A Small UAV Target Tracking Method Based on MIMO Radar

Abstract

The present invention specifically relates to a method for tracking small unmanned aerial vehicle (UAV) targets based on MIMO radar, including: acquiring the beat signals of each antenna array of the target radar; performing range-FFT processing on the beat signals to convert the time-domain signals in the range dimension of the beat signals to the frequency domain; performing RFT processing on the beat signals after range-FFT processing to correct the range migration caused by the movement of distant targets in the beat signals and achieve single-frame coherent accumulation of target energy; performing two-dimensional CFAR processing on the beat signals after RFT processing to achieve target detection of the UAV targets and obtain target detection results; performing angle-FFT processing on the target detection results to extract multi-frame target results; and realizing continuous tracking of the corresponding UAV targets based on the multi-frame target results in combination with the particle filter algorithm. The method for tracking small UAV targets based on MIMO radar in the present invention can ensure the accuracy and effectiveness of small UAV target detection and tracking.

Description

Technical Field

[0001] The present invention relates to the technical field of target detection, and particularly to a method for tracking small unmanned aerial vehicle (UAV) targets based on MIMO radar. Background Art

[0002] Small (rotary-wing) UAVs have been widely used in fields such as education and research, aerial photography, military, agriculture, and transportation, playing a crucial role in our lives. However, due to the lack of unified standards and regulatory control measures in the UAV industry, UAV illegal flight incidents occur frequently, not only violating air traffic control but also endangering national security, people's lives and property safety, and disrupting social order. Therefore, in the face of the potential safety hazards brought by such UAV targets, it is urgent to develop effective UAV tracking and monitoring methods and formulate UAV regulatory control measures to ensure the safety of the low-altitude airspace.

[0003] Existing UAV target detection methods mainly include sound perception technology, optical video detection and tracking technology, passive radar detection and positioning technology, and active radar detection technology. Among them, the Chinese patent with the publication number CN112577481A discloses "A method for positioning ground targets of a rotary-wing UAV", which combines the image position information output by the target detection algorithm with information such as the attitude angle and altitude of the UAV, calculates the relative position information between the target and the UAV in the navigation coordinate system through the coordinate transformation relationship, and combines the UAV GPS longitude and latitude information to calculate the absolute position information of the target; designs the control law of the UAV to achieve the target positioning of the system platform.

[0004] The method for positioning ground targets of the rotary-wing UAV in the above existing solution actually uses optical video detection and tracking technology to achieve the tracking of small UAV targets. The optical video detection and tracking technology uses a camera to obtain the image information of the UAV target, but the blind area of the camera's line of sight cannot be avoided, and it is easily affected by weather conditions such as night, cloud, fog, rain, etc. At the same time, the acquisition radius of the audio of the sound perception technology is relatively small, making it difficult to ensure the detection and tracking effect; while the effective radiation power of the external radiation source of the passive radar detection and positioning technology is relatively low, and the echo formed after irradiating the target becomes very weak and is easily affected by interference and clutter, thus reducing the performance of the detection system.

[0005] The applicant has found that the active radar detection technology uses its own transmitting device to generate electromagnetic waves in a certain frequency band to irradiate the UAV target and receive the target echo signal reflected back. This method has the advantages of high sensitivity, good resolution, the ability to detect targets at a long distance, the ability to provide environmental perception under all weather conditions, and strong anti-interference ability, and can be applied to the tracking of small UAV targets. However, small UAVs belong to the "low, small, and weak" targets, mostly flying in the low-altitude airspace, with a small RCS (radar cross section) and weak radar echo signals received. The UAV target is easily submerged in clutter, making it difficult to detect and track the UAV, and ordinary active radars are difficult to ensure the accuracy of UAV target tracking.

[0006] However, the applicant has found that MIMO radar has decorrelation for target echo signals, making the average received energy of the echo tend to be constant, capable of smoothing the target RCS and improving the RCS fluctuation of the target, thereby improving the ability to suppress clutter interference and the performance of target detection and tracking, and ensuring the accuracy and effectiveness of small UAV target detection and tracking. Therefore, how to design a method for realizing small UAV target tracking based on MIMO radar is a technical problem that needs to be solved urgently. Summary of the Invention

[0007] Aiming at the deficiencies of the above-mentioned existing technologies, the technical problem to be solved by the present invention is: how to provide a method for realizing small UAV target tracking based on MIMO radar, so as to ensure the accuracy and effectiveness of small UAV target detection and tracking.

