Human body activity detection method and system based on wireless signals

Abstract

The invention relates to the technical field of wireless sensing, and particularly discloses a human body activity detection method and system based on wireless signals. Demodulation is carried out based on different time windows through linear frequency modulation wave signals, the method is suitable for detecting the dynamic state and the static state of a target, detection of a static target is optimized, accurate monitoring can be achieved, and then posture changes and tiny physiological activities of a human body are captured and analyzed in a high-sensitivity mode; comprising but not limited to monitoring of respiratory rate, tracking of heartbeat rhythm and real-time detection of spatial position change, and a firmer guarantee is provided for user safety.

Description

Technical Field

[0001] This invention relates to the field of wireless sensing technology, and in particular to a method and system for detecting human activity based on wireless signals. Background Technology

[0002] With the increasing global trend of aging, health monitoring of the elderly has become a focus of social attention. Traditional health monitoring methods, such as camera-based surveillance systems, are greatly limited in their application in privacy-sensitive areas such as bathrooms and bedrooms, and pose potential risks of privacy breaches. Furthermore, existing monitoring products on the market often have limited functionality, failing to meet the diverse health monitoring needs of the elderly, and suffer from problems such as high false alarm rates and poor user experience. Summary of the Invention

[0003] To address the shortcomings of existing health monitoring systems for the elderly, such as insufficient privacy protection, limited functionality, and high false alarm rates, this invention provides a human activity detection method and system based on millimeter-wave radar technology. This system enables functions such as fall detection, respiratory and heart rate monitoring, indoor activity monitoring, bathroom stay time monitoring, bed rest time monitoring, and user behavior analysis, while simultaneously protecting user privacy.

[0004] To achieve the above-mentioned objectives, the present invention provides the following technical solution:

[0005] In a first aspect, the present invention provides a method for detecting human activity based on wireless signals, the method comprising:

[0006] The control radar module transmits a linear frequency modulated wave signal to the area under test, and receives the echo signal of the linear frequency modulated wave signal reflected by the area under test through the radar module;

[0007] The echo signal is accumulated based on the first time window and demodulated to obtain the dynamic point cloud of the area under test. The echo signal is accumulated based on the second time window and demodulated to obtain the static point cloud of the area under test.

[0008] The activity characteristics of the target human body are determined by combining the dynamic point cloud and the static point cloud;

[0009] The activity characteristics include any one or more of the target human body's position, height, speed, acceleration, respiratory rate, heart rate, and heart rate variability.

[0010] According to one specific implementation, in the above detection method, the span of the first time window is 0 to 0.2 s.

[0011] According to one specific implementation, in the above detection method, the span of the second time window is 0.4 to 4 seconds.

[0012] According to one specific implementation, in the above detection method, the linear frequency modulated wave signal adopts a chirp waveform.

[0013] According to a specific implementation, in the above detection method, the echo signal is demodulated and distance is estimated using Fast Fourier Transform, and the angle of arrival is estimated and the target signal is extracted using digital beamforming and constant false alarm rate detection algorithms to obtain the dynamic point cloud and the static point cloud.

[0014] According to one specific embodiment, the above detection method further includes:

[0015] Based on the location of the target human body, a tracking algorithm is used to track the target human body, and the location during the tracking process is recorded to generate the trajectory of the target human body;

[0016] Based on time series analysis and spatial clustering, the layout features of the test area are identified by combining the trajectory of the target human body, including the position of the bed, the layout of the aisle, the orientation of the sofa, and the location of the entrance and exit.

[0017] According to one specific embodiment, the above detection method further includes:

[0018] During the tracking of the target human body, a fall detection is performed when a change in the height of the target human body is detected.

[0019] The fall detection includes:

[0020] The dynamic point cloud, static point cloud, and the activity features of the target human body are input into a machine learning algorithm to extract the height and speed changes of the target during movement for fall recognition.

[0021] During the identification process, the static point cloud is used to determine whether the target human body has respiratory characteristics;

[0022] If so, a warning will be issued upon detecting a fall of the target human body;

[0023] If not, the target human body is detected as non-living and is removed.

