A low, slow, small multi-target detection method and system based on passive radar
Through segmented correlation accumulation algorithm and CFAR detection, combined with ROOT-MUSIC algorithm, the problem of low signal-to-noise ratio and high computational complexity in 'low slow and low' object detection is solved, and accurate detection and high resolution of weak targets are achieved .
Patent Information
- Application Number
- CN202210651990.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-06-09
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2042-06-09
AI Technical Summary
In the target detection, the signal-to-noise ratio is low due to the weak echo signal of the ‘low slow and small’ target, which requires long-term accumulation and high computational complexity, making it difficult to achieve real-time detection.
The segment correlation accumulation algorithm is used to obtain the relative delay spectrum through the segment correlation of the echo signal and the reference signal, and perform incoherent accumulation and CFAR detection. Combined with the ROOT-MUSIC algorithm to estimate the Doppler frequency, reduce the calculation amount and increase the signal-to-noise Compare.
It effectively reduces the memory space required for long signal processing, improves the signal-to-noise ratio of target detection, reduces the calculation amount, and improves the speed resolution, achieving accurate detection of weak targets.
Smart Images

Figure CN115061109B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of passive radar signal processing, and mainly to a target detection method and system proposed for estimating "low, slow and small" targets. Background Art
[0002] In recent years, with the development of science and technology, drones have been widely used in the fields of photography, urban planning, emergency rescue and military reconnaissance. However, the rapid development of drones has not only brought economic and military convenience to cities, but also brought huge challenges to urban low-altitude safety. Among them, "low, slow and small" drones have the characteristics of small size, high concealment and strong operability. Therefore, the technical research on the detection technology of "low, slow and small" targets has become a hot topic in current research.
[0003] As the main method in the current target detection field, radar has made great progress and its technology is very mature. According to whether the detection signal is actively transmitted, radar can be divided into active radar (Active Radar) and passive radar (Passive Radar). Active radar refers to the radar actively transmitting signals and detecting targets by using the coherence between the target echo and the transmitted signal, while passive radar uses commercially available signals for detection. Available platforms include radio stations (FM), global positioning systems (GPS) and communication base stations (BS). Passive radar systems have many advantages over active radar systems. There is no need to consider full-duplex and self-interference cancellation, nor is there a need to prepare transmitters, which greatly reduces costs. For passive bistatic radars, the reference signal emitted by the external radiation source is usually unknown, so an additional channel needs to be configured to receive the reference signal (direct wave signal), which is generally called the reference channel. The directional high-gain antenna is configured to receive the direct path signal emitted by the transmitter in the reference channel. It is assumed that a pure reference signal with a high signal-to-noise ratio and no multipath clutter can be obtained. An omnidirectional monitoring antenna is set in the monitoring channel to receive the echo signal from the target.
[0004] Dual-base passive radar layout Figure 1 As shown, the signal source is located at position g0 = [x0, y0, z0] T , the position of the directional high-gain antenna in the reference channel is g d =[x d ,y d ,z d ] T , the position of the monitoring antenna in the monitoring channel is g s =[x s ,y s ,z s ] T The position coordinates of the kth target are g k =[xk ,y k ,z k ] T , the velocity coordinate is β is the bistatic angle.
