阅读说明：本技术 一种物联网LoRa信号的波束成形多目标感知方法及系统 (Beam forming multi-target sensing method and system for LoRa signals of Internet of things ) 是由 张扶桑 金蓓弘 于 2021-07-13 设计创作，主要内容包括：本发明涉及一种物联网LoRa信号的波束成形多目标感知方法及系统,其方法包括：步骤S1：基于LoRa信号的波束成形信号构建具有M根天线阵列的LoRa网关,可得到复平面波束成形信号；步骤S2：根据复平面波束成形信号,进行角度-频率谱信号处理,得到角度-频率谱图；步骤S3：根据角度-频率谱图,通过频率和角度扫描以检测目标。本发明公开的方法,通过在特定方向加强接收信号,从而将来自其它方向的信号削弱,解决多目标感知干扰问题,实现多个目标的同时感知。(The invention relates to a beamforming multi-target sensing method and system for LoRa signals of the Internet of things, wherein the method comprises the following steps: step S1: constructing an LoRa gateway with M antenna arrays based on the beam forming signals of the LoRa signals, and obtaining complex plane beam forming signals; step S2: processing an angle-frequency spectrum signal according to the complex plane beam forming signal to obtain an angle-frequency spectrum; step S3: and according to the angle-frequency spectrogram, detecting the target through frequency and angle scanning. The method disclosed by the invention weakens signals from other directions by strengthening the received signals in a specific direction, solves the problem of multi-target perception interference and realizes simultaneous perception of a plurality of targets.)

1. A beamforming multi-target sensing method for LoRa signals of the Internet of things is characterized by comprising the following steps:

step S1: constructing an LoRa gateway with M antenna arrays based on the beam forming signals of the LoRa signals, and obtaining complex plane beam forming signals;

step S2: processing an angle-frequency spectrum signal according to the complex plane beam forming signal to obtain an angle-frequency spectrum;

step S3: and according to the angle-frequency spectrogram, detecting the target through frequency and angle scanning.

2. The method for beamforming multi-target sensing of an LoRa signal of an internet of things according to claim 1, wherein the step S1: constructing an LoRa gateway with M antenna arrays based on the beam forming signals of the LoRa signal, so as to obtain complex plane beam forming signals, specifically comprising:

step S11: the path lengths from the detected target to the M antennas are respectively represented as d₁(t)，d₂(t)，...，d_M(t), where t is time, function d_i(t) represents a function of the path length over time; the dynamic vector and the static vector corresponding to the M antennas are respectively expressed as H_d1，H_d2，...，H_dMAnd H_s1，H_s2，...，H_sM(ii) a The signals of the M receiving antennas are expressed as the following formula (1):

wherein k is a sweep frequency,for the carrier frequency error to be the carrier frequency error,for sampling frequency error, A_iIs the dynamic vector intensity on the ith antenna, λ is the LoRa signal wavelength, j represents the complex symbol;

step S12: the phase difference between the dynamic signal arriving at the ith antenna and the first antenna and the phase differences between the dynamic signal arriving at the ith antenna and the other antennas are calculated by the following formulas (2) to (3):

wherein θ is the arrival angle of the signal; ω (θ) is a steering vector;is the phase difference between the ith antenna and the first antenna;

step S13: the received signal in the beamforming direction θ is constructed by calculating a weighted sum of the received signals on all M antennas, and is expressed as the following equation (4):

wherein the static vector is combinedIs shown asEquation (4) can be simplified to equation (5) below:

due to d_i(t)-(i-1)d·sin(θ)＝d₁(t) the 2 nd to mth antennas are aligned with the first antenna; order toI.e., the sum of the energies of the dynamic components on all antennas, equation (5) can be simplified to equation (6) below:

step S14: since random phase rotation may be caused by carrier frequency error (CFO) and sampling frequency error (SFO), it is necessary to eliminate the random phase rotation in equation (6)Multiplying the signal received by each antenna by the weight vector to obtain the following equation (7):

order toIs a 1 × M weight vector, in which there are 2M parameters to be determined, and a 1 × M-dimensional weight vector W needs to be found, so that

Let W be the solution result, obtain the beam nulling signal from equation (7), as shown in equation (8);

the reference signal is denoted asWherein the content of the first and second substances,

to eliminate the unwanted components, y (θ, t) is divided by ref (t) to obtain the complex plane beamforming signal, as shown in equation (9):

wherein the content of the first and second substances,is an estimated static vector;a beam enhanced dynamic vector.

