Clutter map constant false alarm detection method based on DDMA-MIMO

Abstract

The invention belongs to the technical field of radar detection, and particularly relates to a clutter map constant false alarm rate detection method based on DDMA-MIMO. In an actual measurement scene, Doppler and distance dimension 2D-FFT processing is carried out on data of an (n + 1) th frame under a DDMA-MIMO system, and a distance-Doppler spectrum (namely RV-MAP) is obtained. In order to avoid a speed fuzzy phenomenon, the RV-MAP graph is demodulated by a DDMA method based on Empty-band. In the process, in order to reduce the influence of the scene clutter on the detection performance, the clutter map of the nth frame is combined with the demodulation data of the (n + 1) th frame to perform constant false alarm rate (CFAR) detection, and finally, the RV-MAP map of the (n + 1) th frame and the clutter map of the nth frame are combined to update the clutter map of the (n + 1) th frame. Through actual measurement data verification, after a traditional CFAR algorithm is carried out, a target, clutter and noise are reserved in RV-MAP, and after a clutter map constant false alarm detection method is adopted, more clutter data are filtered out, and target data are reserved. And meanwhile, due to filtering of clutter data, the detection rate of the target is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of radar signal processing, and in particular relates to a clutter pattern constant false alarm detection method based on DDMA-MIMO. Background Art

[0002] Constant false alarm detection (CFAR) has good detection performance when the number of reference units matches the uniform width of clutter, and is an important component of radar signal processing systems. However, for clutter with drastic changes in clutter intensity in distance and azimuth and narrow uniform width, such as ground clutter, the constant false alarm loss will increase rapidly as the number of references decreases. At this time, traditional CFAR is no longer applicable. Clutter map CFAR processing can iteratively process the echo sampling values ​​obtained from previous scans in time, and then estimate the current clutter background intensity to achieve constant false alarm detection. Therefore, it is suitable for clutter that is non-uniform in space but stable in time. Clutter map CFAR processing has a wide range of applications in radar moving target detection, which enables radar signal processing systems to have good detection performance. Summary of the invention

[0003] The purpose of the present invention is to provide a DDMA-MIMO-based clutter map constant false alarm detection method for improving the target detection performance of a vehicle-mounted DDMA-MIMO radar system. The method comprises the following steps:

[0004] S1. Initialize system parameters:

[0005] Initialize system target parameters and clutter parameters, the target parameters include: target distance, target speed and target angle; the clutter parameters include: radar height from the ground, clutter introduction range limit.

[0006] S2. Constructing the vehicle-mounted DDMA-MIMO signal model:

[0007] The vehicle-mounted DDMA-MIMO signal model includes target modeling and clutter modeling. According to the target parameters and clutter parameters in step S1, the transmission signal and echo signal of DDMA-MIMO are constructed.

[0008] S3, Range-Doppler 2D-FFT:

[0009] The obtained vehicle-mounted DDMA-MIMO echo signal is processed by 2D-FFT in the range dimension and the Doppler dimension to obtain the range-Doppler spectrum, namely, RV-MAP.

[0010] S4, DDMA-MIMO demodulation:

[0011] Because all radar transmitting antennas transmit signals simultaneously under the DDMA-MIMO system, if demodulation is not performed after 2D-FFT, velocity ambiguity will occur, making it impossible to obtain the correct velocity of the target.

[0012] S5, clutter image constant false alarm detection:

[0013] The Nitzberg clutter map CFAR algorithm is used to filter out clutter and only retain target parameters on the RV-MAP to achieve target detection.

[0014] Furthermore, the specific method of step S1 is:

[0015] Initialize the system target parameters and clutter parameters. The target parameters include: target distance r, target speed v and target angle θ; the clutter parameters include: radar height H from the ground, clutter introduction range R min With R max Between, where R min Indicates the starting distance of introducing clutter, R max Indicates the end distance of introducing clutter.

