Joint diagonalized L-type MIMO radar circular and non-circular hybrid direction finding method

The two-dimensional ESPRIT algorithm for MIMO radar systems effectively estimates and pairs 2D-DOD and 2D-DOA by jointly diagonalizing circular and non-circular signals, addressing the coexistence challenge and achieving reliable angle estimation.

CN111175690BActive Publication Date: 2024-12-06NINGBO UNIV
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN201911035936.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2019-10-29
Publication Date
2024-12-06
Estimated Expiration
2039-10-29

AI Technical Summary

Technical Problem

Existing MIMO radar systems struggle with 2D-DOD and 2D-DOA estimation when circular and non-circular signals coexist, lacking effective methods for addressing this, particularly in the context of non-circular signal characteristics.

Method used

A two-dimensional ESPRIT algorithm is proposed to estimate 2D-DOD and 2D-DOA by jointly diagonalizing the circular and non-circular hybrid direction finding method, utilizing an L-type MIMO radar to construct a new data vector and perform high-resolution parameter estimation.

Benefits of technology

The method achieves automatic angle matching and distinguishes mixed signals, demonstrating good estimation performance through simulation experiments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN111175690B_ABST
    Figure CN111175690B_ABST
Patent Text Reader

Abstract

The present invention belongs to the field of radar signal processing. In order to propose a two-dimensional ESPRIT algorithm to estimate 2D-DOD and 2D-DOA, the automatic matching of angles can be realized, and the problem of angle merging can be solved, and the estimation performance is good. To this end, the technical solution adopted by the present invention is to jointly diagonalize the L-type MIMO radar circular and non-circular hybrid direction finding method. First, a new data vector is constructed using the data vector received by the multi-input multi-output MIMO radar and its conjugate. Then, the high-resolution parameter estimation ESPRIT method is used to estimate the two-dimensional wave departure direction 2D-DOD and the two-dimensional wave arrival direction 2D-DOA based on the joint diagonalization of the direction matrix containing non-circular information. The present invention is mainly used in radar signal processing occasions.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of radar signal processing, and in particular relates to an L-type MIMO radar. Background Art

[0002] MIMO radar designed by using Multiple-Input Multiple-Output (MIMO) technology is a new type of radar system with many potential advantages. In recent years, the joint estimation technology of Direction of Departure (DOD) and Direction of Arrival (DOA) in MIMO radar has received extensive attention and research. For the one-dimensional (1D) DOD and DOA (Direction of Arrival, DOA) estimation problems, most of them directly use the 2D-DOA estimation method in array signal processing. For the two-dimensional (2D) 2D-DOD and 2D-DOA estimation problems, there are fewer discussions. On the other hand, since the non-circular characteristics of non-circular signals can improve the angle estimation accuracy and detect more signals, a series of algorithms are proposed by combining the non-circular characteristics with bistatic MIMO radar. However, the signals received by these algorithms are all under the premise of pure non-circularity, and the coexistence of circular and non-circular signals cannot be estimated. To solve this problem, a literature proposes the ESPRIT algorithm for DOD and DOA estimation when circular and non-circular signals coexist in bistatic MIMO radar. The above work only studies the 1D-DOD and 1D-DOA estimation problems, and requires parameter pairing. Currently, there are still few studies on 2D-DOD and 2D-DOA estimation problems using non-circular characteristics. Summary of the invention

[0003] In order to overcome the shortcomings of the prior art, the present invention aims to propose a two-dimensional ESPRIT algorithm to estimate 2D-DOD and 2D-DOA, which can realize automatic matching of angles, solve the problem of angle merging, and have good estimation performance. To this end, the technical solution adopted by the present invention is to jointly diagonalize the circular and non-circular hybrid direction finding method of the L-type MIMO radar, firstly construct a new data vector using the data vector received by the multi-input multi-output MIMO radar and its conjugate, and then use the high-resolution parameter estimation ESPRIT method to estimate the two-dimensional wave departure direction 2D-DOD and the two-dimensional wave arrival direction 2D-DOA based on the joint diagonalization of the direction matrix containing non-circular information.

