A low-cost, distributed RFID-based human motion recognition method and system

Abstract

The present invention discloses a low-cost, distributed RFID-based human motion recognition method and system. The method includes sending an excitation signal and a coding signal through a transmitter; the excitation signal is used to activate an RFID tag; the coding signal is used to instruct the RFID tag to send a reflection signal; the reflection signal is received by at least one receiver; the reflection signal has different phases after being reflected by different human motions; after eliminating the carrier frequency offset, self-interference and random phase of the reflection signal, the phase information in the reflection signal is extracted; the phase information is input into a trained classifier to obtain the human motion corresponding to the phase information output by the classifier; the classifier stores the mapping relationship between phase information and human motion. The solution provided by the present invention can reduce costs, increase signal sampling rate, and thus improve the detection accuracy of the system.

Description

Technical Field

[0001] The present invention relates to the field of information science and technology, and in particular to a low-cost, distributed RFID-based human motion recognition method and system. Background Art

[0002] In recent years, with the rapid growth of wireless and mobile devices and their applications, the research on human action recognition has developed at an exponential rate. Human action recognition enables users to interact with smart environments, monitor users' daily activities, and infer their health status. For example, by monitoring the activities of the elderly, their health status and behavioral changes can be tracked and timely assistance can be provided in case of accidents. Action recognition has been widely used in various applications in various fields such as healthcare, sports activity recognition, and home rehabilitation. Recent studies have tried to build activity recognition systems using various commercial devices on the market to realize these applications.

[0003] Existing human motion recognition solutions mainly use cameras, motion sensors or other dedicated sensors.

[0004] (1) Disadvantages of wearable sensor-based detection solutions:

[0005] The detection scheme based on wearable sensors requires the sensor to be in direct contact with the body. The sensor must be worn on the body for detection, which has a high reliance on the stability of the device. Moreover, the need to wear sensors all the time will bring inconvenience to users, especially the disabled or the elderly with limited mobility. Secondly, most of the current wearable devices do not have device-side data processing, which requires the wearable devices to maintain an uninterrupted communication connection in order to upload the sensor data to the cloud in real time for processing.

[0006] (2) Disadvantages of camera-based detection solutions:

[0007] The camera-based image detection solution is very dependent on the lighting conditions of the environment. First, in low-light conditions, it is difficult for the camera to obtain clear images, which will reduce the accuracy of human motion detection. Secondly, the camera relies on a straight-line line of sight path. Once there are other objects blocking the camera and the detected target, detection will not be possible. More importantly, the technology based on cameras and deep learning requires large-scale collection of human activity data, and the data collection cost is high. It will also infringe on the privacy of users, and a large amount of video data will pose a serious security threat to users.

[0008] (3) Disadvantages of acoustic-based detection solutions:

[0009] Acoustic-based solutions can achieve non-contact perception and are not affected by lighting and privacy violations. However, sound waves are mechanical waves. When propagating in the air, they decay quickly due to the impact of air molecule collisions and heat conduction. Therefore, the perception distance of human motion detection technology based on acoustics is short. At the same time, microphones can collect sound signals, which may contain personal conversations or other private information, which may cause inconvenience or insecurity to users and also pose certain security risks.

[0010] (4) Disadvantages of detection solutions based on radio frequency equipment:

[0011] At present, RF-based human motion recognition is mainly based on FMCW radar and RFID readers. However, FMCM radar is based on the millimeter wave frequency band for perception. The wavelength of millimeter wave is short, the signal attenuation is fast, and the perception distance is short. In addition, FMCW radar requires complex RF circuit design and special equipment such as high-frequency electronic components, and its versatility is low. Although RFID tags are low in cost, RFID readers use full-duplex communication mode and require complex anti-interference hardware circuits and high-precision ADC (Analog-to-Digital Converter) to process signals, resulting in high reader costs. More importantly, commercial RFID readers use a polling mechanism to achieve multiple transmissions and multiple receptions, which cannot increase the sampling rate of the signal and is of no help in improving the detection accuracy of the system.

[0012] In summary, the motion recognition solution based on wearable sensors requires the human body to wear the wearable device all the time and maintain a real-time network connection. The human motion recognition solution based on images cannot work in dark environments and when the camera is blocked, and it is easy to bring privacy and security risks to people. The acoustic-based solution can achieve non-contact motion recognition and is not affected by lighting conditions. However, when sound waves propagate in the air, their attenuation speed is fast and the transmission distance is short, so the perception distance needs to be improved. The human motion recognition solution based on radio frequency can solve the above problems, but the human motion recognition solution based on radio frequency has technical problems such as high cost and insufficient sampling rate. Summary of the invention

[0013] In order to solve the technical problems of high cost and insufficient sampling rate of existing radio frequency-based human motion recognition solutions, the embodiments of the present invention provide a low-cost, distributed RFID-based human motion recognition method and system.

