WO2020000123A1 - 一种共轭梯度空时自适应处理方法及系统 - Google Patents

一种共轭梯度空时自适应处理方法及系统 Download PDF

Info

Publication number
WO2020000123A1
WO2020000123A1 PCT/CN2018/092550 CN2018092550W WO2020000123A1 WO 2020000123 A1 WO2020000123 A1 WO 2020000123A1 CN 2018092550 W CN2018092550 W CN 2018092550W WO 2020000123 A1 WO2020000123 A1 WO 2020000123A1
Authority
WO
WIPO (PCT)
Prior art keywords
space
conjugate gradient
time
time adaptive
snapshot
Prior art date
Application number
PCT/CN2018/092550
Other languages
English (en)
French (fr)
Inventor
阳召成
汪小叶
何凯旋
刘海帆
黄建军
Original Assignee
深圳大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳大学 filed Critical 深圳大学
Priority to PCT/CN2018/092550 priority Critical patent/WO2020000123A1/zh
Publication of WO2020000123A1 publication Critical patent/WO2020000123A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01SRADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section

Definitions

  • the invention belongs to the technical field of radar signal processing, and in particular relates to a conjugate gradient space-time adaptive processing method and system based on the transform domain L1 norm.
  • STAP Space-Time Adaptive Processing
  • the technical problem to be solved by the present invention is to provide a conjugate gradient space-time adaptive processing method and system based on the L1 norm in the transform domain, which aims to solve the existing technology and significantly improve the convergence performance of the algorithm when snapshot support is limited. , But the problem of high computational complexity.
  • the present invention is implemented in such a way that a conjugate gradient space-time adaptive processing method based on the transform domain L1 norm includes:
  • Step A Determine the input signal of the conjugate gradient space-time adaptive filter based on the L1 norm of the transform domain according to the target space-time steering vector and the interference plus noise component in the distance unit to be detected;
  • Step B establishing an optimization problem model of the conjugate gradient space-time adaptive filter according to the clutter sparse characteristics of the distance unit to be detected;
  • step C a conjugate gradient algorithm based on the L1 norm in the transform domain is used to recursively calculate the optimization problem model to obtain a weight vector of the conjugate gradient space-time adaptive filter;
  • Step D Adaptively process an input signal of the conjugate gradient space-time adaptive filter according to the weight vector.
  • step A the space-time snapshot of the distance unit to be detected is represented by x, then:
  • ⁇ t is a complex number
  • v t represents the target space-time steering vector
  • x u represents the interference plus noise component
  • x u is composed of interference x j , clutter x c and thermal noise x n .
  • the conjugate gradient space-time adaptive filter is expressed as the following optimization problem, namely:
  • step C includes:
  • R represents the interference plus noise covariance matrix
  • R k represents the interference plus noise covariance matrix at the k-th snapshot
  • ⁇ k represents the output power of the balanced filter w H Rw and the angle Doppler at the k-th snapshot.
  • the sparseness regular parameter of the Lele pattern, ⁇ k > 0, k represents the index of the k-th space-time snapshot, and sk represents the value obtained under the k-th snapshot.
  • Step C2 let Calculate sk using the conjugate gradient method based on the L1 norm of the transform domain:
  • sk (l) sk (l-1) + ⁇ k (l) p k (l);
  • ⁇ k (l) and p k (l) represent the adaptive step size and direction vector of the l-th iteration in the k-th space-time snapshot, respectively, and sk (l) represents the The value calculated by the first iteration;
  • Step C3 iteratively calculate sk (l) according to the preset total number of iterations L:
  • ⁇ k (w k ) H R (w k ) / ⁇ -
  • p k (l) r k (l-1) + ⁇ k (l-1) p k (l-1);
  • r k (l) r k (l-1)- ⁇ k (l) B k p k (l);
  • Step C4 Substituting sk (L) obtained by the iterative calculation into (1) to obtain k weight vectors of the conjugate gradient space-time adaptive filter under the space-time snapshot:
  • step C5 it is assumed that the total number of snapshots is K.
  • the above steps C1, C2, C3, and C4 are repeated to obtain the weight vector of the conjugate gradient space-time adaptive filter.
  • An embodiment of the present invention also provides a conjugate gradient space-time adaptive processing system based on a transform domain L1 norm, including:
  • a signal model establishing unit configured to determine an input signal of a conjugate gradient space-time adaptive filter based on a L1 norm in a transform domain according to a target space-time steering vector and interference plus noise components in a distance unit to be detected;
  • a filter optimization problem model establishing unit configured to establish an optimization problem model of the conjugate gradient space-time adaptive filter according to clutter sparse characteristics of the distance unit to be detected;
  • a weight vector calculation unit configured to recursively and iteratively calculate the optimization problem model by using a conjugate gradient algorithm based on a transform domain L1 norm to obtain a weight vector of the conjugate gradient space-time adaptive filter;
  • a signal processing unit is configured to adaptively process an input signal of the conjugate gradient space-time adaptive filter according to the weight vector.
  • ⁇ t is a complex number
  • v t represents the target space-time steering vector
  • x u represents the interference plus noise component
  • x u is composed of interference x j , clutter x c and thermal noise x n .
  • conjugate gradient space-time adaptive filter is expressed as the following optimization problem, namely:
  • 1 and H represent the l 1 norm and the conjugate transpose operator, respectively,
  • 1 and H represent the l 1 norm and the conjugate transpose operator, respectively,
  • weight vector calculation unit is specifically configured to:
  • R represents the interference plus noise covariance matrix
  • R k represents the interference plus noise covariance matrix at the k-th snapshot
  • ⁇ k represents the output power of the balanced filter w H Rw and the angle Doppler at the k-th snapshot.
  • the sparseness regular parameter of the Lele pattern, ⁇ k > 0, k represents the index of the k-th space-time snapshot, and sk represents the value obtained under the k-th snapshot.
  • sk (l) sk (l-1) + ⁇ k (l) p k (l);
  • ⁇ k (l) and p k (l) represent the adaptive step size and direction vector of the l-th iteration in the k-th space-time snapshot, respectively, and sk (l) represents the The value calculated by the first iteration;
  • ⁇ k (w k ) H R (w k ) / ( ⁇ -
  • p k (l) r k (l-1) + ⁇ k (l-1) p k (l-1);
  • r k (l) r k (l-1)- ⁇ k (l) B k p k (l);
  • the present invention has the beneficial effects that, according to the embodiment of the present invention, according to the target space-time steering vector and clutter interference component of the distance unit to be detected, a conjugate gradient space-time adaptive filter based on the transform domain L1 norm Input signal, based on the clutter sparse characteristics of the distance unit to be detected, the optimization problem model of the conjugate gradient space-time adaptive filter is established.
  • the conjugate gradient algorithm of the L1 norm in the transform domain is used to recursively calculate the optimization problem model.
  • To obtain the weight vector of the conjugate gradient space-time adaptive filter and adaptively process the input signal of the wave filter according to the weight vector.
  • An embodiment of the present invention proposes a conjugate gradient adaptive processing method based on the L1 norm in the low-complexity transform domain, which can iteratively calculate the space-time self-contained conjugate gradient based on the L1 norm in the transform domain under some simple assumptions. Adapting the filter's weight vector without inverting the covariance matrix reduces the computational complexity by about 10 times, while maintaining good performance.
  • FIG. 1 is a flowchart of a conjugate gradient space-time adaptive processing method based on a transform domain L1 norm according to an embodiment of the present invention
  • FIG. 2 is a schematic structural diagram of a conjugate gradient space-time adaptive processing system based on a transform domain L1 norm according to an embodiment of the present invention
  • 3 is a schematic diagram showing the relationship between the computational complexity of DoFs and the space-time freedom of the system according to an embodiment of the present invention
  • FIG. 4 is a schematic diagram showing the relationship between the drying ratio of the output signal and the number of space-time snapshots provided by an embodiment of the present invention
  • FIG. 5 is a schematic diagram showing a relationship between an output signal drying ratio and a normalized Doppler frequency according to an embodiment of the present invention
  • FIG. 6 is a schematic diagram of a relationship between a detection probability P d and a target signal-to-noise ratio SNR according to an embodiment of the present invention.
  • FIG. 1 shows a conjugate gradient space-time adaptive processing method based on the transform domain L1 norm according to an embodiment of the present invention, including:
