Distributed Radar Data Registration and Fusion Detection Method Based on Low-Communication Measurement

By designing initial thresholds and quantization values ​​in a distributed radar system, processing low-communication-volume measurement points at local radar stations, and performing adaptive data registration and fusion detection at the fusion center, the problems of measurement mismatch and communication burden are solved, and the engineering application of adaptive data registration and fusion detection is realized.

CN117192497BActive Publication Date: 2026-11-10XIDIAN UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202311117717.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-08-31
Publication Date
2026-11-10
Estimated Expiration
2043-08-31

AI Technical Summary

Technical Problem

In distributed radar systems, the measurements from various local radar stations are mismatched in the spatiotemporal dimensions, making it difficult for the fusion center to complete data registration and fusion detection, and also causing excessive communication burden. Existing technologies have not been able to effectively solve this problem.

Method used

By designing initial thresholds, quantization thresholds, and quantization values, local radar stations perform low-communication-volume measurement point processing, and adaptive data registration and generalized likelihood ratio-based fusion detection are performed at the fusion center, reducing the communication burden and achieving adaptive data registration.

Benefits of technology

It effectively reduces the communication burden of distributed radar systems, realizes adaptive data registration and fusion detection of local radar stations, and facilitates practical engineering applications.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117192497B_ABST
    Figure CN117192497B_ABST
Patent Text Reader

Abstract

This invention proposes a distributed radar data registration and fusion detection method based on low-bandwidth measurements, primarily addressing the problem that existing fusion detection algorithms are difficult to apply in engineering due to their failure to consider measurement mismatch. The implementation scheme is as follows: A local raw test statistic is constructed based on the echo signal; initial thresholds, quantization thresholds, and quantization values ​​are designed based on the initial data throughput and the number of quantization intervals, constructing low-bandwidth measurement points; the fusion center adaptively constructs a detection window based on the low-bandwidth measurement points and constructs a global test statistic based on the generalized likelihood ratio criterion; the global test statistic is compared with the detection threshold to determine the presence of the target. Simulation results show that this invention can effectively complete data registration under extremely low communication and computational burdens, maintain a good detection probability, facilitate engineering implementation, and can be used for distributed radar data registration and target detection under bandwidth-constrained conditions.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of radar technology, and specifically relates to a distributed radar data registration and fusion detection method, which can be used for distributed radar data registration and target detection under limited communication bandwidth. Background Technology

[0002] With the rapid development of technology, stealth material manufacturing and hardware integration technologies have emerged rapidly, significantly reducing the signal-to-noise ratio (SNR) of potential target echoes in radar detection environments. Simultaneously, the rapid advancements in radar and electromagnetic technologies have made today's radar detection environments increasingly complex and volatile. Traditional single-radar detection systems face problems such as weak echo SNR, poor survivability, and low flexibility. Therefore, there is an urgent need to research more effective target detection technologies to ensure that radar systems can detect targets in a timely manner. A distributed radar system comprises multiple geographically dispersed local radar stations and a data fusion center. Multiple local radar stations can observe targets from different perspectives and at different times, transmitting local observation information to the fusion center for fusion decision-making. Because distributed radar systems can achieve higher spatial diversity gain, SNR gain, and geometric gain compared to single radar stations, they can achieve higher target detection performance. Furthermore, multiple geographically dispersed local radar stations can improve system security and flexibility. Therefore, distributed radar detection systems can not only greatly enhance the system's detection power but also improve its survivability and degrees of freedom.

[0003] However, target detection in distributed radar systems faces two main challenges. First, the extremely high communication bandwidth required for processing by transmitting observation information directly from each local radar station to the fusion center places a significant hardware burden on radar systems with limited resources. Second, the different local radar stations observe the target from different angles and at different times, and measurement noise exists, causing a mismatch in the measurements across the spatiotemporal dimensions, making it difficult for the fusion center to complete fusion detection. Therefore, completing the data registration of measurements from each local radar station is a crucial prerequisite for further fusion detection.

[0004] To address the high communication burden in distributed fusion detection, the literature [J. Lu, S. Zhou, Z. Shao, and H. Liu, “Distributed detection with local least squares quantization,” in 2019 IEEE International Conference on Signal and Information and Data Processing, 2019.] studies a minimum mean square error quantization method for local test statistics. At local radar stations, this method reduces the system communication burden by performing low-bit minimum mean square error quantization on the local test statistics and transmitting only the quantized test statistics to the fusion center. At the fusion center, the Neyman-Pearson criterion is used to fuse the quantized test statistics from each local radar station and complete the fusion decision.

[0005] The literature [Shenghua Zhou, Yidi Chen, Hongwei Liu, Qinzhen Hu, Ding Cao, and Leilei Xu. Distributed detection with initial thresholds in local sensors. In IEEE Radar Conference, pages 244–248, Johnsburg, South Africa, March 2015.] studies a distributed detection technique based on threshold-based screening statistics. At local radar stations, setting an initial threshold eliminates most false alarms, and only transmits statistics exceeding the initial threshold to the fusion center, effectively reducing the system's communication burden. At the fusion center, the Neyman-Pearson criterion is used to perform fusion decisions on the statistics filtered out by the initial threshold.