[0008] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0009] A method for tracking small UAV targets based on MIMO radar, comprising the following steps:

[0010] S1: Obtain the beat signals of each antenna array of the target radar for detecting the corresponding UAV target;

[0011] S2: Perform range-FFT processing on the beat signals to convert the time-domain signals in the distance dimension of the beat signals to the frequency domain;

[0012] S3: Perform RFT processing on the beat signals after range-FFT processing to correct the range migration caused by the movement of targets at a long distance and achieve single-frame coherent accumulation of target energy;

[0013] S4: Perform two-dimensional CFAR processing on the beat signals after RFT processing to achieve the detection of UAV targets and obtain the corresponding target detection results;

[0014] S5: Perform angle-FFT processing on the target detection result to extract multi-frame target results containing distance information, speed information, and angle information of the UAV target;

[0015] S6: Based on the multi-frame target results, combine the particle filter algorithm to achieve continuous tracking of the corresponding UAV target.

[0016] Preferably, in step S1, the target radar is a MIMO radar system composed of two transmitting antennas with a spacing of 2λ and four receiving antennas with a spacing of λ / 2; moreover, the transmitting angle and receiving angle of the target radar are equal, so that an eight-element virtual antenna array with a spacing of λ / 2 can be formed, where λ represents the wavelength of the target radar.

[0017] Preferably, the beat signal is expressed by the following formula:

[0018]

[0019] In the formula: S b represents the beat signal; c represents the speed of light; N represents the number of chirps; T c represents the chirp period; k represents the frequency modulation slope; v represents the speed of the UAV target relative to the target radar; R represents the distance between the UAV target and the target radar; f 0 represents the carrier frequency; n represents the number of receiving antennas; d represents the antenna array spacing; θ represents the azimuth angle of the UAV target relative to the target radar.

[0020] Preferably, in step S3, the formula for RFT processing is as follows:

[0021]

[0022] In the formula: ε represents a known constant relative to f(t, r + vt);

[0023] The linear equation of range walk caused by the movement of the UAV target is as follows:

[0024] r s = r + vt, t ∈ [-T / 2, T / 2];

[0025] In the formula: r represents the slant range; v represents the radial velocity at t = 0; T represents the coherent integration time.

[0026] Preferably, during RFT processing, in the complex signal domain, jointly utilize the amplitude and phase information to achieve continuous coherent integration, and introduce a pair of Doppler compensation functions H v (t) and H θ (t) as Fourier integral components to cancel the phase fluctuations between different pulses, generate the final coherent peak, and correct the range walk caused by the movement of the UAV target.

[0027] Preferably, a pair of Doppler compensation functions H v (t) and H θ (t) are respectively expressed as:

[0028]

[0029] In the formula: v represents the radial velocity at the moment of t = 0; λ represents the wavelength; θ represents the polar angle, which represents the counterclockwise angle from the range migration line to the t-axis on the t-r s plane; c represents the speed of light; j represents the imaginary part of a complex number; g represents the wave path difference introduced by the array interval.

[0030] Preferably, in step S4, the two-dimensional CFAR processing includes a reference cell, a guard cell, and a detection cell. The reference cell is used to estimate the noise power, the guard cell is used to make the average interference power estimation value more accurate, and the detection cell is used to judge the range-Doppler dimensional data of the target.

[0031] Preferably, the two-dimensional CFAR threshold factor T during the two-dimensional CFAR processing is expressed as:

[0032]

[0033] In the formula: P fa represents the preset false alarm probability; 2N refer represents the length of the reference cell.

[0034] Preferably, when the product of the two-dimensional CFAR threshold factor T and the output Z is less than the value of the detection cell, the target peak is detected; when the product of the two-dimensional CFAR threshold factor T and the output Z is greater than the value of the detection cell, there is no target peak.

[0035] Preferably, in step S6, the particle weight in the particle filter algorithm is expressed as:

[0036]

[0037] In the formula: i represents the i-th particle; k represents the k-th moment; R represents the measurement noise covariance matrix; represents the prediction of the observation value of the i-th particle at the k-th moment; Z g (k) represents the observation value at the k-th moment.