  • S101 Determine an input signal of a conjugate gradient space-time adaptive filter based on a L1 norm in a transform domain according to a target space-time steering vector and an interference plus noise component in a distance unit to be detected;
  • an implementation method of conjugate gradient space-time adaptive based on the L1 norm of the transformation domain is proposed for the STAP algorithm with sparse constraints, which is subject to the low complexity of the traditional conjugate gradient algorithm CG (Conjugate Gradient Methods). Inspiration, the embodiment of the present invention first defines a new weight vector, and then derives an iterative algorithm for calculating a new weight vector under some simple assumptions of the CG algorithm based on the L1 norm of the transform domain. Finally, the proposed calculation of the STAP weight vector is obtained without the low complexity of the matrix inverse matrix. Under some simple assumptions, the embodiment of the present invention derives a conjugate gradient space-time adaptive processing method based on the L1 norm of the transform domain, which has low calculation complexity and avoids inversion of the covariance matrix.
  • Step 1 Establish a signal model:
  • x u is an interference plus noise vector consisting of interference x j , clutter x c and thermal noise x n .
  • Step 2 Considering the sparseness of the angular Doppler pattern of the clutter within the distance unit to be detected, the STAP filter (SC-STAP) under sparse constraints is usually designed as:
  • 1 and H represent the l 1 norm and the conjugate transpose operator, respectively,
  • 1 and H represent the l 1 norm and the conjugate transpose operator, respectively,
  • Step 3 Transform the CG algorithm based on the transform domain L1 norm to recursively calculate the weight vector w in formula (2):
  • ⁇ k (l) and p k (l) are the adaptive step size and corresponding direction vector of the l-th iteration in the k-th space-time snapshot, respectively.
  • the direction vector p k (l) is defined as:
  • r k (l-1) represents the residual vector and is defined as:
  • the k k orthogonality of k p k (l-1) and vector p k (l) is:
  • r k (l) r k (l-1) - ⁇ k (l) B k p k (l) (12)
  • FIG. 2 shows an adaptive processing system based on a conjugate gradient provided by an embodiment of the present invention, including:
  • a signal model establishing unit 201 is configured to determine an input signal of a conjugate gradient space-time adaptive filter based on a L1 norm in a transform domain according to a target space-time steering vector and an interference plus noise component in a distance unit to be detected;
  • a filter optimization problem model establishing unit 202 is configured to establish an optimization problem model of the conjugate gradient space-time adaptive filter according to the clutter sparse characteristics of the distance unit to be detected;
  • a weight vector calculation unit 203 configured to recursively and iteratively calculate the optimization problem model by using a conjugate gradient algorithm based on a transform domain L1 norm to obtain a weight vector of the conjugate gradient space-time adaptive filter;
  • the signal processing unit 204 is configured to perform adaptive processing on an input signal of the conjugate gradient space-time adaptive filter according to the weight vector.
  • ⁇ t is a complex number
  • v t represents the target space-time steering vector
  • x u represents the interference plus noise component
  • x u is composed of interference x j , clutter x c and thermal noise x n .
  • conjugate gradient space-time adaptive filter is expressed as the following optimization problem, namely:
  • weight vector calculation unit 203 is specifically configured to:
  • R represents the interference plus noise covariance matrix
  • R k represents the interference plus noise covariance matrix at the k-th snapshot
  • ⁇ k represents the output power of the balanced filter w H Rw and the angle Doppler at the k-th snapshot.
  • the sparseness regular parameter of the Lele pattern, ⁇ k > 0, k represents the index of the k-th space-time snapshot, and sk represents the value obtained under the k-th snapshot.
  • sk (l) sk (l-1) + ⁇ k (l) p k (l);
  • ⁇ k (l) and p k (l) represent the adaptive step size and direction vector of the l-th iteration in the k-th space-time snapshot, respectively, and sk (l) represents the The value calculated by the first iteration;
  • ⁇ k (w k ) H R (w k ) / ( ⁇ -
  • p k (l) r k (l-1) + ⁇ k (l-1) p k (l-1);
  • r k (l) r k (l-1)- ⁇ k (l) B k p k (l);
  • CG-SC-STAP is used to describe the adaptive processing method based on conjugate gradient provided by the embodiment of the present invention.
  • the following describes the performance of the embodiment of the present invention in terms of algorithm complexity and performance of the CG-SC-STAP algorithm through simulation data Beneficial effect.
  • the complexity of the proposed algorithm is O (D (MN) 2 ) (including D (2 (MN) 2 + 5MN + 3) multiplication and 2 (MN) 2 + 4MN-3) addition, each space-time snapshot has D iterations. Based on numerical simulations, it is found that the proposed algorithm converges every sampling iteration, and the number of iterations is related to the number of pulses M and the antenna element N. Therefore, compared with an SC-STAP algorithm with a complexity of O ((MN) 3 ) (including (MN) 3 +2 (MN) 2 + 2MN multiplication and (MN) 3 -1 addition), an embodiment of the present invention proposes The algorithm provides lower complexity.
  • MN space-time degree of freedom
  • the embodiment of the present invention provides simulation results to illustrate the performance of the CG-SC-STAP algorithm, and compares its performance with sparse sensing beamformer, space-time multi-beam (STMB), local area joint processing (JDL), and principal component (PC) Compare methods.
  • STMB space-time multi-beam
  • JDL local area joint processing
  • PC principal component Compare methods.
  • ⁇ m and ⁇ m are zero-mean Gaussian random variables of 0.005 and 0.005 * ⁇ / 2, respectively, and the rank of the PC is 50;
  • FIG. 4 shows the SINR loss versus The relationship between the number of snapshots. It has been observed that CG-SC-STAP shows comparable performance to the SC-STAP algorithm. The SINR loss at the normalized Doppler frequency is shown in Figure 5. The number of snapshots is set to 50. It can be seen that when the target normalized Doppler frequency is less than 0.15, CG-SC-STAP has similar SINR performance as SC-STAP and outperforms all other algorithms with lower SINR than SC-STAP.
  • SINR signal-to-interference and noise ratio
  • An embodiment of the present invention proposes a sparsely constrained low complexity space-time adaptive processing method based on conjugate gradient technology. Under some simple assumptions, iteratively calculates a CG-based weight vector without using a covariance matrix to invert. Reduce the computational complexity by about 10 times, while maintaining good performance.
  • the embodiments of the present invention can be applied to clutter suppression and moving target detection in an airborne radar system, reduce the computational complexity, and improve the clutter suppression level and target detection capability of the radar system.
  • the CG-SC-STAP provided by the embodiment of the present invention is used in the field of radar signal processing, and a low-complexity space-time adaptive processing method based on the conjugate gradient technology is proposed.
  • the embodiment of the present invention based on some simple assumptions, iteratively calculates a CG-based weight vector without inverse of a covariance matrix, and derives a weight vector for calculating an adaptation.
  • the embodiment of the present invention has the performance equivalent to the iterative least squares sparse constraint STAP (SC-STAP) with the support of limited samples, but the calculation complexity is relatively low.
  • An embodiment of the present invention further provides a terminal, including a memory, a processor, and a computer program stored on the memory and running on the processor.
  • the processor executes the computer program, the L1 norm based on the transformation domain shown in FIG. 1 is implemented.
  • Each step in the conjugate gradient space-time adaptive processing method of numbers is implemented.
  • An embodiment of the present invention also provides a readable storage medium having a computer program stored thereon.
  • the computer program is executed by a processor, the conjugate gradient space-time based on the transformation domain L1 norm shown in FIG. 1 is implemented. Steps in adaptive processing.
  • each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist separately physically, or two or more modules may be integrated into one module.
  • the above integrated modules can be implemented in the form of hardware or software functional modules.
  • the integrated module When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium.
  • the technical solution of the present invention essentially or part that contributes to the existing technology or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium Including a plurality of instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention.
  • the foregoing storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disks or optical disks and other media that can store program codes .