[0004] The specific steps are as follows:

[0005] Step 1: Expand the data vector Perform singular value decomposition and get

[0006] Step 2: Construct the direction matrix set according to formula (12):

[0007] G l =(K l2 U S ) + K l1 U S =EΘ l E H ,l=1,2,3,4 (12)

[0008] Among them, K l1 For a strictly non-circular mixed signal, about the angle θ k1 The selection matrix, K l2 is the circular mixed signal about the angle θ k1 The selection matrix of θ is kl The eigenvector of

[0009] Step 3 Perform joint diagonalization to obtain a unitary matrix

[0010] Step 4: Calculate the eigenvalue according to formula (14) Then, we estimate from formula (15)

[0011]

[0012]

[0013] Each circular signal is considered as two virtual strictly non-circular signals with equal angles, so θ kl There are K' angles, but in reality, there are only K incident signals. According to the number of repeated angles, circular and non-circular signals are distinguished, and θ is obtained by averaging. kl,c ;

[0014] Step 5 estimates the 2D-DOD and 2D-DOA of the circular signal through equation (16).

[0015]

[0016] The specific steps are as follows:

[0017] (a) Signal reception model

[0018] A MIMO radar consisting of an L-shaped transmitting array and an L-shaped receiving array is used. The number of array elements of the transmitting array is M=M1+M2-1, where M1 and M2 array elements are located on the X and Y axes respectively. The number of array elements of the receiving array is N=N1+N2-1, where N1 and N2 array elements are located on the X' and Y' axes respectively. The four subarrays are uniform linear arrays, and the array element spacing is equal and set to d, taking d=λ / 2, λ is the wavelength of the signal. Assume that there are K signals s k (t), k = 1, 2, ..., K is incident on the array, K = K n +K c , the direction of the kth signal is recorded as (θ k1 ,θ k2 ,θ k3 ,θ k4 ), where (θ k1 ,θ k2 ) is the 2D-DOD of the kth signal, (θ k3 ,θ k4 ) is the 2D-DOA of the kth signal, and the incident signal includes K n Non-circular signal s n,k (t),k=1,2,…,K n and K c Circle signal s c,k (t),k=1,2,…,K c , the output vector x(t) of the array at sampling t is expressed as:

[0019] x(t)=C(θ i1 ,θ i2 ,θ i3 ,θ i4 )s(t)+n(t) (1)

[0020] where x(t)=[x1(t),…,x MN (t)] T Represents the receiving data matrix of the array; C = [c1, c2, ..., c K ] T represents the popularity matrix of the extended virtual array, is the extended virtual array manifold vector, a k and b k are the M×1 dimensional transmit and N×1 dimensional receive array populous vectors, respectively, expressed as n(t)=[n1(t),…,n MN (t)] T represents the additive Gaussian white noise matrix, whose mean is zero and variance is σ 2 ; s(t)=[s1(t),…,s K (t)]T represents the mixed incident source signal vector, and s(t) is re-expressed as

[0021]

[0022] in is the non-circular phase of the non-circular signal, Φ is a K×K'-dimensional matrix, K'=K n +2K c , a K'×1-dimensional vector containing K n Strictly non-circular signal s n (t) and the circular signal s c K of (t) c Real part and K c Imaginary part

[0023] Rewrite the manifold matrix C as

[0024] C=[C1(θ k1,n ,θ k2,n ,θ k3,n ,θ k4,n ) C2(θ k1,c ,θ k2,c ,θ k3,c ,θ k4,c )]=[C1 C2] (3)

[0025] Where C1 is MN×K n The steering matrix for a strictly non-circular signal, C1, is MN×K c Steering matrix for circular signals;

[0026] Substituting equation (2) and equation (3) into equation (1), we can obtain

[0027]

[0028] In order to simplify the representation, some parts of the following text will omit the angle pair (θ k1 ,θ k2 ,θ k3 ,θ k4 ) and time t;

[0029] (b) Four-dimensional parameter estimation

[0030] In order to utilize the strictly non-circular characteristics of strictly non-circular signals and the virtual strictly non-circular characteristics of circular signals, the data vector x and its conjugate corresponding coupled data vector are combined into a new data vector y

[0031]

[0032] Among them, γ MNis a MN×MN dimensional switching matrix, is the 2MN×K' extended steering vector, which is expressed as

[0033]

[0034] is a 2MN×1 noise vector,

[0035] Performing singular value decomposition on Y yields:

[0036]

