KR20180113267A - Method and apparatus for estimating direction of arrival using generation of virtual received signals based on uniform linear array antenna - Google Patents

Abstract

Disclosed are an improved direction of arrival (DOA) estimation method and apparatus which employ a linear prediction extension technique, thereby having excellent resolution even with a small number of antennas. In an ULA antenna in which N actual antennas are spaced apart by an equal interval d in a row, a conversion matrix indicating the relationship between received signals received from i^th to (N-2+i)^th antennas and a received signal received from an (N-1+i)^th antenna is obtained. The obtained conversion matrix is multiplied by values corresponding to received signals received from (i+1)^th to (N-1+i)^th antennas to generate a virtual received signal received from a new virtual antenna corresponding to an (N+i)^th antenna. As incrementing, by one, the number i of signals received from the virtual antenna, obtaining the conversion matrix and generating the virtual received signal are repeated until the number i reaches a preset number M. By using a preset number M of virtual received signals and signals directly received from an N number of actual antennas together, a DOA estimation algorithm such as the Bartlett algorithm is performed to calculate a DOA. Because the DOA estimation method and apparatus can be used even when antenna intervals are narrow, a grating lobe is not generated while maintaining a high resolution. It is not required to set an initial sector. The number of virtual antennas can be increased three times or more than the number of actual antennas.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for estimating an arrival angle using a virtual reception signal based on a uniform linear array antenna,

The present invention relates to a technique related to a uniform linear array (ULA) antenna, and more particularly, to a technique of generating a virtual received signal based on a ULA antenna and estimating a direction of arrival (DOA) And more particularly,

One of the most important problems in autonomous vehicle systems is human safety. Many sensors such as cameras, radio detection and ranging (RADAR) and light detection and ranging (LIDAR) are used in autonomous vehicles to prevent unexpected car accidents. Among various target detection sensors, radar that can be used in bad weather is widely studied.

For example, when two cars in front are adjacent to each other at the same distance from the radar sensor, the radar must be able to recognize that there are two cars ahead of them. If the velocities and ranges of the targets are similar, it is important to accurately estimate the DOA of the received signal received from the target to separate multiple targets. Therefore, high-resolution received signal DOA estimation algorithms such as the multiple signal classification (MUSIC) algorithm and the estimation of signal parameters via rotational invariance technique (ESPRIT) algorithms have been studied for decades come. Also, in recent years, the Bartlett algorithm, which is less affected by the signal-to-noise ratio (SNR) than the MUSIC algorithm, has received much attention and is being widely used.

When searching for a target using an ULA-type antenna, a narrow main beam width and a low side lobe are required to achieve precise resolution and high accuracy. When the aperture of the antenna array is wide, the beam width of the main lobe is narrow. The narrow beam width allows for the identification of closely adjacent targets. However, side lobes and grating lobes can occur and prevent positioning of the target.

Also, the wide aperture of this antenna array occupies a space that is too large.

Several methods have been proposed to solve these problems at once, but they have not yet been evaluated as a satisfactory solution.

It is better to increase the number of antennas in order to increase the resolution when measuring the arrival of the received signal of the radar. However, increasing the number of actual physical antennas increases the space occupied by them. Instead of increasing the number of actual antennas, a virtual antenna using the antenna can be used to increase the total number of antennas. The existing method is limited because the number of virtual antennas that can be increased is influenced by the actual number of antennas.

The present invention has the advantage that the number of virtual antennas that can be extended in the ULA antenna environment is increased compared to the prior art, the narrow beam width and the low side lobe are limited and the grating lobe is not generated at the limited antenna aperture size, The present invention also provides a method and an apparatus for estimating a received signal arrival angle using a virtual received signal generation based on a ULA antenna.

According to embodiments of the present invention, a received signal DOA estimation method using ULA antenna-based virtual receive signal generation is a method of estimating a received signal DOA using a ULA antenna in which N actual antennas are spaced apart by a uniform distance d , Obtaining a transformation matrix representing a relationship between received signals received by the (N-2) th antenna and received signals received by the (N-1 + i) th antenna; The transformed matrix obtained is multiplied by a value corresponding to the received signals received from the (i + 1) th to the (N-1 + i) th antennas, and the virtual received signal received by the new virtual antenna corresponding to the (N + ≪ / RTI > And a step of obtaining the transformation matrix until the number i reaches a preset number M while increasing the number i of signals received by the virtual antenna by one and generating the virtual reception signal recursively .

In one embodiment of the received signal DOA estimation method, the predetermined number M is determined in consideration of at least a signal-to-noise ratio (SNR) and an interval of the antennas, and a performance improvement of a received signal arrival angle (DOA) The maximum number of virtual antennas that can be received.

In one embodiment of the received signal DOA estimation method, the preset number M may be at least three times the actual number N of antennas.

In one embodiment of the received signal DOA estimation method, the antenna spacing d may not be greater than the wavelength? Of the received signal.

In one embodiment of the received signal DOA estimation method, the transform matrix may be obtained using a linear least squares (LLS) method based on a relationship between received signals.