[0006] The above detection methods do not take into account the mismatch problem of measurements from different local radar stations, and therefore cannot complete adaptive data registration and fusion detection for different local radar stations, making them difficult to apply in engineering projects in actual distributed radar systems. Summary of the Invention

[0007] The purpose of this invention is to address the shortcomings of the existing technologies by proposing a distributed radar data registration and fusion detection method based on low-traffic measurement. This method aims to reduce the system's communication burden while completing adaptive data registration and fusion detection for each local radar station, thus facilitating practical engineering applications.

[0008] The technical solution to achieve the objective of this invention is as follows: Based on the communication bandwidth requirements, the initial threshold, quantization threshold, and quantization value are calculated. Each local radar station constructs low-communication-capacity measurement points based on these values ​​and transmits them to the fusion center. The fusion center completes adaptive data registration based on the received low-communication-capacity measurement points from the local radar stations and fuses the local test statistics of low communication-capacity data using a generalized likelihood ratio detector to determine whether the target exists. Specific steps include the following:

[0009] (1) For a distributed radar system including a fusion center and N local radar stations, the original local test statistic l is constructed based on the echo signal of the nth radar station. n,ρ Where ρ = 1, 2, ..., C n Indicates the resolvable cell index, C n This represents the number of resolution units of the nth radar station, where n = 1, 2, ..., N;

[0010] (2) Based on the system communication bandwidth requirements, set the initial data throughput α of the nth local radar station. n and the number of quantization intervals M n Or the number of quantization bits b n =log2M n Based on the distribution function of the original local test statistic, an initial threshold γ is designed. 1,n ;

[0011] (3) Based on the given initial threshold γ 1,n and the number of quantization intervals M n Design quantization threshold τ n Quantization value v n ;

[0012] (4) Construct local measurement points with low communication traffic and transmit them to the fusion center:

[0013] (4a) Based on the initial threshold γ 1,n Quantization threshold τ n Quantization value v n and the original local test statistic l n,ρ Construct a local test statistic q with low communication volume n,ρ ;

[0014] (4b) Each local radar station will exceed the initial threshold γ 1,n The distance and angle measurements of the corresponding resolution units are converted to the global rectangular coordinate system to obtain the measured abscissa and ordinate after coordinate transformation.

[0015] (4c) The local radar station constructs a set of low-communication measurement points based on the measurement abscissa and ordinate after coordinate transformation, and the low-communication test statistics corresponding to the measurement points, and transmits it to the fusion center.

[0016] (5) The fusion center constructs a detection window based on the received low-communication-volume measurement points:

[0017] (5a) Use a set to collect the measurement points received by the fusion center. This indicates that, based on the c-th measurement point Construct the area where the target may exist in This indicates the number of measurement points in the input fusion center. The superscript I(c) indicates that the measurement point comes from the I(c)th radar station.

[0018] (5b) Areas where the target may exist Discretize to obtain a set of discrete points.

[0019] (5c) For any discrete point Construct its detection window after data registration in

[0020] (5d) Use a set to collect all detection windows Represented as:

[0021]

[0022] (6) The fusion center constructs a global test statistic based on the detection window and determines whether the target exists:

[0023] (6a) For any detection window A global test statistic is constructed based on the generalized likelihood ratio criterion.

[0024] (6b) Based on the given global false alarm probability P fa Offline creation of detection window dictionary Detection threshold dictionary γ and randomization probability factor dictionary κ;

[0025] (6c) For any detection window Based on the global test statistic Detection Window Dictionary The detection threshold dictionary γ and the randomization probability factor dictionary κ are used to determine whether the target exists.

[0026] (6c1) Search Detection Window Corresponding to the detection window dictionary Index in

[0027] (6c2) The global test statistic With detection threshold Compare and determine if the target exists:

[0028] like Greater than the detection threshold Then the target is determined to exist;

[0029] like Less than the detection threshold Then the target is determined to not exist;

[0030] like Equal to detection threshold Then proceed to step (6c3);

[0031] (6c3) Generate a random number u that follows a uniform distribution from 0 to 1, and combine it with the randomization probability factor. Comparison:

[0032] like Then the target is determined to exist;

[0033] Otherwise, the target is determined to not exist.

[0034] Compared with the prior art, the present invention has the following advantages:

[0035] First, this invention provides a low-communication-volume processing method for local radar station data. Specifically, it first performs threshold screening on local test statistics to eliminate the vast majority of noise points. Then, based on preset quantization thresholds and quantization values, it quantizes statistics that exceed the initial threshold to further reduce communication volume. Finally, the local radar station only transmits low-communication-volume measurement points to the fusion center, thus significantly reducing the communication burden of the distributed radar system and facilitating the hardware implementation of the system in engineering.

[0036] Secondly, this invention proposes an adaptive data registration algorithm based on measurement points with low communication volume. That is, by adaptively constructing a detection window through the location information and statistical information of the measurement points, the distributed radar system can achieve adaptive data registration with low communication volume, laying the foundation for subsequent fusion detection and facilitating engineering implementation.

[0037] Third, for the low-communication-volume measurement and data registration method of each local radar station, the present invention constructs a detector and decision criterion based on the generalized likelihood ratio criterion, which can realize the fusion decision of low-communication-volume measurement points of each local radar station in the distributed radar system, which is beneficial to engineering implementation. Attached Figure Description

[0038] Figure 1 This is a flowchart illustrating the overall implementation of the present invention;

[0039] Figure 2 This is a sub-flowchart of the present invention for solving the quantization threshold and quantization value based on the minimum mean square error quantization method;

[0040] Figure 3 This is a graph showing the changes in the multi-station measurement registration probability and the multi-station target measurement input probability as a function of the signal-to-noise ratio.