[0037] Among them, 2MN×K' dimension U S and 2MN×(2MN-K') dimension U N are the left singular signal subspace and the noise subspace, respectively, L×K' dimension V S and L×(2MN-K') dimension V N are the right singular signal subspace and the noise subspace respectively, L is the number of snapshots, Σ S =diag(λ1,λ2,…,λ K' ) and Σ N =diag(λ K'+1 ,λ K'+2 ,…,λ 2MN ) represent the diagonal matrices consisting of K' and 2NM-K' eigenvalues ​​respectively;

[0038] Define two selection matrices as follows

[0039]

[0040]

[0041] Strictly non-circular and circular mixed signals about the angle θ k1 The selection matrix is

[0042]

[0043]

[0044] Among them are

[0045] Similarly, we get the angle θ kl The selection matrix (K l1 ,K l2 );

[0046] Based on the selection matrix K l1 and K l2 Based on the principle of the non-circular ESPRIT algorithm, we define the angle θ klThe direction matrix G l ;

[0047] G l =(K l2 U S ) + K l1 U S =EΘ l E H (12)

[0048] Where E is a K'×K' unitary matrix, and the diagonal matrix Θ containing the angle information l The expression is

[0049]

[0050] G in formula (12) l Satisfy the joint diagonalization condition. Define a set G = {G1, G2, G3, G4}, and obtain a unitary matrix E = [e1, e2, …, e K' ], E is θ kl The eigenvector of kl The eigenvalues ​​of θ maintain a one-to-one correspondence during the joint diagonalization process, without the need for angle pairing. kl The eigenvalue of is calculated by the following formula

[0051]

[0052] Then, it is easy to get from formula (14)

[0053]

[0054] Each circular signal is considered as two virtual strictly non-circular signals with equal angles, so θ kl There are K' angles. However, in reality, there are only K incident signals, so the circular and non-circular signals are distinguished according to the number of repeated angles. Since the two estimated angles of the circular signal are reliable, the final non-circular signal angle θ is obtained by averaging. kl,c (l=1,2,3,4),

[0055]

[0056] in and are the first and second angles estimated for non-circular signals, respectively.

[0057] The characteristics and beneficial effects of the present invention are:

[0058] The present invention is based on L-type MIMO radar. Considering that the radar sends circular and non-circular signals at the same time, the non-circular information is fully utilized to realize the joint estimation and automatic pairing of 2D DOD and 2D DOA, and the mixed signals can be distinguished. This conclusion has been verified by simulation experiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 Scatter plots of 2D DOD and 2D DOA estimates.

[0060] Figure 2 Flow chart of the MIMO radar circular and non-circular hybrid direction finding method of the present invention. DETAILED DESCRIPTION

[0061] The invention belongs to the field of radar signal processing, and specifically relates to a joint diagonalization-ESPRIT direction finding algorithm based on an L-type MIMO radar, which jointly estimates and automatically pairs a two-dimensional (2D) angle of departure (DOD) and a 2D angle of arrival (DOA) in the case of coexistence of circular and non-circular signals.

[0062] The present invention proposes a two-dimensional ESPRIT algorithm to estimate 2D-DOD and 2D-DOA based on the joint diagonalization technology when circular and non-circular signals coexist in a dual-base MIMO radar. The invention can realize automatic angle matching and solve the problem of angle merging, and has good estimation performance.

[0063] The technical solution adopted by the present invention is: by making full use of the L-type MIMO array structure to obtain an extended virtual array on the receiving array, firstly, a new data vector is constructed using the received data vector and its conjugate, and then the ESPRIT method is used to estimate 2D-DOD and 2D-DOA based on the joint diagonalization of the direction matrix containing non-circular information.

[0064] Explanation of symbols: (·) * , (·) T , (·) + and(·) H They represent conjugate, transpose, pseudo-inverse and conjugate transpose respectively. diag(·) represents diagonalization operation; blkdiag(·) represents the generation of block diagonal matrix; Represents the Kron product operation; I k represents the k-dimensional identity matrix; arg(·) represents the phase angle of the complex number; represents the estimated value of the symbol θ.