3. The method for beamforming multi-target sensing of an LoRa signal of an internet of things according to claim 1, wherein the step S2: according to the complex plane beam forming signal, processing an angle-frequency spectrum signal, specifically comprising:

step S21: adopting a Fast Fourier Transform (FFT) algorithm to each angle of the complex plane beam forming signal to obtain a result with positive and negative frequencies;

step S22: summing the results of the positive and negative frequencies to obtain the actual corresponding energy of the detection target, so as to obtain an angle-frequency spectrogram; and acquiring an approximate region where the detection target is located according to the angle-frequency spectrogram.

4. The method for beamforming multi-target sensing of an LoRa signal of an internet of things according to claim 1, wherein the step S3: according to the angle-frequency spectrogram, detecting a target through frequency and angle scanning specifically comprises the following steps:

step S31: selecting signals with strong periodicity and large fluctuation in the angle-frequency spectrogram as candidate frequencies through frequency scanning;

step S32: and taking the angle with the energy in the candidate frequency higher than a dynamic threshold as a final detection target through angle scanning, wherein the dynamic threshold is obtained through dynamic calculation according to the maximum energy and the minimum energy.

5. The beamforming multi-target sensing system for the LoRa signals of the Internet of things is characterized by comprising the following modules:

the beam forming signal constructing module is used for constructing an LoRa gateway with M antenna arrays based on the beam forming signals of the LoRa signals, so that complex plane beam forming signals can be obtained;

an angle-frequency spectrogram constructing module, configured to perform angle-frequency spectrogram signal processing according to the complex plane beam forming signal to obtain an angle-frequency spectrogram;

and the frequency and angle scanning module is used for scanning through frequency and angle according to the angle-frequency spectrogram so as to detect the target.

Technical Field

The invention relates to the field of intelligent sensing, in particular to a method and a system for multi-target sensing of beam forming of LoRa signals of the Internet of things.

Background

In recent years, wireless sensing has received a great deal of attention from both academic and industrial communities. Various wireless technologies for data communication are used for wireless non-contact sensing. These wireless sensing technologies include WiFi, RFID, 60GHz, UWB, ultrasound and visible light. Non-contact sensing applications range from fine-grained respiration monitoring, finger tracking to coarse-grained fall detection, gait recognition and daily activity recognition. Despite the broad prospects, a key problem remains: wireless sensing has difficulty in dealing with multiple goals. Because the wireless sensing relies on the reflected signal from the sensed object to obtain information about the direction and speed of motion. When there are multiple sensing targets, the signals of the multiple reflecting targets are mixed and superposed at the receiving end, and thus cannot be used for sensing any single target. This limitation restricts the wireless sensing system to work only when there is one sensing target, which severely hinders the wide use of the system when there are multiple targets in the real scene. For example, a couple may perform constant breath tests at night, assess sleep quality, and apnea and other abnormal events. A signal reflected from one person can significantly interfere with the detection of another person. Therefore, sensing multiple targets is an important challenge for wireless sensing.

The LoRa signal is subjected to non-contact sensing and has the characteristic of wide sensing range, and the LoRa is originally designed to be used for remote communication among the devices of the Internet of things. The communication range is further than traditional indoor WiFi signals. In open areas, the transmission range can reach several kilometers [11 ]. The key to long-range communications is the use of chirped spread spectrum modulation (Chrip) and its high coding gain. Compared to WiFi, 0-5dB signal strength is required for decoding, and LoRa can still decode even if the signal-to-noise ratio is as low as-20 dB. Therefore, using the LoRa signal to significantly increase the sensing range, a wide range of sensing at 25-50 meters can be achieved. Although greater distance is an important step towards practical use in wireless sensing, the problem of multi-target sensing interference introduced in a large range becomes more serious.