[0024] According to one specific embodiment, the above detection method further includes:

[0025] Based on the identified location of the bed in the test area, a virtual safe zone is set at the location of the bed. When the target human body moves to the safe zone during sleep and is accompanied by a change in height, the fall detection is performed.

[0026] According to one specific embodiment, the above detection method further includes:

[0027] By combining the trajectory of the target human body and the layout characteristics of the area to be tested, the key health indicators of the target human body are analyzed and obtained. The key health indicators include: activity level, bed rest time, number of times to urinate at night, and the duration of each time to urinate at night.

[0028] Based on the acquired activity characteristics of the target human body, combined with the key health indicators, a personalized health analysis report of the target human body is generated.

[0029] According to one specific embodiment, the above detection method further includes:

[0030] The personalized health analysis report is uploaded to the cloud platform, and when an abnormality is found in the personalized health analysis report, family members or medical personnel are notified through the terminal device. The notification method includes one or more of the following: APP reminder, SMS reminder, and telephone reminder; the abnormality includes sleep apnea and heart disease.

[0031] According to one specific embodiment, the above detection method further includes:

[0032] By collecting the height of the target human body at different positions in the area to be tested, error data of the target human body height at different distances and angles are obtained;

[0033] The height of the target human body is dynamically adjusted and calibrated based on the error data.

[0034] Secondly, the present invention provides a human activity detection system based on wireless signals, the system comprising:

[0035] A radar module comprising at least one transmitting antenna and at least one receiving antenna, for transmitting a linear frequency modulated wave signal toward a region under test, and for receiving an echo signal of the linear frequency modulated wave signal reflected from the region under test;

[0036] A processing module is used to control the radar module and perform human activity detection using a human activity detection method based on wireless signals as described in any of the above-mentioned methods.

[0037] A cloud platform for receiving personalized health analysis reports;

[0038] The terminal device is used to notify family members and / or medical personnel when an abnormality is found in the personalized health analysis report.

[0039] According to one specific implementation, in the above detection system, the system includes an angle sensor for detecting the installation angle and automatically calibrating the installation error; the radar module is installed by side mounting on the wall or top mounting on the roof, and the height is not less than 2 meters. In the case of side mounting, the angle between the radar module and the horizontal plane is adjustable.

[0040] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0041] Based on the technical solution provided in the first aspect above, the present invention transmits a linear frequency modulated wave signal and demodulates the accumulated echo signal based on a preset time window, thereby simultaneously obtaining the dynamic point cloud and static point cloud of the area to be measured. It is suitable for detecting the dynamic and static states of targets, optimizes the detection of stationary targets, and can achieve accurate monitoring. In addition, it can capture and analyze the changes in human body posture and subtle physiological activities with high sensitivity. Specifically, it includes, but is not limited to, monitoring of respiratory rate, tracking of heart rhythm, and real-time detection of spatial position changes, providing a more solid guarantee for user safety. Attached Figure Description

[0042] Figure 1 A schematic diagram of a human activity detection system based on wireless signals provided in an embodiment of the present invention;

[0043] Figure 2 This is a flowchart illustrating a human activity detection method based on wireless signals provided in an embodiment of the present invention.

[0044] Figure 3 This is a schematic diagram of user trajectory analysis provided in an embodiment of the present invention;

[0045] Figure 4 This is a schematic diagram of a fall detection process provided in an embodiment of the present invention;

[0046] Figure 5 This is a schematic diagram of respiratory characteristic waveforms provided in an embodiment of the present invention;

[0047] Figure 6 This is a schematic diagram of heart rate characteristic waveforms provided in an embodiment of the present invention;

[0048] Figure 7 This is a flowchart illustrating the behavioral habit analysis provided in an embodiment of the present invention. Detailed Implementation

[0049] The present invention will be further described in detail below with reference to experimental examples and specific embodiments. However, this should not be construed as limiting the scope of the above-mentioned subject matter of the present invention to the following embodiments; all technologies implemented based on the content of the present invention fall within the scope of the present invention.

[0050] The terms "first," "second," etc., used in the specification, embodiments, claims, and drawings of this invention are for distinguishing purposes only and should not be construed as indicating or implying relative importance or order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, such as including a series of steps or units. A method, system, product, or apparatus is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or apparatuses.