In one embodiment of the method for estimating a received signal DOA, the method includes the steps of: N real antennas receiving radio signals in a time domain reflected back from a target ahead; Converting a radio signal of the received time domain into a signal of a frequency domain; And converting the signal corresponding to the bit frequency of the signal of the converted frequency domain into a signal of the time domain. Also, the signal used to obtain the transformation matrix may be a time domain signal corresponding to a signal corresponding to a bit frequency.

In one embodiment of the method for estimating a received signal DOA, the step of converting to a frequency domain signal includes: converting the received time domain radio signal into a complex signal form using Hilbert transform; Converting the complex-valued signal into a frequency-domain signal for performing Fourier transform; And calibrating the signal in the transformed frequency domain.

In one embodiment of the received signal DOA estimation method, the step of converting the signal corresponding to the 'bit frequency into the signal in the time domain' includes extracting a signal corresponding to a bit frequency among the signals in the converted frequency domain ; And converting the signal corresponding to the extracted bit frequency into an inverse Fourier transformed signal and a time domain signal.

In one embodiment of the reception signal DOA estimation method, a predetermined reception signal DOA estimation algorithm is performed by using the virtual reception signal of the predetermined number (M) and the reception signal directly received by the N actual antennas together And calculating the received signal DOA.

In one embodiment of the received signal DOA estimation method, the predetermined received signal DOA estimation algorithm may be a Bartlett algorithm.

According to another aspect of the present invention, there is provided a ULA receiving antenna including N actual antennas spaced apart from each other by an equal distance d and receiving radio signals incident from a front side, A receiver for extracting a predetermined signal from N received signals received through the antennas of the ULA receiving antenna and converting the signals into N digital signals; And a controller configured to receive the N digital signals provided by the receiver to generate M preset virtual reception signals received through a virtual antenna, to generate M virtual reception signals and N digital signals And a signal processing unit for calculating a reception signal DOA estimation value by using the received signal DOA estimation value. In the received signal DOA estimation device, the M virtual reception signals are (a) received signals received from the i-th to (N-2 + i) (B) multiplying the obtained transformation matrix by a value corresponding to the reception signals received from the (i + 1) th to the (N-1 + i) th antennas, (A) and (b) of a process of generating a virtual reception signal received by a new virtual antenna corresponding to the (N + i) -th antenna, the number i of signals to be received by the virtual antenna is incremented by one, By recursively repeating the process until it reaches the end.

In one embodiment of the received signal DOA estimator, the predetermined number M may be at least three times the actual number N of antennas.

In one embodiment of the received signal DOA estimator, the signal processor may obtain the transform matrix using a linear least squares (LLS) method based on a relationship between received signals.

(C) a process in which N actual antennas receive radio signals in a time domain reflected back from a target in front of the target, (d) (E) a signal used to obtain the transform matrix through a process of converting a signal corresponding to a bit frequency of the received signal of the transformed frequency domain into a signal in the time domain, Can be created.

In one embodiment of the reception signal DOA estimating apparatus, the signal processing unit may further include: (f) a signal processing unit for converting the 'received wireless signal in the time domain' into ' (G) a process of converting the complex number signal into a signal of a frequency domain for performing a Fourier transform, and (h) a process of calibrating a signal of the converted frequency domain. can do.

In the embodiment of the reception signal DOA estimating apparatus, the signal processing unit may be configured to (i) convert the signal corresponding to the bit frequency into a signal in the time domain, Extracting a signal corresponding to a bit frequency from the signal, and (j) performing inverse Fourier transform on a signal corresponding to the extracted bit frequency to convert the signal into a time domain signal.

The present invention creates a virtual antenna through extrapolation based on the relationship between the actual received signals reflected back from the target. The improved received signal DOA estimation technique using the transformed vector according to the present invention has a significantly higher resolution even when the antenna aperture size is small.

Even when a small number of antennas are used, the present invention can achieve the same effect as when there are many antennas. According to the present invention, unlike the conventional extrapolation method, the number of virtual antennas can be increased to at least three times the actual number of physical antennas. According to the present invention, the number of virtual antennas can be greatly increased by using the linear prediction extension, and the degree of resolution improvement is further increased.

Further, unlike the conventional interpolation method, the present invention is much simpler than the conventional interpolation method because it is not necessary to set the initial detection target sector when estimating the received signal DOA.

 The present invention has the advantage of not generating a grating lobe while maintaining a high resolution because it can be used even when the antenna interval is narrow. Since there is no grating lobe, the grating lobe is not misjudged as an actual target. That is, since a narrow beam width, a low side lobe and a low grating lobe are simultaneously provided at a limited antenna aperture size, high resolution can be realized.

The present invention can be applied to a radar for an autonomous vehicle. The enhanced resolution allows multiple vehicles near the front to be recognized as accurately as the actual number of vehicles without being mistaken for one. As a result, it is possible to precisely observe the front side and effectively prevent a vehicle collision accident or the like.