[0005] Considering that the speed of a small target is slow, it can be regarded as stationary during a long-term accumulation period. Assuming that the baseband signal emitted by the signal source is s(t), the baseband reference signal received by channel 1 (reference channel) can be expressed as:
[0006]
[0007] Where b0 is the complex amplitude of the reference signal, is the direct path delay, c is the signal propagation speed, φ1 is the phase offset caused by the carrier, which obeys the uniform distribution on [0,2π], z1(t) has a mean of 0, and the equation is σ 2 Complex Gaussian white noise, according to the Friis equation, b0 can be expressed as:
[0008]
[0009] Where EIRP is the equivalent isotropic radiated power of the transmitter, G r,direct is the antenna gain of the direct path, λ is the signal wavelength
[0010] The baseband echo signal received by channel 2 (monitoring channel) can be expressed as:
[0011]
[0012] where b k is the complex amplitude of the echo signal scattered by the kth target. According to the bistatic radar equation, b k It is expressed as:
[0013]
[0014] EIRP is the equivalent isotropic radiated power of the signal source, G r,echo is the antenna gain of the echo path, σ RCS,k (β) is the bistatic radar cross section of the target, R 1k =||g0-g k ||2 and R 2k =||g s -g k ||2 are the radial distances of the target relative to the signal source and the radar receiver, φ2 is the phase offset caused by the carrier, which obeys the uniform distribution on [0,2π], τ k and f k are the time delay and Doppler information of the kth target, respectively, which are expressed by the following formula:
[0015]
[0016]
[0017] In order to ensure the target detection effect, the receiving end needs to ensure a certain signal-to-noise ratio. The signal-to-noise ratio of the k-th target echo received by the receiver is expressed as follows:
[0018]
[0019] where k B is the Boltzmann constant, T is the thermodynamic temperature of the receiver, B is the signal bandwidth, and NF is the noise factor.
[0020] Range-Doppler two-dimensional coherent processing (Range-Doppler processing) is a commonly used method for calculating the range-Doppler response of radar signals. The reference signal within the coherent accumulation time T0 is subjected to a series of time delays and Doppler frequency shifts and correlated with the target echo signal. The expression is as follows:
[0021]
[0022] The target distance resolution and speed resolution are expressed by the following equations:
[0023]
[0024]
[0025] Since the energy of the scattered echo signal of the target with a small RCS is weak, the accumulation time required by the Range-Doppler processing algorithm is very long. Therefore, the signal processing during the entire accumulation time requires a lot of calculations, which is not conducive to real-time target detection. In order to calculate more efficiently, the reference signal and the echo can be segmented. When the segment length is less than a certain value, the phase accumulation generated by the Doppler frequency can be considered to exist only between segments and not within the segment. FFT is used on each segment to implement matched filtering to obtain related signals at different delays, and then FFT between segments at a specific delay is used to obtain Doppler information. However, this algorithm requires channel estimation in advance and still has a certain computational complexity. Summary of the invention
[0026] In view of the deficiencies of the prior art, the present invention proposes a segmented correlation accumulation target detection method.
[0027] The present invention proposes a similar segmented correlation accumulation algorithm to achieve more effective long-term integration. The algorithm does not require channel estimation and can use non-cooperative OFDM signals to estimate the distance and Doppler velocity information of the target.
[0028] The present invention also provides a low, slow, small multi-target detection system based on passive radar.
[0029] Terminology explanation:
[0030] 1. Range-Doppler processing refers to range Doppler two-dimensional coherent processing;
[0031] 2. IFFT refers to inverse fast Fourier transform;
[0032] 3. FFT refers to fast Fourier transform;
[0033] 4. DFT refers to discrete Fourier transform
[0034] 5. CFAR refers to the constant false alarm detection algorithm;
[0035] 6. ROOT-MUSIC algorithm refers to the root-finding MUSIC algorithm.
[0036] The technical solution of the present invention is:
[0037] A segmented correlation accumulation target detection method, comprising:
[0038] Firstly, the relative delay spectrum of the echo signal is obtained by the segmented correlation between the echo signal and the reference signal;
[0039] Then, the correlation values of each segment are incoherently accumulated to obtain the relative delay spectrum of the target echo signal after noise reduction;
[0040] Thirdly, the CFAR detection program is used to detect the target in the range domain, and the target delay information is used to obtain the segmented correlation signal sampling at the relative delay.
[0041] Finally, the ROOT-MUSIC algorithm is used to estimate the target Doppler with high resolution frequency.