[0016] Furthermore, the specific method of step S2 is:

[0017] First, construct the transmission signal based on the DDMA-MIMO system. In the DDMA-MIMO system, all transmitting antennas transmit at the same time, and the signal of each transmitting antenna is Doppler frequency offset so that the signals of different transmitting antennas can be separated in the Doppler domain. Assuming that the MIMO radar has M transmit and N receive, and the waveform uses linear frequency modulation continuous wave LFM, then the waveform of the mth transmitting antenna is

[0018]

[0019] Where k represents the kth sweep cycle, with a total of K sweep cycles, f0 represents the center frequency of the radar, μ represents the modulation slope of the LFM signal, T represents the sweep cycle of the LFM signal, and f m Represents the Doppler frequency offset of the mth transmitting antenna.

[0020] The DDMA waveform is a waveform that can be orthogonalized after Doppler processing at the receiving end. m The change of (k-1)T achieves orthogonality in the Doppler domain. Consider dividing the entire Doppler pulse repetition frequency PRF into M orthogonal sub-repetition frequency channels, and the bandwidth △f of each channel is Then f m for

[0021]

[0022] When constructing the target echo signal, the echo signal received by the nth receiving antenna is the sum of the echoes of the transmitted signals of the M transmitting antennas. Without considering the amplitude attenuation, the echo signal received by the nth receiving antenna is

[0023]

[0024] Where t represents the echo delay of the radar signal transmitted by the target.

[0025] The modeling of radar clutter signals is based on the introduction range of radar clutter as R min To R max The radar antenna angle corresponding to each clutter unit is calculated based on the radar height H, and the ground units are divided one by one based on this. Assuming that the number of clutter units after division is N, the echo data of the clutter unit can be expressed as

[0026]

[0027] Where t i represents the distance between the clutter unit and the center of the radar array. Then the clutter can be modeled as

[0028]

[0029] Finally, the radar clutter signal is superimposed on the target echo data.

[0030]

[0031] Furthermore, the specific method of step S3 is:

[0032] After obtaining the target and clutter echo data based on the DDMA-MIMO mechanism, first perform FFT on each pulse to generate distance dimension data, and then perform FFT on the data on each distance unit, that is, perform FFT in the Doppler dimension to obtain velocity dimension data. After completing such a two-dimensional FFT, the range-Doppler spectrum of the environment and target obtained by the radar in this frame of time, namely RV-MAP, can be obtained.

[0033] Furthermore, the specific method of step S4 is:

[0034] In order to solve the problem of DDMA-MIMO Doppler ambiguity, an Empty-band based DDMA demodulation algorithm can be used. The so-called Empty-band DDMA is to divide the Doppler frequency domain into M+E when setting the Doppler frequency offset of each transmitting antenna. bandA simple linear form of equal-width channels is used. Based on the number of transmitting antennas, one or two Empty subbands are added. The number of Empty-bands to be introduced should be selected based on the number of transmitting antennas M and the number of Doppler domain sampling points currently configured. It is necessary to ensure that the number of Doppler units allocated to each subband is an integer. At this time, the Doppler frequency offset f of the mth transmitting antenna is m for

[0035]

[0036] The introduction of Empty-band evenly divides the entire RV-MAP into M+E band sub-bands. When performing target mapping, the correct target position will be used as the starting point, and the target will appear in M-1 consecutive sub-bands on the right (if it exceeds the M+E band sub-bands, it will continue to appear from the first sub-band). The left side of the correct target is continuous E band The subband where no target falls (if it exceeds the first subband, it will continue to appear from the last subband), and the energy of the subband where no target falls must be lower than the energy of the remaining M subbands where target signals fall. Then according to this rule, assuming that the energy of each subband is X i , for M+E band The energy of each subband group is obtained by performing a cyclic shift of length M.

[0037]

[0038] Find the maximum subband group energy Z i , according to the index number i of the maximum value, find the i-th subband where the target is located.