1 shows a wireless signal reception of a ULA antenna in which N antennas are arranged in the form of a ULA.
Figure 2 illustrates the configuration of a system for implementing the method according to the present invention.
3 is a flow chart illustrating an estimation procedure of a received signal DOA based on the method according to the present invention.
Fig. 4 shows a concrete execution procedure of step S200 of the flowchart of Fig.
FIG. 5 shows a concrete execution procedure of step S300 of the flowchart of FIG.
FIG. 6 shows a concrete execution procedure of step S400 of the flowchart of FIG.
Fig. 7 shows a calculation procedure for the received signal DOA estimation based on the method according to the invention in the DSP of the system of Fig.
FIG. 8 conceptually shows a method of increasing a virtual antenna receive signal using a virtual antenna according to the present invention.
FIG. 8 conceptually shows a method of the present invention for calculating a reception signal of a virtual antenna using a reception signal of an actual antenna.
Fig. 9 shows a situation in which two vehicles in front are recognized by applying the algorithm of the present invention to a radar system for a vehicle.
FIG. 10 illustrates a Bartlett pseudospectrum for an actual antenna array and an extended array using a virtual antenna, respectively.
11A and 11B are diagrams for explaining the algorithm according to the present invention for the case where the detection target sector of the radar apparatus is narrowed to [-5 deg., 5 deg.] And when it is wide to [-15 deg., 15 deg. And a Bartlet pseudo spectrum in a case where a Bartlet algorithm is applied by generating a virtual reception signal by a conventional interpolation method.
12 is a graph showing a change in a root mean square error (RMSE) with respect to the estimation performance of a target position according to a change in the number of virtual antennas generated by an extrapolation algorithm according to the present invention.
13 is a graph showing a change in an RMSE error with an increase in the number of antennas.
14 shows the received signal strength for each antenna when 64 antennas are used in a ULA array.
15 shows a change in the component value of the conversion vector u as the positional change of the virtual antenna and the increase in the number of antennas.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

는 다음과 같이 표현 될 수 있다.FIG. 1 shows that N antennas receive a radio signal reflected from a target ahead in a ULA antenna spaced apart by an equal distance d in a row. In such an ULA type antenna,

Can be expressed as follows.

......(1)

......(One)

,

는 전치 연산자, N 은 안테나의 수를 나타낸다.

는 스티어링 행렬(steering matrix)이다. 스티어링 행렬은 다음과 같이 주어진 스티어링 벡터

로 구성된다.here,

,

Denotes a permutation operator, and N denotes the number of antennas.

Is a steering matrix. The steering matrix is given by:

.

......(2)

......(2)

이고, 여기서 d 는 안테나 간격이다.

는 시간 t에서의 Lx1 입사 신호 벡터를 나타내며,

는 Nx1 제로 평균 백색 가우시안 잡음 벡터(zero mean white Gaussian noise vector)를 나타내고,

는 ULA 안테나 배열의 연장선과 직각을 이루는 직선과 각 안테나로 수신되는 입사 신호 벡터가 이루는 각을 나타낸다.Here, L is the number of the target ,? Is the wavelength of the received signal of the antenna, d _i is the distance from the first antenna to the i- th antenna, ULA is used in the embodiment of the present invention

, Where d is the antenna spacing.

Represents the Lx1 incident signal vector at time t,

Denotes a Nx1 zero mean white Gaussian noise vector,

Represents the angle formed by a straight line perpendicular to the extension line of the ULA antenna array and an incident signal vector received by each antenna.

The autocorrelation matrix of the received signal can be estimated as follows.

......(3)

(3)

는 기대값(Expectation)을 나타내고,

는 수신신호

의 상관행렬을 나타내며,

는 공액 전치 연산(conjugate transpose operation)을 나타낸다. 만약 신호가 에르고드적(ergodic: 상당한 기간이 지난 후, 하나의 체계가 최초의 상태와 거의 비슷한 상태로 돌아가는 조건 하에 있는)인 경우, 앙상블 평균(ensemble average)은 시간 평균으로 표현할 수 있으므로 시간 평균을 사용하여 자기 상관 행렬

를 계산할 수 있다.here,

Represents an expected value,

Lt; RTI ID =

≪ / RTI >

Represents a conjugate transpose operation. If the signal is ergodic, then the ensemble average can be expressed as a time average, so that if the system is under a condition that returns to a state almost similar to the original state, Using Autocorrelation Matrices

Can be calculated.

......(4)

......(4)

Here, K is the number of time samples.

On the other hand, the goal of the Bartlett Algorithm is to determine a weight vector that maximizes the power of the received signal while keeping the magnitude of the noise constant. The array output may be expressed as a value obtained by multiplying the received signal by a weight vector w .

......(5)

(5)

Where w is the Nx1 weight vector and y (t) is the weighted output of the received signal. If there is a signal coming from the angle θ, the array output can be expressed as:

......(6)

(6)

이고,

은 잡음 분산을 나타낸다. 잡음 성분의 크기를 일정하게 유지하기 위해,

로 설정한다. 따라서 식 (6)의 해는 다음과 같다.here

ego,

Represents the noise variance. To keep the size of the noise component constant,

. Therefore, the solution of equation (6) is as follows.