Disclosure of Invention

In order to solve the technical problem, the invention provides a method and a system for multi-target sensing of beam forming of LoRa signals of the Internet of things.

The technical solution of the invention is as follows: a beamforming multi-target sensing method for LoRa signals of the Internet of things comprises the following steps:

step S1: constructing an LoRa gateway with M antenna arrays based on the beam forming signals of the LoRa signals, and obtaining complex plane beam forming signals;

step S2: processing an angle-frequency spectrum signal according to the complex plane beam forming signal to obtain an angle-frequency spectrum;

step S3: and according to the angle-frequency spectrogram, detecting the target through frequency and angle scanning.

Compared with the prior art, the invention has the following advantages:

the invention discloses a multi-target sensing method for beam forming of LoRa signals of the Internet of things, which is characterized in that a beam forming method based on LoRa multi-antennas is designed by utilizing long-distance and low-power-consumption LoRa signals, a phase calibration method of signals is provided by searching reference signals without dynamic target influence, and all antenna signals are aligned and superposed, so that good directivity is given to waveforms, and interference is eliminated. On the basis, a signal angle-frequency spectrum generation method is provided for multi-target identification by using complex plane representation of signals, and targets are jointly estimated by frequency scanning (frequency domain) and angle scanning (airspace) in combination with signal initial phase information (time), so that the problem of multi-target perception interference is solved, and simultaneous perception of a plurality of targets is realized.

Drawings

Fig. 1 is a flowchart of a beamforming multi-target sensing method for an internet of things LoRa signal in an embodiment of the present invention;

fig. 2 is a specific flow diagram of a beamforming multi-target sensing method for an internet of things LoRa signal according to an embodiment of the present invention;

fig. 3 is a diagram of a LoRa beam in accordance with an embodiment of the present invention;

fig. 4 is a diagram illustrating LoRa beamforming according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of beam nulling in an embodiment of the present invention;

FIG. 6 is a schematic view of an angle-frequency spectrum according to an embodiment of the present invention;

FIG. 7 is a diagram illustrating a multi-target breath detection scenario in an embodiment of the present invention;

FIG. 8 is a schematic view of an angle-frequency spectrum according to an embodiment of the present invention;

FIG. 9 is a schematic diagram of frequency sweep thresholds in an embodiment of the present invention;

FIG. 10 is a schematic diagram illustrating the effect of real-time visualization of multi-target respiration monitoring in an embodiment of the present invention;

fig. 11 is a structural block diagram of a beamforming multi-target sensing system for an LoRa signal of the internet of things in the embodiment of the present invention.

Detailed Description

The invention provides a beam forming multi-target sensing method for LoRa signals of the Internet of things, which weakens signals from other directions by strengthening received signals in a specific direction, thereby solving the problem of multi-target sensing interference and realizing the simultaneous sensing of a plurality of targets.

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings.

The method according to the invention is directed to the detection of objects or persons having a frequency, for example, vibrating objects having a vibration frequency or persons having a breathing frequency. The following embodiments take the example of human respiratory rate as an example, and perform multi-target sensing.

Example one

As shown in fig. 1, the beamforming multi-target sensing method for the LoRa signal of the internet of things provided by the embodiment of the present invention includes the following steps:

step S1: constructing an LoRa gateway with M antenna arrays based on the beam forming signals of the LoRa signals, and obtaining complex plane beam forming signals;

step S2: processing an angle-frequency spectrum signal according to the complex plane beam forming signal to obtain an angle-frequency spectrum;

step S3: and according to the angle-frequency spectrogram, detecting the target through frequency and angle scanning.