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

本发明适用于雷达信号处理领域,提供共轭梯度空时自适应处理方法,包括:根据待检测距离单元内目标空时导向向量和干扰加噪音分量,确定基于变换域L1范数的共轭梯度空时自适应滤波器的输入信号;根据待检测距离单元的杂波稀疏特性,建立共轭梯度空时自适应滤波器的优化问题模型;采用基于变换域L1范数的共轭梯度算法,递归迭代地计算优化问题模型,得到共轭梯度空时自适应滤波器的权向量;根据权向量对该输入信号进行自适应处理。本发明实施例提出了一种低复杂度的基于变换域L1范数的共轭梯度空时自适应处理方法,能够在一些简单假设条件下,迭代地计算滤波器的权向量而不用协方差矩阵求逆,降低了10倍左右计算的复杂度,同时保持良好的性能。

Description

一种共轭梯度空时自适应处理方法及系统 技术领域
本发明属于雷达信号处理技术领域,尤其涉及一种基于变换域L1范数的共轭梯度空时自适应处理方法及系统。
背景技术
空时自适应处理(STAP,Space-Time Adaptive Processor)是机载雷达系统中杂波抑制和动目标检测的经典技术。对STAP技术的研究一直紧紧围绕着如何在保持STAP算法的杂波抑制性能的同时,加快低训练快拍下的收敛性和降低系统计算的复杂性是STAP技术的关键问题。
在STAP的发展历程中,为了解决STAP技术的关键问题,先后提出了三种主要类型的算法,即降维和降秩STAP技术,基于先验信息的STAP技术,基于稀疏度的STAP技术。这三种算法在STAP技术的发展中是按时间先后顺序提出的。首先,降维和降秩STAP技术是采用降低系统自由度来加快低训练快拍下的收敛性,之后基于先验信息的STAP技术是他们通过将一些先验知识结合到STAP设计中来加快低快拍下的收敛性。最后对于基于稀疏度的STAP技术是利用问题的稀疏本质,如杂波或权向量等的稀疏性,以加速STAP方法在低快拍下的收敛性。
虽然这些方法在快拍支持有限的情况下显着提高了滤波器杂波抑制和目标检测的性能,但高计算复杂度仍然是一个问题。
发明内容
本发明所要解决的技术问题在于提供一种基于变换域L1范数的共轭梯度空时自适应处理方法及系统,旨在解决现有技术在快拍支持有限的情况下显著 提高算法的收敛性能,但计算复杂度高的问题。
本发明是这样实现的,一种基于变换域L1范数的共轭梯度空时自适应处理方法,包括:
步骤A,根据待检测距离单元内目标空时导向向量和干扰加噪音分量,确定基于变换域L1范数的共轭梯度空时自适应滤波器的输入信号;
步骤B,根据所述待检测距离单元的杂波稀疏特性,建立所述共轭梯度空时自适应滤波器的优化问题模型;
步骤C,采用基于变换域L1范数的共轭梯度算法,递归迭代地计算所述优化问题模型,得到所述共轭梯度空时自适应滤波器的权向量;
步骤D,根据所述权向量对所述共轭梯度空时自适应滤波器的输入信号进行自适应处理。
进一步地,在所述步骤A中,所述待检测距离单元的空时快拍以x表示,则有:
x=α tv t+x u
其中,其中α t为复数,v t表示所述目标空时导向向量,x u表示所述干扰加噪音分量,
Figure PCTCN2018092550-appb-000001
Figure PCTCN2018092550-appb-000002
表示多普勒频率,
Figure PCTCN2018092550-appb-000003
表示空间频率,
Figure PCTCN2018092550-appb-000004
表示克罗内克积,x u由干扰x j、杂波x c和热噪声x n组成。
进一步地,在所述步骤B中,所述共轭梯度空时自适应滤波器表示为以下优化问题,即:
Figure PCTCN2018092550-appb-000005
其中,||·|| 1、H分别表示l 1范数和共轭转置算子,
Figure PCTCN2018092550-appb-000006
表示干扰加噪声协方差矩阵,E[·]表示随机变量的期望值,
Figure PCTCN2018092550-appb-000007
表示具有范围在-0.5到0.5的多普勒频率的f d,j(j=1,...,N d)和具有空间频率f s,i(i=1,...,N s)的空时导向矩阵,N d表示多普勒频率总个数,N s表示空间频率总个数,λ表示平衡滤波器输出功率w HRw和角度多普勒方向图的稀疏性的正则参数, λ=w HRw/(ρ-||w HQ|| 1)。
进一步地,所述步骤C包括:
步骤C1,令s k=(R kkQ∏(w k)Q H) -1v t,求得所述共轭梯度空时自适应滤波器的权向量w k
Figure PCTCN2018092550-appb-000008
其中,R表示干扰加噪声协方差矩阵,R k表示第k个快拍下的干扰加噪声协方差矩阵,λ k表示第k个快拍下的平衡滤波器输出功率w HRw和角度多普勒方向图的稀疏性的正则参数,λ k>0,k表示第k个空时快拍的索引,s k表示第k个快拍下所求得的值,
Figure PCTCN2018092550-appb-000009
步骤C2,令
Figure PCTCN2018092550-appb-000010
采用基于变换域L1范数的共轭梯度法计算s k
s k(l)=s k(l-1)+μ k(l)p k(l);
μ k(l)和p k(l)分别表示第k个空时快拍中的第l次迭代的自适应步长和方向向量,s k(l)表示第k次空时快拍中的第l次迭代计算得到的值;
步骤C3,根据预置的迭代总次数L,迭代地计算s k(l):
B k=R kkQ∏(w k)Q H
Figure PCTCN2018092550-appb-000011
λ k=(w k) HR(w k)/ρ-||(ρ-||(w k) HQ|| 1);
Figure PCTCN2018092550-appb-000012
Figure PCTCN2018092550-appb-000013
Figure PCTCN2018092550-appb-000014
p k(l)=r k(l-1)+τ k(l-1)p k(l-1);
Figure PCTCN2018092550-appb-000015
r k(l)=r k(l-1)-μ k(l)B kp k(l);
步骤C4,将迭代计算得到的s k(L)代入(1),得到k个空时快拍下所述共轭梯度空时自适应滤波器的权向量:
Figure PCTCN2018092550-appb-000016
步骤C5,假定总快拍数为K,对每一个快拍,重复上述步骤C1、C2、C3和C4,即得到所述共轭梯度空时自适应滤波器的权向量。
本发明实施例还提供了一种基于变换域L1范数的共轭梯度空时自适应处理系统,包括:
信号模型建立单元,用于根据待检测距离单元内目标空时导向向量和干扰加噪音分量,确定基于变换域L1范数的共轭梯度空时自适应滤波器的输入信号;
滤波器优化问题模型建立单元,用于根据所述待检测距离单元的杂波稀疏特性,建立所述共轭梯度空时自适应滤波器的优化问题模型;
权向量计算单元,用于采用基于变换域L1范数的共轭梯度算法,递归迭代地计算所述优化问题模型,得到所述共轭梯度空时自适应滤波器的权向量;