[0051] Furthermore, any embodiment or design described as "exemplary" or "for example" in the embodiments of the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner for ease of understanding.

[0052] Please refer to Figure 1 The diagram illustrates a structural schematic of a wireless signal-based human activity detection system according to an embodiment of the present invention. The system includes:

[0053] A radar module comprising at least one transmitting antenna and at least one receiving antenna, for transmitting a linear frequency modulated wave signal toward a region under test, and for receiving an echo signal of the linear frequency modulated wave signal reflected from the region under test;

[0054] The processing module is used to control the radar module and to perform human activity detection using a wireless signal-based human activity detection method.

[0055] In one possible implementation, the radar module employs a millimeter-wave radar and is designed with various linear frequency modulated wave waveforms to adapt to the acquisition time window, further enhancing the detection capability of moving and stationary targets. In particular, it optimizes the detection of stationary targets through short-time window accumulation and long-time window downsampling accumulation.

[0056] Furthermore, the system includes an angle sensor for detecting the installation angle and automatically calibrating installation errors. Specifically, the angle sensor and the radar chip, or the communication module, are on the same hardware. The angle sensor detects the installation angle of the radar chip and transmits this precisely monitored value directly to the radar signal processor. Through a built-in automatic calibration mechanism, the radar signal processor can instantly adjust and compensate for angle deviations caused by installation errors, thereby achieving high-precision sensing by the radar system. This system can calibrate errors during the installation process by reading the values ​​from the angle sensor on the hardware. Because there may be errors in the angle during installation, these errors may lead to errors in the target height obtained by the radar. This calibration is specifically performed to compensate and calibrate when calculating the target height. This design not only simplifies the installation process but also significantly improves the overall performance and reliability of the radar system.

[0057] Furthermore, the radar module can be installed by side mounting on a wall or top mounting on a roof, with a height of not less than 2 meters. In the case of side mounting, the angle between the radar module and the horizontal plane is adjustable.

[0058] Furthermore, the processing module can be composed of high-performance artificial intelligence computing and processing devices such as a central processing unit (CPU), a graphics processing unit (GPU), and a field programmable gate array (FPGA) to realize the function of human body detection.

[0059] The following is a detailed description and explanation of a human activity detection method based on wireless signals provided by the embodiments of the present invention, with reference to specific implementation methods.

[0060] Please refer to Figure 2 The diagram illustrates a flowchart of a human activity detection method based on wireless signals provided by an embodiment of the present invention. The method includes:

[0061] Step 100: Control the radar module to transmit a linear frequency modulated wave signal toward the area to be measured;

[0062] Step 200: Receive the echo signal of the linear frequency modulated wave signal reflected from the area under test by the radar module;

[0063] Step 300: Demodulate the echo signal accumulated based on the first time window to obtain the dynamic point cloud of the area under test; Demodulate the echo signal accumulated based on the second time window to obtain the static point cloud of the area under test.

[0064] Step 400: Combine the dynamic point cloud and the static point cloud to determine the activity characteristics of the target human body.

[0065] In one possible implementation, the activity characteristics include one or more of the target human body's position, height, velocity, acceleration, respiratory rate, heart rate, and heart rate variability.

[0066] It should be noted that the transmitted signal involved in this embodiment of the invention is a continuous frequency modulated wave, and the received signal is a copy of the transmitted signal after a time delay due to spatial propagation. Key parameters such as the reflected energy distribution, time of flight, and phase characteristics in the received signal are analyzed through signal processing. Through multi-antenna signal processing, the delay of signals received by different antennas can be analyzed, enabling target angle measurement. This system can highly sensitively capture and analyze subtle physiological activities and postural changes in the elderly, specifically including but not limited to monitoring respiratory rate, tracking heart rhythm, and real-time detection of spatial position changes.

[0067] Furthermore, this invention, by transmitting linear frequency modulated wave signals and leveraging their evenly spaced characteristics, accumulates the demodulated return signals over a short time window, enabling the generation of both dynamic and static point clouds. This makes it suitable for detecting both walking and stationary people. Stationary people require a longer detection window, while moving people require a shorter one.