[0039] Furthermore, the specific method of step S5 is:

[0040] In order to filter the influence of fixed and stationary clutter in the scene on the detected target, different from the traditional constant false alarm detection algorithm, this method introduces the clutter map method. The main idea is to extract and record the clutter data in the scene in real time by combining the sub-band data in the RV-MAP of each frame in the radar processing process, and finally use it for the detection judgment of the current processing frame. Therefore, the dimension of the clutter map is always the same as the sub-band dimension in the RV-MAP after step S4.

[0041] For the first frame data, the subband data set q1 in the RV-MAP obtained after the processing in step S4 is assigned to the first frame clutter map data q1, and the processing of the second frame is entered.

[0042] Suppose the subband data set q in the RV-MAP obtained after the nth (n>1) frame data is processed in step S4n , the clutter map formed by the subband data in the first n-1 frames of RV-MAP is p n-1 , assuming that the set q n The amplitude of the kth point in is q n (k), clutter map p n-1 The amplitude of the kth point in is p n-1 (k), first perform constant false alarm detection, for q n The kth point q n (k) Calculate its detection threshold T·p n-1 (k), where T represents the threshold coefficient, which is determined by the following formula:

[0043]

[0044] Where P fa is the false alarm probability, n is the number of frames currently processed and w is the forgetting factor. n (k) and T·p n-1 (k) is the magnitude, if q n (k) is greater than T·p n-1 (k) then it is determined that the target exists in the sub-band data, otherwise it is determined that the target does not exist.

[0045] After the above steps, the target detection process of the clutter map constant false alarm algorithm is completed. The next step is to combine the previous n-1 frames of clutter map p n-1 and the subband data set q in the nth frame RV-MAP n As well as the detection result of the nth frame, iteratively update the clutter map p of the nth frame n , when the kth point is not detected as a target in the nth frame, the update method is as follows:

[0046] p n (k) = (1-w)p n-1 (k)+wq n (k)

[0047] If it is detected as a target, the update method is as follows:

[0048] p n (k) = p n-1 (k)

[0049] The obtained clutter map p n The subband data set q in the RV-MAP obtained after processing in step S4 of the n+1th frame is used n+1 Detection and judgment.

[0050] Compared with the prior art, compared with the traditional constant false alarm detection algorithm, the present invention provides a clutter map constant false alarm detection method based on DDMA-MIMO. When using traditional CFAR for detection, in an environment where the clutter in distance and azimuth has strong changes, the constant false alarm loss will increase rapidly with the decrease in the number of references, and the obtained CFAR result has non-target clutter points. The clutter map constant false alarm detection method based on DDMA-MIMO provided by the present invention can iteratively process the echo sampling values ​​obtained by previous scanning in the context of vehicle-mounted DDMA-MIMO in time, estimate the current clutter background intensity, and then filter out the clutter background in the echo signal, thereby improving the detection performance of the CFAR algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 The present invention is a flow chart of a DDMA-MIMO-based clutter pattern constant false alarm detection method.

[0052] Figure 2 This is a DDMA-MIMO signal transmission diagram of step S2.

[0053] Figure 3 FIG. 4 is a schematic diagram of clutter simulation in step S2.

[0054] Figure 4 This is a flow chart of the range-Doppler 2D-FFT of step S3.

[0055] Figure 5 It is the DDMA-MIMO speed ambiguous RV-MAP of step S4.

[0056] Figure 6 It is the RV-MAP after DDMA-MIMO demodulation in step S4.

[0057] Figure 7 It is the clutter diagram constant false alarm detector diagram of step S5.

[0058] Figure 8 It is the RV-MAP using traditional CFAR.

[0059] Fig. 9 The present invention adopts the RV-MAP of the clutter map constant false alarm detection method. DETAILED DESCRIPTION

[0060] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0061] In the specific implementation, the AWR2243 millimeter wave radar cascade platform produced by TI was used, which adopted a 3-transmit 4-receive MIMO system and an empty-band number of 1. Figure 1 As shown, a flow chart of a clutter image constant false alarm detection method based on DDMA-MIMO includes the following steps:

[0062] S1. Initialize system parameters:

[0063] Initialize the system target parameters and clutter parameters. The target parameters include: the target distance is about 2.7 meters, the target speed is about 6 meters / second, and the target angle is 0 degrees; the clutter parameters include: the height of the radar from the ground is 60 centimeters, and the clutter is related to the actual measurement environment on site.