As shown in fig. 2, a flow chart of the method provided by the present invention is shown.

In one embodiment, the step S1: constructing an LoRa gateway with M antenna arrays based on the beam forming signals of the LoRa signal, so as to obtain complex plane beam forming signals, specifically comprising:

step S11: the path lengths from the detected target to the M antennas are respectively represented as d₁(t),d₂(t),…,d_M(t), where t is time, function d_i(t) represents a function of the path length over time; the dynamic vector and the static vector corresponding to the M antennas are respectively expressed as H_d1,H_d2,…,H_dMAnd H_s1,H_s2,…,H_sM(ii) a The signals of the M receiving antennas are expressed as the following formula (1):

…

where k is the sweep frequency, phi_cIs the carrier frequency error, phi_sFor sampling frequency error, A_iThe strength of the dynamic vector on the ith antenna is shown, lambda is LoRa signal wavelength, and j is a complex symbol;

as shown in fig. 3, a LoRa beam diagram is shown, wherein the gray solid arrows represent dynamic vectors and the black solid arrows represent static vectors. To obtain stronger signal reception in a particular direction, the path delays of the signals are compensated and the multi-antenna signals are summed. Assuming that the linear antenna array has a distance d, d is 16 cm for the 915MHz LoRa signal used in the embodiment of the present invention. The detection target and antenna array distance can be tens of meters, much greater than d. Thus, the angle of the target to each antenna can be considered to be the same.

The signals received by the 2 nd to m th antennas can be regarded as delayed versions of the signals received by the first antenna. Therefore, there is an extra signal propagation path length in air, and accordingly an extra phase rotation angle. And when the arrival angle of the signal is theta, calculating the phase difference between the m-th antenna and the first antenna of the dynamic signal and the phase differences between the first antenna and the rest antennas.

Step S12: the phase difference between the dynamic signal arriving at the ith antenna and the first antenna and the phase differences between the dynamic signal arriving at the ith antenna and the other antennas are calculated by the following formulas (2) to (3):

wherein θ is the arrival angle of the signal; ω (θ) is a steering vector;is the phase difference between the ith antenna and the first antenna;

step S13: the received signal in the beamforming direction θ is constructed by calculating a weighted sum of the received signals on all M antennas, and is expressed as the following equation (4):

wherein the static vector is combinedIs shown asEquation (4) can be simplified to equation (5) below:

due to d_i(t)-(i-1)d·sin(θ)＝d₁(t) the 2 nd to mth antennas are aligned with the first antenna; order toI.e., the sum of the energies of the dynamic components on all antennas, equation (5) can be simplified to equation (6) below:

the beamformed signals are composed of static and dynamic components. By choosing the correct theta, the reflected signals (dynamic components) of the targets on the M antennas add up to enhance, while the static components add up randomly. As shown in fig. 4, the LoRa beam forming diagram shows that after beam forming, the target information contained in the dynamic vector is greatly enhanced, while the static vector is not changed much (due to random addition). Thus, in vector space, the motion vector is rotated relative to the static vector, producing a stronger signal than before. However, carrier frequency error (CFO) and sampling frequency error (SFO) corrupt the signal variations due to their presence. To eliminate the CFO and SFO induced random phase rotation, this term needs to be removedThe method provided by the invention can eliminate the unwanted signals, mainly random phase offsets CFO and SFO, by finding a reference signal. It is also necessary to ensure that the dynamic component containing the target perceptual information is not affected in this process. To achieve this, the method performs the inverse operation of the beamforming process for summing up the dynamic component cancellation acquired by the multiple antennas at the receiving end, which is called "beam nulling" as shown in fig. 5.

