 4 can also acquire the I(t) signal and the Q(t) signal at a location separate from the subject 7 and the millimeter-wave circuit 2 via various communication means such as the Internet. For example, a millimeter-wave circuit 2 placed in a hospital acquires and outputs the I(t) and Q(t) signals of a patient 7, who is the subject of measurement, and these signals can be acquired by a signal processing device 4 placed in another location via a network such as the Internet. In this way, it becomes possible to process the biological signals of the subject 7 who is in a remote location. Furthermore, millimeter-wave circuits 2 can be arranged in multiple locations, and one signal processing device 4 can perform biosignal processing for each I(t) signal and Q(t) signal transmitted from each millimeter-wave circuit 2.

[0073] In the embodiments and variants described above, when an I(t) signal and a Q(t) signal are output from the millimeter wave circuit 2, an original signal (output signal y) from which a biosignal can be appropriately extracted in real time is output, and the biosignal is acquired. On the other hand, it is also possible to perform acquisition of data for biological signal processing and output of the output signal y based on the acquired data (and further acquisition of the biological signal) at different times. That is, at a predetermined timing, the signal processing device 4 transmits millimeter waves to the subject 7 and receives reflected waves, thereby acquiring the I(t) signal and Q(t) signal output from the millimeter wave circuit 2. Thereafter, the signal processing device 4 functioning as claim 1 acquires m×n phase signals from the I(t) signal and Q(t) signal stored in the storage means A, and performs the biological signal processing described in FIG. The IQ data may be stored in a predetermined storage means A from the millimeter wave circuit 2, and the signal processing device 4 may acquire the IQ data (phase signal) from the storage means A. In this case, the storage means A may be provided in either the biological signal processing device 1 or the signal processing device 4, or in another device (for example, a test data storage server, etc.). In this way, the signal processing device 4 acquires the IQ data output from the millimeter wave circuit 2, and the manner of acquisition may include direct acquisition, indirect acquisition, real-time acquisition, or later acquisition.

[0074] In the embodiment and modified examples described above, the case where s pieces of IQ data arranged in the time direction from the radar cube 432 created in step 12 are averaged (frame averaged) (step 14) has been described. On the other hand, all chirps may be used for CMA (ensemble averaging) without performing frame averaging (stepping) in the time direction. When frame averaging is performed, the weight w is updated every ensemble average number, whereas when frame averaging is not performed, the ensemble average is based on the IQ data within the frame, so the weight w is updated every frame. When using frame averaging and ensemble averaging, for example, if the subject 7 is moving and tracking ability needs to be improved, it is better to update the weights in detail, so it is preferable to apply ensemble averaging.

[0075] In the embodiment and modified examples described above, the weight w is calculated by the steepest descent method using the update formula of FIG. 6(b) and the weight information 434 is updated. Alternatively, the weight w may be calculated and the weight information 434 may be updated by other methods such as the Marquardt method, which minimizes the error using the gradient of a function. The weight w calculated using the update formula W based on the steepest descent method is calculated for each frame, and the weight information 434 is updated. The weight w of the updated weight information 434 gradually approaches an optimal value by repeating the update for each frame. In contrast to this, in the case of the Marquardt algorithm, it is determined whether the weight w calculated for each frame has converged (whether Q has become the smallest), and if the weight w has converged, the weight information 434 is updated. Therefore, for a frame before the weight information 434 is updated, step 22 is processed using the weight w before the update stored in the weight information 434.

[0076] In the embodiment and modified examples described above, the biological signal processing device can be configured as follows. (Configuration 1) An acquisition means for acquiring n phase signals per s chirps of a transmission wave from a transmission means, which is frequency-continuously modulated and transmitted to a subject, and its reflected wave, for each of m channels; a selection means for selecting a predetermined number (m' x n') of target phase signals as selected phase signals from the radar cube in which the (m x n) phase signals acquired by the acquisition means are arranged; By applying the CMA with each of the selected phase signals as an input signal x, weight update means for determining and updating each weight w such that the distortion component of the envelope in the output signal y synthesized by multiplying the weight w corresponding to each input signal x is minimized; Output means for outputting an output signal y obtained by synthesizing the input signal x using each updated weight w; A biological signal processing device, characterized by comprising the above. (Configuration 2) Further comprising transmission means for transmitting the transmission wave; The biological signal processing device according to Configuration 1, characterized by the above. (Configuration 3) The selection means selects m'×n' phase signals, where at least one of m' and n' is 2 or more and m'≠m and n'≠n, as the selected phase signals; The biological signal processing device according to Configuration 1 or Configuration 2, characterized by the above. (Configuration 4) The selection means selects a phase signal corresponding to the distance between the measured person and the reference point as at least one of the n' phase signals out of the n phase signals where 1≦n'<n; The biological signal processing device according to Configuration 1, Configuration 2, or Configuration 3, characterized by the above. (Configuration 5) The selection means selects n' phase signals before and after including the phase signal of the peak value obtained by FFT processing on the n phase signals in a predetermined one of the m channels as the phase signal corresponding to the distance; The biological signal processing device according to any one of Configurations 1 to 4, characterized by the above. (Configuration 6) The selection means acquires phase signals acquired in m' (1≦m'<m) channels out of the m channels as the selected phase signals; The biological signal processing device according to any one of Configurations 1 to 5, characterized by the above. (Configuration 7) The selection means determines m' channels based on the signal-to-noise ratio; The biological signal processing device according to any one of Configurations 1 to 6, characterized by the above. (Configuration 8) The weight update means calculates each weight w for each of the input signals x by applying a CMA (Constant Modulus Algorithm) to the input signals x. 8. The biological signal processing device according to any one of configurations 1 to 7, wherein: (Configuration 9) The weight updating means applies CMA to the input signal x using an evaluation function Q to which a restriction term A for the direction of signal arrival is added. 9. The biological signal processing device according to any one of configurations 1 to 8, wherein: (Configuration 10) A biological signal acquisition means for acquiring a biological signal from the synthesized output signal y; 10. The biological signal processing device according to any one of configurations 1 to 9, comprising: (Configuration 11) An acquisition function for acquiring n phase signals per s chirps of a transmission wave from a transmission means, which is frequency-continuously modulated and transmitted to a subject, and the reflected wave, for each of m channels; a selection function for selecting a predetermined number (m' x n') of target phase signals as selected phase signals from the radar cube in which the (m x n) phase signals acquired by the acquisition means are arranged; a weight update function that determines and updates each weight w that minimizes the distortion component of the envelope in an output signal y obtained by multiplying each input signal x by a weight w corresponding to the input signal x and combining the input signals x by applying CMA to each of the selected phase signals; an output function that outputs an output signal y obtained by combining the input signals x using the updated weights w; A biosignal processing program that enables a computer to achieve this. (Configuration 12) An acquisition step of acquiring n phase signals per s chirps of a transmission wave from a transmission means, which is frequency-continuously modulated and transmitted to a subject, and the reflected wave, for each of m channels; a selection step of selecting a predetermined number (m' x n') of target phase signals as selected phase signals from the radar cube in which the (m x n) phase signals acquired by the acquisition means are arranged; a weight update step of determining and updating each weight w that minimizes the distortion component of the envelope in an output signal y obtained by multiplying each of the selected phase signals x by a corresponding weight w by applying CMA to the input signal x; an output step of outputting an output signal y obtained by combining the input signals x using the updated weights w; A biological signal processing method comprising: [Explanation of symbols]