[0064] S2. Constructing the vehicle-mounted DDMA-MIMO signal model:

[0065] Step S2: The vehicle-mounted DDMA-MIMO signal model includes target modeling and clutter modeling. The radar signal emission diagram is as follows: Figure 2 As shown, the clutter simulation diagram is as follows Figure 3 As shown, a DDMA-MIMO transmission signal is constructed.

[0066] S3, Range-Doppler 2D-FFT:

[0067] The flow chart of the distance-Doppler 2D-FFT in step S3 is as follows: Figure 4 As shown, the obtained vehicle-mounted DDMA-MIMO signal model is subjected to 2D-FFT processing in the range dimension and the Doppler dimension to obtain the RV-MAP.

[0068] S4, DDMA-MIMO demodulation:

[0069] Since all radar transmitting antennas transmit signals simultaneously in the DDMA-MIMO system, if demodulation is not performed after 2D-FFT, the following will occur: Figure 5 The speed fuzzy RV-MAP shown in the figure shows three speeds of the target. After using the DDMA demodulation algorithm based on Empty-band, since the real target is in sub-band 3, the data of sub-band 3 is retained, and the demodulated RV-MAP is as follows: Figure 6 shown.

[0070] S5, clutter image constant false alarm detection:

[0071] The clutter diagram constant false alarm detector of step S5 is shown in FIG. Figure 7 As shown, the Nitzberg clutter map CFAR algorithm is used to filter out the clutter and only retain the target parameters on the RV-MAP.

[0072] The RV-MAP after using the clutter map constant false alarm detection method is as follows Fig. 9 As shown in the figure, compared with the conventional CFAR algorithm, Figure 8 Compared with the traditional CFAR algorithm, the target, clutter and noise are retained in the RV-MAP, while the clutter map constant false alarm detection method is used to filter out more clutter data and retain the target data. At the same time, due to the filtering of clutter data, the target detection rate is improved, that is, the number of targets divided by the sum of the number of non-targets and the number of targets.