Step S14: since random phase rotation may be caused by carrier frequency error (CFO) and sampling frequency error (SFO), it is necessary to eliminate the random phase rotation in equation (6)Every dayThe signal received by the line is multiplied by the weight vector to yield the following equation (7):

order toIs a 1 × M weight vector, in which there are 2M parameters to be determined, and a 1 × M-dimensional weight vector W needs to be found, so that

Since there is no prior knowledge of the magnitude of the dynamic vector for each antenna, the weight vector W cannot be solved directly. It is therefore expressed as an optimization problem, in particular, defining the energy ratio of the dynamic component frequencies for a given W. That is, in the frequency spectrum, the proportion of the total energy in the frequency spectrum of the energy caused by the motion of the object. Specifically, the calculation of the energy ratio includes: obtaining a weighted sum signal of the dynamic frequency components over a time window by performing Fast Fourier Transform (FFT) on the signal; 2) the sum of the energies of the FFT peak and its neighbors is divided by the sum of the energies of all FFT points. And (4) applying a random optimization algorithm to iteratively optimize and search to find the best W. Let W be the solution result, obtain the beam nulling signal from equation (7), as shown in equation (8);

therefore, the components which only contain static components and need to be eliminated can be quickly obtained; reference signal Ref (t) is expressed asWherein the content of the first and second substances,

to eliminate the unwanted components, y (θ, t) is divided by ref (t) to obtain the complex plane beamforming signal, as shown in equation (9):

wherein the content of the first and second substances,is an estimated static vector;a beam enhanced dynamic vector.

From equation (9), it can be seen that the varying random phase offset is eliminated. The remaining part is the dynamic path length d₁(t) and some constants (i.e., static vectors), the dynamic vectors are enhanced while preserving perceptual information.

In one embodiment, the step S2: according to the complex plane beam forming signal, the angle-frequency spectrum signal processing is carried out, and the method specifically comprises the following steps:

step S21: adopting a Fast Fourier Transform (FFT) algorithm to each angle of the complex plane beam forming signal to obtain a result with positive and negative frequencies;

step S22: summing the results of the positive frequency and the negative frequency to obtain the actual corresponding energy of the detection target, so as to obtain an angle-frequency spectrogram; and acquiring an approximate region where the detection target is located according to the angle-frequency spectrogram.

Taking four people breathing in a room for example, as shown in fig. 7, they are sitting at-40, 0, 20 and 40 degrees relative to the LoRa device, with corresponding breathing rates of 0.23Hz, 0.37Hz, 0.23 Hz. The energy distribution of the respiratory signals of the four detection targets in each direction is calculated, and an angle-frequency spectrogram as shown in fig. 8 is obtained, wherein three areas with prominent highlights are clearly displayed. The first human with a breathing rate of-40 degrees at 0.23Hz and the second target at 0 degrees at 0.37 Hz. Since the targets 3 and 4 have the same breathing frequency and direction of approach, a distinction needs to be made by subsequent processing.

In the angle-frequency spectrogram, the identification target has two features: angle and frequency. Due to the side lobes of beamforming, the same target can be detected from multiple angles. If there is only one target, the direction in which the signal is stronger can simply be selected. However, when there are multiple targets, the two targets may have the same angle. Therefore, a simple threshold cannot be employed to detect the target.

The invention provides a frequency and angle scanning mode based on a dynamic threshold value to detect the accurate position of a target. It is considered that even if multiple targets have the same breathing rate, their breathing rhythms may not be completely synchronized. Thus, the curvature of the respiration waveform can be calculated, and using the time difference of the peaks in the curvature, the target can be distinguished even at the same rate and the same angle. By dynamically adjusting the energy threshold, frequency scanning and angle scanning are sequentially performed, thereby eliminating false alarms caused by side lobes and detecting multiple targets.

In one embodiment, the step S3: according to the angle-frequency spectrogram, the method detects the target through frequency and angle scanning, and specifically comprises the following steps:

step S31: selecting signals with strong periodicity and large fluctuation in an angle-frequency spectrogram as candidate frequencies through frequency scanning;

in the frequency sweep, a signal having strong periodicity and large fluctuation is regarded as a candidate target. The maximum energy in all angles is selected for each frequency in the angle-frequency spectrogram. As fig. 9 shows the frequency scan results for 4 targets detected, in the example, the maximum and minimum energies are 3.566 and 0.502, and the embodiment of the present invention sets the dynamic threshold to 1.726. Therefore, 0.23Hz and 0.37Hz can be selected as candidate frequencies, thereby distinguishing the object 2 from other objects.