[0014] The technical solution of the embodiment of the present invention is achieved as follows:

[0015] The embodiment of the present invention provides a low-cost, distributed RFID-based human motion recognition method, the method comprising:

[0016] An excitation signal and a coding signal are sent through a transmitter; the excitation signal is used to activate an RFID tag; the coding signal is used to instruct the RFID tag to send a reflection signal; the reflection signal is received by at least one receiver; the reflection signal has different phases after being reflected by different human body movements; after eliminating the carrier frequency offset, self-interference and random phase of the reflection signal, the phase information in the reflection signal is extracted; the phase information is input into a trained classifier to obtain the human body movement corresponding to the phase information output by the classifier; the classifier stores the mapping relationship between the phase information and the human body movement.

[0017] The embodiment of the present invention also provides a low-cost, distributed RFID-based human motion recognition system, the system comprising a transmitter, at least one receiver, a phase extraction module and a motion recognition module: the transmitter comprises a first development board, a first radio frequency chip and a first transmitting antenna; the receiver comprises a first module and a receiving antenna; the first development board is used to send instructions after programming; the first radio frequency chip is used to control the first transmitting antenna to send signals under the instructions of the first development board; the first transmitting antenna is used to send excitation signals and coding signals; the first module is used to control the receiving antenna to receive signals; the receiving antenna is used to receive reflected signals; the reflected signals have different phases after being reflected by different human motions; the phase extraction module is used to extract the phase information in the reflected signal after performing carrier frequency offset, self-interference and random phase elimination operations on the reflected signal; the motion recognition module is used to input the phase information into a trained classifier to obtain the human motion corresponding to the phase information output by the classifier; the classifier stores the mapping relationship between phase information and human motion.

[0018] The low-cost, distributed RFID-based human motion recognition method and system provided by the embodiment of the present invention sends an excitation signal and a coding signal through a transmitter; the excitation signal is used to activate the RFID tag; the coding signal is used to instruct the RFID tag to send a reflection signal; the reflection signal is received by at least one receiver; the reflection signal has different phases after being reflected by different human motions; after eliminating the carrier frequency offset, self-interference and random phase of the reflection signal, the phase information in the reflection signal is extracted; the phase information is input into a trained classifier to obtain the human motion corresponding to the phase information output by the classifier; the classifier stores the mapping relationship between the phase information and the human motion. The solution provided by the present invention decouples the RFID reader into a transmitter and a receiver, and uses low-cost equipment to construct the transmitter and the receiver, which can reduce the system cost. At the same time, this embodiment uses at least one receiver to receive the reflection signal, which can improve the signal sampling rate, thereby improving the detection accuracy of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A schematic diagram of a flow chart of a low-cost, distributed RFID-based human motion recognition method according to an embodiment of the present invention;

[0020] Figure 2 Schematic diagram of the signal transmission process of an embodiment of the present invention;

[0021] Figure 3 This is a schematic diagram of the module workflow of an embodiment of the present invention;

[0022] Figure 4 This is a schematic diagram of PIE encoding according to an embodiment of the present invention;

[0023] Figure 5 This is a schematic diagram of the original signal of the physical layer according to an embodiment of the present invention;

[0024] FIG6 (a) is a schematic diagram of an IQ diagram of an RN16 preamble sequence according to an embodiment of the present invention;

[0025] FIG6( b ) is a schematic diagram of a phase diagram of an RN16 preamble sequence according to an embodiment of the present invention;

[0026] FIG. 7 ( a ) is a schematic diagram of an IQ diagram without CFO influence according to an embodiment of the present invention;

[0027] FIG7( b ) is a schematic diagram of a phase diagram without CFO influence according to an embodiment of the present invention;

[0028] Figure 8 A schematic diagram of an IQ diagram for eliminating self-interference and unknown phase offset according to an embodiment of the present invention;

[0029] Fig. 9A schematic diagram of continuous monitoring phases when standing up and sitting down according to an embodiment of the present invention;

[0030] Fig.10 Schematic diagram of a signal transmission model according to an embodiment of the present invention; wherein: Fig.10 (a) is a schematic diagram of the channel transmission path. Fig.10 (b) is a schematic diagram of channel transmission data processing;

[0031] FIG11( a ) is a schematic diagram of an IQ graph before and after removing a static path according to an embodiment of the present invention;

[0032] FIG11( b ) is a schematic diagram of a phase diagram after removing the static path according to an embodiment of the present invention;

[0033] Fig.12 A schematic diagram of the data misalignment problem caused by action interception of four antennas according to an embodiment of the present invention;

[0034] FIG13 (a) is a schematic diagram of linear interpolation of missing data according to an embodiment of the present invention;

[0035] FIG13( b ) is a schematic diagram showing a state where the length of data after interpolation remains consistent according to an embodiment of the present invention;

[0036] Fig.14 Schematic diagram of a data fusion method for performing weighted summation on antenna data according to an embodiment of the present invention; wherein: Fig.14 (a) is the time phase diagram of data 1, Fig.14 (b) is the time phase diagram of data 2, Fig.14 (c) is the time phase diagram of data 3, Fig.14 (d) is the time phase diagram of data 4, Fig.14 (e) is the time phase diagram of the fused data. Fig.14 (f) is a schematic diagram of the correlation comparison of data 1, data 2, data 3, and data 4;

[0037] Fig.15 The following is a schematic diagram of the hardware design process of an embodiment of the present invention;

[0038] Fig.16 This is a schematic diagram of signal transmission according to an embodiment of the present invention;

[0039] Fig.17 1 is a diagram showing the internal structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The present invention will be described in further detail below with reference to the accompanying drawings and embodiments.