[0042] Preferably, according to the present invention, obtaining the relative delay spectrum of the echo signal by segmented correlation between the echo signal and the reference signal includes:
[0043] (1) The digital baseband reference signal sample data received by the reference channel, i.e., the reference signal s ref Divided into I segments, each segment has a data length of L1, and the nth sample of the reference signal in the i-th segment As shown in formula (I):
[0044]
[0045] In formula (I), is a rectangular window with a length of L1;
[0046] (2) The digital baseband echo signal sample data received by the monitoring channel, i.e., the echo signal s ech The overlapping segments are divided into I segments, and the length of each overlapping segment is L2. The nth sample of the i-th segment echo signal As shown in formula (II):
[0047]
[0048] In formula (II), is a rectangular window with a length of L2;
[0049] (3) The time delay of the target relative to the direct path is obtained by the delay unit r corresponding to the correlation peak between the reference signal and the echo signal. The correlation process between the echo signal and the reference signal in the time domain is shown in formula (III):
[0050]
[0051] In formula (III), ψ i (r) refers to the correlation peak at delay unit r, are the sample values of the i-th echo signal and the reference signal at the delay unit r respectively.
[0052] Further preferably, in formula (III), discrete Fourier transform and inverse discrete Fourier transform are used to simplify the correlation process, and fast Fourier transform is used to accelerate the calculation, as shown in formula (IV):
[0053]
[0054] In formula (IV), for Execute N FFT The result of the point-length discrete Fourier transform, for Execute N FFT The result of taking the conjugate after the point length discrete Fourier transform, for The sample of the anti-corrugation signal, N FFT Should satisfy N FFT ≥L2.
[0055] Preferably, according to the present invention, performing incoherent accumulation on each segment correlation value to obtain the relative delay spectrum of the target echo signal after noise reduction means: performing incoherent accumulation on the segment correlation results of multiple segments in step (3) to obtain the relative delay distribution of the target, that is, the relative delay spectrum of the target echo signal after noise reduction. The process is shown in formula (V):
[0056]
[0057] In formula (V), Ψ(r) is the correlation peak value of the target relative delay distribution corresponding to delay unit r.
[0058] Preferably, according to the present invention, a CFAR detection program is used to detect a target in a range domain, and target delay information is used to obtain segmented correlation signal sampling at a relative delay; comprising:
[0059] (4) By setting the false alarm probability p f The CFAR algorithm is used to detect the target from the relative delay spectrum RDP of the target echo signal after noise reduction; the false alarm probability refers to the probability that the CFAR determines the existence of the target when there is no target signal;
[0060] When the relative delay spectrum Ψ(r) of the target echo signal after noise reduction is greater than the threshold T0, it is regarded as a target or a group of targets that cannot be identified by delay alone and need to be further distinguished by Doppler information. When the value of the relative delay spectrum RDP of the target echo signal after noise reduction at the target delay unit is not greater than the threshold, it is regarded as no target signal is detected, that is, the target is missed. The threshold T0 is calculated based on the false alarm probability and the background noise distribution. For noise that obeys exponential distribution, T0=μ0ln(p f ), μ0 is the estimated value of background noise power;
[0061] The kth target signal delay is calculated by formula (VI):
[0062]
[0063] In formula (VI), is the kth target signal delay estimate, It is the target delay unit estimate;
[0064] (5) Estimate the target delay unit in step (4) Assign the value to r in formula (III) to get the target delay unit estimate The correlation peak at the relative time delay is the sampling value of the segment correlation signal The sampling values of the segment-related signal at the relative time delay are collected into a vector, denoted as Indicates that the vector dimension is Ix1, S k As represented by formula (VII):
[0065]
[0066] In formula (VII), S k (i) refers to S k The i-th vector value in .
[0067] Further preferably, p f =10 -4 .