Step S32: and taking the angle with the energy in the candidate frequency higher than a dynamic threshold value as a final detection target through angle scanning, wherein the dynamic threshold value is dynamically calculated according to the maximum energy and the minimum energy.

For the candidate frequencies, angles with energies above a threshold are selected as candidates. Here, the dynamic threshold frequency is also calculated from the maximum and minimum energy at that time. By using only the angle information, the object 1 can be separated from the objects 3, 4. However, when the two targets (3 and 4) have similar breathing frequencies and are also close to each other physically, the angular information is not sufficient to separate them. The peak times on the curvature lines will be further compared. If the peaks of two candidate targets are separated in time, it is considered that they are likely to correspond to different targets. Therefore, even if the target 3 and the target 4 are located at the same position, the angle information cannot be separated and can be distinguished with a clear timing relationship.

The embodiment of the invention adopts a LoRa sensing platform formed based on LoRa nodes and gateways, wherein the LoRa nodes are Semtech SX1276 connected to Arduino Uno. The LoRa node transmits signals with a bandwidth of 125kHz over the 915MHz frequency band. The LoRa gateway is based on USRP X310 and is equipped with four directional antennas (TX 900-PB-1313). The LoRa gateway is connected to a notebook computer with an Intel Core i7 CPU and 16GB RAM. Samples were taken at 1000 samples per second and processed in real time using MATLAB. To monitor the breathing of multiple targets, signal streams on four antennas are acquired, each time processing the signal streams in a sliding window of 30 seconds. For each scan angle, a signal representation in the complex plane is calculated and the outliers are removed with a band pass filter. By analyzing the breathing pattern, the number of targets and their corresponding breathing frequency can be determined. The respiration monitoring band NUL-236 is used to provide a true value for respiration detection. Fig. 10 shows a schematic diagram of the effect of real-time visualization of multi-target respiration monitoring.

The method provided by the invention can be used for sensing multiple targets by using angle information (different spatial orientations), frequency information (different breathing frequencies) and time information (initial phases caused by breathing of different targets, namely curvature peak value time difference).

The invention discloses a multi-target sensing method for beam forming of LoRa signals of the Internet of things, which is a beam forming method based on LoRa multi-antenna design by utilizing long-distance and low-power-consumption LoRa signals. By searching for a reference signal without dynamic target influence, a signal phase calibration method is provided, and each antenna signal is aligned and superposed, so that good directivity is given to a waveform, and interference is eliminated. On the basis, a signal angle-frequency spectrum generation method is provided for multi-target identification by using complex plane representation of signals, and targets are jointly estimated by frequency scanning (frequency domain) and angle scanning (airspace) in combination with signal initial phase information (time), so that the problem of multi-target perception interference is solved, and simultaneous perception of a plurality of targets is realized.

Example two

As shown in fig. 11, an embodiment of the present invention provides a beamforming multi-target sensing system for an LoRa signal of an internet of things, including the following modules:

a beam forming signal constructing module 41, configured to construct an LoRa gateway having M antenna arrays based on a beam forming signal of the LoRa signal, so as to obtain a complex plane beam forming signal;

an angle-frequency spectrogram constructing module 42, configured to perform angle-frequency spectrogram signal processing according to the complex plane beam forming signal to obtain an angle-frequency spectrogram;

and a frequency and angle scanning module 43, configured to scan through frequency and angle to detect the target according to the angle-frequency spectrogram.

The above examples are provided only for the purpose of describing the present invention, and are not intended to limit the scope of the present invention. The scope of the invention is defined by the appended claims. Various equivalent substitutions and modifications can be made without departing from the spirit and principles of the invention, and are intended to be within the scope of the invention.