[0041] The embodiment of the present invention provides a low-cost, distributed RFID-based human motion recognition method, such as Figure 1 As shown, the method includes:

[0042] Step 101: sending an excitation signal and a coding signal through a transmitter; the excitation signal is used to activate the RFID tag; the coding signal is used to instruct the RFID tag to send a reflection signal;

[0043] Step 102: receiving the reflected signal by at least one receiver; the reflected signal has different phases after being reflected by different human body movements;

[0044] Step 103: extracting phase information from the reflected signal after performing carrier frequency offset, self-interference and random phase elimination operations on the reflected signal;

[0045] Step 104: input the phase information into a trained classifier to obtain a human body motion corresponding to the phase information output by the classifier; the classifier stores a mapping relationship between the phase information and the human body motion.

[0046] The method of this embodiment utilizes the different reflections of human body movements on RFID tag signals to identify different movements, and obtains the mapping relationship between different movements and RFID signal models, so as to effectively detect different human body movements. The whole process does not require the human body to wear any equipment, is not affected by lighting conditions, and does not infringe on privacy.

[0047] Specifically, see Figure 2 and Figure 3 , this embodiment may include an RFID transceiver subsystem, a phase extraction module and an action recognition module. Among them, the RFID transceiver subsystem may include a transmitter (i.e., a transmitting antenna) and multiple receivers (i.e., receiving antennas). The transmitting antenna first sends an excitation signal to activate the RFID tag. After the tag is activated, the transmitting antenna simultaneously sends a coded signal (which may be called a Query signal frame), and the tag will reflect a reflected signal (which may be called an RN16 signal frame or a Round Number 16 signal frame) after receiving the Query signal. After the RN16 signal frame is received by the receiving antenna, the phase extraction module will perform carrier frequency offset, self-interference and random phase elimination on the RN16 signal to extract the phase information of the signal. Finally, the action recognition module analyzes the different phase features caused by different human body movements and trains a suitable classifier to detect human body movements. In this embodiment, there may be multiple receiving antennas, and the reflected signals received by multiple receiving antennas are used to identify human body movements, which can improve the signal sampling rate and thus provide the accuracy of human body movement recognition.

[0048] The excitation signal and the coding signal in this embodiment are pulse width coding frame structures that support the RFID protocol. In order to reduce costs, in one embodiment, a radio frequency chip that does not support pulse width coding (ie, PIE coding) but supports on-off keying coding (ie, OOK coding) may be selected. For example, the RFM69HW radio frequency chip. The RFM69HW radio frequency chip has the advantage of low cost. At the same time, it can operate on general frequency bands including the UHF ISM band (such as 915 MHz). And its transmitting module provides an interface that can be used to configure most of the main radio frequency communication parameters (such as carrier frequency, transmission power and link frequency).

[0049] Therefore, considering the above situation, in order to improve the versatility of this solution, this embodiment provides a signal conversion method for this purpose.

[0050] Specifically, see Figure 4 In the RFID standard communication protocol, the Query signal is encoded by Pulse Interval Modulation (PIE). Figure 4 As shown in FIG. 1 , when data 0 is sent, it is represented by a high level with a short duration and a low level with a short duration. When data 1 is sent, it is represented by a high level with a long duration and a low level with a short duration.

[0051] In OOK encoding, data 1 sends a high level, and data 0 sends a low level, so this embodiment needs to implement PIE encoding through the OOK encoding module. Based on this, this embodiment designs the Query signal as follows: data 1 is represented by 1110, and data 0 is represented by 10. The duration of different levels is expressed by different bits of data, so that PIE encoding is expressed through OOK encoding, which is compatible with the RFID standard communication protocol.

[0052] That is, in one embodiment, the sending of the excitation signal and the coding signal through the transmitter includes: determining the pulse width coding frame structure of the excitation signal and the coding signal; converting the pulse width coding frame structure of the excitation signal and the coding signal using a preset conversion relationship between pulse width coding and on-off keying coding to obtain the on-off keying coding frame structure of the excitation signal and the coding signal; sending the on-off keying coding frame structure of the excitation signal and the coding signal through the transmitter; the preset conversion relationship between pulse width coding and on-off keying coding is that 1 in the pulse width coding is represented by on-off keying code 1110, and 0 in the pulse width coding is represented by on-off keying code 10.