[0068] According to the preferred embodiment of the present invention, using the ROOT-MUSIC algorithm to estimate the target Doppler with high resolution frequency means: S obtained in step (5) k Perform ROOT-MUSIC high-resolution frequency estimation to obtain the Doppler frequency estimate of the target signal, including:
[0069] A. Take vector S k The N values in are taken as the observed sample values S k (0),S k (1),…,S k (N-1), estimate the M-order autocorrelation matrix
[0070] B. Yes Perform eigendecomposition to obtain the eigenvector u corresponding to the MK smallest eigenvalues K+1 ,u K+2 ,...u K+M , construct the matrix G = [u K+1 u K+2 ...u K+M ], define the vector a(z)=[1 z -1 …z -(M-1) ],z=e jω , ω is the complex signal frequency;
[0071] C. Solve equation a T (z (-1) )GG H a(z)=0, find the K roots closest to the unit circle. The phases of these K roots are the Doppler frequency estimates of the target signal.
[0072] A segmented correlation accumulation target detection system, comprising:
[0073] The relative delay spectrum obtaining module of the echo signal is configured to obtain the relative delay spectrum of the echo signal through the segmented correlation between the echo signal and the reference signal;
[0074] The module for obtaining the relative time delay spectrum of the target echo signal after noise reduction is configured to perform incoherent accumulation of the correlation values of each segment to obtain the relative time delay spectrum of the target echo signal after noise reduction;
[0075] The segment-correlated signal sampling module at the relative time delay is configured to detect the target in the range domain using the CFAR detection procedure and obtain the segment-correlated signal sampling at the relative time delay using the target time delay information;
[0076] The target Doppler estimation module is configured to estimate the target Doppler at a high resolution frequency using a ROOT-MUSIC algorithm.
[0077] The beneficial effects of the present invention are:
[0078] 1. The present invention processes the reference signal and the long-term accumulated echo signal in segments, effectively reducing the memory space required for long signal processing;
[0079] 2. Use segmented correlation and accumulation to improve the signal-to-noise ratio when searching for targets, overcoming the performance degradation caused by the frequency shift caused by target movement in the long correlation operation between the echo signal and the reference signal;
[0080] 3. The present invention does not need to set the frequency point during the target detection process, and only processes the data at the relative time delay of the target, which effectively reduces the amount of calculation;
[0081] 4. The present invention uses the ROOT-MUSIC high-resolution algorithm to estimate the target Doppler frequency, effectively improving the velocity resolution of the target. BRIEF DESCRIPTION OF THE DRAWINGS
[0082] Figure 1 It is a schematic diagram of the system structure based on the dual-base passive radar of the present invention;
[0083] Figure 2 It is a flow chart of the target detection method of segmented correlation accumulation of the present invention;
[0084] Figure 3 Schematic diagram of CFAR target relative delay detection results in the embodiment. DETAILED DESCRIPTION
[0085] The present invention will be further defined below in conjunction with the accompanying drawings and embodiments, but is not limited thereto.
[0086] Example 1
[0087] The communication base station transmits OFDM signals as an external radiation source, and the radar receiver is set up with dual channels. Channel 1 is set up with a directional high-gain antenna to receive the reference signal, and channel 2 is set up with an omnidirectional antenna to receive the target echo signal. The OFDM signal parameters and radar system parameters used are shown in Table 1, where f c is the carrier frequency, △f is the subcarrier spacing, N sub is the number of subcarrier spacing, N sym is the number of OFDM symbols that need to be processed in one long-term accumulation.