[0041] Figure 4 This is a graph showing the change in the number of detection windows and the number of input measurement points as a function of the initial data throughput of this invention.

[0042] Figure 5 This is a graph showing the change in detection probability as a function of signal-to-noise ratio according to the present invention. Detailed Implementation

[0043] The specific embodiments and effects of the present invention will be described in further detail below with reference to the accompanying drawings. It should be noted that the embodiments are only used to illustrate the present invention and do not constitute any limitation on the present invention.

[0044] Reference Figure 1 The implementation steps for this example are as follows:

[0045] Step 1: Construct the original local test statistic.

[0046] (1.1) For a distributed radar system comprising a signal fusion center and N local radar stations, assume that the nth radar station contains C in the search area. n There are n resolution cells, where n = 1, 2, ..., N. The echo signal e from the ρth resolution cell of the nth radar station is... n,ρ The model is as follows:

[0047]

[0048] Where: ρ = 1, 2, ..., C n Represents the resolvable cell index, ρ n This represents the resolution cell index of the target at the nth radar station;

[0049] z n,ρ This represents noise echoes, which follow a pattern with a mean of 0 and a variance of . The complex Gaussian distribution of s; n,ρ The target echo represents a signal with a mean of 0 and a variance of . The complex Gaussian distribution;

[0050] This indicates that there are no assumptions about the objective. This indicates that the objective is based on an assumption;

[0051] (1.2) Based on the echo signal e of the ρth resolution unit of the nth radar station n,ρ The original local test statistic l n,ρ Build as:

[0052]

[0053] in This is the set of reference cell indices containing the ρ-th resolution cell of the n-th radar station. Let n be the number of reference elements for the nth radar station. c represents the index of the k-th reference cell. n,k This represents the echo signal of the k-th reference cell, which has a mean of 0 and a variance of . The complex Gaussian distribution.

[0054] Step 2: Design the initial threshold.

[0055] (2.1) Based on the echo signal e n,ρ parameter K n , The original local test statistic l is obtained n,ρ The probability density function is given by the formula:

[0056]

[0057] in This represents the signal-to-noise ratio of the nth radar station;

[0058] (2.2) Based on the initial data throughput α of the nth radar station given by the system n and the original local test statistic l n,ρ The probability density function is used to obtain the initial data throughput α. n With the initial threshold γ 1,n Relationship:

[0059]

[0060] (2.3) Based on the initial data pass rate α n With the initial threshold γ 1,n The relationship is used to solve for the initial threshold γ of the nth radar station. 1,n :

[0061]

[0062] Step 3: Design the quantization threshold and quantization value for the nth radar station.

[0063] Existing methods for designing quantization thresholds and quantization values ​​include minimum mean square error quantization, maximum f-divergence quantization, maximum Fisher information quantization, maximum entropy quantization, and uniform quantization. This example uses, but is not limited to, the minimum mean square error quantization method.

[0064] Reference Figure 2 The implementation of this step includes the following:

[0065] (3.1) Set the number of loops Nc Set the precision factor Δ p Initialize loop count flag s = 1, initialize quantization threshold. Initialize quantization noise power Where T represents transpose. This is the initialization threshold value for the i-th quantization threshold of the n-th radar station, where i = 1, 2, ..., M. n -1; let Where f = 1, 2, ..., N c ;

[0066] (3.2) Based on the assumed signal-to-noise ratio of the nth radar station given by the system Calculate the assumed probability of passing the target measurement.

[0067]

[0068] (3.3) Measure the probability of passing based on the assumed target. Calculate the cutoff probability density function

[0069]

[0070] (3.4) Given the quantization threshold value after the (s-1)th optimization and the truncation probability density function Solve for the j-th quantization value after the s-th optimization.

[0071]

[0072] (3.5) Given the quantization value after the s-th optimization Solve for the i-th quantization threshold after the s-th optimization.

[0073]

[0074] (3.6) Given the quantization threshold value after the s-th optimization. and the quantization value after the sth optimization Solve for the quantization noise power after the s-th optimization.

[0075]

[0076] (3.7) Based on the quantized noise power after the (s-1)th optimization quantization noise power after the s-th optimization Determine if the loop terminates to obtain the final optimized quantization threshold and quantization value:

[0077] if or s>N c Then let τ n,0 =0, let The final optimized quantization threshold value is obtained. and quantization value

[0078]

[0079]

[0080] in

[0081] Otherwise, increment the value of the loop count flag s by 1 and return (3.4).

[0082] Step 4: Construct local measurement points with low communication traffic and transmit them to the fusion center.