[0053] Specifically, see Table 1, which shows a Query frame structure that is fully compatible with the RFID protocol.

[0054] Table 1

[0055]

[0056] The frame structure after conversion based on the conversion relationship is shown in Table 2 below.

[0057] Table 2

[0058]

[0059] In addition, in order to accurately control the RFID tag to send the reflection signal, indication information is set in the coded signal. When the indication information is 1, the coded signal instructs the RFID tag to send the reflection signal; when the indication information is 0, the coded signal instructs the RFID tag not to send the reflection signal.

[0060] In addition, unlike commercial RFID readers in the prior art, this embodiment uses a distributed architecture to decouple the functions of a single RFID reader into two parts: the transmitter (RFM69HW module) and the receiver (RTL-SDR dongle module). These devices are manufactured by different suppliers and are equipped with different oscillators. Therefore, in practical applications, it is difficult to strictly synchronize the clock frequencies of the transmitter and receiver. For example, during down-conversion, the local oscillator signal of the receiver may have a frequency inconsistency with the transmitted carrier signal, resulting in a frequency offset between the transmitter and the receiver, such as Figure 5 shown.

[0061] Therefore, it is necessary to eliminate this carrier frequency offset.

[0062] This embodiment proposes a method for eliminating carrier frequency offset based on the expression formula of the received signal y(t).

[0063] Specifically, equation (1) is the mathematical expression of the signal down-conversion process, and the received signal y(t) is the superposition of the carrier signal (i.e., the coded signal) and the tag reflection signal.

[0064] Calculation formula (1)

[0065] After sorting:

[0066] Calculation formula (2)

[0067] in, It is the carrier signal sent, which provides energy to the tag during the tag response process. is the tag reflection signal RN16; x(t) is the indication information in the coded signal. When x(t)=1, the tag reflects the carrier signal (reflection state), and when x(t)=0, the tag does not reflect the carrier signal (absorption state); The carrier frequency offset (CFO) is caused by the inconsistent clock frequency of the transceiver. and represent the initial phase offset caused by the transmitter and the tag, respectively, and Represent the carrier frequency of modulation and demodulation respectively, A cw and A tag represents the amplitude of the signal, j is the sign of the complex number, and t is the time.

[0068] As shown in Figure 6 (a) and Figure 6 (b), the carrier frequency offset is reflected in the IQ diagram as the phase of the RN16 preamble sequence will rotate evenly. The degree of phase rotation is related to the degree of CFO offset. CFO will cause the phase information of the signal to be unable to be correctly extracted. Therefore, the influence of CFO must be eliminated before subsequent phase measurement can be performed.

[0069] In this embodiment, it is considered that the CFO will not change much in the short time from the start of carrier signal transmission to the end of tag response. That is to say, in a signal transmission process (about 5ms in actual transmission), Should remain relatively stable.

[0070] Therefore, this embodiment uses the CFO during the carrier signal transmission period to offset the influence of the CFO during the RN16 transmission period during one signal transmission period. That is, the RN16 reflection signal received when the indication information is 1 (here, the RN16 reflection signal is the superposition of the coded signal and the tag reflection signal) is divided by the RN16 reflection signal received when the indication information in the same period is 0 (here, the RN16 reflection signal is only the coded signal) to eliminate the influence of CFO.

[0071] The calculation formula is as follows:

[0072] Calculation formula (3)

[0073] in, To eliminate the reflected signal after the carrier frequency is offset, is the reflected signal without eliminating the carrier frequency offset, CW is the coded signal, represents the phase of a randomly intercepted CW signal, It is the carrier signal sent, which provides energy to the tag during the tag response process. is the tag reflection signal RN16; x(t) is the indication information in the coded signal. When x(t)=1, the tag reflects the carrier signal (reflection state), and when x(t)=0, the tag does not reflect the carrier signal (absorption state); The carrier frequency offset (CFO) is caused by the inconsistent clock frequency of the transceiver. and represent the initial phase offset caused by the transmitter and the tag, respectively, and Represent the carrier frequency of modulation and demodulation respectively, A cw and A tag represents the amplitude of the signal, j is the sign of the complex number, and t is the time.

[0074] Through the above CFO elimination method, CFO can be successfully removed. At this time, the IQ diagram and phase diagram of the signal are drawn, as shown in Figure 7 (a) and Figure 7 (b).

[0075] From the above calculation formula (3), it can be seen that there are still two unknown terms: (1) the self-interference term affecting RN16; (2) the unknown phase offset . Unknown phase offset Represents the unknown initial signal phase and changes with the selected carrier.

[0076] Since the self-interference term in the above calculation formula (3) is a constant term, therefore, in this embodiment, = 0. That is, the RN16 reflected signal received when the indication information is 0 is the self-interference term. Subtracting the RN16 reflected signal received when the indication information is 0 from the reflected signal with the carrier frequency offset eliminated is the reflected signal with the self-interference term eliminated.