[0068] In one possible implementation, the aforementioned linear frequency modulated wave signal is configured as a chirp waveform, the duration, time interval, bandwidth, and start and end frequencies of which can be adjusted according to scenario requirements to meet the requirements of different distance resolutions and maximum measurement distances. The echo signal is demodulated using a fast Fourier transform for distance estimation, and the angle of arrival is estimated and the target signal is extracted using digital beamforming and constant false alarm rate (CFAR) algorithms to obtain the dynamic point cloud and the static point cloud. When calculating the dynamic point cloud, this embodiment of the invention selects a short cumulative time window (first time window) to construct a data matrix for the chirp, typically with a time span of less than 0.2 s; when calculating the static point cloud, a slightly longer cumulative time window (second time window) is selected using interval sampling to construct a data matrix for the chirp, typically with a time span of 0.4-4 s. It is understood that this embodiment of the invention generates two types of point cloud results simultaneously when generating point cloud data. For example, a dynamic point cloud is calculated based on data accumulated over a short time window of 48 chirp intervals (i.e., the first time window), and a static point cloud is calculated based on data accumulated over a long time window of 6 or 10 chirp intervals (i.e., the second time window). These two types of point cloud data are output simultaneously. When the target is moving, a rich dynamic point cloud can be generated, and the result of the dynamic point cloud is more reliable. When the target is stationary, no dynamic point cloud can be generated, but the static point cloud accumulated over a long time window can still capture the target's features. This is because although the target is not moving, the chest rise and fall caused by breathing can generate static point cloud features.

[0069] It should be understood that the above implementation method is only an example. Any method of obtaining the dynamic point cloud and static point cloud of the target human body by adjusting the duration, time interval, bandwidth and start and end frequency of the linear frequency modulated wave signal and then setting the corresponding time window for cumulative demodulation should fall within the scope of this invention.

[0070] Specifically, in the process of demodulating the signal, embodiments of the present invention provide an example of a demodulated signal, including:

[0071] The transmitted signal is as follows: Formula 1:

[0072]

[0073] The received signal is as shown in Formula 2:

[0074]

[0075] The demodulated signal is shown in Formula 3:

[0076]

[0077] Among them, S IF(t) represents the demodulated signal at time t, A TX A represents the amplitude of the transmitted continuous-frequency modulated wave signal. RX The amplitude of the echo signal is represented by w0, and the starting angular frequency is represented by A. b T represents the frequency change during signal transmission. d The flight time refers to the interval between the time of transmission and the time of reception.

[0078] Signal processing is performed on the demodulated signal obtained from Formula 3, including extracting the target's range, phase information, etc.

[0079] Let represent the time of flight of the signal, c be the speed of light, and d represent the distance from the target to the radar. Therefore, the frequency of the decoded signal reflected back by the target is... Where S represents the slope of the frequency change of the transmitted signal. The intermediate frequency signal in Formula 3, after passing through a low-pass filter, yields the low-frequency component shown below.

[0080]

[0081] Where k represents multiple different reflected signals, because the same radar may receive multiple reflected signals simultaneously. A k This represents the signal strength of each reflected signal. This indicates the phase of each signal.

[0082]

[0083] The true distance to the target is shown in Equation 5. The relationship between the frequency of the intermediate frequency signal and the target distance is f. IF =S*T d .

[0084] In radar signal processing, the frequency of the intermediate frequency signal is obtained through Fourier transform. The distance to the target is calculated.

[0085] The result of the first Fourier transform of the intermediate frequency (IF) signal is a complex number containing two components, I and Q. It's important to note that the radar uses an orthogonal dual-receiver system. The I component represents the portion of the signal that is in phase with the transmitted signal, indicating the portion aligned with the reference signal. The Q component represents the portion of the signal that is 90 degrees out of phase with the transmitted signal, indicating the portion out of phase with the reference signal (a quarter of a period). The amplitude and phase can be calculated from the I and Q components. Therefore, the phase of the IF signal can be calculated using the following formula:

[0086]

[0087] After obtaining the phase φ(t) of the demodulated signal, a Fourier transform is performed on the phase to obtain the target's velocity.