[0065] The specific technical solutions are as follows:

[0066] (a) Signal reception model

[0067] Consider a MIMO radar consisting of an L-shaped transmitting array and an L-shaped receiving array. The number of elements in the transmitting array is M = M1 + M2-1, where M1 and M2 elements are located on the X and Y axes respectively. The number of elements in the receiving array is N = N1 + N2-1, where N1 and N2 elements are located on the X' and Y' axes respectively. The four subarrays are uniform linear arrays, and the spacing between the elements is equal and is set to d, where d = λ / 2, where λ is the wavelength of the signal. Assume that there is K (K = K n +K c ) signals k (t), k = 1, 2, ..., K is incident on the array, and the direction of the kth signal is recorded as (θ k1 ,θ k2 ,θ k3 ,θ k4 ), where (θ k1 ,θ k2 ) is the 2D-DOD of the kth signal, (θ k3 ,θ k4 ) is the 2D-DOA of the kth signal, and the incident signal includes K n Non-circular signal s n,k (t),k=1,2,…,K n and K c Circle signal s c,k (t),k=1,2,…,K c , the output vector x(t) of the array at sampling t is expressed as

[0068] x(t)=C(θ i1 ,θ i2 ,θ i3 ,θ i4 )s(t)+n(t) (1)

[0069] where x(t)=[x1(t),…,x MN (t)] T Represents the receiving data matrix of the array; C = [c1, c2, ..., c K ] T represents the popularity matrix of the extended virtual array, is the extended virtual array manifold vector, a k and b k are the M×1 dimensional transmit and N×1 dimensional receive array populous vectors, which can be expressed as n(t)=[n1(t),…,n MN (t)] T represents the additive Gaussian white noise matrix, whose mean is zero and variance is σ 2 ; s(t)=[s1(t),…,s K(t)] T represents the mixed incident source signal vector, s(t) can be re-expressed as

[0070]

[0071] in is the non-circular phase of the non-circular signal, Φ is a K×K'-dimensional matrix, K'=K n +2K c , a K'×1-dimensional vector containing K n Strictly non-circular signal s n (t) and the circular signal s c K of (t) c Real part and K c Imaginary part

[0072] Rewrite the manifold matrix C as

[0073] C=[C1(θ k1,n ,θ k2,n ,θ k3,n ,θ k4,n ) C2(θ k1,c ,θ k2,c ,θ k3,c ,θ k4,c )]=[C1 C2] (3)

[0074] Where C1 is MN×K n The steering matrix for a strictly non-circular signal, C1, is MN×K c Steering matrix for circular signals.

[0075] Substituting equation (2) and equation (3) into equation (1), we can obtain

[0076]

[0077] In order to simplify the representation, some parts of the following text will omit the angle pair (θ k1 ,θ k2 ,θ k3 ,θ k4 ) and time t.

[0078] (b) Four-dimensional parameter estimation

[0079] In order to utilize the strictly non-circular characteristics of strictly non-circular signals and the virtual strictly non-circular characteristics of circular signals, the data vector x and its conjugate corresponding coupled data vector are combined into a new data vector y

[0080]

[0081] Among them, γ MNis a MN×MN dimensional switching matrix, is the 2MN×K' extended steering vector, which is expressed as

[0082]

[0083] is a 2MN×1 noise vector,

[0084] Performing singular value decomposition on Y yields

[0085]

[0086] Among them, 2MN×K' dimension U S and 2MN×(2MN-K') dimension U N are the left singular signal subspace and the noise subspace, respectively, L×K' dimension V S and L×(2MN-K') dimension V N are the right singular signal subspace and noise subspace respectively, and L is the number of snapshots. S =diag(λ1,λ2,…,λ K' ) and Σ N =diag(λ K'+1 ,λ K'+2 ,…,λ 2MN ) represent the diagonal matrices consisting of K' and 2NM-K' eigenvalues ​​respectively.

[0087] Define two selection matrices as follows

[0088]

[0089]

[0090] Strictly non-circular and circular mixed signals about the angle θ k1 The selection matrix is

[0091]

[0092]

[0093] Among them are

[0094] Similarly, the angle θ can be obtained kl The selection matrix (K) of (l=2,3,4) l1 ,K l2 ),l=2,3,4.

[0095] Based on the selection matrix K l1 and K l2Based on the principle of the non-circular ESPRIT algorithm, we define the angle θ kl The direction matrix G of (l=1,2,3,4) l

[0096] G l =(K l2 U S ) + K l1 U S =EΘ l E H ,l=1,2,3,4 (12)

[0097] Where E is a K'×K' unitary matrix, and the diagonal matrix Θ containing the angle information l The expression is

[0098]

[0099] Obviously, G in formula (12) l Satisfy the joint diagonalization condition. Define a set G = {G1, G2, G3, G4}, and obtain a unitary matrix E = [e1, e2, …, e K' ], E is θ kl The eigenvector of (l=1,2,3,4), θ kl The eigenvalues ​​of θ maintain a one-to-one correspondence during the joint diagonalization process, without the need for angle pairing. kl The eigenvalue of can be calculated by the following formula