[0077] In addition, for unknown phase offset , because when the indication information is 0, there is also an unknown phase offset in the RN16 reflected signal received Therefore, this embodiment eliminates the unknown phase offset in the reflected signal by dividing the reflected signal with the self-interference term eliminated by the RN16 reflected signal received when the indication information is 0. .

[0078] The calculation formula is as follows:

[0079] Calculation formula (4)

[0080] in, To eliminate the reflected signal after the carrier frequency is offset, is the reflected signal without eliminating the carrier frequency offset, Represents the phase of the randomly intercepted CW signal; x(t) is the indication information in the coded signal. When x(t)=1, the tag reflects the carrier signal (reflection state), and when x(t)=0, the tag does not reflect the carrier signal (absorption state); and Represent the initial phase offset caused by the transmitter and the tag, A cw and A tag represents the amplitude of the signal, j is the sign of the complex number, and t is the time.

[0081] At this point, CFO, self-interference and unknown phase offset have been effectively eliminated. Now draw the phase diagram after algorithm processing, as shown in Figure 8 As shown, CFO is effectively eliminated and self-interference is also effectively eliminated.

[0082] Once the interference is eliminated, phase measurement can be performed. Fig. 9 The phase characteristics of different actions can be seen in Figure 1 (standing up and sitting down twice). However, the amplitude difference caused by the action is only about 0.5 rad, which is caused by the static path interference.

[0083] from Fig.10 It can be seen from the channel transmission model that due to the influence of the static path, after the human body reflection signal is superimposed on the static reflection signal, the phase change ( ) compared to the actual phase change ( ) is not obvious and is not used for subsequent classification and recognition.

[0084] In order to eliminate the influence of the static path, it is necessary to subtract the static path signal to extract the signal containing only human body reflection. The traditional static path elimination method estimates the center of the superimposed signal and uses the estimated center position vector as the static path for elimination. However, the premise of this method is to assume that the superimposed signal is a circular arc, which is often not true in practice. In this system, the static path interference is eliminated by eliminating the static path in real time. That is, at each action sampling, the superimposed signal generated by the action and the static path signal before the action are obtained. Draw the IQ diagram of the signal, as shown in Figure 11 (a). Subtract the static path signal from the superimposed signal (that is, the reflected signal received when there is human body movement reflection minus the reflected signal received when there is no human body movement reflection), and the dynamic path signal containing only human body reflection can be extracted. After subtracting the static path, the signal caused by human body activity is distributed around the center of the circle. Draw the phase diagram of the signal again. As can be seen from Figure 11 (b), the phase difference of the original signal increases from about 0.5rad to about 3.1rad. Although the static path will change with the change of the environment, the present invention ensures accurate elimination of the static path by updating the static path in real time, and can adapt to different environments. So far, this embodiment can accurately extract the phase characteristics generated by human body movements.

[0085] Since the present invention needs to continuously detect the action, a sliding window can be set to extract the signal containing the phase change of the action. Here, an empirical threshold is set, which is three times the standard deviation of the phase value in a static environment. When the phase standard deviation in three consecutive windows exceeds the threshold, it proves that an action has occurred, and the action signal begins to be intercepted at this time (the intercepted signal at this time is the reflection signal when there is human body movement reflection), and then when the phase standard deviation in three consecutive windows is lower than the threshold, the interception of the action signal ends. When the above conditions are not met, the intercepted signal is the reflection signal when there is no human body movement reflection. In the above manner, the reflection signal received when there is human body movement reflection and the reflection signal received when there is no human body movement reflection can be accurately obtained, and then the reflection signal received when there is human body movement reflection is subtracted from the reflection signal received when there is no human body movement reflection to obtain a dynamic path signal containing only human body reflection.

[0086] Based on the phase features generated by the above-mentioned different actions, thanks to the reliability and accuracy of the Transformer model in sequence prediction, the present invention uses the Transformer model to judge the multiple types of actions collected (the present invention collects eight types of actions) to achieve real-time detection of human activities. However, considering that the deep learning method requires a large number of data sets for training, although a large amount of data is collected, more data sets can further improve the recognition accuracy of the model. Based on this, the present invention uses a data enhancement method to expand our data set. Unlike the traditional data enhancement method, this embodiment uses time series data enhancement technology (Time Series Data Augmentation, TSDA), and the model uses TimeGAN. TimeGAN is different from simply cropping, translating and rotating the data. TimeGAN extracts the time information of the data, so it is very important for action detection. Because the action in this embodiment is continuous, the phase features generated are also continuous, with time characteristics, and are not isolated phase values. Therefore, TimeGAN can generate data that is more similar to the data set, which means that the generated data is more real, which plays a very important role in improving the accuracy of the neural network.