[0088] If the radar module contains multiple antennas, they can be arranged into an array antenna. Utilizing the phase difference relationship between the multiple antennas, the target's angle can be calculated using Fourier transform. The target's angle can include both elevation and horizontal angles. After obtaining the target's elevation angle, the target's altitude relative to the radar can be calculated by combining this with the target's distance. The formula for the target's altitude relative to the radar is as follows:

[0089] d1=d*sin(θ ele ),

[0090] If the installation height of the radar is known, the height of the target can be calculated.

[0091] d3 = d2 - d1,

[0092] d2 is the radar's installation height, and d3 is the target's height.

[0093] The system processes the acquired frequency signals, including point cloud generation, target trajectory management for continuous tracking of dynamic objects, target vital signs calculation, and accurate calculation of fall characteristics. Through the built-in machine learning inference engine, it can analyze the processed data in real time and quickly provide feedback on the target's fall status, achieving efficient real-time monitoring and response.

[0094] The specific signal processing flow is as follows: First, obtain the target's position, including distance, angle, and height. Then, track the target based on its position information using the Extended Kalman Array (EKA) tracking algorithm. The tracking algorithm provides real-time position information of the target during its movement; this recorded position information is called the target's trajectory. During target tracking, if a significant change in the target's height is detected, a machine learning algorithm is invoked. For example, if the height decreases by 0.5 meters, a machine learning algorithm is called to determine if a fall has occurred. Machine learning algorithms include logistic regression, random forest, decision tree, and SVM. One or more of these algorithms can be selected to infer whether a fall has occurred.

[0095] Machine learning algorithms perform feature analysis on a large amount of pre-collected data and train the model based on the extracted features, integrating the resulting parameters into the algorithm. Key features include changes in the target's height, movement speed, and acceleration, as well as temporal feature analysis of height changes. Using only a single or a few features to determine fall detection results in poor performance; fusing multiple features can improve the algorithm's performance. Leveraging the learning ability of machine learning algorithms on large amounts of data allows for the acquisition of optimal parameters. Without learning from large datasets, the algorithm's compatibility and generalization ability are relatively poor.

[0096] Furthermore, long-term data on the target individual is analyzed, and the algorithm's accuracy is improved through iterative machine learning modeling to enable the development of personalized health management plans. This analysis includes three core functions: automatic room layout recognition, personalized health management, and accurate fall detection.

[0097] In one possible implementation, this invention innovatively utilizes machine learning algorithms, particularly time-series analysis and spatial clustering methods, to perform in-depth analysis of a target's long-term activity trajectory within a room. The tracking algorithm obtains the target's real-time location information during its movement; this location information is recorded as the target's trajectory. By identifying the target's dwelling patterns and movement paths in different areas, key layout features of the room are automatically inferred, including but not limited to the specific location of the bed, the layout of the aisles, the orientation of the sofa, and the precise location of entrances and exits. The trajectory information is input into the algorithm to analyze the statistics of the time spent in different locations within the room. Based on these statistics, the dwell time can be inferred. For example, the location of the bed can be determined based on the dwell time at night. Figure 3 As shown in the image. Furthermore, based on the trajectory information, the location where the target appeared and left the room can be determined, and the location of the room's entrance and exit can be determined from this location. This function greatly simplifies the process of users manually entering room configuration information, improving the system's usability and intelligence.

[0098] In one possible implementation, this embodiment of the invention further provides a health monitoring function. By continuously tracking the target's activity trajectory within a room, and using classification and regression algorithms from machine learning, it deeply analyzes key health indicators such as the target's activity level and bed rest time. When this embodiment of the invention collects sleep time information of the user over a period of time, it can determine the distribution of sleep time within specific time periods, as well as the range of respiratory rate, heart rate, and heart rate variability during sleep. Based on these analysis results, it can automatically generate personalized health analysis reports, providing users with customized health management suggestions and early warning services, thus facilitating efficient health management and disease prevention. For example, by accumulating the respiratory and heart rate data of the monitored person during sleep, if an abnormality in respiratory rate or heart rate is detected on a particular day or period, the user will be promptly alerted. Specifically, this abnormality refers to an excessively high or low respiratory rate, or an excessively high or low heart rate.