[0100]

[0101] Then, it is easy to get from formula (14)

[0102]

[0103] Each circular signal is considered as two virtual strictly non-circular signals with equal angles, so θ kl There are K' angles. However, in reality, there are only K incident signals, so the circular and non-circular signals can be distinguished based on the number of repeated angles. Since the two estimated angles of the circular signal are reliable, the final non-circular signal angle θ can be obtained by averaging. kl,c (l=1,2,3,4),

[0104]

[0105] in and are the first and second angles estimated for non-circular signals, respectively.

[0106] To verify the resolution performance of the present invention, the number of array elements of the L-type MIMO array is set to M1=M2=N1=N2=5, and the array element spacing d is set to half the wavelength λ / 2. There are 2 strictly non-circular (BPSK) and 2 circular (QPSK) signals. The angular parameters of the BPSK signal are (85°, 70°, 100°, 105°) and (90°, 95°, 115°, 100°), and the angular parameters of the QPSK signal are (105°, 85°, 80°, 80°) and (75°, 90°, 95°, 75°). The signal-to-noise ratio SNR=10dB, the number of snapshots is 500, and M c =200. Figure 1 It can be seen from the 2D-DOD and 2D-DOA distributions of the mixed signal that the present invention can successfully estimate the two-dimensional angle of circular and non-circular mixed signals, and can effectively distinguish BPSK and QPSK signals.

[0107] Step 1: Expand the data vector Perform singular value decomposition and get

[0108] Step 2: Construct the direction matrix set according to formula (12):

[0109] Step 3 Perform joint diagonalization to obtain a unitary matrix

[0110] Step 4: Calculate the eigenvalue according to formula (14) Then, we estimate from formula (15)

[0111] Step 5 estimates the 2D-DOD and 2D-DOA of the circular signal through equation (16).

Claims

1. A joint diagonalized L-type MIMO radar circular and non-circular hybrid direction finding method, characterized in that: First, a new data vector is constructed using the data vector received by the multi-input multi-output MIMO radar and its conjugate. Then, the high-resolution parameter estimation ESPRIT method is used to estimate the two-dimensional direction of departure 2D-DOD and the two-dimensional direction of arrival 2D-DOA based on the joint diagonalization of the direction matrix containing non-circular information. The specific steps are as follows: Step 1: Expand the data vector Perform singular value decomposition and get Step 2: Construct the direction matrix set according to formula (12): MR l =(K l2 IN S ) + K l1 IN S =EΘ l E H ,l=1,2,3,4 (12) Among them, K l1 For a strictly non-circular mixed signal, about the angle θ k1 The selection matrix, K l2 is the circular mixed signal about the angle θ k1 The selection matrix of θ is kl The eigenvector of Step 3 Perform joint diagonalization to obtain a unitary matrix Step 4: Calculate the eigenvalue according to formula (14) Then, we estimate from formula (15) Each circular signal is considered as two virtual strictly non-circular signals with equal angles, so θ kl There are K' angles, but in reality, there are only K incident signals. According to the number of repeated angles, circular and non-circular signals are distinguished, and θ is obtained by averaging. kl,c ; Step 5: Estimate the 2D-DOD and 2D-DOA of the circular signal by equation (16):