[0083] (4.1) Based on the initial threshold γ 1,n Quantization threshold τ n Quantization value v n and the original local test statistic l n,ρ Calculate the local test statistic q for low communication volume. n,ρ :

[0084]

[0085] Where τ n,m v is the m-th quantization threshold for the n-th radar station. n,m This is the m-th quantization value of the n-th radar station, where m = 1, 2, ..., M n ;

[0086] (4.2) Each local radar station will exceed the initial threshold γ 1,n The distance and angle measurements of the corresponding resolution units are converted to the global rectangular coordinate system to obtain the measured abscissa and ordinate after coordinate transformation. The formula is as follows:

[0087]

[0088]

[0089] Where: x n,ζ y n,ζ These represent the x-coordinate and y-coordinate of the distance and angle measurements of the ζ-th resolution unit at the n-th radar station, after coordinate transformation. This indicates that the original test statistic exceeds the initial threshold γ. 1,n The index of measurement points;

[0090] This represents the angle between the polar axis of the nth radar station and the x-axis of the global rectangular coordinate system. This represents the x-coordinate of the nth radar station in the global Cartesian coordinate system. This represents the ordinate of the nth radar station in the global rectangular coordinate system;

[0091] r n,ζ φ n,ζ The distance and angle measurements for the nth radar station in the ζth resolution cell are respectively expressed as:

[0092]

[0093] In the formula: The radial distance of the target relative to radar station n, Let n be the azimuth angle of the target relative to radar station n;

[0094] It is distance-dependent additive Gaussian noise with a mean of 0 and a standard deviation of . Gaussian distribution, It is azimuth additive Gaussian noise with a mean of 0 and a standard deviation of . Gaussian distribution;

[0095] U(r min ,r max ) indicates that the range of values ​​is from r min to r max The uniform distribution, U(φ) min ,φ max ) indicates the range of values ​​from φ min To φ max The uniform distribution of r min r represents the minimum distance measurement value. max φ represents the maximum distance measurement value. min φ represents the minimum value of the azimuth measurement. max This indicates the maximum value of the azimuth measurement;

[0096] (4.3) The nth radar station is based on the measured abscissa x after coordinate transformation. n,ζ and measuring the ordinate y n,ζ And the low-communication test statistic q corresponding to the measurement point. n,ζ Construct a set of measurement points with low communication traffic. It is then transmitted to the fusion center, and the formula is as follows:

[0097]

[0098] Where q n,ζ Let be the local test statistic with low communication volume corresponding to the ζ-th resolution unit.

[0099] Step 5: The fusion center constructs a detection window based on the received low-communication-volume measurement points.

[0100] (5.1) The measurement points received by the fusion center are set using a set. The formula is as follows:

[0101]

[0102] (5.2) Given the c-th measurement point Construct the area where the target may exist The implementation steps include the following:

[0103] (5.2.1) Based on the low-communication local test statistic q at the c-th measurement point c Estimating signal-to-noise ratio

[0104]

[0105] Where: τ I(c),m Let v represent the m-th quantization threshold of the I(c)-th radar station. I(c),m Let m be the quantization value of the i-th radar station (c), where m = 1, 2, ..., M. I(c) ;

[0106] M I(c) K represents the number of quantization intervals for the I(c)th radar station. I(c) γ represents the number of reference elements for the I(c)th radar station. 1,I(c) This is the initial threshold for the I(c)th radar station;

[0107] λ I(c),min λ is the preset minimum signal-to-noise ratio for the I(c)th radar station. I(c),max The preset maximum signal-to-noise ratio for the I(c)th radar station;

[0108] (5.2.2) Based on the estimated signal-to-noise ratio Solve for the standard deviation of distance dimension measurement noise and the standard deviation of azimuth measurement noise

[0109]

[0110]

[0111] Where c = 3 × 10 8 m / s represents the speed of light, B I(c) Let I be the receiver bandwidth of the I(c)th radar station. The azimuth beamwidth of the I(c)th radar station is 3 dB.

[0112] (5.2.3) Based on the x-coordinate of the c-th measurement point c y-axis c Distance dimension measurement noise standard deviation and the standard deviation of azimuth measurement noise Calculate the possible areas where the target may exist.

[0113]

[0114] in: x represents the lower bound of the distance to the region where the target may exist. c =(x c ,y c ) T Let c be the coordinates of the measurement point. Let I(c) be the coordinates of the i-th radar station. Let x be the x-coordinate of the i(c)th radar station. Let be the ordinate of the I(c)th radar station. Indicates taking The 2-norm, β w For the scale factor of the associated region;

[0115] This indicates the upper bound of the distance to the region where the target may exist;

[0116] This indicates the lower bound of the azimuth angle of the area where the target may exist;

[0117] This indicates the upper azimuth bound of the area where the target may exist;

[0118] (5.3) Areas where the target may exist Discretize to obtain a set of discrete points.

[0119]

[0120] Where E r E a These are the discrete coefficients for the distance dimension and the orientation dimension, respectively;

[0121] (5.4) For any discrete point Construct its detection window after data registration in The steps include the following:

[0122] (5.4.1) Calculate discrete points Radial distance from other local radar stations and azimuth And order Represents the relationship between the e-th radar station and discrete points. The relative positions of the points, where e = 1, 2, ..., N, e ≠ I(c);

[0123] (5.4.2) Based on discrete points and relative position points Constructing a detection window after data registration

[0124]

[0125] in The detection window for the nth local radar station is expressed as:

[0126]

[0127] (5.5) Use a set to collect all detection windows Represented as:

[0128]

[0129] Step 6: The fusion center constructs a global test statistic based on the detection window and determines whether the target exists.

[0130] (6.1) For any detection window A global test statistic is constructed based on the generalized likelihood ratio criterion. The implementation steps include the following:

[0131] (6.1.1) Based on the detection window of the nth radar station and measurement point set Estimate the target's resolvable cell index at the nth local radar station

[0132]

[0133] in This includes all data transmitted from the nth radar station to the fusion center within the local detection windows. The set of measurement point indices within;

[0134] (6.1.2) Based on the estimated target's resolution cell index at the nth radar station and measurement point set Estimate the signal-to-noise ratio of the nth radar station.