[0087] In addition, in order to give full play to the advantages of the distributed system in the present application, the present embodiment needs to make full use of the data collected by the distributed receiver. That is, how the phase feature put into the classifier should contain the characteristic information of all antenna data. In this regard, the present embodiment proposes a data fusion method that assigns different weights to different antenna data. The starting point of this method is: because the action has different impacts on the paths of different receiving antennas (for example, when human body action occurs in the direct path and non-direct path of signal transmission), the degree of impact on the phase is also different. For antenna data with obvious changes in phase characteristics, it can be intuitively thought that the signal characteristics it contains are also more obvious, so a greater weight is given to it. For antenna data with relatively insignificant changes in phase characteristics, although the characteristic changes are relatively insignificant, the data also contains characteristic information of the signal and should not be discarded, so a smaller weight is given to it. Finally, the data of each antenna is multiplied by its corresponding weight and added together to obtain phase characteristic data containing all antenna data. The specific implementation is as follows:

[0088] Calculation formula (5)

[0089] Calculation formula (6)

[0090] Where W i Represents the weight of each antenna data, which is based on the peak-to-peak value A of the antenna i We get n, which represents the number of receiving antennas. A is calculated based on the weight W.i And each antenna data The phase information obtained by weighted summation contains the characteristics of all antenna data.

[0091] However, in the process of using this data fusion method, data alignment is a problem that needs to be faced. When using the action interception algorithm to intercept the action data of each antenna, it cannot be guaranteed that the start time and length of the data interception for each antenna are completely consistent, so the problem of data misalignment occurs.

[0092] Considering that the distributed antennas of this system are turned on at the same time when collecting signals, it means that the antennas are aligned at the beginning of data collection. Therefore, when the action signal is intercepted, its corresponding time information is also recorded and saved. Then, the data is aligned and retained by comparing the time to ensure that the data of each antenna is aligned in time. Fig.12 shown.

[0093] Although the signals are aligned in time, some data may be missing at certain time points due to signal acquisition reasons, which also poses a challenge to data fusion. In this regard, this application proposes a clever method: for data missing at certain time points, a linear interpolation method is used to complete the data. For example, in multi-antenna data, the data of a certain antenna There is a missing at a certain time point, such as shown in Figure 13 (a), which is recorded as .So The corresponding time point Then the average time of other antennas is obtained: , k represents the number of other antennas with data at this time point. Then assume that the antenna is at time point The previous time point was (The corresponding phase is ), the next time point is (The corresponding phase is ), then the missing phase data Figure 13 (b) shows the data of the four antennas before and after data completion.

[0094] After using this method to fill in the missing data points, the final data is obtained through the above data fusion method as the input of the classifier. Fig.14 This is a result diagram of data fusion for a certain fall action, and the correlation between the fused data and the data of each antenna is calculated. It can be seen that the average correlation coefficient between the fused data and the data of each antenna is above 0.89, which proves that the fused data has good retention of the data information of each antenna.

[0095] The low-cost, distributed RFID-based human motion recognition method provided by the embodiment of the present invention sends an excitation signal and a coding signal through a transmitter; the excitation signal is used to activate the RFID tag; the coding signal is used to instruct the RFID tag to send a reflection signal; the reflection signal is received by at least one receiver; the reflection signal has different phases after being reflected by different human motions; after eliminating the carrier frequency offset, self-interference and random phase of the reflection signal, the phase information in the reflection signal is extracted; the phase information is input into a trained classifier to obtain the human motion corresponding to the phase information output by the classifier; the classifier stores the mapping relationship between the phase information and the human motion. The solution provided by the present invention decouples the RFID reader into a transmitter and a receiver, and uses low-cost equipment to construct the transmitter and the receiver, which can reduce the system cost. At the same time, this embodiment uses at least one receiver to receive the reflection signal, which can improve the signal sampling rate, thereby improving the detection accuracy of the system.

[0096] The embodiment of the present invention also provides a low-cost, distributed RFID-based human motion recognition system, the system comprising a transmitter, at least one receiver, a phase extraction module and a motion recognition module: the transmitter comprises a first development board, a first radio frequency chip and a first transmitting antenna; the receiver comprises a first module and a receiving antenna; the first development board is used to send instructions after programming; the first radio frequency chip is used to control the first transmitting antenna to send signals under the instructions of the first development board; the first transmitting antenna is used to send excitation signals and coding signals; the first module is used to control the receiving antenna to receive signals; the receiving antenna is used to receive reflected signals; the reflected signals have different phases after being reflected by different human motions; the phase extraction module is used to extract the phase information in the reflected signal after performing carrier frequency offset, self-interference and random phase elimination operations on the reflected signal; the motion recognition module is used to input the phase information into a trained classifier to obtain the human motion corresponding to the phase information output by the classifier; the classifier stores the mapping relationship between phase information and human motion.

[0097] This embodiment analyzes the standard EPC C1G2 communication protocol and uses low-cost fully programmable commercial radio components to build a low-cost, distributed RFID-based human motion recognition system. Compared with the existing human motion recognition solutions, it can achieve non-contact and dark environment detection, with a longer perception distance and low cost.