2. The circular and non-circular hybrid direction finding method of the joint diagonalized L-type MIMO radar as claimed in claim 1, characterized in that: The specific steps are as follows: (a) Signal reception model A MIMO radar consisting of an L-shaped transmitting array and an L-shaped receiving array is used. The number of array elements of the transmitting array is M=M1+M2-1, where M1 and M2 array elements are located on the X and Y axes respectively. The number of array elements of the receiving array is N=N1+N2-1, where N1 and N2 array elements are located on the X' and Y' axes respectively. The four subarrays are uniform linear arrays, and the array element spacing is equal and set to d, taking d=λ / 2, λ is the wavelength of the signal. Assume that there are K signals s k (t), k = 1, 2,, K is incident on the array, K = K n +K c , the direction of the kth signal is recorded as (θ k1 ,θ k2 ,θ k3 ,θ k4 ), where (θ k1 ,θ k2 ) is the 2D-DOD of the kth signal, (θ k3 ,θ k4 ) is the 2D-DOA of the kth signal, and the incident signal includes K n Non-circular signal s n,k (t),k=1,2,,K n and K c Circle signal s c,k (t),k=1,2,,K c , the output vector x(t) of the array at sampling t is expressed as: x(t)=C(θ i1 ,i i2 ,i i3 ,i i4 )s(t)+n(t) (1) where x(t)=[x1(t),,x MN (t)] T Represents the receiving data matrix of the array; C = [c1, c2,, c K ] T represents the popularity matrix of the extended virtual array, is the extended virtual array manifold vector, a k and b k are the M×1 dimensional transmit and N×1 dimensional receive array populous vectors, respectively, expressed as n(t)=[n1(t),,n MN (t)] T represents the additive Gaussian white noise matrix, whose mean is zero and variance is σ 2 ; s(t)=[s1(t),,s K (t)] T represents the mixed incident source signal vector, and s(t) is re-expressed as in is the non-circular phase of the non-circular signal, Φ is a K×K'-dimensional matrix, K'=K n +2K c , a K'×1-dimensional vector containing K n Strictly non-circular signal s n (t) and the circular signal s c K of (t) c Real part and K c Imaginary part Rewrite the manifold matrix C as C=[C1(θ k1,n ,i k2,n ,i k3,n ,i k4,n )C2(θ k1,c ,i k2,c ,i k3,c ,i k4,c )]=[C1C2] (3) Where C1 is MN×K n The steering matrix for strictly non-circular signals, C2, is MN×K c Steering matrix for circular signals; Substituting equation (2) and equation (3) into equation (1), we can obtain In order to simplify the representation, some parts of the following text will omit the angle pair (θ k1 ,θ k2 ,θ k3 ,θ k4 ) and time t; (b) Four-dimensional parameter estimation In order to utilize the strictly non-circular characteristics of strictly non-circular signals and the virtual strictly non-circular characteristics of circular signals, the data vector x and its conjugate corresponding coupled data vector are combined into a new data vector y Among them, γ MN is a MN×MN dimensional switching matrix, is the 2MN×K' extended steering vector, which is expressed as Performing singular value decomposition on Y yields: Among them, 2MN×K' dimension U S and 2MN×(2MN-K') dimension U N are the left singular signal subspace and the noise subspace, respectively, L×K' dimension V S and L×(2MN-K') dimension V N are the right singular signal subspace and the noise subspace respectively, L is the number of snapshots, Σ S =diag(λ1,λ2,,λ K' ) and Σ N =diag(λ K'+1 ,λ K'+2 ,,λ 2MN ) represent the diagonal matrices consisting of K' and 2NM-K' eigenvalues ​​respectively; Define two selection matrices as follows Strictly non-circular and circular mixed signals about the angle θ k1 The selection matrix is Among them are Similarly, we get the angle θ kl The selection matrix (K l1 ,K l2 ); Based on the selection matrix K l1 and K l2 Based on the principle of the non-circular ESPRIT algorithm, we define the angle θ kl The direction matrix G l ; MR l =(K l2 IN S ) + K l1 IN S =EΘ l E H (12) Where E is a K'×K' unitary matrix, and the diagonal matrix Θ containing the angle information l The expression is G in formula (12) l Satisfying the joint diagonalization condition, define a set G = {G1, G2, G3, G4}, and obtain a unitary matrix E = [e1, e2, …, e] based on the joint diagonalization method K' ], E is θ kl The eigenvector of kl The eigenvalues ​​of θ maintain a one-to-one correspondence during the joint diagonalization process, without the need for angle pairing. kl The eigenvalue of is calculated by the following formula Then, it is easy to get from formula (14) Each circular signal is considered as two virtual strictly non-circular signals with equal angles, so θ kl There are K' angles, but in reality, there are only K incident signals, so the circular and non-circular signals are distinguished according to the number of repeated angles. Since the two estimated angles of the circular signal are reliable, the final non-circular signal angle θ is obtained by averaging. kl,c (l=1,2,3,4), in and are the first and second angles estimated for non-circular signals, respectively.

Citation Information

Patent Citations

  • Low-complexity receiving and transmitting angle joint estimation method for non-circular signal double-base MIMO radar

    CN103983952A

  • High-order cumulant based bistatic MIMO (Multiple Input Multiple Output) radar parameter estimation method

    CN104330783A

  • A low-complexity single-base MIMO radar non-circular signal angle estimation method

    CN107576940A