信号处理单元,用于根据所述权向量对所述共轭梯度空时自适应滤波器的输入信号进行自适应处理。
进一步地,所述待检测距离单元的空时快拍以x表示,则有:
x=α tv t+x u
其中,其中α t为复数,v t表示所述目标空时导向向量,x u表示所述干扰加噪音分量,
Figure PCTCN2018092550-appb-000017
Figure PCTCN2018092550-appb-000018
表示多普勒频率,
Figure PCTCN2018092550-appb-000019
表示空间频率,
Figure PCTCN2018092550-appb-000020
表示克罗内克积,x u由干扰x j、杂波x c和热噪声x n组成。
进一步地,所述共轭梯度空时自适应滤波器表示为以下优化问题,即:
Figure PCTCN2018092550-appb-000021
其中,||·|| 1、H分别表示l 1范数和共轭转置算子,
Figure PCTCN2018092550-appb-000022
表示干扰加噪声协方差矩阵,E[·]表示随机变量的期望值,
Figure PCTCN2018092550-appb-000023
表示具 有范围在-0.5到0.5的多普勒频率的f d,j(j=1,...,N d)和具有空间频率f s,i(i=1,...,N s)的空时导向矩阵,λ表示平衡滤波器功率w HRw和角度多普勒方向图的稀疏性的正则参数,λ=w HRw/(ρ-||w HQ|| 1)。
进一步地,所述权向量计算单元具体用于:
首先,令s k=(R kkQ∏(w k)Q H) -1v t,求得所述共轭梯度空时自适应滤波器的权向量w k
Figure PCTCN2018092550-appb-000024
其中,R表示干扰加噪声协方差矩阵,R k表示第k个快拍下的干扰加噪声协方差矩阵,λ k表示第k个快拍下的平衡滤波器输出功率w HRw和角度多普勒方向图的稀疏性的正则参数,λ k>0,k表示第k个空时快拍的索引,s k表示第k个快拍下所求得的值,
Figure PCTCN2018092550-appb-000025
其次,令
Figure PCTCN2018092550-appb-000026
采用基于变换域L1范数的共轭梯度法计算sk:
s k(l)=s k(l-1)+μ k(l)p k(l);
μ k(l)和p k(l)分别表示第k个空时快拍中的第l次迭代的自适应步长和方向向量,s k(l)表示第k次空时快拍中的第l次迭代计算得到的值;
然后,根据预置的迭代总次数L,迭代地计算s k(l):
B k=R kkQ∏(w k)Q H
Figure PCTCN2018092550-appb-000027
λ k=(w k) HR(w k)/(ρ-||(w k) HQ|| 1);
Figure PCTCN2018092550-appb-000028
Figure PCTCN2018092550-appb-000029
Figure PCTCN2018092550-appb-000030
p k(l)=r k(l-1)+τ k(l-1)p k(l-1);
Figure PCTCN2018092550-appb-000031
r k(l)=r k(l-1)-μ k(l)B kp k(l);
然后,将迭代计算得到的s k(L)代入(2),得到k个空时快拍下所述共轭梯度空时自适应滤波器的权向量:
Figure PCTCN2018092550-appb-000032
最后,假定总快拍数为K,对每一个快拍,重复上述步骤,即得到所述共轭梯度空时自适应滤波器的权向量。
本发明与现有技术相比,有益效果在于:本发明实施例根据待检测距离单元的目标空时导向矢量和杂波干扰分量,基于变换域L1范数的共轭梯度空时自适应滤波器的输入信号,根据待检测距离单元的杂波稀疏特性建立该共轭梯度空时自适应滤波器的优化问题模型,采用变换域L1范数的共轭梯度算法,递归迭代地计算该优化问题模型,得到该共轭梯度空时自适应滤波器的权向量,根据该权向量对波滤器的输入信号进行自适应处理。本发明实施例提出了一种基于低复杂度变换域L1范数的共轭梯度自适应处理方法,能够在一些简单假设条件下,迭代地计算基于变换域L1范数的共轭梯度空时自适应滤波器的权向量而不用协方差矩阵求逆,降低了10倍左右计算的复杂度,同时保持良好的性能。
附图说明
图1是本发明实施例提供的基于变换域L1范数的共轭梯度空时自适应处理方法的流程图;
图2是本发明实施例提供的基于变换域L1范数的共轭梯度空时自适应处理系统的结构示意图;
图3是本发明实施例提供的针对DoFs的计算复杂度与系统空时自由度的关系示意图;
图4是本发明实施例提供的输出信干燥比与空时快拍数的关系示意图;
图5是本发明实施例提供的输出信干燥比与归一化多普勒频率的关系示意图;
图6是本发明实施例提供的检测概率P d与目标信噪比SNR的关系示意图。
具体实施方式
为了使本发明的目的、技术方案及优点更加清楚明白,以下结合附图及实施例,对本发明进行进一步详细说明。应当理解,此处所描述的具体实施例仅仅用以解释本发明,并不用于限定本发明。
图1示出了本发明实施例提供的一种基于变换域L1范数的共轭梯度空时自适应处理方法,包括:
S101,根据待检测距离单元内目标空时导向向量和干扰加噪音分量,确定基于变换域L1范数的共轭梯度空时自适应滤波器的输入信号;
S102,根据所述待检测距离单元的杂波稀疏特性,建立所述共轭梯度空时自适应滤波器的优化问题模型;
S103,采用基于变换域L1范数的共轭梯度算法,递归迭代地计算所述优化问题模型,得到所述共轭梯度空时自适应滤波器的权向量;
S104,根据所述权向量对所述共轭梯度空时自适应滤波器的输入信号进行自适应处理。
本发明实施例中,对具有稀疏约束的STAP算法提出一种基于变换域L1范数的共轭梯度空时自适应的实现方法,受到传统共轭梯度算法CG(Conjugate Gradient Methods)的低复杂度的启发,本发明实施例首先定义一种新的权向量,然后在基于变换域L1范数的CG算法的一些简单假设条件下导出一种计算新的权向量的迭代算法。最后,所提出的STAP权向量计算在不需要矩阵逆矩阵的低复杂度条件下求得。本发明实施例在一些简单的假设条件下,推导了一种基于变换域L1范数的共轭梯度空时自适应处理方法,计算复杂度低,避免了协 方差矩阵求逆。
下面,具体阐述本发明的步骤如下:
步骤一、建立信号模型:
假设一个正侧视脉冲多普勒机载雷达,其N个阵元的均匀线阵(ULA)在一个相干处理间隔(CPI,Coherent Process Interval)中接收具有固定脉冲重复频率(PRF,Pulse Recurrence Frequency)f r的M个相干脉冲,假设目标区域内只有一个目标,雷达接收信号可以表示为长度NM×1的空时快拍向量x=α tv t+x u,其中α t为复数,
Figure PCTCN2018092550-appb-000033
是空间频率
Figure PCTCN2018092550-appb-000034
和多普勒频率
Figure PCTCN2018092550-appb-000035
的目标空时导向向量。
Figure PCTCN2018092550-appb-000036