[0098] This embodiment can solve the following technical problems: realize low-cost, distributed RFID transmission system design, solve the problem of insufficient sampling rate; overcome the problems of carrier frequency offset, self-interference and random phase caused by hardware heterogeneity; realize Packet-In-Packet (PIP) communication, solve the compatibility problem between low-cost radio frequency modules and standard RFID communication protocols; propose a set of data enhancement solutions for RFID-based human motion recognition.

[0099] First, the system may include an RFID transceiver subsystem, a phase extraction module, and an action recognition module. The RFID transceiver subsystem may include a transmitter (i.e., a transmitting antenna) and multiple receivers (i.e., a receiving antenna). Fig.15 and Fig.16 As shown, the transmitter in this embodiment is composed of a low-cost Arduino Uno development board, an RFM69HW radio frequency chip and a transmitting antenna. The receiver is composed of a low-cost RTL-SDR dongle module and a receiving antenna. The transmitting antenna first sends an excitation signal to activate the RFID tag. After the tag is activated, the transmitting antenna simultaneously sends a coded signal (i.e., a Query signal frame), and the tag reflects a reflected signal (i.e., an RN16 signal frame) after receiving the Query signal. After the RN16 signal frame is received by the receiver, the system extracts the phase information of the signal by eliminating the carrier frequency offset, self-interference and random phase of the RN16 signal. Finally, by analyzing the different phase features caused by different human body movements, a suitable classifier is trained to realize the detection of human body movements.

[0100] In order to achieve flexible programming, in this embodiment, the IC pin of the RF module can be connected to a low-cost Arduino Uno R3 development board, and the sending and receiving of the Query signal designed above can be achieved by programming the Arduino development board.

[0101] In addition, many low-cost SDR receivers (e.g., RTL-SDR, NESDR Mini 2+, etc.) can be used to receive the reflected signal from the tag. In the present invention, we chose the low-cost RTL-SDRdongle with RTL2832 ADC chip to build the RFID receiver. The present invention uses the same circularly polarized antenna as the transmitting antenna and the receiving antenna. More importantly, the present invention can deploy multiple receivers to simultaneously receive the reflected signal of the tag without collision. At the same time, our design is not limited to a specific type of SDR receiver, and many high-end SDR receivers are likely to provide better performance.

[0102] This embodiment uses low-cost radio equipment to decouple the RFID reader's transmission and reception into two parts: a transmitter and a receiver. At the same time, at least one receiver is set to achieve distributed signal transmission and reception. That is, a one-transmit-multiple-receive mode improves the sampling rate of the system, thereby improving the detection accuracy of the system. At the same time, the system has a simple structure, a small size, and is easy to meet the requirements for test equipment and conditions. It has high versatility and can be used for monitoring in daily environments.

[0103] The above-mentioned system provided in this embodiment belongs to the same concept as the above-mentioned method embodiment. The specific implementation process can be found in the method embodiment in detail and will not be repeated here.

[0104] In order to implement the method of the embodiment of the present invention, the embodiment of the present invention also provides a computer program product, which includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps of the above method.

[0105] Based on the hardware implementation of the above program modules, and in order to implement the method of the embodiment of the present invention, the embodiment of the present invention further provides an electronic device (computer device). Specifically, in one embodiment, the computer device may be a terminal, and its internal structure diagram may be as follows: Fig.17 As shown. The computer device includes a processor A01, a network interface A02, a display screen A04, an input device A05 and a memory (not shown in the figure) connected through a system bus. Among them, the processor A01 of the computer device is used to provide computing and control capabilities. The memory of the computer device includes an internal memory A03 and a non-volatile storage medium A06. The non-volatile storage medium A06 stores an operating system B01 and a computer program B02. The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 in the non-volatile storage medium A06. The network interface A02 of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor A01, the method of any one of the above embodiments is implemented. The display screen A04 of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device A05 of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0106] Those skilled in the art will understand that Fig.17The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0107] The device provided by the embodiment of the present invention includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, the method of any one of the above embodiments is implemented.

[0108] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0109] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0110] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0111] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1The steps for the functions specified in one or more boxes.

[0112] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0113] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0114] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0115] It can be understood that the memory of the embodiment of the present invention can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM); the magnetic surface memory can be a disk memory or a tape memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM, SyncLink Dynamic Random Access Memory), and direct RAMbus random access memory (DRRAM, Direct Rambus Random Access Memory).The memories described in the embodiments of the present invention are intended to include, but are not limited to, these and any other suitable types of memories.