[0088] Table 1
[0089]
[0090] Multiple targets are randomly distributed in the 120-degree sector of the base station. Three targets are randomly generated within the range of 100-1000m from the base station. The location and speed information of the targets are shown in Table 2:
[0091] Table 2
[0092]
[0093]
[0094] A segmented correlation accumulation target detection method, such as Figure 2 As shown, including:
[0095] First, the relative delay spectrum of the echo signal is obtained by segmented correlation between the echo signal and the reference signal; including:
[0096] (1) The digital baseband reference signal sample data received by the reference channel, i.e., the reference signal s ref Divide into I segments, take I = 640, the length of each segment is L1, L1 = 5824, the nth sample of the i-th reference signal As shown in formula (I):
[0097]
[0098] In formula (I), is a rectangular window with a length of L1;
[0099] (2) The digital baseband echo signal sample data received by the monitoring channel, i.e., the echo signal s ech The overlap is divided into I segments, and the length of each overlap segment is L2, L2 = 16384. The overlap length L2-L1 determines the maximum estimable delay in the sample. Therefore, the number of echo signal samples in the monitoring channel should be larger than the number of reference signal samples (τ max -τ0)*F s , where F s is the sampling rate. The nth sample of the i-th echo signal As shown in formula (II):
[0100]
[0101] In formula (II), is a rectangular window with a length of L2; when the phase accumulation in the echo signal segment is much smaller than 2π, the Doppler frequency in the echo signal segment can be ignored.
[0102] (3) The time delay of the target relative to the direct path is obtained by the delay unit r corresponding to the correlation peak between the reference signal and the echo signal. The correlation process between the echo signal and the reference signal in the time domain is shown in formula (III):
[0103]
[0104] In formula (III), ψ i (r) refers to the correlation peak at delay unit r, are the sample values of the i-th echo signal and the reference signal at the delay unit r, respectively. In formula (III), discrete Fourier transform (DFT) and inverse discrete Fourier transform (IDFT) are used to simplify the correlation process, and fast Fourier transform (FFT) is used to accelerate the calculation, as shown in formula (IV):
[0105]
[0106] In formula (IV), for Execute N FFT The result of the point-length discrete Fourier transform (DFT), for Execute N FFT The result of taking the conjugate after the point-length discrete Fourier transform (DFT) is for The sample of the anti-corrugation signal, N FFT Should satisfy N FFT ≥L2. The number of FFT and IFFT points are both N FFT =16384.
[0107] Then, the 640-segment correlation values are incoherently accumulated to obtain the relative delay spectrum of the target echo signal after noise reduction; that is, the segmented correlation results of multiple segments in step (3) are incoherently accumulated to obtain the relative delay distribution (RDP) of the target, that is, the relative delay spectrum of the target echo signal after noise reduction. The incoherent accumulation process can improve the signal-to-noise ratio of target detection. The process is shown in formula (V):
[0108]
[0109] In formula (V), Ψ(r) is the correlation peak value of the target relative delay distribution corresponding to delay unit r.
[0110] Thirdly, the CFAR detection program is used to detect the target in the range domain, and the target delay information is used to obtain the segmented correlation signal sampling at the relative delay; including:
[0111] (4) By setting the false alarm probability p f The CFAR algorithm is used to detect the target from the relative delay spectrum RDP of the target echo signal after noise reduction; the false alarm probability refers to the probability that the CFAR determines the existence of the target when there is no target signal;
[0112] When the relative delay spectrum Ψ(r) of the target echo signal after noise reduction is greater than the threshold T0, it is regarded as a target or a group of targets that cannot be identified by delay alone and need to be further distinguished by Doppler information. When the value of the relative delay spectrum RDP of the target echo signal after noise reduction at the target delay unit is not greater than the threshold, it is regarded as no target signal is detected, that is, the target is missed. The threshold T0 is calculated based on the false alarm probability and the background noise distribution. For noise that obeys exponential distribution, T0=μ0 ln(p f ), μ0 is the estimated value of background noise power;
[0113] The kth target signal delay is calculated by formula (VI):
[0114]
[0115] In formula (VI), is the kth target signal delay estimate, It is the target delay unit estimate;
[0116] (5) Estimate the target delay unit in step (4) Assign the value to r in formula (III) to get the target delay unit estimate The correlation peak at the relative time delay is the sampling value of the segment correlation signal The sampling values of the segment-related signal at the relative time delay are collected into a vector, denoted as Indicates that the vector dimension is Ix1, S k As represented by formula (VII):
[0117]
[0118] In formula (VII), S k (i) refers to S k The i-th vector value in .