......(7)

(7)

The output spectrum of the Bartlett algorithm can be expressed as:

......(8)

......(8)

2 shows an exemplary configuration of a ULA antenna based radar system for implementing the method according to the present invention. The radar system 10 includes a receiving unit 30 connected to a ULA receiving antenna 60, a transmitting unit 20 connected to a ULA transmitting antenna 70, a digital signal processing unit connected to the transmitting unit 20 and the receiving unit 30, 40). The radar system 10 may further include a user interface 50 connected to the signal processing unit 40.

The ULA receive antenna 60 and the ULA transmit antenna 70 may each include a plurality of antennas. In particular, the ULA receiving antenna 60 may have a ULA type antenna array in which a plurality of antennas are arranged in a line at equal intervals. The ULA receiving antenna 60 can receive a reflected signal that is reflected from a target ahead of the radio frequency signal transmitted through the transmitting antenna 70 and returned.

The transmission unit 20 wirelessly transmits the radar signal to the target ahead through the transmission antenna 70. [ According to one embodiment, the transmitter 20 may include a waveform generator 24, an oscillator 26, and a power amplifier (PA) 28. The waveform generator 24 can generate a signal having an analog waveform of a desired period and shape based on the digital transmission signal provided by the signal processing unit 40. [ For example, the signal generator 24 may provide a triangular waveform modulated signal (triangular wave) to the oscillator 26 as a transmit signal. The oscillator 26 converts the transmission signal generated by the waveform generator 24 into a high frequency radio frequency (RF) signal for wireless transmission. The oscillator 26 may, for example, perform frequency modulation of the transmit signal provided by the waveform generator 24. [ The oscillator 26 may also provide the converted RF signal to the mixer 34 of the receiver 30 as a reference signal. The power amplifier 28 amplifies the RF signal output from the oscillator 26 to an output required for transmission and provides the amplified RF signal to the transmission antenna 70. The oscillator 26 may be composed of, for example, a voltage controlled oscillator (VCO).

The receiving unit 30 receives the RF signal reflected from the target after being transmitted from the transmitting antenna 70 and reflected back to the target through the receiving antenna 60. The receiving unit 30 down-converts the RF signal based on the reference signal provided from the oscillator 26 of the transmitting unit 20, converts the RF signal into a digital signal, and provides the digital signal to the signal processing unit 40. The receiver 30 includes a low-noise amplifier (LNA) 32, a mixer 34, and an analog-to-digital converter 34 for each antenna constituting the receiving antenna 60. The low- -Digital Converter: ADC, 38).

The low noise amplifier 32 is connected to a corresponding antenna of the ULA receiving antenna 60 to amplify a weak reception signal captured by the ULA receiving antenna 60. The received signal amplified by the low-noise amplifier 32 is supplied to the mixer 34. [ The mixer 34 may downconvert the received signal based on the frequency difference between the amplified received signal and the RF signal provided from the oscillator 26 of the transmitter 20. That is, in the mixer 34, the amplified received signal and the RF signal provided from the oscillator 26 of the transmitter 20 are mixed, the frequency difference between the two signals is calculated, and the calculated difference frequency is converted into a frequency A beat signal is obtained. The bit signal obtained in the mixer 34 is converted into a digital signal by the ADC 38. The digital received signal thus obtained is provided to the signal processing unit 40. The receiving unit 30 may further include a low-pass filter (LPF) 36 for removing a low-frequency component included in the bit signal output from the mixer 34.

According to one embodiment, the signal processing unit 40 can control the overall operation of the transmitting unit 20, the receiving unit 30, and the user interface 50. The signal processing unit 40 receives the digital information corresponding to the signal reflected from the forward target from the receiving unit 30 and performs arithmetic processing according to the method described below to estimate the DOA of the received signal received from the forward target. In addition, the signal processing unit 40 generates information to be transmitted as a target through the transmitting antenna 70 through signal processing, and provides the information to the transmitting unit 20. The signal processing unit 40 may be implemented by, for example, a digital signal processor (DSP), a microcomputer, or the like.

The user interface (UI) 50 displays the processing result of the signal processing unit 40 or delivers a user's instruction to the signal processing unit 40. [

The configuration of the radar system 10 shown in FIG. 2 is merely an example, and it is needless to say that the radar system 10 may have a different configuration depending on the demodulation method and the like of the radio signal. The radar system 10 may be installed in, for example, a vehicle and used as a radar apparatus for a vehicle.

The method of generating a virtual antenna signal according to the present invention can be implemented by a program. The program can be embedded in the signal processing unit 40 and executed thereby.

3 is a flow chart illustrating an estimation procedure of a received signal DOA based on the method according to the present invention. Fig. 7 shows the calculation process for the received signal DOA estimation based on the method according to the invention in the signal processing section 40 of the radar system 10 of Fig.

A method for estimating a received signal DOA according to the present invention will be schematically described in the radar system 10. In the radar system 10, the transmitting unit 20 generates an RF signal based on the digital transmission signal provided by the signal processing unit 40, and transmits the RF signal through the transmission antenna 70. [ The RF signal transmitted from the transmission antenna 70 as a target is reflected on the target in front of the target. Each antenna of the ULA receiving antenna 60 can receive the RF signal reflected back to the target and transmit it to the receiving unit 30. The receiving unit 30 generates a beat signal based on the frequency difference between the RF signal received from each antenna 60 and the RF signal provided from the oscillator 26 and converts it into a digital signal And provides it to the signal processing unit 40 (step S100).