[0116] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0117] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A low-cost, distributed RFID-based human motion recognition method, characterized in that: The method comprises: An excitation signal and a coding signal are sent through a transmitter; the excitation signal is used to activate the RFID tag; the coding signal is used to instruct the RFID tag to send a reflection signal; The reflected signal is received by at least one receiver; the reflected signal has different phases after being reflected by different continuous movements of the human body; After performing carrier frequency offset, self-interference and random phase elimination operations on the reflected signal, extracting phase information from the reflected signal; Inputting the phase information into a trained classifier to obtain a human body motion corresponding to the phase information output by the classifier; the classifier stores a mapping relationship between the phase information and the human body motion; The step of sending the excitation signal and the coding signal through the transmitter includes: Determining a pulse width coding frame structure of the excitation signal and the coding signal; Using a preset conversion relationship between pulse width coding and on-off keying coding, the pulse width coding frame structure of the excitation signal and the coding signal is converted to obtain the on-off keying coding frame structure of the excitation signal and the coding signal; The excitation signal and the on-off keying coding frame structure of the coding signal are sent by a transmitter.

2. The low-cost, distributed RFID-based human motion recognition method according to claim 1, characterized in that: The conversion relationship between the preset pulse width code and the on-off keying code is that 1 in the pulse width code is represented by the on-off keying code 1110, and 0 in the pulse width code is represented by the on-off keying code 10.

3. The low-cost, distributed RFID-based human motion recognition method according to claim 1, characterized in that: The coded signal includes indication information; when the indication information is 1, the coded signal instructs the RFID tag to send a reflection signal; when the indication information is 0, the coded signal instructs the RFID tag not to send a reflection signal.

4. The low-cost, distributed RFID-based human motion recognition method according to claim 3, characterized in that: Performing operations of eliminating carrier frequency offset, self-interference and random phase on the reflected signal, including: Acquire a first reflection signal when the indication information is 0 and a second reflection signal when the indication information in the same period is 1; Dividing the second reflected signal by the first reflected signal to obtain a third reflected signal with carrier frequency offset eliminated; Subtracting the first reflected signal from the third reflected signal to obtain a fourth reflected signal with the self-interference term eliminated; The fourth reflected signal is divided by the first reflected signal to obtain a fifth reflected signal with the unknown phase offset eliminated.

5. The low-cost, distributed RFID-based human motion recognition method according to claim 1, characterized in that: Extracting phase information from the reflected signal, comprising: Get the preset threshold value; Extracting from the reflected signal a first segment signal whose phase values ​​are all higher than the preset threshold and a second segment signal whose phase values ​​are all lower than the preset threshold; Subtracting the second segment signal from the first segment signal to obtain a human body reflection signal with static path interference eliminated; Phase information in the human body reflection signal is extracted.

6. The low-cost, distributed RFID-based human motion recognition method according to claim 1, characterized in that: When the number of the receivers is greater than 1, extracting phase information from the reflected signal includes: Determine the weight of each reflected signal according to the peak value in the reflected signal received by each receiver; The phase information in the reflected signal is obtained by weighted summing using the weight of each reflected signal.

7. The low-cost, distributed RFID-based human motion recognition method according to claim 1, characterized in that: The weight of each reflected signal is determined according to a peak value in the reflected signal received by each receiver, including: The weight of each reflected signal is determined using the following calculation formula: ； Among them, W i is the weight of each reflected signal, A i is the peak value of the reflected signal, and n is the number of reflected signals.

8. The low-cost, distributed RFID-based human motion recognition method according to claim 1, characterized in that: The phase information in the reflected signal is obtained by weighted summing using the weight of each reflected signal, including: The phase information in the reflected signal is obtained using the following calculation formula: ； Among them, A is based on the weight W i and the reflected signal received by each receiver The phase information containing all reflected signal characteristics is obtained by weighted summation, W i is the weight of each antenna data, and n is the number of antenna data.

9. A low-cost, distributed RFID-based human motion recognition system, characterized in that: The system includes a transmitter, at least one receiver, a phase extraction module and an action recognition module: the transmitter includes a first development board, a first radio frequency chip and a first transmitting antenna; the receiver includes a first module and a receiving antenna; The first development board is used to send instructions after programming; The first radio frequency chip is used to control the first transmitting antenna to send signals under the instruction of the first development board; The first transmitting antenna is used to send an excitation signal and a coding signal; The first module is used to control the receiving antenna to receive signals; The receiving antenna is used to receive the reflected signal; the reflected signal has different phases after being reflected by different continuous movements of the human body; The phase extraction module is used to extract the phase information in the reflected signal after performing carrier frequency offset, self-interference and random phase elimination operations on the reflected signal; The action recognition module is used to input the phase information into a trained classifier to obtain the human body action corresponding to the phase information output by the classifier; the classifier stores a mapping relationship between the phase information and the human body action; Among them, the first transmitting antenna is also used to determine the pulse width coding frame structure of the excitation signal and the coding signal; use the preset conversion relationship between pulse width coding and on-off keying coding to convert the pulse width coding frame structure of the excitation signal and the coding signal to obtain the on-off keying coding frame structure of the excitation signal and the coding signal; and send the on-off keying coding frame structure of the excitation signal and the coding signal through a transmitter.

Images