[0119] From S k The Doppler information of the estimated target.
[0120] p f =10 -4 .
[0121] Finally, the ROOT-MUSIC algorithm is used to estimate the target Doppler with high-resolution frequency. This algorithm does not need to set the frequency point to realize the Doppler estimation process, and only processes the data at the relative time delay of the target, which significantly reduces the amount of calculation. It means: S obtained in step (5) k Perform ROOT-MUSIC high-resolution frequency estimation to obtain the Doppler frequency estimate of the target signal. Usually, there is only one target at a specific delay unit, so it is relatively easy to estimate S kEmbedded Doppler information in; specifically including:
[0122] A. Take vector S k The N values in are taken as the observed sample values S k (0),S k (1),…,S k (N-1), estimate the M-order autocorrelation matrix
[0123] Estimate the M-order autocorrelation matrix The process is prior art and will not be described in detail here.
[0124] B. Yes Perform eigendecomposition to obtain the eigenvector u corresponding to the MK smallest eigenvalues K+1 ,u K+2 ,...u K+M , construct the matrix G = [u K+1 u K+2 ...u K+M ], define the vector a(z)=[1 z -1 ...z -(M-1) ],z=e jω , ω is the complex signal frequency;
[0125] C. Solve equation a T (z (-1) )GG H a(z)=0, find the K roots closest to the unit circle. The phases of these K roots are the Doppler frequency estimates of the target signal.
[0126] The multi-target detection results are as follows Figure 3 As shown; According to the detection results, the relative delay units of the five targets are shown in Table 3; According to formula (V), the estimated value of the target distance R1+R2 can be calculated, and the distance estimation error can be obtained by comparing the actual value of the target distance. It can be seen that the algorithm proposed in the present invention can also have a good detection effect for weak target signals with SNR=--40.8532dB, and the distance estimation error is within 1m.
[0127] Substitute the relative distance unit estimates into ψ i (r) in which the set of sampling signal vectors of the segment correlation signal at the target relative time delay is obtained = {S k |k=1,…K}.
[0128] By performing ROOT-MUSIC high-resolution frequency estimation on the signal vectors in set S, the target Doppler velocity can be calculated. By comparing the true value of the target Doppler velocity, the Doppler velocity estimation error can be obtained. The estimation results are shown in Table 3. According to the estimation results, it can be seen that when the SNR is high, the result of estimating the target Doppler velocity using the ROOT-MUSIC algorithm is more accurate, and the estimation error is within 2m / s.
[0129] Table 3
[0130]
[0131] Example 2
[0132] A segmented correlation accumulation target detection system, comprising:
[0133] The relative delay spectrum obtaining module of the echo signal is configured to obtain the relative delay spectrum of the echo signal through the segmented correlation between the echo signal and the reference signal;
[0134] The module for obtaining the relative time delay spectrum of the target echo signal after noise reduction is configured to perform incoherent accumulation of the correlation values of each segment to obtain the relative time delay spectrum of the target echo signal after noise reduction;
[0135] The segment-correlated signal sampling module at the relative time delay is configured to detect the target in the range domain using the CFAR detection procedure and obtain the segment-correlated signal sampling at the relative time delay using the target time delay information;
[0136] The target Doppler estimation module is configured to estimate the target Doppler at a high resolution frequency using a ROOT-MUSIC algorithm.