表示克罗内克积,x u是由干扰x j、杂波x c和热噪声x n组成的干扰加噪声向量。
步骤二、考虑到待检测距离单元内杂波的角度多普勒方向图的稀疏性,通常将稀疏约束下的STAP滤波器(SC-STAP)设计为:
Figure PCTCN2018092550-appb-000037
其中,||·|| 1、H分别表示l 1范数和共轭转置算子,
Figure PCTCN2018092550-appb-000038
表示干扰加噪声协方差矩阵,E[·]表示随机变量的期望值,
Figure PCTCN2018092550-appb-000039
表示具有范围在-0.5到0.5的多普勒频率的f d,j(j=1,...,N d)和具有空间频率f s,i(i=1,...,N s)的时空导向矩阵,λ是平衡滤波器功率w HRw和角度多普勒方向图的稀疏性的正则参数,这里λ=w HRw/(ρ-||w HQ|| 1),且λ>0。
就拉格朗日乘数而言,最佳的加权向量则是:
Figure PCTCN2018092550-appb-000040
其中
Figure PCTCN2018092550-appb-000041
|w HQ| i(i=1,...,N dN s)是||w HQ|| 1的第i项,ε是一个很小的常数。
注意到计算式(2)中的权向量w的计算复杂度非常高,所需存储(用于矩阵操作)空间大。因此,在实际应用中尤其困难,特别是对于大型DoFs雷 达系统。而且,由于∏(w)是w的函数,所以不能直接应用传统CG技术来计算权向量。
步骤三、基于变换域L1范数的变换CG算法递归计算公式(2)中的权向量w:
假设:
s k=(R kkQ∏(w k)Q H) -1v t  ( 3)
其中k表示第k个空时快拍的索引,s k表示第k个空时快拍下所求得的值,那么自适应的权向量为:
Figure PCTCN2018092550-appb-000042
为了利用变换域L1范数的变换CG算法递归导出s k,需要去解决以下等价的最小化问题:
Figure PCTCN2018092550-appb-000043
其中,
Figure PCTCN2018092550-appb-000044
表示复数的实部,因为R k,∏(w k)和λ k的估计方式会影响算法的性能,所以,在本申请中,采用指数衰减的数据窗来估计R k可以提供与递归最小二乘(RLS,Recursive least squares)算法来估计相当的性能。干扰协方差矩阵
Figure PCTCN2018092550-appb-000045
其中β是遗忘因子。同时,利用权向量在一个空时快拍中,不会有巨大的改变的事实,∏(w k)通过使用||(w k-1) HQ|| 1来近似,并且λ k=(w k) HR kw k/(ρ-||(w k) HQ|| 1),所以公式(5)的解的计算为:
s k(l)=s k(l-1)+μ k(l)p k(l)  (6)
其中μ k(l)和p k(l)分别是第k次空时快拍中的第l次迭代的自适应步长和相应的方向向量。方向向量p k(l)被定义为:
p k(l)=r k(l-1)+τ k(l-1)p k(l-1)  (7)
其中,r k(l-1)表示残差向量,被定义为:
Figure PCTCN2018092550-appb-000046
并且
B k=R kkQ∏(w k)Q H  (9)
τ k(l-1)为方向向量p k(l)提供B k正交性,然后从B kp k(l)=B kr k(l-1)+τ k(l-1)B kp k(l-1)和向量p k(l)的B k正交性得:
Figure PCTCN2018092550-appb-000047
(p k(l)) HB kr k(l-1)=(p k(l)) HB kp k(l)  (11)
通过在公式(6)的两边应用B k并减去v t,得到:
r k(l)=r k(l-1)-μ k(l)B kp k(l)  (12)
综合残差向量和公式(11)的正交性,得到:
Figure PCTCN2018092550-appb-000048
从等式r k(l-1)=r k(l-2)-μ k(l-1)B kp k(l-1)和(r k(l-1)) Hr k(l-2)=0可以得到:
(r k(l-1)) Hr k(l-1)=-μ k(l-1)(r k(l-1)) HB kp k(l-1)  (14)
和(r k(l-2)) Hr k(l-2)=μ k(l-1)(r k(l-2)) HB kp k(l-1)  (15)
将公式(14)和公式(15)代入公式(10)得到:
Figure PCTCN2018092550-appb-000049
最后,通过迭代计算公式(9)、(8)、(13)、(6)、(12)、(16)、(7)和(4)来得到公式(2)中权向量的值。
图2示出了本发明实施例提供的一种基于共轭梯度的自适应处理系统,包括:
信号模型建立单元201,用于根据待检测距离单元内目标空时导向向量和干扰加噪音分量,确定基于变换域L1范数的共轭梯度空时自适应滤波器的输入信号;
滤波器优化问题模型建立单元202,用于根据所述待检测距离单元的杂波 稀疏特性,建立所述共轭梯度空时自适应滤波器的优化问题模型;
权向量计算单元203,用于采用基于变换域L1范数的共轭梯度算法,递归迭代地计算所述优化问题模型,得到所述共轭梯度空时自适应滤波器的权向量;
信号处理单元204,用于根据所述权向量对所述共轭梯度空时自适应滤波器的输入信号进行自适应处理。
进一步地,所述待检测距离单元的空时快拍以x表示,则有:
x=α tv t+x u
其中,其中α t为复数,v t表示所述目标空时导向向量,x u表示所述干扰加噪音分量,
Figure PCTCN2018092550-appb-000050
Figure PCTCN2018092550-appb-000051
表示多普勒频率,
Figure PCTCN2018092550-appb-000052
表示空间频率,
Figure PCTCN2018092550-appb-000053
表示克罗内克积,x u由干扰x j、杂波x c和热噪声x n组成。
进一步地,所述共轭梯度空时自适应滤波器表示为以下优化问题,即:
Figure PCTCN2018092550-appb-000054
其中,||·|| 1、H分别表示l 1范数和共轭转置算子,
Figure PCTCN2018092550-appb-000055
表示干扰加噪声协方差矩阵,E[·]表示随机变量的期望值,
Figure PCTCN2018092550-appb-000056
表示具有范围在-0.5到0.5的多普勒频率的f d,j(j=1,...,N d)和具有空间频率f s,i(i=1,...,N s)的空时导向矩阵,λ表示平衡滤波器功率w HRw和角度多普勒方向图的稀疏性的正则参数,λ=w HRw/(ρ-||w HQ|| 1)。
进一步地,权向量计算单元203具体用于:
首先,令s k=(R kkQ∏(w k)Q H) -1v t,求得所述共轭梯度空时自适应滤波器的权向量w k
Figure PCTCN2018092550-appb-000057