[0038] Compared with the prior art, the UAV target tracking method in the present invention has the following beneficial effects:

[0039] In the present invention, a method for small unmanned aerial vehicle (UAV) target tracking based on MIMO radar utilizes the decorrelation of the target echo signals by the MIMO radar, which makes the average received energy of the echoes tend to be constant, enables smoothing of the target RCS and improves the RCS fluctuation of the target. Furthermore, it enhances the ability to suppress clutter interference and the target detection and tracking performance, thus ensuring the accuracy and effectiveness of small UAV target detection and tracking. Meanwhile, the present invention processes signals through RFT, conducts joint search along the distance and velocity directions of the UAV target, and realizes continuous long-time coherent integration through a Doppler filter bank to compensate for the phase fluctuations between different sampling pulses and compensate for the range cell migration, achieving single-frame coherent integration of target energy. Furthermore, without changing the target radar hardware system, it significantly improves the performance of small UAV target detection and tracking. In addition, the present invention performs range-FFT processing on the signals before RFT processing, and performs two-dimensional CFAR processing and angle-FFT processing on the signals after RFT processing, enabling simple and intuitive extraction of signal frequency-related information. Furthermore, it can ensure the signal-to-noise ratio of the overall signal, making the anti-interference ability stronger and ensuring the effect of detecting long-distance moving targets. Two-dimensional CFAR processing makes a decision for each detection unit in the range-Doppler matrix composed of range-Doppler dimension data, thereby realizing target detection, and can maximize the target detection probability under the constraint of a constant false alarm probability. The angle-FFT processing extracts the multi-frame target results containing the distance information, velocity information, and angle information of the UAV target, which can effectively obtain the distance, velocity, and angle of the small UAV target, thus ensuring the accuracy and effectiveness of small UAV target detection and tracking. Finally, based on the particle filter algorithm, continuous tracking of the UAV target can be effectively achieved. The particle filter has advantages in dealing with noise that cannot be compared by other filters, that is, for any linear or non-linear system model, Gaussian or non-Gaussian noise model, the particle filter can be effectively applied and processed. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] To make the objectives, technical solutions, and advantages of the invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings, where:

[0041] Figure 1 is the logic block diagram of the small UAV target tracking method;

[0042] Figure 2 is the schematic diagram of the small UAV target tracking method;

[0043] Figure 3 is the schematic diagram of performing RFT processing;

[0044] Figure 4 is the flowchart of the particle filter algorithm. Detailed implementation manners

[0045] The following is a further detailed description through specific implementation manners:

[0046] Example:

[0047] In this example, a method for tracking small unmanned aerial vehicle (UAV) targets based on a MIMO radar is disclosed

[0048] As Figure 1 and Figure 2 shown, the method for tracking small UAV targets based on a MIMO radar includes:

[0049] S1: Obtain the beat signals of each antenna array of the target radar for detecting the corresponding UAV target;

[0050] S2: Perform range-FFT (Fast Fourier Transform in the range dimension) processing on the beat signals to convert the time-domain signals in the range dimension of the beat signals to the frequency domain;

[0051] S3: Perform RFT (Radon-Fourier transform) processing on the beat signals after range-FFT processing to correct the range walk caused by the movement of distant targets in the beat signals and achieve single-frame coherent accumulation of target energy;

[0052] S4: Perform two-dimensional CFAR (Constant False-Alarm Rate) processing on the beat signals after RFT processing to achieve target detection of UAV targets and obtain the corresponding target detection results;

[0053] S5: Perform angle-FFT processing on the target detection results to extract multi-frame target results containing the range information, speed information, and angle information of the UAV target;

[0054] S6: Based on the multi-frame target results, combine the particle filter algorithm to achieve continuous tracking of the corresponding UAV target.

[0055] In the present invention, for the method of small UAV target tracking based on MIMO radar, the decorrelation of the target echo signal by the MIMO radar is utilized, so that the average received energy of the echo tends to be constant, the RCS of the target can be smoothed and the RCS fluctuation of the target can be improved, thereby enhancing the clutter interference suppression ability and the target detection and tracking performance, and ensuring the accuracy and effectiveness of small UAV target detection and tracking. At the same time, in the present invention, the signal is processed by RFT, jointly searched along the distance and speed directions of the UAV target, and the phase fluctuation between different sampling pulses is compensated by the Doppler filter bank to compensate for the cross-range cell migration, realizing the coherent integration of target energy in a single frame, and thereby significantly improving the performance of small UAV target detection and tracking without changing the target radar hardware system. In addition, the present invention performs range-FFT processing on the signal before RFT processing, and performs two-dimensional CFAR processing and angle-FFT processing on the signal after RFT processing, enabling the simple and intuitive extraction of signal frequency-related information, thereby ensuring the signal-to-noise ratio of the overall signal, making the anti-interference ability stronger, and ensuring the effect of long-distance moving target detection. The two-dimensional CFAR processing makes a decision on each detection unit in the range-Doppler matrix composed of range-Doppler dimension data, thereby realizing target detection, and can maximize the target detection probability under the constraint of a constant false alarm probability. The angle-FFT processing extracts the multi-frame target results containing the distance information, speed information and angle information of the UAV target, can effectively obtain the distance, speed and angle of the small UAV target, and thereby ensure the accuracy and effectiveness of small UAV target detection and tracking. Finally, the continuous tracking of the UAV target can be effectively realized based on the particle filter algorithm. The particle filter has advantages that other filters cannot match in dealing with noise, that is, for any linear or non-linear system model, Gaussian or non-Gaussian noise model, the particle filter can be effectively applied and processed.