[0077] 1. Biosignal processing device 2 Millimeter-wave circuits 3. Control device 4. Signal Processing Device 7 Person to be measured 21 Transmitter 23 Transmitting Antenna 24 Distribution phase shift section 25 receiving antenna 26, 27 mixer 41 CPU 42 ROM 43 RAM 431 IF data 432 Radar Cube 433 Selected IQ Data 434 Weight Information 435 Vital Information 44 Input / output section 45 Storage device 451 Biosignal Processing Program 452 CMA Program 453 Vital Information w weight

Claims

1. an acquisition means for acquiring n phase signals per s chirps of a transmission wave from a transmission means which is frequency-continuously modulated and transmitted to a subject, and the reflected wave, for each of m channels; a selection means for selecting a predetermined number (m' x n') of target phase signals as selected phase signals from the radar cube in which the (m x n) phase signals acquired by the acquisition means are arranged; a weight update means for determining and updating weights w that minimize distortion components of an envelope in an output signal y obtained by multiplying each of the selected phase signals x by a corresponding weight w by applying CMA to the input signal x; an output means for outputting an output signal y obtained by combining the input signals x using the updated weights w; A biological signal processing device comprising:

2. Further, a transmitting means for transmitting the transmission wave is provided. The biological signal processing device according to claim 1 .

3. the selecting means selects m' x n' phase signals as selected phase signals, where at least one of m' and n' is 2 or more and m' = m and n' = n is not true; The biological signal processing device according to claim 1 .

4. the selecting means selects a phase signal corresponding to the distance between the subject and a reference point as at least one phase signal from among n' phase signals, where 1≦n'<n. The biological signal processing device according to claim 1 .

5. the selecting means selects n' phase signals before and after the phase signal of the peak value obtained by FFT processing of the n phase signals in a predetermined channel out of the m channels as phase signals corresponding to the distance; 5. The biological signal processing device according to claim 4.

6. the selection means acquires, as selected phase signals, phase signals acquired from m' (1≦m'<m) channels out of m channels; The biological signal processing device according to claim 1 .

7. the selecting means determines m' channels based on the signal-to-noise ratio; 7. The biological signal processing device according to claim 6.

8. the weight update means applies a constant modulus algorithm (CMA) to the input signal x to obtain each weight w for each input signal x; The biological signal processing device according to claim 1 .

9. the weight updating means applies CMA to the input signal x using an evaluation function Q to which a restriction term A for the direction of signal arrival has been added; The biological signal processing device according to claim 8 .

10. a biosignal acquisition means for acquiring a biosignal from the synthesized output signal y; 10. The biological signal processing device according to claim 1, further comprising:

11. an acquisition function for acquiring, for each of m channels, n phase signals per s chirps of a transmission wave from a transmission means that is frequency-continuously modulated and transmitted to a subject and its reflected wave; a selection function for selecting a predetermined number (m' x n') of target phase signals as selected phase signals from the radar cube in which the (m x n) phase signals acquired by the acquisition means are arranged; a weight updating function that determines and updates each weight w that minimizes distortion components of an envelope in an output signal y obtained by multiplying each input signal x by a weight w corresponding to the input signal x and combining the output signal y by applying CMA to each of the selected phase signals as an input signal x; an output function that outputs an output signal y obtained by combining the input signals x using the updated weights w; A biosignal processing program that enables a computer to achieve this.

12. an acquisition step of acquiring n phase signals per s chirps of a transmission wave from a transmission means, which is frequency-continuously modulated and transmitted to a subject, and the reflected wave, for each of m channels; a selection step of selecting a predetermined number (m' x n') of target phase signals as selected phase signals from the radar cube in which the (m x n) phase signals acquired by the acquisition means are arranged; a weight updating step of determining and updating weights w that minimize distortion components of an envelope in an output signal y obtained by multiplying each of the selected phase signals x by a weight w corresponding to the input signal x and combining the input signals x using CMA; an output step of outputting an output signal y obtained by combining the input signal x using the updated weights w; A biological signal processing method comprising:

Images