[0135]

[0136] in For the first Local test statistics with low communication volume corresponding to each resolution unit;

[0137] (6.1.3) Based on the estimated target's resolution cell index at the nth radar station and signal-to-noise ratio Calculate the global test statistic

[0138]

[0139] in Indicates in the detection window The set of radar station indices with a greater than 0 local measurement points input internally. The Dirac function is expressed as:

[0140]

[0141] (6.2) Based on the given global false alarm probability P fa Offline creation of detection window dictionary Detection threshold dictionary γ and randomization probability dictionary κ:

[0142] (6.2.1) Set the maximum number of measurement points K within the detection window. max Enumerate the number of measurement points within the detection window of each radar station, given N radar stations, ranging from 1 to K. max The possible combinations of time, and record all possible combinations as a detection window dictionary.

[0143] (6.2.2) For any detection window Estimate the target's resolvable cell index at the nth radar station

[0144]

[0145] in This represents the number of samples in the dictionary. This represents the local detection window of the nth radar station. This includes all data transmitted from the nth radar station to the fusion center within the local detection windows. The set of measurement point indices within;

[0146] (6.2.3) Based on the estimated target's resolution cell index at the nth radar station Estimating signal-to-noise ratio

[0147]

[0148] in For the first Local test statistics with low communication volume corresponding to each resolution unit;

[0149] (6.2.4) Based on the estimated target's resolution cell index at the nth radar station and signal-to-noise ratio Calculate the global test statistic

[0150]

[0151] in Indicates in the detection window Internal input is the set of radar station indexes with a local measurement point count greater than 0;

[0152] (6.2.5) Under the assumption that the objective does not exist, the global test statistic is generated offline using Matlab software. of t represents a set of random numbers. l ,in Indicates to Round up;

[0153] (6.2.6) Take the generated random number t l A vector is a collection of unique random numbers sorted in ascending order. Set the loop count flag r = 1;

[0154] (6.2.7) Take The r-th element Determine t l Medium to large The number of elements R 1,l,r Sum equals The number of elements R 2,l,r Solve for t separately l Medium to large Element proportion Sum equals Element proportion

[0155] (6.2.8) Based on the given global false alarm probability P fa t l Medium to large The proportion of elements P 1,l,r Sum equals The proportion of elements P 2,l,r Determine if the loop has terminated and calculate the detection window. The corresponding detection threshold γ l and randomization probability κ l :

[0156] If P1,l,r >P fa Increment the value of the loop count marker r by 1, and return to step (6.2.7);

[0157] If P 1,l,r =P fa ,make Let κ l =0, proceed to step (6.2.9);

[0158] If P 1,l,r <P fa ,make make Perform step (6.2.9);

[0159] (6.2.9) Based on the calculated detection threshold γ l and randomization probability factor κ l The detection threshold dictionary γ and the randomization probability factor dictionary κ are obtained respectively:

[0160]

[0161]

[0162] (6.3) For any detection window Based on the global test statistic Detection Window Dictionary The detection threshold dictionary γ and the randomization probability factor dictionary κ are used to determine whether the target exists.

[0163] (6.3.1) Search Detection Window Corresponding to the detection window dictionary Index in

[0164] (6.3.2) The global test statistic With detection threshold Compare and determine if the target exists:

[0165] like Greater than the detection threshold Then the target is determined to exist;

[0166] like Less than the detection threshold Then the target is determined to not exist;

[0167] like Equal to detection threshold Then proceed to step (6.3.3);

[0168] (6.3.3) Generate a random number u that follows a uniform distribution from 0 to 1, and combine it with the randomization probability factor. Comparison:

[0169] like Then the target is determined to exist;

[0170] Otherwise, the target is determined to not exist.

[0171] The numbers used in the above steps are only for the purpose of more clearly illustrating the implementation scheme of the present invention, and their order is not limited.

[0172] The technical effects of the present invention will be further illustrated by the following simulation experiments.

[0173] I. Simulation conditions:

[0174] Suppose there are N = 2 radar stations, and the coordinates of the first radar station are (0,0). T The location coordinates of the second radar station are (25000, 0). T The search area is a 5km × 5km rectangle, with the horizontal coordinate ranging from -2500 to 2500m and the vertical coordinate ranging from 60000 to 65000m. The radar station has 32 reference elements, a receiver bandwidth of 1MHz, a 3dB azimuth beamwidth of 3°, 3 local quantization bits, and a range dimension dispersion coefficient E. r =3, azimuth dimension dispersion coefficient E a =3, the estimated minimum signal-to-noise ratio of the target is 0dB, and the estimated maximum signal-to-noise ratio of the target is 15dB. The global false alarm probability is P. fa =10 -4 Number of iterations N c =1000.

[0175] Hardware environment: CPU is Intel Xeon E5-2667 v4 with a clock speed of 3.20GHz, memory is 128GB, and operating system is 64-bit.

[0176] Software environment: Microsoft Windows 10 Professional Edition, MATLAB R2022b simulation software.