Claims
1. A segmented correlation accumulation target detection method, It is characterized in that Used to realize low, slow, small and multiple target detection based on passive radar, including: Firstly, the relative delay spectrum of the echo signal is obtained by the segmented correlation between the echo signal and the reference signal; Then, the correlation values of each segment are incoherently accumulated to obtain the relative delay spectrum of the target echo signal after noise reduction; Thirdly, the CFAR detection program is used to detect the target in the range domain, and the target delay information is used to obtain the segmented correlation signal sampling at the relative delay. Finally, the ROOT-MUSIC algorithm is used to estimate the target Doppler at high resolution frequency; Using the ROOT-MUSIC algorithm to estimate the target Doppler at high resolution frequency means: k Perform ROOT-MUSIC high-resolution frequency estimation to obtain the Doppler frequency estimate of the target signal, S k It refers to a vector that collects the sampled values of the segment-related signals at relative time delays, including: A. Take vector S k The N values in are taken as the observed sample values S k (0),S k (1),…,S k (N-1), estimate the M-order autocorrelation matrix B. Yes Perform eigendecomposition to obtain the eigenvector u corresponding to the MK smallest eigenvalues K+1 ,u K+2 ,...u K+M , construct the matrix G = [u K+1 u K+2 ...u K+M ], define the vector a(z)=[1z -1 …z -(M-1) ],z=e jω , ω is the complex signal frequency; C. Solve equation a T (z (-1) )GG H a(z)=0, find the K roots closest to the unit circle. The phase of these K roots is the Doppler frequency estimate of the target signal; The relative delay spectrum of the echo signal is obtained by segment correlation between the echo signal and the reference signal, including: (1) The digital baseband reference signal sample data received by the reference channel, i.e., the reference signal s ref Divided into I segments, each segment has a data length of L 1 , the nth sample of the i-th reference signal As shown in formula (I): In formula (I), The length is L 1 Rectangular window; (2) The digital baseband echo signal sample data received by the monitoring channel, i.e., the echo signal s ech The overlapping segments are divided into I segments, and the length of each overlapping segment is L 2 , the nth sample of the i-th echo signal As shown in formula (II): In formula (II), The length is L 2 Rectangular window; (3) The time delay of the target relative to the direct path is obtained by the delay unit r corresponding to the correlation peak between the reference signal and the echo signal. The correlation process between the echo signal and the reference signal in the time domain is shown in formula (III): In formula (III), ψ i (r) refers to the correlation value at delay unit r, are the sample values of the i-th echo signal and the reference signal at the delay unit r respectively; In formula (III), discrete Fourier transform and inverse discrete Fourier transform are used to simplify the correlation process, and fast Fourier transform is used to accelerate the calculation, as shown in formula (IV): In formula (IV), for Execute N FFT The result of the point-length discrete Fourier transform, for Execute N FFT The result of taking the conjugate after the point length discrete Fourier transform, for The sample of the anti-corrugation signal, N FFT Should satisfy N FFT ≥L 2 .
2. According to the target detection method of segmented correlation accumulation of claim 1, It is characterized in that Performing non-coherent accumulation on the segmented correlation values to obtain the relative delay spectrum of the target echo signal after noise reduction means: performing non-coherent accumulation on the segmented correlation results of multiple segments in step (3) to obtain the relative delay distribution of the target, that is, the relative delay spectrum of the target echo signal after noise reduction. The process is shown in formula (V): In formula (V), Ψ(r) is the correlation value of the target relative delay distribution corresponding to the delay unit r.
3. The target detection method of segmented correlation accumulation according to claim 1, It is characterized in that Use the CFAR detection program to detect the target in the range domain, and use the target delay information to obtain the segmented correlation signal sampling at the relative delay; including: (4) By setting the false alarm probability p f The CFAR algorithm is used to detect the target from the relative delay spectrum RDP of the target echo signal after noise reduction; the false alarm probability refers to the probability that the CFAR determines the existence of the target when there is no target signal; When the relative delay spectrum Ψ(r) of the target echo signal after noise reduction is greater than the threshold T 0 When the target is detected by the delay, it is regarded as a target or a group of targets that cannot be identified by delay alone and need to be further distinguished by Doppler information. When the value of the relative delay spectrum RDP of the target echo signal after noise reduction at the target delay unit is not greater than the threshold, it is regarded as no target signal is detected, that is, the target is missed; the threshold T 0 It is calculated based on the false alarm probability and background noise distribution. For noise that obeys exponential distribution, T 0 =μ 0 ln(p f ), μ 0 is the estimated value of background noise power; The kth target signal delay is calculated by formula (VI): In formula (VI), is the kth target signal delay estimate, It is the target delay unit estimate; (5) Estimate the target delay unit in step (4) Assign the value to r in formula (III) to get the target delay unit estimate The correlation peak at the relative time delay is the sampling value of the segment correlation signal The sampling values of the segment-related signal at the relative time delay are collected into a vector, denoted as Indicates that the vector dimension is Ix1, S k As represented by formula (VII): In formula (VII), S k (i) refers to S k The i-th vector value in .