The reception signal provided by the reception unit 30 to the signal processing unit 40 is a time domain signal. And converts it into a frequency domain signal (step S200). The flow chart of FIG. 4 shows a concrete execution procedure of step S200 of the flowchart shown in FIG. The signal in the time domain provided by the receiver 30 may be converted into a complex signal form using a Hilbert Transform algorithm in the signal processor 40 in operation S210. Then, the transformed complex signal can be converted into a frequency domain signal using Fast Fourier Transform (FFT) algorithm (operation S220). The converted frequency domain signal can be calibrated (step S230).

A signal corresponding to a beat frequency among the signals converted into the frequency domain can be converted into a signal in the time domain again (operation S300). The flow chart of FIG. 5 shows a concrete execution procedure of step S300. In operation S310, a bit frequency signal is extracted from the frequency domain signal obtained through the Fourier transform. And performs inverse fast Fourier transform (IFFT) on the extracted bit frequency signal to convert it into a time domain signal in step S320.

When a time domain signal is obtained in this way, a virtual reception signal received through a virtual antenna is generated using the converted time domain signal (step S400).

When a virtual reception signal of a desired number of virtual antennas is obtained, an algorithm for estimating a reception signal DOA is performed using the virtual reception signal and a signal received through an actual physical antenna together at step S500.

The present invention proposes an improved received signal DOA estimation algorithm, for example, to increase the resolution of the Bartlett algorithm and to lower the level of lattice lobe and side lobes. In the conventional array interpolation algorithm, the sector indicating the visual range should be set. If a target is present outside the sector, conventional interpolation algorithms can not accurately estimate the position of the target. You also need to perform additional steps to set up the initial sector. However, the method according to the present invention does not need to set a sector.

The flowchart of FIG. 6 shows the procedure of step S400 of generating a virtual received signal in more detail. 8 is a diagram illustrating a case where a ULA receiving antenna 60 is composed of four actual antennas arranged in the ULA form and a virtual antenna is generated using the received signals of the four actual antennas, And conceptually illustrates the method of the invention. The process of generating a virtual received signal with reference to these drawings (step S400) will be described in more detail.

It is assumed that there are N actual antennas constituting the ULA receive antenna 60 (Fig. 8 is a case where N = 4). (Steps S420 and S430) of generating a virtual reception signal received through the generated virtual antenna while repeating steps (S420 and S430) while increasing the number i of reception signals received via the virtual antenna by one recursively. The process of generating such a virtual received signal (steps S420 and S430) is repeated until the number i of virtual antennas becomes equal to the number M of the preset maximum virtual antenna (or the virtual received signal received through it) (Step S440).

Specifically, a value of 1 is assigned to the number i of virtual antennas (step S410). In order to generate a virtual reception signal, a transformation matrix (Transformation Matrix) representing a relationship between a reception signal received from the first to (N-1) th antennas and a reception signal received at the Nth antenna is obtained (step S420 when i = 1). FIG. 8A illustrates the case where N = 4, where signals X ₁ , X ₂ , and X ₃ received by the first to third antennas and signal X ₄ calculates a transformation matrix U ₁ represents the relationship between. The transform matrix U ₁ is a vector that multiplies the received signals of antennas 1, 2, and 3 by a weight to generate a received signal of antenna 4, which can be obtained using the following equation.

......(9)

(9)

......(10)

(10)

(N + 1) -th virtual reception signal by multiplying the obtained transformation matrix U ₁ by a value corresponding to a reception signal received from the second through N-th N-1 antennas (step S430 when i = 1 ). This virtual received signal corresponds to a virtual signal received through the newly generated (N + 1) th antenna (this is the first virtual antenna). Figure 8 (b) illustrates this. That is, as shown in equation (11), the conversion matrix U ₁ obtained previously is multiplied by the signals received by the antennas 2, 3, and 4 to generate a reception signal X ₅ of the fifth antenna (i.e., the first virtual antenna) Lt; / RTI >

......(11)

(11)

Steps S420 and S430 are repeatedly performed until the number i of the received signals of the virtual antennas thus formed reaches the number M of the preset maximum virtual antennas (or the virtual reception signals received therethrough). Here, the number M of the maximum virtual antennas (or the virtual reception signals received through the virtual antennas) can be defined as the maximum number of virtual antennas (virtual reception signals) capable of improving the performance of the received signal DOA resolution by applying the method of the present invention . The value of M may vary depending on the SNR in the actual use environment, the interval between the antennas constituting the ULA antenna 60, and the like. According to actual tests, the number M of virtual antennas can be approximately three times or more the number of actual antennas.