其中,R表示干扰加噪声协方差矩阵,R k表示第k个快拍下的干扰加噪声协方差矩阵,λ k表示第k个快拍下的平衡滤波器输出功率w HRw和角度多普勒方向图的稀疏性的正则参数,λ k>0,k表示第k个空时快拍的索引,s k表示第k个 快拍下所求得的值,
Figure PCTCN2018092550-appb-000058
其次,令
Figure PCTCN2018092550-appb-000059
采用基于变换域L1范数的共轭梯度法计算s k
s k(l)=s k(l-1)+μ k(l)p k(l);
μ k(l)和p k(l)分别表示第k个空时快拍中的第l次迭代的自适应步长和方向向量,s k(l)表示第k次空时快拍中的第l次迭代计算得到的值;
然后,根据预置的迭代总次数L,迭代地计算s k(l):
B k=R kkQ∏(w k)Q H
Figure PCTCN2018092550-appb-000060
λ k=(w k) HR(w k)/(ρ-||(w k) HQ|| 1);
Figure PCTCN2018092550-appb-000061
Figure PCTCN2018092550-appb-000062
Figure PCTCN2018092550-appb-000063
p k(l)=r k(l-1)+τ k(l-1)p k(l-1);
Figure PCTCN2018092550-appb-000064
r k(l)=r k(l-1)-μ k(l)B kp k(l);
然后,将迭代计算得到的s k(L)代入(17),得到k个空时快拍下所述共轭梯度空时自适应滤波器的权向量:
Figure PCTCN2018092550-appb-000065
最后,假定总快拍数为K,对每一个快拍,重复上述步骤,即得到所述共轭梯度空时自适应滤波器的权向量。
以CG-SC-STAP表示本发明实施例提供的基于共轭梯度的自适应处理方法,下面,通过仿真数据来说明本发明实施例在算法复杂度和CG-SC-STAP算法的性能两方面的有益效果。
①算法复杂度:
所提出的算法复杂度是O(D(MN) 2)(包括D(2(MN) 2+5MN+3)乘法和2(MN) 2+4MN-3)加法,每个空时快拍具有D次迭代。在数值仿真的基础上,发现所提出的算法每次采样迭代收敛,迭代次数与脉冲数M和天线阵元N有关。因此,与复杂度为O((MN) 3)(包括(MN) 3+2(MN) 2+2MN乘法和(MN) 3-1加法)的SC-STAP算法相比,本发明实施例提出的算法提供更低的复杂度。另外,两个算法的每个空时快拍的复数乘法次数是DoFs雷达的空时自由度(MN)函数,如图3所示,其中选择D=7。可以看到,本发明实施例所提出的CG-SC-STAP显示了计算复杂度比SC-STAP算法少10倍左右。
②CG-SC-STAP算法的性能:
本发明实施例提供仿真结果来说明CG-SC-STAP算法的性能,并将其性能与稀疏感知波束形成器,空时多波束(STMB),局域联合处理(JDL)和主分量(PC)方法相比较。考虑以下模拟场景:平台速度125m/s,平台高度8km,载频1.2GHz,N=12,,M=12,,f r=2kHz,两个干扰机的干扰噪声比均为30dB,方位角分别为-45°和60°,杂波噪声比(CNR)为45dB。空间导向向量模型为c⊙v s(f s),阵列增益和相位误差c=[c 1,...,c M] T
Figure PCTCN2018092550-appb-000066
此处ε m和φ m分别为0.005和0.005*π/2的零均值高斯随机变量,PC的秩为50;JDL使用三个角度单元和三个多普勒单元(3×3);一个目标单元,四个角度单元和八个多普勒单元(8+4+1)用于STMB;β=0.9998,ε=10 -6,
Figure PCTCN2018092550-appb-000067
以及D=7。
在第一个仿真中,将这些算法在信号干扰加噪声比(SINR)损失进行比较。图4显示了SINR损失与
Figure PCTCN2018092550-appb-000068
的快拍数量之间的关系。据观察,CG-SC-STAP显示出与SC-STAP算法相当的性能。在归一化的多普勒频率上的SINR损失如图5所示。快拍数量设置为50。可以看到,当目标归一化多普勒频率小于0.15时,CG-SC-STAP具有与SC-STAP相似的SINR性能,并且胜过所有其他比SC-STAP具有更低SINR的算法。
在第二个仿真中,给出虚警率Pfa=0.001的检测概率(P d)与目标信噪比 (SNR)的关系,如图6所示。目标归一化多普勒频率为0.25,训练快拍数为50。阈值和Pd两者估计使用10/PFA蒙特卡洛运行。结果表明,本发明实施例提供的方法具有与SC-STAP几乎相同的P d,但高于其他算法。
本发明实施例提出了一种基于共轭梯度技术的稀疏约束低复杂度空时自适应处理方法,在一些简单的假设条件下,迭代地计算基于CG的权向量而不用协方差矩阵求逆,降低了10倍左右计算的复杂度,同时保持良好的性能。本发明实施例可以应用于机载雷达系统中杂波抑制和动目标检测,降低计算的复杂度,提高雷达系统杂波抑制水平与目标检测能力。
具体地,本发明实施例提供的CG-SC-STAP使用于雷达信号处理领域,提出了一种基于共轭梯度技术的低复杂度的空时自适应处理方法。本发明实施例基于一些简单的假设条件下,迭代地计算基于CG的权向量而不用协方差矩阵求逆,导出用于计算自适应的权向量。本发明实施例在有限样本支持下具有与迭代最小二乘稀疏约束的STAP(SC-STAP)相当的性能,但计算复杂度相对较低。
本发明实施例还提供了一种终端,包括存储器、处理器及存储在存储器上且在处理器上运行的计算机程序,处理器执行计算机程序时,实现如图1所示的基于变换域L1范数的共轭梯度空时自适应处理方法中的各个步骤。
本发明实施例中还提供一种可读存储介质,其上存储有计算机程序,所述计算机程序被处理器执行时,实现如图1所示的基于变换域L1范数的共轭梯度空时自适应处理方法中的各个步骤。
另外,在本发明各个实施例中的各功能模块可以集成在一个处理模块中,也可以是各个模块单独物理存在,也可以两个或两个以上模块集成在一个模块中。上述集成的模块既可以采用硬件的形式实现,也可以采用软件功能模块的形式实现。
所述集成的模块如果以软件功能模块的形式实现并作为独立的产品销售或使用时,可以存储在一个计算机可读取存储介质中。基于这样的理解,本发明 的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的全部或部分可以以软件产品的形式体现出来,该计算机软件产品存储在一个存储介质中,包括若干指令用以使得一台计算机设备(可以是个人计算机,服务器,或者网络设备等)执行本发明各个实施例所述方法的全部或部分步骤。而前述的存储介质包括:U盘、移动硬盘、只读存储器(ROM,Read-Only Memory)、随机存取存储器(RAM,Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。
以上所述仅为本发明的较佳实施例而已,并不用以限制本发明,凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等,均应包含在本发明的保护范围之内。