4. The target detection method of segmented correlation accumulation according to claim 3, It is characterized in that p f =10 -4 。 5. A segmented correlation accumulation target detection system, It is characterized in that include: The relative delay spectrum obtaining module of the echo signal is configured to obtain the relative delay spectrum of the echo signal through the segmented correlation between the echo signal and the reference signal; The module for obtaining the relative time delay spectrum of the target echo signal after noise reduction is configured to perform incoherent accumulation of the correlation values of each segment to obtain the relative time delay spectrum of the target echo signal after noise reduction; The segment-correlated signal sampling module at the relative time delay is configured to detect the target in the range domain using the CFAR detection procedure and obtain the segment-correlated signal sampling at the relative time delay using the target time delay information; The target Doppler estimation module is configured to estimate the target Doppler at a high resolution frequency using a ROOT-MUSIC algorithm; Using the ROOT-MUSIC algorithm to estimate the target Doppler at high resolution frequency means: k Perform ROOT-MUSIC high-resolution frequency estimation to obtain the Doppler frequency estimate of the target signal, S k It refers to a vector that collects the sampled values of the segment-related signals at relative time delays, including: A. Take vector S k The N values in are taken as the observed sample values S k (0),S k (1),…,S k (N-1), estimate the M-order autocorrelation matrix B. Yes Perform eigendecomposition to obtain the eigenvector u corresponding to the MK smallest eigenvalues K+1 ,u K+2 ,...u K+M , construct the matrix G = [u K+1 u K+2 ...u K+M ], define the vector a(z)=[1z -1 …z -(M-1) ],z=e jω , ω is the complex signal frequency; C. Solve equation a T (z (-1) )GG H a(z)=0, find the K roots closest to the unit circle. The phase of these K roots is the Doppler frequency estimate of the target signal; The relative delay spectrum of the echo signal is obtained by segment correlation between the echo signal and the reference signal, including: (1) The digital baseband reference signal sample data received by the reference channel, i.e., the reference signal s ref Divided into I segments, each segment has a data length of L 1 , the nth sample of the i-th reference signal As shown in formula (I): In formula (I), The length is L 1 Rectangular window; (2) The digital baseband echo signal sample data received by the monitoring channel, i.e., the echo signal s ech The overlapping segments are divided into I segments, and the length of each overlapping segment is L 2 , the nth sample of the i-th echo signal As shown in formula (II): In formula (II), The length is L 2 Rectangular window; (3) The time delay of the target relative to the direct path is obtained by the delay unit r corresponding to the correlation peak between the reference signal and the echo signal. The correlation process between the echo signal and the reference signal in the time domain is shown in formula (III): In formula (III), ψ i (r) refers to the correlation value at delay unit r, are the sample values of the i-th echo signal and the reference signal at the delay unit r respectively; In formula (III), discrete Fourier transform and inverse discrete Fourier transform are used to simplify the correlation process, and fast Fourier transform is used to accelerate the calculation, as shown in formula (IV): In formula (IV), for Execute N FFT The result of the point-length discrete Fourier transform, for Execute N FFT The result of taking the conjugate after the point length discrete Fourier transform, for The sample of the anti-corrugation signal, N FFT Should satisfy N FFT ≥L 2 .
Citation Information
Cited By
FIR-IAA target Doppler frequency estimation method based on whitening in interference environment
CN120993330A