8 (c), a transformation matrix U ₂ indicating the relationship between X ₂ , X ₃ , X ₄ and X ₅ is obtained in the same manner as described above, and then X ₃ , X ₄ , and X ₅ are multiplied by a transformation matrix U ₂ and i = conceptually shown to produce two new virtual received signal X ₆ of the time of. As shown in FIG. 8C, the reception signals X ₂ , X ₃ , and X ₄ of the three actual antennas before the last signal among the five reception signals are multiplied by weights in the same manner as described above, for obtaining the transformation matrix U ₂ to make the reception signal X ₅ is multiplied by the following, X _3, X _4, transform matrix U ₂ to X ₅ conceptually to the generation of the i = 2 the received signal X of the new virtual antenna ₆ when the .

In this manner, the transformation matrix U _i is obtained while increasing the i value by 1, and the process of generating the virtual reception signal using the obtained transformation matrix U _i is repeated until M virtual reception signals are generated. Figure 8 (d) illustrates this. The M virtual antennas receiving the M virtual signals are respectively placed at the left and right sides of the physical ULA antenna 60, that is, in the order of M / 2. It is of course possible that the M virtual antennas are located on either the left or right side of the physical real ULA antenna 60 or the number of virtual antennas arranged on the left and right sides are different.

The method of the present invention for generating a virtual received signal while increasing the virtual antenna is generalized using the following equation.

The transformation matrix U _i may be represented by a transformation vector u . The transform vector u Can be obtained using the Linear Least Square (LLS) method. The conversion vector is obtained by using the correlation matrix of the received signal in the existing linear prediction method. In the conventional linear prediction method, the number of virtual antennas can be increased up to twice the number of actual antennas.

와 다른 안테나의 수신신호

의 관계를 이용하여 제1 안테나로부터 Nd 만큼 떨어진 N+1 번째 안테나의 수신신호를 예측할 수 있다.On the other hand, the present invention directly calculates the transform vector u using the relation between the received signals, so that it is possible to generate more virtual antennas than the conventional method. A virtual antenna of at least three times the actual number of antennas can be generated. If there are N ULA antennas, the received signal of the Nth antenna

And the reception signal of the other antenna

The reception signal of the (N + 1) th antenna that is Nd away from the first antenna can be predicted.

If you use an array antenna with a large antenna spacing, the resolution (resolution) that distinguishes multiple targets at similar points in the forward direction is better. However, when the antenna interval is larger than 0.5 ?, a somewhat large side lobe is generated. If the antenna spacing is greater than 1.0 [lambda], some grating lobes having a size similar to the size of the main lobe are created. It is difficult to distinguish the grating lobe from the target because the grating lobe is close to the center and is similar in size to the main lobe when it occurs.

However, using the ULA-based virtual receive signal generation method according to the present invention, a high-resolution beam can be formed. Even when the spacing between the antennas of the ULA antenna 60 is shorter than 1.0 ?, the resolution is higher than that of the existing Bartlett algorithm using antenna spacings larger than 1.0 ?. It has the advantage of being able to distinguish the targets without causing a grating lobe. Based on the above description, the i th estimated signal can be obtained using the following equation.

......(12)

(12)

는 (i+N-1) 번째 위치에 안테나가 있을 때 수신 될 것으로 예상되는 선형 예측 확장에 의해 생성된 (i-1) 번째 수신신호를 나타내고, 즉

는 이미 결정한 수신신호 벡터를 나타낸다.here,

Represents an (i-1) -th received signal generated by a linear prediction extension that is expected to be received when there is an antenna at the (i + N-1) -th position,

Indicates a previously determined received signal vector.

......(13)

(13)

와

사이의 차이의 노름(norm)을 최소화하는 변환 벡터

를 얻는다. 이는 다음과 같이 주어진다.Using the linear least squares method in Eq. (13)

Wow

A transformation vector that minimizes the norm of the difference between

. This is given by

......(14)

(14)

를 사용하여 (i+N-1) 번째 안테나가 존재한다고 가정하면, 그 안테나로 들어오는 수신신호는 다음과 같이 예측 될 수 있다.(14) < / RTI >

Assuming that (i + N-1) th antenna exists, the received signal to the antenna can be estimated as follows.

......(15)

(15)

는 실제 안테나가 있는 위치에서 생성된 신호이므로 더 정확한 신호

를 사용한다. The received signals of the i + N + 1, i + N + 2, ..., (N + M) th virtual antennas can be predicted using the above method. M is the number of virtual antennas.

Is a signal generated at a position where an actual antenna exists, so that a more accurate signal

Lt; / RTI >

When a desired number of reception signals of the virtual antennas are obtained through the above process, an algorithm for estimating the reception signal DOA is performed. That is, the received signal DOA estimation algorithm is performed using the signals received by the N actual antennas and the M estimated reception signals generated by these signals. The received signal DOA estimation algorithm may be, for example, a known Bartlet algorithm or a new known received signal DOA estimation algorithm which may be developed in the future or other known algorithms.