[0177] II. Simulation Content and Results

[0178] Simulation 1: Under the above conditions, the simulation of this invention is performed with an initial data throughput of 0.01 and a scale factor of β for the associated regions. w The curves showing the multi-station measurement registration probability as a function of signal-to-noise ratio (SNR) for values ​​of 1, 2, and 3 are presented, along with simulations of the multi-station target measurement input probability as a function of SNR. The results are as follows: Figure 3 .

[0179] from Figure 3It is evident that as the signal-to-noise ratio increases, the probability of multi-station measurement registration also increases; as the scale factor β of the associated region increases... w As the scale factor β of the associated region increases, the probability of multi-station measurement registration also increases. w When the value is 3, the multi-station measurement registration probability is approximately equal to the multi-station target measurement input probability, indicating that the present invention can achieve data registration well while greatly reducing the system communication burden, and can be applied in engineering.

[0180] Simulation 2: Under the above conditions, the simulation of this invention is performed with a target signal-to-noise ratio of 19.24 dB and a correlation region scale factor of β. w When the input measurement points are 1, 2, or 3, the curves showing the change in the number of detection windows with the initial data throughput are simulated, along with the curves showing the change in the number of input measurement points with the initial data throughput. Figure 4 As shown.

[0181] from Figure 4 It is evident that as the initial data throughput increases, the number of input measurement points and the number of detection windows also increase; as the scale factor β of the associated region increases... w As the scale factor β of the associated region increases, the number of detection windows also increases; w When the number of input measurement points is 3, comparing the number of input measurement points with the number of detection windows shows that the number of detection windows and the number of input measurement points are approximately linearly related, indicating that the computational complexity of this invention is approximately linear, which is beneficial for engineering implementation.

[0182] Simulation 3: Under the above conditions, the simulation shows the curve of the detection probability as a function of the signal-to-noise ratio when the local detection window includes 5 input measurement points. The results are as follows: Figure 5 As shown.

[0183] from Figure 5 It is evident that as the signal-to-noise ratio and initial data throughput increase, the target detection probability also increases. When the initial data throughput is 0.1, the signal-to-noise ratio loss caused by data compression is minimal. With a detection probability of 0.62 as a reference, the signal-to-noise ratio loss caused by data compression is less than 0.6 dB, indicating that the present invention can greatly reduce the system communication burden while still enabling the system to maintain good detection performance.

[0184] In summary, this invention can effectively complete data registration under extremely low communication and computational burdens, and maintain a good detection probability, which is beneficial for engineering implementation and application.

[0185] The above description is merely a specific example of the present invention and is not intended to limit the present invention in any way. Obviously, those skilled in the art, after understanding the content and principles of the present invention, may make various modifications and changes in form and detail without departing from the principles and structure of the present invention. However, these modifications and changes based on the ideas of the present invention are still within the scope of protection of the claims of the present invention.

Claims

1. A distributed data registration and fusion detection method based on low-traffic measurement, characterized in that, Including the following: (1) For a distributed radar system including a fusion center and N local radar stations, the original local test statistic l is constructed based on the echo signal of the nth radar station. n,ρ Where ρ = 1, 2, ..., C n Indicates the resolvable cell index, C n This represents the number of resolution units of the nth radar station, where n = 1, 2, ..., N; (2) Based on the system communication bandwidth requirements, set the initial data throughput α of the nth local radar station. n and the number of quantization intervals M n Or the number of quantization bits b n =log2M n Based on the distribution function of the original local test statistic, an initial threshold γ is designed. 1,n ; (3) Based on the given initial threshold γ 1,n and the number of quantization intervals M n Design quantization threshold τ n Quantization value v n ; (4) Construct local measurement points with low communication traffic and transmit them to the fusion center: (4a) Based on the initial threshold γ 1,n Quantization threshold τ n Quantization value v n and the original local test statistic l n,ρ Construct a local test statistic q with low communication volume n,ρ ; (4b) Each local radar station will exceed the initial threshold γ 1,n The distance and angle measurements of the corresponding resolution units are converted to the global rectangular coordinate system to obtain the measured abscissa and ordinate after coordinate transformation; (4c) The local radar station constructs a set of low-communication measurement points based on the measurement abscissa and ordinate after coordinate transformation, and the low-communication test statistics corresponding to the measurement points, and transmits it to the fusion center. (5) The fusion center constructs a detection window based on the received low-communication-volume measurement points: (5a) Use a set to collect the measurement points received by the fusion center. This indicates that, based on the c-th measurement point Construct the area where the target may exist in This indicates the number of measurement points in the input fusion center. The superscript I(c) indicates that the measurement point comes from the I(c)th radar station. (5b) Areas where the target may exist Discretize to obtain a set of discrete points. (5c) For any discrete point Construct its detection window after data registration in (5d) Use a set to collect all detection windows Represented as: (6) The fusion center constructs a global test statistic based on the detection window and determines whether the target exists: (6a) For any detection window A global test statistic is constructed based on the generalized likelihood ratio criterion. (6b) Based on the given global false alarm probability P fa Offline creation of detection window dictionary Detection threshold dictionary γ and randomization probability factor dictionary κ; (6c) For any detection window Based on the global test statistic Detection Window Dictionary The detection threshold dictionary γ and the randomization probability factor dictionary κ are used to determine whether the target exists. (6c1) Search Detection Window Corresponding to the detection window dictionary Index in (6c2) The global test statistic With detection threshold Compare and determine if the target exists: like Greater than the detection threshold Then the target is determined to exist; like Less than the detection threshold Then the target is determined to not exist; like Equal to detection threshold Then proceed to step (6c3); (6c3) Generate a random number u that follows a uniform distribution from 0 to 1, and combine it with the randomization probability factor. Comparison: like Then the target is determined to exist; Otherwise, the target is determined to not exist.