[0056] In the specific implementation process, in step S1, the target radar is a MIMO (Multiple-Input-Multiple-Output) radar system composed of two transmitting antennas with a spacing of 2λ and four receiving antennas with a spacing of λ / 2; moreover, the transmitting angle and the receiving angle of the target radar are equal, enabling the formation of an eight-element virtual antenna array with a spacing of λ / 2, where λ represents the wavelength of the target radar.

[0057] The MIMO radar configuration mode is a staggered mode, by selecting and to obtain the virtual array:

[0058]

[0059] Where: x T,m represents the position of the m-th transmitting antenna; x R,n represents the position of the n-th receiving antenna; λ represents the wavelength of the target radar; M represents the number of transmitting antennas; N represents the number of receiving antennas.

[0060] The virtual array is a uniform array with an interval of λ / 2, and the interval λ / 2 is selected to avoid spatial aliasing.

[0061] In the present invention, two transmitting antennas of the MIMO radar transmit mutually orthogonal frequency-modulated continuous waves, and four receiving antennas use waveform orthogonality to extract the signals of each transmitting antenna to synthesize an eight-aperture virtual antenna array, which can improve the angular resolution of detecting small unmanned aerial vehicle targets, significantly improve the weak target detection performance, and thus can enhance the tracking effect of small unmanned aerial vehicle targets.

[0062] In the specific implementation process, the beat signal is expressed by the following formula:

[0063]

[0064] Where: S b represents the beat signal; c represents the speed of light; N represents the number of chirps; T c represents the chirp period; k represents the frequency modulation slope; v represents the speed of the unmanned aerial vehicle target relative to the target radar; R represents the distance between the unmanned aerial vehicle target and the target radar; f 0 represents the carrier frequency; n represents the number of receiving antennas; d represents the antenna array spacing; θ represents the azimuth angle of the unmanned aerial vehicle target relative to the target radar.

[0065] The MIMO radar transmitter sends a linear frequency-modulated signal chirp in a periodic manner, and the transmitted signal is expressed as:

[0066]

[0067] The target echo received by the radar receiver is the time delay of the transmitted signal, and the received signal is expressed as:

[0068]

[0069] Where: represents the round-trip time of the N+1-th cycle.

[0070] The transmitted signal and the received signal are output by a mixer to obtain the corresponding beat signal.

[0071] In the present invention, the beat signal of the MIMO radar antenna array contains the distance, speed and angle information of the target. Through subsequent algorithms, the distance, speed and angle of the small unmanned aerial vehicle target can be obtained, thereby ensuring the accuracy and effect of detecting and tracking the small unmanned aerial vehicle target.

[0072] In the specific implementation process, the formula for RFT processing is as follows:

[0073]

[0074] In the formula: ε represents a known constant relative to f(t, r + vt);

[0075] The linear equation of range walk caused by the target movement of the UAV is as follows:

[0076] r s = r + vt, t ∈ [-T / 2, T / 2];

[0077] In the formula: r represents the slant range; v represents the radial velocity at t = 0; T represents the coherent integration time.

[0078] Combined with Figure 3 As shown, during RFT processing, in the complex signal domain, the amplitude and phase information are jointly utilized to achieve continuous coherent integration, and a pair of Doppler compensation functions H v (t) and H θ (t) are introduced as Fourier integral components to cancel the phase fluctuations between different pulses, generate the final coherent peak, and correct the range walk caused by the UAV target movement.