The ULA-based virtual receive signal generation method according to the present invention can be applied to a radar device for target detection, so that the received signal DOA can be estimated with a high resolution. FIG. 9 shows a situation in which two vehicles 200a and 200c ahead are recognized by applying a ULA-based virtual reception signal generation algorithm of the present invention to, for example, a radar device 10 for an autonomous vehicle 100. FIG. FIG. 10 is a diagram showing the Bartlett pseudospectrum of the case where the Bartlett algorithm applying the existing interpolation method is applied to the actual antenna array (antenna spacing d = 0.8λ and 2.0λ) (hereinafter referred to as 'Case 1' In the case of applying the Bartlet algorithm for an extended antenna array using an actual antenna (antenna spacing d = 0.8?) And a virtual antenna generated according to the algorithm of the present invention (extrapolation method) together (hereinafter referred to as Case 2) Illustrate the Bartlett pseudospectrum. In the simulation, ULA antenna with four antennas was used. In all simulations, the SNR is set to 10dB and the black dotted line indicates the position of the two targets. The angle between the radar device 10 and the two target vehicles 200a and 200b is [-3 DEG, 3.5 DEG]. The number of virtual antennas (M) is 16. The positions of the actual antenna and the virtual antenna are the same in both cases.

Consider the case where two vehicles 200a and 200b are adjacent to each other at substantially the same distance from the radar device 10 (a situation where the angle? Between two targets is small). Conventional radar apparatuses using only the signals received through the actual antennas and applying the Bartlet algorithm have a poor resolution, and mistakenly believe that there is one vehicle ahead of the two vehicles (see FIG. 9 (b) and FIG. 10 (See the blue graph in the previous section). Using a ULA antenna array having a large antenna interval (for example, when antenna spacing d = 2?), The resolution for distinguishing a vehicle is improved, but a grating lobe is generated to cause another problem See the black graph on the right). However, the radar apparatus 10 applying the Bartlett algorithm using the received signal of the virtual antenna by applying the algorithm of the present invention may be applied to a ULA antenna array having a small antenna interval (for example, when the antenna interval d = 0.8λ) So that the two vehicles 200a and 200b on the front side can be recognized as separate vehicles (refer to the red graph 310 of FIG. 10). Therefore, the algorithm of the present invention is advantageous in that even if the antenna interval d of the ULA antenna array is set to a value not larger than the wavelength? Of the received signal of the antenna, the resolution for the target can be improved without generating the grating lobe.

There is a difference in the resolution of the estimation of the received signal DOA according to the range of the detection target sector of the radar device. 11A shows a case in which a reception signal of a virtual antenna is generated by an algorithm according to the present invention when the detection target sector of the radar apparatus is narrowed to [-5 DEG, 5 DEG], and the Bartlet algorithm is applied, And a Bartlett pseudo spectrum when the Bartlet algorithm is applied. The blue graph 330 represents the former, and the red graph 340 represents the latter. The antenna spacing d is 0.8?, The actual number of antennas N is 4, and the number of virtual antennas M is 16. The positions of the actual antenna and the virtual antenna are the same in both cases. If the range of the target sector to be detected is narrowed, the received signal DOA estimation error shows that the case of applying the conventional interpolation method is larger than that of the method of the present invention.

11 (b) shows a case in which the reception signal of the virtual antenna is generated by the algorithm according to the present invention and the Bartlet algorithm is applied when the detection target sector of the radar device is wide [-15 degrees, 15 degrees] And a Bartlett pseudo spectrum when the Bartlet algorithm is applied. The 350 blue graph corresponds to the former case, and the 360 red graph corresponds to the latter case. It can be seen that the conventional interpolation method algorithm significantly degrades the angular estimation performance when the sector is widened as compared with the algorithm of the present invention. Also, the conventional interpolation algorithm can see that when the sector is widened, the grating lobes occur. If the target is not within that sector range, performance will be worse.

On the other hand, how much the number of virtual antennas can be increased and the validity of the increased number of virtual antennas will be described. The performance of estimating the position of the target varies as the number of antennas generated by extrapolation increases. The performance may be analyzed using a root-mean-square error (RMSE).

......(16)

(16)

는 m 번째 시뮬레이션에서 k 번째 표적의 추정된 각도를 나타낸다. 같은 조건에서 1000번의 시뮬레이션을 수행하였다. N _s represents the number of simulations,

Represents the estimated angle of the k-th target in the m-th simulation. 1000 simulations were performed under the same conditions.

12 is a graph illustrating a change in RMSE according to a change in the number of virtual antennas generated by the extrapolation algorithm according to the present invention. If the target is located at 3 ° and -3 ° in case of SNR = 10dB, the resolution is not good at first and the RMSE is high because the two targets can not be separated. However, as the number of antennas generated by the linearly predicted extension increases, the RMSE gradually decreases and saturation is observed within a certain range from a certain moment.

13 shows a change in RMSE with an increase in the number of antennas. As shown in FIG. 13, as the number of virtual antennas generated by the expansion increases, the error of the signal received by the newly generated virtual antenna increases. However, if the total number of antennas exceeds about 20, the strength of the power is weakened as the new received signal of the virtual antenna is generated, so that the error value converges to one.

14 shows the received signal strength for each antenna when 64 antennas are used in a ULA array. The red graph 370 is 60 real antenna arrays and the blue graph 380 is the intensity of the received signal of 60 extended antenna arrays obtained by performing the method of the present invention using four real antennas. As shown in Fig. 14, when the number of virtual antennas increases by a certain number, the power of the received signal becomes too small. The error value converges to 1 as shown in Fig. 13, and the RMSE value does not decrease as shown in Fig.