2. The method according to claim 1, characterized in that, The original local test statistic l constructed in step (1) n,ρ Its expression is: Where e n,ρ This represents the echo signal of the ρ-th resolution unit of the nth local radar station. This is a set of reference cell indices containing the ρ-th resolution cell of the nth radar station. Let n be the number of reference elements for the nth radar station. c represents the index of the k-th reference unit. n,k This represents the echo signal of the k-th reference cell, which has a mean of 0 and a variance of . The complex Gaussian distribution.

3. The method according to claim 1, characterized in that, The initial threshold γ designed in step (2) 1,n Its formula is: Where n = 1, 2, ..., N represents the radar station index, and K... n This represents the number of reference units for the nth radar station.

4. The method according to claim 1, characterized in that, In step (3), the initial threshold γ is given. 1,n and the number of quantization intervals M n Design quantization threshold τ n Quantization value v n Its expression is as follows: in, T Indicates transpose, τ n,i Let i be the quantization threshold for the nth radar station, where i = 1, 2, ..., M n -1, v n,j This is the j-th quantization value of the n-th radar station, where j = 1, 2, ..., M n .

5. The method according to claim 1, characterized in that, In step (4a), based on the initial threshold γ 1,n Quantization threshold τ n Quantization value v n and the original local test statistic l n,ρ Construct a local test statistic q with low communication volume n,ρ The calculation formula is as follows: Where n = 1, 2, ..., N represents the radar station index, τ n,m v is the m-th quantization threshold for the n-th radar station. n,m This is the m-th quantization value of the n-th radar station, where m = 1, 2, ..., M n .

6. The method according to claim 1, characterized in that, In step (4b), the measured abscissa and ordinate after coordinate transformation are obtained, as shown below: Where: x n,ζ y n,ζ These are the coordinate transformation of the distance and angle measurements of the ζ-th resolution unit at the n-th radar station, representing the abscissa and ordinate of the measurements, respectively. This indicates that the original test statistic exceeds the initial threshold γ. 1,n The index of measurement points; This represents the angle between the polar axis of the nth radar station and the x-axis of the global rectangular coordinate system. This represents the x-coordinate of the nth radar station in the global Cartesian coordinate system. This represents the ordinate of the nth radar station in the global rectangular coordinate system; r n,ζ T n,ζ The distance and angle measurements for the nth radar station in the ζth resolution cell are respectively expressed as: Where: ρ n This represents the resolution cell index of the target at the nth radar station. This indicates that the objective is based on an assumption; The radial distance of the target relative to radar station n, Let n be the azimuth angle of the target relative to radar station n; It is distance-dependent additive Gaussian noise with a mean of 0 and a standard deviation of . Gaussian distribution, It is azimuth additive Gaussian noise with a mean of 0 and a standard deviation of . Gaussian distribution; U(r min ,r max ) indicates that the range of values ​​is from r min to r max The uniform distribution of U(φ) min ,φ max ) indicates the range of values ​​from φ min To φ max The uniform distribution of r min r represents the minimum distance measurement value. max φ represents the maximum distance measurement value. min φ represents the minimum value of the azimuth measurement. max This indicates the maximum value of the azimuth measurement.

7. The method according to claim 1, characterized in that, In step (4c), the local radar station constructs a set of measurement points with low communication throughput based on the measured abscissa and ordinate after coordinate transformation, and the low communication throughput test statistic corresponding to the measurement point. The formula is as follows: in T Indicates transpose. Let x be the set of low-communication measurement points for the nth radar station. n,ζ y n,ζ These are the x-coordinate and y-coordinate of the distance and angle measurements after coordinate transformation, respectively, for the ζ-th resolution unit. n,ζ Let be the low-communication local test statistic corresponding to the ζ-th resolution unit of the n-th radar station. This indicates that the original test statistic exceeds the initial threshold γ. 1,n The index of measurement points.

8. The method according to claim 1, characterized in that, In step (5a), based on the c-th measurement point Construct the area where the target may exist The implementation steps include the following: (5a1) Based on the low-communication local test statistic q of the c-th measurement point c Estimating signal-to-noise ratio in: T Indicates transpose, τ I(c),m Let v represent the m-th quantization threshold of the I(c)-th radar station. I(c),m Let m be the quantization value of the i-th radar station (c), where m = 1, 2, ..., M. I(c) ; M I(c) K represents the number of quantization intervals for the I(c)th radar station. I(c) γ represents the number of reference elements for the I(c)th radar station. 1,I(c) This is the initial threshold for the I(c)th radar station; λ I(c),min λ is the preset minimum signal-to-noise ratio for the I(c)th radar station. I(c),max The preset maximum signal-to-noise ratio for the I(c)th radar station; (5a2) Based on the estimated signal-to-noise ratio Solve for the standard deviation of distance dimension measurement noise and the standard deviation of azimuth measurement noise Where c = 3 × 10 8 m / s represents the speed of light, B I(c) Let I be the receiver bandwidth of the I(c)th radar station. The azimuth beamwidth of the I(c)th radar station is 3 dB; (5a3) Based on the x-coordinate of the c-th measurement point c y-axis c Distance dimension measurement noise standard deviation and the standard deviation of azimuth measurement noise Calculate the possible area where the target may exist. in: x represents the lower bound of the distance to the region where the target may exist. c =(x c ,y c ) T Let c be the coordinates of the measurement point. Let I(c) be the coordinates of the i-th radar station. Let x be the x-coordinate of the i(c)th radar station. The ordinate of the I(c)th radar station, Indicates taking The 2-norm, β w For the scale factor of the associated region; This indicates the upper bound of the distance to the region where the target may exist; This indicates the lower bound of the azimuth angle of the area where the target may exist; This indicates the upper azimuth bound of the area where the target may exist.