Claims (8)

  1. 一种基于变换域L1范数的共轭梯度空时自适应处理方法,其特征在于,包括:
    步骤A,根据待检测距离单元内目标空时导向向量和干扰加噪音分量,确定基于变换域L1范数的共轭梯度空时自适应滤波器的输入信号;
    步骤B,根据所述待检测距离单元的杂波稀疏特性,建立所述共轭梯度空时自适应滤波器的优化问题模型;
    步骤C,采用基于变换域L1范数的共轭梯度算法,递归迭代地计算所述优化问题模型,得到所述共轭梯度空时自适应滤波器的权向量;
    步骤D,根据所述权向量对所述共轭梯度空时自适应滤波器的输入信号进行自适应处理。
  2. 如权利要求1所述的共轭梯度空时自适应处理方法,其特征在于,在所述步骤A中,所述待检测距离单元的空时快拍以x表示,则有:
    x=α tv t+x u
    其中,其中α t为复数,v t表示所述目标空时导向向量,x u表示所述干扰加噪音分量,
    Figure PCTCN2018092550-appb-100001
    表示多普勒频率,
    Figure PCTCN2018092550-appb-100002
    表示空间频率,
    Figure PCTCN2018092550-appb-100003
    表示克罗内克积,x u由干扰x j、杂波x c和热噪声x n组成。
  3. 如权利要求2所述的共轭梯度空时自适应处理方法,其特征在于,在所述步骤B中,所述共轭梯度空时自适应滤波器表示为以下优化问题,即:
    Figure PCTCN2018092550-appb-100004
    其中,||·|| 1、H分别表示l 1范数和共轭转置算子,
    Figure PCTCN2018092550-appb-100005
    表示干扰加噪声协方差矩阵,E[·]表示随机变量的期望值,
    Figure PCTCN2018092550-appb-100006
    表示具有范围在-0.5到0.5的多普勒频率的f d,j(j=1,...,N d)和具有空间频率f s,i(i=1,...,N s)的空时导向矩阵,N d表示多普勒频率总个数,N s表示空间频率总个数,λ表示平衡滤波器输出功率w HRw和角度多普勒方向图的稀疏性的正则参数, λ=w HRw/(ρ-||w HQ|| 1)。
  4. 如权利要求3所述的共轭梯度空时自适应处理方法,其特征在于,所述步骤C包括:
    步骤C1,令
    Figure PCTCN2018092550-appb-100007
    求得所述共轭梯度空时自适应滤波器的权向量w k
    Figure PCTCN2018092550-appb-100008
    其中,R表示干扰加噪声协方差矩阵,R k表示第k个快拍下的干扰加噪声协方差矩阵,λ k表示第k个快拍下的平衡滤波器输出功率w HRw和角度多普勒方向图的稀疏性的正则参数,λ k>0,k表示第k个空时快拍的索引,s k表示第k个快拍下所求得的值,
    Figure PCTCN2018092550-appb-100009
    步骤C2,令
    Figure PCTCN2018092550-appb-100010
    采用基于变换域L1范数的共轭梯度法计算s k
    s k(l)=s k(l-1)+μ k(l)p k(l);
    μ k(l)和p k(l)分别表示第k个空时快拍中的第l次迭代的自适应步长和方向向量,s k(l)表示第k次空时快拍中的第l次迭代计算得到的值;
    步骤C3,根据预置的迭代总次数L,迭代地计算s k(l):
    B k=R kkQ∏(w k)Q H
    Figure PCTCN2018092550-appb-100011
    λ k=(w k) HR(w k)/(ρ-||(w k) HQ|| 1);
    Figure PCTCN2018092550-appb-100012
    Figure PCTCN2018092550-appb-100013
    Figure PCTCN2018092550-appb-100014
    p k(l)=r k(l-1)+τ k(l-1)p k(l-1);
    Figure PCTCN2018092550-appb-100015
    r k(l)=r k(l-1)-μ k(l)B kp k(l);
    步骤C4,将迭代计算得到的s k(L)代入(1),得到k个空时快拍下所述共轭梯度空时自适应滤波器的权向量:
    Figure PCTCN2018092550-appb-100016
    步骤C5,假定总快拍数为K,对每一个快拍,重复上述步骤C1、C2、C3和C4,即得到所述共轭梯度空时自适应滤波器的权向量。
  5. 一种基于变换域L1范数的共轭梯度空时自适应处理系统,其特征在于,包括:
    信号模型建立单元,用于根据待检测距离单元内目标空时导向向量和干扰加噪音分量,确定基于变换域L1范数的共轭梯度空时自适应滤波器的输入信号;
    滤波器优化问题模型建立单元,用于根据所述待检测距离单元的杂波稀疏特性,建立所述共轭梯度空时自适应滤波器的优化问题模型;
    权向量计算单元,用于采用基于变换域L1范数的共轭梯度算法,递归迭代地计算所述优化问题模型,得到所述共轭梯度空时自适应滤波器的权向量;
    信号处理单元,用于根据所述权向量对所述共轭梯度空时自适应滤波器的输入信号进行自适应处理。
  6. 如权利要求5所述的共轭梯度空时自适应处理系统,其特征在于,所述待检测距离单元的空时快拍以x表示,则有:
    x=α tv t+x u
    其中,其中α t为复数,v t表示所述目标空时导向向量,x u表示所述干扰加噪音分量,
    Figure PCTCN2018092550-appb-100017
    表示多普勒频率,
    Figure PCTCN2018092550-appb-100018
    表示空间频率,
    Figure PCTCN2018092550-appb-100019
    表示克罗内克积,x u由干扰x j、杂波x c和热噪声x n组成。
  7. 如权利要求6所述的共轭梯度空时自适应处理系统,其特征在于,所述共轭梯度空时自适应滤波器表示为以下优化问题,即:
    Figure PCTCN2018092550-appb-100020
    其中,||·|| 1、H分别表示l 1范数和共轭转置算子,
    Figure PCTCN2018092550-appb-100021
    表示干扰加噪声协方差矩阵,E[·]表示随机变量的期望值,
    Figure PCTCN2018092550-appb-100022
    表示具有范围在-0.5到0.5的多普勒频率的f d,j(j=1,...,N d)和具有空间频率f s,i(i=1,...,N s)的空时导向矩阵,λ表示平衡滤波器功率w HRw和角度多普勒方向图的稀疏性的正则参数,λ=w HRw/(ρ-||w HQ|| 1)。
  8. 如权利要求7所述的共轭梯度空时自适应处理系统,其特征在于,所述权向量计算单元具体用于:
    首先,令s k=(R kkQ∏(w k)Q H) -1v t,求得所述共轭梯度空时自适应滤波器的权向量w k
    Figure PCTCN2018092550-appb-100023
    其中,R表示干扰加噪声协方差矩阵,R k表示第k个快拍下的干扰加噪声协方差矩阵,λ k表示第k个快拍下的平衡滤波器输出功率w HRw和角度多普勒方向图的稀疏性的正则参数,λ k>0,k表示第k个空时快拍的索引,s k表示第k个快拍下所求得的值,
    Figure PCTCN2018092550-appb-100024
    其次,令
    Figure PCTCN2018092550-appb-100025
    采用基于变换域L1范数的共轭梯度法计算s k
    s k(l)=s k(l-1)+μ k(l)p k(l);
    μ k(l)和p k(l)分别表示第k个空时快拍中的第l次迭代的自适应步长和方向向量,s k(l)表示第k次空时快拍中的第l次迭代计算得到的值;
    然后,根据预置的迭代总次数L,迭代地计算s k(l):
    B k=R kkQ∏(w k)Q H
    Figure PCTCN2018092550-appb-100026
    λ k=(w k) HR(w k)/(ρ-||(w k) HQ|| 1);
    Figure PCTCN2018092550-appb-100027
    Figure PCTCN2018092550-appb-100028
    Figure PCTCN2018092550-appb-100029
    p k(l)=r k(l-1)+τ k(l-1)p k(l-1);
    Figure PCTCN2018092550-appb-100030
    r k(l)=r k(l-1)-μ k(l)B kp k(l);
    然后,将迭代计算得到的s k(L)代入(2),得到k个空时快拍下所述共轭梯度空时自适应滤波器的权向量:
    Figure PCTCN2018092550-appb-100031
    最后,假定总快拍数为K,对每一个快拍,重复上述步骤,即得到所述共轭梯度空时自适应滤波器的权向量。
PCT/CN2018/092550 2018-06-25 2018-06-25 一种共轭梯度空时自适应处理方法及系统 WO2020000123A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2018/092550 WO2020000123A1 (zh) 2018-06-25 2018-06-25 一种共轭梯度空时自适应处理方法及系统