[0099] In one possible implementation, this invention provides an innovative automatic height calibration model to enhance the accuracy and reliability of fall detection. This invention collects height information of the same target at different locations within a room, obtaining height error data at different distances and angles. Since the height of the same target does not change significantly during normal walking within a room, an error matrix is ​​obtained after filtering out particularly large or small outliers. Furthermore, machine learning techniques are used, combined with the target's natural movement trajectory within the room, to dynamically adjust and calibrate the user's height parameters. Through this mechanism, this invention can more accurately identify whether the target is in an abnormal posture (such as lying on the ground), thus effectively supplementing and improving traditional fall detection algorithms. The application of this model significantly improves the sensitivity and specificity of fall detection, providing a more robust guarantee for user safety. By tracking changes in the user's height at different locations and distances during daily activities, height errors are compensated and calibrated based on machine learning algorithm training.

[0100] The error in height measurement originates from the error in radar elevation angle measurement, which is a systematic error and is also related to errors in the radar module installation process. Without height error calibration, system performance will significantly degrade.

[0101] For fall detection, please refer to Figure 4 This diagram illustrates a fall detection process provided by an embodiment of the present invention. The embodiment of the present invention analyzes changes in human movement characteristics and body shape features, and uses a machine learning model to identify fall patterns. When the model predicts a high-risk fall event, it issues a warning.

[0102] Specifically, the input data for fall detection consists of point cloud data detected by radar and target information. Point cloud data is a set of three-dimensional spatial data acquired by radar, reflecting information such as the target's position, speed, and shape. Noise reduction and clustering processes are applied to the point cloud data to improve its quality and accuracy. Target information includes the target's position, velocity, acceleration, and height. Fall detection is particularly effective by extracting features related to changes in height and velocity during the target's movement. After feature extraction, these features are input into a machine learning algorithm.

[0103] This embodiment employs advanced feature extraction technology, carefully selecting 130 predefined features from the preprocessed point cloud data. These features cover multiple dimensions such as point cloud descent rate, displacement trajectory changes, dynamic evolution of human body contours, and posture angle changes. They are all carefully designed based on extensive experimental verification and in-depth data analysis, and can comprehensively and accurately capture subtle changes and key features in fall events.

[0104] Specifically, this invention incorporates powerful machine learning models such as logistic regression and SVM. These models, through training, can efficiently identify fall patterns hidden in point cloud data and accurately distinguish suspected fall events. Furthermore, the system integrates respiratory feature monitoring as an innovative verification method. Upon identifying a suspected fall, this invention immediately checks whether the target exhibits respiratory characteristics. If the monitoring results show no respiratory activity, it is determined to be a false alarm, and this invention will automatically remove the ghost image to avoid unnecessary alarm reporting. Conversely, if the target is confirmed to have respiratory characteristics, the fall event is immediately reported to ensure timely response and subsequent processing.

[0105] In bed fall detection, this embodiment of the invention utilizes radar to continuously monitor changes in the spatial position of the bedside. By setting a virtual "safe zone," once a user moves outside the bed boundary while sleeping, accompanied by a change in height, it is determined to be a fall from bed and an alarm is triggered.

[0106] Understandably, in bed fall detection, this embodiment of the invention presets a virtual 'safety boundary' in the area to be tested, accurately defining the safe area around the bed. Once it detects that a user crosses this boundary during sleep, accompanied by significant height changes (i.e., a fall event is detected), the system immediately determines it as a bed fall risk event and quickly triggers an alarm mechanism to ensure timely response and intervention. Bed fall detection is not limited to edge computing or cloud computing. This system can obtain the bed's position and height through user trajectory analysis. When it detects that a target's height drops below the bed's height, a warning is triggered. This function relies on the system analyzing the bed's position and height information.

[0107] In respiratory and heart rate monitoring, the radar module provided in this embodiment of the invention captures minute chest movements using high-frequency waves to accurately calculate respiration and heart rate. The calculation of respiration and heart rate is based on the phase information of the intermediate-frequency signal, obtained through phase feature analysis. The phase calculation is shown in Formula 6. For example, applying a low-pass filter of 0.1Hz to 0.5Hz to the phase signal yields the respiratory waveform. Applying a band-pass filter of 0.5Hz to 2Hz to the phase signal yields psychological characteristics. Any apnea or abnormal heart rate will trigger a system alert, and the data will be recorded for medical personnel reference. Long-term recording and analysis of sleep data for the elderly will trigger timely health alerts when abnormalities in respiration or heart rate are detected, or when there are significant differences from historical data, prompting users to undergo physical examinations.