[0079] Specifically, a pair of Doppler compensation functions H v (t) and H θ (t) are respectively expressed as:

[0080]

[0081] In the formula: v represents the radial velocity at t = 0; λ represents the wavelength; θ represents the polar angle, representing the counterclockwise angle from the range walk line on the t - r s plane to the t - axis; c represents the speed of light; j represents the imaginary part of a complex number; g represents the wave path difference introduced by the array spacing.

[0082] In the present invention, by means of RFT processing the signal, joint search is carried out along the distance and velocity directions of the UAV target, and continuous long - time coherent integration is achieved through a Doppler filter bank to compensate for the phase fluctuations between different sampling pulses, compensate for the cross - range cell walk, realize the single - frame coherent accumulation of target energy, and thus, without changing the target radar hardware system, significantly improve the performance of detecting and tracking small UAV targets.

[0083] In the specific implementation process, two-dimensional CFAR processing includes a reference cell, a guard cell, and a detection cell. The reference cell is used to estimate the noise power; the guard cell is used to make the average interference power estimate more accurate. Guard cells are set around the cell to be detected, and these guard cells are not substituted into the operation when estimating the average interference power; the detection cell is used to judge the range-Doppler dimensional data of the target.

[0084] Specifically, the two-dimensional CFAR threshold factor T during two-dimensional CFAR processing is expressed as:

[0085]

[0086] In the formula: P fa represents the preset false alarm probability; 2N refer represents the length of the reference cell.

[0087] For two-dimensional CFAR detection, it is first necessary to accurately estimate the system noise level and determine the reference window. In this case, the estimation of the unknown noise level depends on the arithmetic mean of all random variables within the two-dimensional reference window:

[0088]

[0089] In the formula: X m,n represents the random variable within the two-dimensional reference window near the detection cell; m represents the index in the range dimension, and n represents the index in the Doppler dimension.

[0090] If all the X m,n samples within the two-dimensional reference window are independent and identically distributed, then the estimation of the obtained arithmetic mean will be the best result; however, if the target to be detected occupies the reference cells within the two-dimensional reference window, the estimation result will be invalid; therefore, it is necessary to design a two-dimensional guard cell near the detection cell, and these guard cells will be excluded when estimating the arithmetic mean.

[0091] Specifically, when the product of the two-dimensional CFAR threshold factor T and the output Z is less than the value of the detection cell, a target peak is detected; when the product of the two-dimensional CFAR threshold factor T and the output Z is greater than the value of the detection cell, there is no target peak.

[0092] In the specific implementation process, the flowchart of the particle filter algorithm is as Figure 4 shown.

[0093] The particle weights in the particle filter algorithm are expressed as:

[0094]

[0095] In the formula: i represents the i-th particle; k represents the k-th moment; R represents the measurement noise covariance matrix; Denote the prediction of the observation value of the $i$-th particle at time $k$; $Z$ g $(k)$ represents the observation value at time $k$.

[0096] Weight calculation is the core of the particle filter algorithm. According to the weight size, a large number of "high-quality" particles can be replicated, and an elimination system can be implemented for "low-quality" particles. In addition, after weight calculation, it is also the basis for re-guiding the spatial distribution of particles, and the weight ultimately affects the filtering result.

[0097] Steps for calculating weights:

[0098] First, substitute each particle representing the state $X$ i $(k)$ at time $k - 1$ into the state equation $X(k)=f(X(k - 1),W(k))$ to obtain the one-step prediction value where $i = 1,2,\cdots,N$;

[0099] Due to the factors of the particle set, the obtained is also a set. Substitute each value in this set into the observation equation $Z(k)=h(X(k),V(k))$ to calculate the prediction of the observation value

[0100] At the current time, that is, time $k$, the measurement system can uniquely collect an observation value $Z$ g $(k)$. Then the absolute value of the deviation between the observation prediction of the particle set at time $k$ and the measured value:

[0101]

[0102] Then, according to the standard form of the Gaussian distribution the weight calculation formula can be obtained.

[0103] In the present invention, the particle filter algorithm is an approximate Bayesian filtering algorithm based on Monte Carlo simulation. Its core idea is to approximate the probability density function of the system random variable with some discrete random sampling points, and use the sample mean to replace the integral operation to obtain the minimum variance estimate of the state, so as to effectively realize the continuous tracking of the UAV target.