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2018/092550 WO2020000123A1 (zh) 2018-06-25 2018-06-25 一种共轭梯度空时自适应处理方法及系统

Publications (1)

Publication Number Publication Date
WO2020000123A1 true WO2020000123A1 (zh) 2020-01-02

Family

ID=68985294

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/092550 WO2020000123A1 (zh) 2018-06-25 2018-06-25 一种共轭梯度空时自适应处理方法及系统

Country Status (1)

Country Link
WO (1) WO2020000123A1 (zh)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107219511A (zh) * 2017-06-06 2017-09-29 深圳大学 波束‑多普勒方向图稀疏约束的stap方法及装置
CN107544061A (zh) * 2017-09-11 2018-01-05 深圳大学 基于知识的稀疏恢复空时自适应处理方法及系统

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107219511A (zh) * 2017-06-06 2017-09-29 深圳大学 波束‑多普勒方向图稀疏约束的stap方法及装置
CN107544061A (zh) * 2017-09-11 2018-01-05 深圳大学 基于知识的稀疏恢复空时自适应处理方法及系统

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
YANG, ZHAOCHENG ET AL.: "Theory and Methods of Sparsity-Based Space-Time Adaptive Processing", INFORMATION & TECHNOLOGY, 15 October 2014 (2014-10-15), pages 1136 - 129, ISSN: 1674-022X *
YANG, ZHAOCHENG: "Theory . and Methods of Sparsity-Based Space-Time Adaptive Processing", INFORMATION & TECHNOLOGY, 15 October 2014 (2014-10-15), China, pages 1136 - 129, ISSN: 1674-022X *

Similar Documents

Publication Publication Date Title
CN109031227B (zh) 一种共轭梯度空时自适应处理方法及系统
WO2018094565A1 (zh) 脉冲噪声下的波束成形方法及装置
Zhuang et al. Interference cancellation beamforming robust to pointing errors
Landau et al. Robust adaptive beamforming algorithms using the constrained constant modulus criterion
CN107219511B (zh) 波束-多普勒方向图稀疏约束的stap方法及装置
Mohammadzadeh et al. Adaptive beamforming based on theoretical interference‐plus‐noise covariance and direction‐of‐arrival estimation
Li et al. Distributed MIMO radar based on sparse sensing: Analysis and efficient implementation
Chen et al. Distributed algorithms for array signal processing
Meng et al. Robust adaptive beamforming for coprime array with steering vector estimation and covariance matrix reconstruction
Mohammadzadeh et al. Robust adaptive beamforming for fast moving interference based on the covariance matrix reconstruction
Mohammadzadeh et al. Modified robust capon beamforming with approximate orthogonal projection onto the signal-plus-interference subspace
Li et al. Parameter estimation based on fractional power spectrum density in bistatic MIMO radar system under impulsive noise environment
Dakulagi et al. Smart antenna system for DOA estimation using single snapshot
Santiago et al. Noise subspace-based iterative technique for direction finding
CN114726385A (zh) 基于功率估计的卫星导航接收机空域抗干扰方法
Smith et al. A new approach to moving target detection using unit circle roots constrained adaptive matched filter
Ruan et al. Low‐complexity robust adaptive beamforming algorithms exploiting shrinkage for mismatch estimation
Yang et al. Robust beamforming with imprecise array geometry using steering vector estimation and interference covariance matrix reconstruction
Yang et al. Hybrid direct data domain sigma-delta space–time adaptive processing algorithm in non-homogeneous clutter
WO2020000123A1 (zh) 一种共轭梯度空时自适应处理方法及系统
Pinilla et al. Banraw: Band-limited radar waveform design via phase retrieval
CN114844544B (zh) 一种基于低管秩张量分解的互质阵列波束成形方法、系统及介质
Zhang et al. Robust adaptive beamforming algorithm based on damped singular value decomposition regularization
Morgan Adaptive algorithms for solving generalized eigenvalue signal enhancement problems
Smith Unit circle roots based sensor array signal processing

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18924077

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 28/04/2021)

122 Ep: pct application non-entry in european phase

Ref document number: 18924077

Country of ref document: EP

Kind code of ref document: A1