[0108] The respiratory characteristic waveforms detected in this embodiment of the invention are as follows: Figure 5 As shown, the waveform and respiratory rate of respiration are recorded. The heart rate characteristic waveform detected in this embodiment of the invention is as follows: Figure 6 As shown, it records heart rate, heart rate variability, etc.

[0109] In the analysis of behavioral habits, please refer to Figure 7 This diagram illustrates a flowchart of the behavioral habit analysis provided by an embodiment of the present invention. Specifically, by monitoring users' activity patterns over a long period, this embodiment of the present invention can learn and predict users' behavioral habits, such as peak activity times and frequently used activity areas, thereby providing users with personalized health advice and warnings. This includes a warning for exceeding the time limit in the restroom.

[0110] The input data for behavioral habit analysis is target information detected by radar. By recording the target's location every second for several consecutive days, we can create a detailed map of the target's daily activity patterns. This data not only reveals the target's daily activity level but also allows us to accurately identify peak activity periods, duration of bed rest, and key lifestyle indicators such as time spent in the bathroom area. This meticulous analysis provides strong data support for understanding the target's daily activity patterns.

[0111] In the automatic room layout recognition, this invention can automatically learn and analyze users' behavior patterns at home, thereby automatically identifying and marking key information such as the bed's location, room size, and entrance / exit locations. This process eliminates the need for users to manually input information on an app or webpage, greatly improving the user experience and the system's intelligence.

[0112] In health monitoring, this invention continuously records users' vital signs (such as respiration, heart rate, and HRV), bed rest time, number of nighttime urinations, and duration of each nighttime urination, and performs in-depth analysis and data mining. When abnormal changes in a user's health data are detected, the system can automatically trigger an early warning mechanism to alert the user or their family and prompt appropriate action. Furthermore, the system has long-term monitoring capabilities, providing users with comprehensive health reports and trend analyses, such as those for sleep apnea and heart disease.

[0113] In the early warning and communication mechanism, this embodiment of the invention enables remote notification. The aforementioned abnormal events can be obtained through the cloud and notified to family members or medical personnel, ensuring timely response. Notification methods include one or more of the following: app notification, SMS notification, and telephone notification.

[0114] In this invention, the transmitted linear frequency modulated wave signal is received as a copy of the transmitted signal after a time delay due to spatial propagation. By analyzing key parameters such as reflected energy distribution, flight time, and phase characteristics in the received signal through signal processing, and by processing multiple signals, the delay of signals received by different antennas can be analyzed, enabling target angle measurement. This allows for the high-sensitivity capture and analysis of subtle physiological activities and body posture changes in the human body, including but not limited to monitoring respiratory rate, tracking heart rhythm, and real-time detection of spatial position changes. Furthermore, this invention transmits a linear frequency modulated wave signal through a radar module and simultaneously performs cumulative demodulation based on a preset time window, making it suitable for detecting people walking or stationary, thus achieving accurate monitoring.

[0115] For example, when the processing module provided in the embodiments of the present invention is used to implement the method provided in the embodiments of this application, the processor can be used in steps 100 to 400. It should be understood that the processor in the embodiments of the present invention can be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method embodiments can be completed by the integrated logic circuit of the hardware in the processor or by instructions in the form of software. The processor mentioned above can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor can be a microprocessor or any conventional processor, etc. The steps of the method disclosed in the embodiments of this application can be directly embodied as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, and other mature storage media in the art. The storage medium is located in the memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above method.

[0116] In embodiments of the present invention, the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or device.

[0117] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting human activity based on wireless signals, characterized in that, The method includes: The control radar module transmits a linear frequency modulated wave signal to the area under test, and receives the echo signal of the linear frequency modulated wave signal reflected by the area under test through the radar module; The echo signal is accumulated based on the first time window and demodulated to obtain the dynamic point cloud of the area under test. The echo signal is accumulated based on the second time window and demodulated to obtain the static point cloud of the area under test. The activity characteristics of the target human body are determined by combining the dynamic point cloud and the static point cloud; The activity characteristics include any one or more of the target human body's position, height, speed, acceleration, respiratory rate, heart rate, and heart rate variability.