[0104] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not restrictive. Although the present invention has been described by referring to the preferred embodiments of the present invention, those of ordinary skill in the art should understand that various changes can be made in form and details without departing from the spirit and scope of the present invention defined by the appended claims. At the same time, common knowledge such as the specific structure and characteristics in the embodiments is not described in detail here. Finally, the scope of protection required by the present invention should be based on the content of its claims, and the specific implementation manners described in the specification can be used to interpret the content of the claims.

Claims

1. A method for tracking small unmanned aerial vehicle (UAV) targets based on MIMO radar, characterized in that, it includes the following steps: S1: Obtain the beat signals of each antenna array of the target radar for detecting the corresponding UAV target; In step S1, the target radar is a MIMO radar system composed of two transmitting antennas spaced 2λ apart and four receiving antennas spaced λ / 2 apart; and, the transmitting angle and receiving angle of the target radar are equal, so that eight virtual antenna arrays spaced λ / 2 apart can be formed, where λ represents the wavelength of the target radar; The beat signal is represented by the following formula: Where: S b represents the beat signal; c represents the speed of light; N represents the number of chirps; T c represents the chirp period; k represents the frequency modulation slope; v represents the speed of the UAV target relative to the target radar; R represents the distance between the UAV target and the target radar; f 0 represents the carrier frequency; n represents the number of receiving antennas; d represents the antenna array spacing; θ represents the azimuth angle of the UAV target relative to the target radar; S2: Perform range-FFT processing on the beat signal to convert the time-domain signal in the distance dimension of the beat signal to the frequency domain; S3: Perform RFT processing on the beat signal after range-FFT processing to correct the range walk caused by the movement of the long-distance target in the beat signal and achieve single-frame coherent accumulation of target energy; During RFT processing, in the complex signal domain, the amplitude and phase information are jointly utilized to achieve continuous coherent integration, and a pair of Doppler compensation functions H v (t) and H θ (t) are used as Fourier integral components to cancel the phase fluctuations between different pulses, generate the final coherent peak, and correct the range migration caused by the movement of the UAV target; A pair of Doppler compensation functions H v (t) and H θ (t) are respectively expressed as: Where: v represents the radial velocity at t = 0; λ represents the wavelength; θ represents the polar angle, which represents the counterclockwise angle from the range migration line to the t-axis on the t-r s plane; c represents the speed of light; j represents the imaginary part of a complex number; g represents the wave path difference introduced by the array interval; S4: Perform two-dimensional CFAR processing on the beat signal after RFT processing to achieve target detection of the UAV target and obtain the corresponding target detection result; S5: Perform angle-FFT processing on the target detection result to extract multi-frame target results containing the distance information, speed information, and angle information of the UAV target; S6: Based on the multi-frame target results, combine the particle filter algorithm to achieve continuous tracking of the corresponding UAV target.

2. The method for tracking small UAV targets based on MIMO radar according to claim 1, characterized in that: In step S3, the formula for RFT processing is as follows: In the formula: ε represents a known constant relative to f(t, r + vt); The linear equation of the range walk caused by the movement of the UAV target is as follows: r s = r + vt, where t ∈ [-T / 2, T / 2]; In the formula: r represents the slant range; v represents the radial velocity at t = 0; T represents the coherent integration time.

3. The method for tracking small UAV targets based on MIMO radar according to claim 1, characterized in that: In step S4, the two-dimensional CFAR processing includes a reference unit, a protection unit, and a detection unit; the reference unit is used to estimate the noise power, the protection unit is used to make the average interference power estimation value more accurate, and the detection unit is used to judge the range-Doppler dimension data of the target.

4. The method for tracking small UAV targets based on MIMO radar according to claim 3, characterized in that, The two-dimensional CFAR threshold factor T during two-dimensional CFAR processing is expressed as: Where: P fa represents the preset false alarm probability; 2N refer represents the length of the reference unit.

5. The method for tracking small UAV targets based on MIMO radar according to claim 4, characterized in that: When the product of the two-dimensional CFAR threshold factor T and the output Z is less than the value of the detection unit, a target peak is detected; When the product of the two-dimensional CFAR threshold factor T and the output Z is greater than the value of the detection unit, there is no target peak.

6. The method for tracking small UAV targets based on MIMO radar according to claim 1, characterized in that, In step S6, the particle weight in the particle filter algorithm is expressed as: Where: i represents the i-th particle; k represents the k-th moment; R represents the measurement noise covariance matrix; represents the prediction of the observation value of the i-th particle at the k-th moment; Z g (k) represents the observation value at the k-th moment.

Images