9. The method according to claim 1, characterized in that, In step (5b), the area where the target may exist is... Discretization yields a set of discrete points. It is expressed as follows: in T Indicates transpose, r 1,c express The target may have a lower bound on the distance to the region, r 2,c express The target may exist in an area with an upper bound on the distance φ. 1,c express The target may exist in a region with a lower azimuth limit, φ 2,c express The target in the middle may have an upper azimuth bound in the region, E r E a These are the distance dimension and the orientation dimension discrete coefficients, respectively.

10. The method according to claim 1, characterized in that, In step (5c), for any discrete point Construct its detection window after data registration The steps include the following: (5c1) Based on the measurement points Estimate the signal-to-noise ratio (5c2) Based on the estimated signal-to-noise ratio Solve for the standard deviation of the distance dimension measurement noise. and the standard deviation of azimuth measurement noise (5c3) Calculate discrete points Radial distance from other local radar stations and azimuth And order Represents the relationship between the e-th radar station and discrete points. The relative position points, where T represents the transpose, e = 1, 2, ..., N, e ≠ I(c); (5c4) Based on discrete points and relative position points Constructing a detection window after data registration in The detection window for the nth local radar station is expressed as: Where β w This is the scale factor for the associated region.

11. The method according to claim 1, characterized in that, In step (6a), any detection window A global test statistic is constructed based on the generalized likelihood ratio criterion. The implementation steps include the following: (6a1) Based on the detection window of the nth radar station and measurement point set Estimate the target's resolvable cell index at the nth local radar station in This includes all data transmitted from the nth radar station to the fusion center within the local detection windows. The set of measurement point indices within; (6a2) Based on the estimated target's resolution cell index at the nth radar station and measurement point set Estimate the signal-to-noise ratio of the nth radar station. in: For the first Local test statistics with low communication volume corresponding to each resolution unit; τ n,m v represents the m-th quantization threshold of the n-th radar station. n,m This is the m-th quantization value of the n-th radar station, where m = 1, 2, ..., M. n ; M n K represents the number of quantization intervals for the nth radar station. n The number of reference elements for the nth radar station; λ n,min λ is the preset minimum signal-to-noise ratio for the nth radar station. n,max The preset maximum signal-to-noise ratio for the nth radar station; (6a3) Based on the estimated target's resolution cell index at the nth radar station and signal-to-noise ratio Calculate the global test statistic in Indicates in the detection window The set of radar station indices with a greater than 0 local measurement points input internally. The Dirac function is expressed as:

12. The method according to claim 1, characterized in that, In step (6b), the given global false alarm probability P is used... fa Offline creation of detection window dictionary The detection threshold dictionary γ and the randomization probability factor dictionary κ are implemented through the following steps: (6b1) Set the maximum number of measurement points K in the detection window max Enumerate the number of measurement points within the detection window of each radar station, given N radar stations, ranging from 1 to K. max The possible combinations of time, and record all possible combinations as a detection window dictionary. (6b2) For any detection window Calculate the global test statistic in: This represents the number of samples in the dictionary. This represents the local detection window of the nth radar station; Indicates in the detection window Internal input is the set of radar station indexes with a local measurement point count greater than 0; The index of the target's resolvable cell at the nth local radar station is given. The signal-to-noise ratio of the nth radar station is estimated. The Dirac function is expressed as: (6b3) Using Matlab software to generate global test statistics offline under the assumption that the objective does not exist. of t represents a set of random numbers. l ,in Indicates to Round up; (6b4) Take the generated random number t l A vector is a collection of unique random numbers sorted in ascending order. Set the loop count flag r = 1; (6b5) take The r-th element Determine t l Medium to large The number of elements R 1,l,r Sum equals The number of elements R 2,l,r Solve for t separately l Medium to large Element proportion Sum equals Element proportion (6b6) Based on the given global false alarm probability P fa t l Medium to large The proportion of elements P 1,l,r Sum equals The proportion of elements P 2,l,r Determine if the loop has terminated and calculate the detection window. The corresponding detection threshold γ l and randomization probability factor κ l : If P 1,l,r >P fa Increment the value of the loop count marker r by 1 and return (6b5); If P 1,l,r =P fa ,make Let κ l =0, execute (6b7); If P 1,l,r <P fa ,make make Execute (6b7); (6b7) Based on the calculated detection threshold γ l and randomization probability factor κ l The detection threshold dictionary γ and the randomization probability factor dictionary κ are obtained respectively: in T This indicates transpose.

Citation Information

Patent Citations

  • Method and device for compressing and expanding image intended for encrypted communication, program, and storage medium

    CN105830134A

  • Scheduling the same resource in radio access networks

    CN106797641A