2. The method for detecting human activity based on wireless signals according to claim 1, characterized in that, The span of the first time window is 0 to 0.2 seconds.

3. The method for detecting human activity based on wireless signals according to claim 1, characterized in that, The second time window spans from 0.4 to 4 seconds.

4. The method for detecting human activity based on wireless signals according to claim 1, characterized in that, The linear frequency modulated wave signal adopts the chirp waveform.

5. The method for detecting human activity based on wireless signals according to claim 1, characterized in that, The echo signal is demodulated and distance is estimated using Fast Fourier Transform. The angle of arrival is estimated and the target signal is extracted using digital beamforming and constant false alarm rate detection algorithms to obtain the dynamic point cloud and the static point cloud.

6. The method for detecting human activity based on wireless signals according to claim 1, characterized in that, The method further includes: Based on the location of the target human body, a tracking algorithm is used to track the target human body, and the location during the tracking process is recorded to generate the trajectory of the target human body; Based on time series analysis and spatial clustering, the layout features of the test area are identified by combining the trajectory of the target human body, including the position of the bed, the layout of the aisle, the orientation of the sofa, and the location of the entrance and exit.

7. The method for detecting human activity based on wireless signals according to claim 6, characterized in that, The method further includes: During the tracking of the target human body, a fall detection is performed when a change in the height of the target human body is detected. The fall detection includes: The dynamic point cloud, static point cloud, and the activity features of the target human body are input into a machine learning algorithm to extract the height and speed changes of the target during movement for fall recognition. During the identification process, the static point cloud is used to determine whether the target human body has respiratory characteristics; If so, a warning will be issued upon detecting a fall of the target human body; If not, the target human body is detected as non-living and is removed.

8. The method for detecting human activity based on wireless signals according to claim 7, characterized in that, The method further includes: Based on the identified location of the bed in the test area, a virtual safe zone is set at the location of the bed. When the target human body moves to the safe zone during sleep and is accompanied by a change in height, the fall detection is performed.

9. The method for detecting human activity based on wireless signals according to claim 6, characterized in that, The method further includes: By combining the trajectory of the target human body and the layout characteristics of the area to be tested, the key health indicators of the target human body are analyzed and obtained. The key health indicators include: activity level, bed rest time, number of times to urinate at night, and the duration of each time to urinate at night. Based on the acquired activity characteristics of the target human body, combined with the key health indicators, a personalized health analysis report of the target human body is generated.

10. A method for detecting human activity based on wireless signals according to claim 9, characterized in that, The method further includes: The personalized health analysis report is uploaded to the cloud platform, and when an abnormality is found in the personalized health analysis report, family members or medical personnel are notified through the terminal device. The notification method includes one or more of the following: APP reminder, SMS reminder, and telephone reminder; the abnormality includes sleep apnea and heart disease.

11. The method for detecting human activity based on wireless signals according to claim 1, characterized in that, The method further includes: By collecting the height of the target human body at different positions in the area to be tested, error data of the target human body height at different distances and angles are obtained; The height of the target human body is dynamically adjusted and calibrated based on the error data.

12. A human activity detection system based on wireless signals, characterized in that, The system includes: A radar module comprising at least one transmitting antenna and at least one receiving antenna, for transmitting a linear frequency modulated wave signal toward a region under test, and for receiving an echo signal of the linear frequency modulated wave signal reflected from the region under test; A processing module is used to control the radar module and perform human activity detection using a human activity detection method based on wireless signals as described in any one of claims 1 to 11. A cloud platform for receiving personalized health analysis reports; The terminal device is used to notify family members and / or medical personnel when an abnormality is found in the personalized health analysis report.

13. A human activity detection system based on wireless signals according to claim 12, characterized in that, The system includes an angle sensor for detecting the installation angle and automatically calibrating the installation error. The radar module is installed either side-mounted on a wall or top-mounted on a roof, and the installation height of the radar module is not less than 2 meters. In the side-mounted case, the angle between the radar module and the horizontal plane is adjustable.

Images