Claims

1. A clutter pattern constant false alarm detection method based on DDMA-MIMO, characterized in that: The following steps are involved: S1. Initialize system parameters: Initialize the system target parameters and clutter parameters. The target parameters include: target distance r, target speed v and target angle θ; the clutter parameters include: radar height H from the ground, clutter introduction range R min With R max Between, where R min Indicates the starting distance of introducing clutter, R max Indicates the end distance of introducing clutter; S2. Constructing the vehicle-mounted DDMA-MIMO signal model: First, construct the transmission signal based on the DDMA-MIMO system. In the DDMA-MIMO system, all transmitting antennas transmit at the same time, and the signal of each transmitting antenna is Doppler frequency offset so that the signals of different transmitting antennas can be separated in the Doppler domain. Assuming that the MIMO radar has M transmit and N receive, and the waveform adopts linear frequency modulation continuous wave LFM, then the waveform of the mth transmitting antenna is: Where k represents the kth sweep cycle, with a total of K sweep cycles, f0 represents the center frequency of the radar, μ represents the modulation slope of the LFM signal, T represents the sweep cycle of the LFM signal, and f m represents the Doppler frequency offset of the mth transmitting antenna; The DDMA waveform is orthogonal after Doppler processing at the receiving end, relying on f m The change of (k-1)T achieves orthogonality in the Doppler domain; consider dividing the entire Doppler pulse repetition frequency PRF into M orthogonal sub-repetition frequency channels, and the bandwidth △f of each channel is f m for: When constructing the target echo signal, the echo signal received by the nth receiving antenna is the sum of the echoes of the transmitted signals of the M transmitting antennas; without considering the amplitude attenuation, the echo signal received by the nth receiving antenna is: Where t represents the echo delay of the radar signal transmitted by the target; The modeling of radar clutter signals is based on the introduction range of radar clutter as R min To R max The radar antenna angle corresponding to each clutter unit is calculated based on the radar height H, and the ground units are divided one by one based on this. Assuming that the number of clutter units after division is N, the echo data of the clutter unit is expressed as: Where t i Represents the distance between the clutter unit and the center of the radar array; the clutter is modeled as: Finally, the radar clutter signal is superimposed on the target echo data: S3, Range-Doppler 2D-FFT: After obtaining the target and clutter echo data based on the DDMA-MIMO mechanism, first perform FFT on each pulse to generate distance dimension data, and then perform FFT on the data on each distance unit, that is, perform FFT on the Doppler dimension to obtain velocity dimension data; after completing such a two-dimensional FFT, the range-Doppler spectrum of the environment and target obtained by the radar in this frame time is obtained, which is defined as RV-MAP; S4, DDMA-MIMO demodulation: Using the Empty-band based DDMA demodulation algorithm, the Doppler frequency domain is divided into M+E when setting the Doppler frequency offset of each transmitting antenna. band A simple linear form of equal-width channels is used. On the basis of the number of transmitting antennas, one or two Empty subbands are added. The number of Empty-bands introduced should be selected based on the number of transmitting antennas M and the number of Doppler domain sampling points currently configured. It is necessary to ensure that the number of Doppler units allocated to each subband is an integer. At this time, the Doppler frequency offset f of the mth transmitting antenna is m for: The introduction of Empty-band evenly divides the entire RV-MAP into M+E along the Doppler domain. band sub-bands. When performing target mapping, the correct target position will be taken as the starting point, and the target will appear in M-1 consecutive sub-bands on the right, and the left side of the correct target will be continuous E band The subband where no target falls is bound to have lower energy than the remaining M subbands where target signals fall. Based on this rule, the energy of each subband is defined as X. i , for M+E band The energy of each subband group is obtained by performing a cyclic shift of length M. Find the maximum subband group energy Z i , according to the index number i of the maximum value, find the i-th subband where the target is located; S5, clutter image constant false alarm detection: The Nitzberg clutter map CFAR algorithm is used to filter out clutter and only retain target parameters on the RV-MAP to achieve target detection.

2. A clutter pattern constant false alarm detection method based on DDMA-MIMO according to claim 1, characterized in that: The specific method of S5 is to extract and record the clutter data in the scene in real time in combination with the sub-band data in the RV-MAP of each frame in the radar processing process, and finally use it for detection and judgment of the current processing frame. The dimension of the clutter map is always the same as the sub-band dimension of the RV-MAP after processing in step S4, specifically: For the first frame data, the subband data set q1 in the RV-MAP obtained after the processing in step S4 is assigned to the first frame clutter map data q1, and the processing of the second frame is started; Suppose the subband data set q in the RV-MAP obtained after the nth frame data is processed in step S4 n , n>1, the clutter map formed by the subband data in the first n-1 frames of RV-MAP is p n-1 , define the set q n The amplitude of the kth point in is q n (k), clutter map p n-1 The amplitude of the kth point in is p n-1 (k), first perform constant false alarm detection, for q n The kth point q n (k) Calculate its detection threshold T·p n-1 (k), where T represents the threshold coefficient: Where P fa is the false alarm probability, n is the number of frames currently processed, and w is the forgetting factor; compare q n (k) and T·p n-1 (k) is the magnitude, if q n (k) greater than T·p n-1 (k) determining that the target exists in the sub-band data, otherwise determining that the target does not exist; After completing the target detection, the next step is to combine the previous n-1 frames of clutter map p n-1 and the subband data set q in the nth frame RV-MAP n As well as the detection result of the nth frame, iteratively update the clutter map p of the nth frame n , when the kth point is not detected as a target in the nth frame, the update method is as follows: p n (k)=(1-w)p n-1 (k)+wq n (k) If it is detected as a target, the update method is as follows: p n (k)=p n-1 (k) The obtained clutter map p n The subband data set q in the RV-MAP obtained after processing in step S4 of the n+1th frame is used n+1 Detection and judgment.

Images