Generally, approximately 16 extended virtual antenna receive signals are sufficient to improve the position of the target, the received signal DOA estimation performance, although it is affected by SNR and antenna spacing.

15 shows a change in the component value of the conversion vector u as the positional change of the virtual antenna and the increase in the number of antennas. Since a ULA with a relatively narrow antenna spacing is used, the correlation between the signals received by each antenna is high. Therefore, it can be seen that the value of the transformation vector u remains constant even if the number of antennas increases greatly. If the elements of the transformation vector u continue to change from the beginning, a significant amount of computation is required to obtain the new u . However, u is not changed until the number of virtual antennas increases by more than 40, which is very effective in terms of calculation.

The improved received signal DOA estimation method using the conversion vector u according to the present invention described above can be implemented as a program. The program can be embedded in the signal processing unit 40 and implemented.

The improved received signal DOA estimation method using the transform vector u according to the present invention has a considerably high resolution even when the antenna aperture size is small. In addition, this algorithm can achieve the same effect as when there are many antennas even when a small number of antennas are used. The procedure is simpler than the existing interpolation method because there is no need to set up a sector. Since there is no grating lobe, there is no case where the grating lobe is misjudged as an actual target.

The present invention can be applied to a radar system that recognizes a target ahead. For example, it is effective when applied to a radar for an autonomous vehicle. For example, the method of the present invention can be embodied as a program and embedded in a signal processing unit 40 employed in a radar apparatus of an autonomous vehicle.

10: ULA antenna-based radar system 20:
24: Waveform generator 26: Oscillator
28: Power amplifier (PA) 30: Receiver
32: Low Noise Amplifier (LNA) 34: Mixer
36: Low-pass filter (LPF) 38: AD converter
40: a signal processor 50: a user interface
60: ULA receiving antenna 70: ULA transmitting antenna

Claims (16)

In a ULA antenna in which N actual antennas are spaced apart by an equal distance d in a row,
obtaining a conversion matrix indicating a relationship between received signals received from an i-th antenna to an (N-2 + i) -th antenna and a received signal received through an (N-1 + i) -th antenna;
The transformed matrix obtained is multiplied by a value corresponding to the received signals received from the (i + 1) th to the (N-1 + i) th antennas, and the virtual received signal received by the new virtual antenna corresponding to the (N + ≪ / RTI > And
Obtaining the transformation matrix until the number i reaches a predetermined number M while incrementing the number i of signals received by the virtual antenna by one, and recursively performing the step of generating the virtual reception signal And estimating a received signal arrival angle (DOA) using a ULA-based virtual receive signal generation. 2. The method of claim 1, wherein the predetermined number M is determined in consideration of at least a signal-to-noise ratio (SNR) and an interval of the antennas, wherein a maximum number of virtual antennas And estimating a received signal DOA using a ULA-based virtual receive signal generation. 3. The method of claim 2, wherein the predetermined number M is at least three times the number N of the actual antennas. The method of claim 1, wherein the antenna spacing d is not greater than a wavelength? Of the received signal. 2. The method of claim 1, wherein the transform matrix is obtained using a linear least squares (LLS) method based on a relationship between received signals. Way. 2. The method of claim 1, further comprising: receiving N time-domain radio signals reflected back from a target in the time domain; Converting a radio signal of the received time domain into a signal of a frequency domain; And converting a signal corresponding to a bit frequency of the signal of the converted frequency domain into a signal of a time domain,
Wherein the signal used to obtain the transform matrix is a time domain signal corresponding to a signal corresponding to a bit frequency. 7. The method of claim 6, wherein the step of converting the frequency domain signal comprises:
Converting the received radio signal of the time domain into a complex signal form using Hilbert transform; Converting the complex-valued signal into a frequency-domain signal for performing Fourier transform; And calibrating a signal of the transformed frequency domain. 2. The method of claim 1, further comprising: The method of claim 6, wherein the converting the signal corresponding to the bit frequency into a signal in the time domain includes:
Extracting a signal corresponding to a bit frequency among the signals of the converted frequency domain; And transforming a signal corresponding to the extracted bit frequency into an inverse Fourier transformed signal to obtain a time domain signal. 2. The method of claim 1, further comprising calculating a received signal DOA by performing a predetermined reception signal DOA estimation algorithm using the predetermined number M of virtual reception signals and the reception signals directly received by the N actual antennas And estimating a received signal DOA using a ULA antenna based virtual receive signal generation. 10. The method of claim 9, wherein the predetermined reception signal DOA estimation algorithm is a Bartlett algorithm. A ULA receive antenna including N actual antennas spaced apart in an array with an equal distance d and receiving a forwardly received radio signal;
A receiver for extracting a predetermined signal from N received signals received through the antennas of the ULA receiving antenna and converting the signals into N digital signals; And
The M reception apparatus comprising: a reception unit configured to receive the N digital signals provided by the reception unit to generate M preset virtual reception signals received through a virtual antenna, and to generate the M virtual reception signals and the N digital signals provided by the reception unit, And a signal processing unit for calculating a reception signal DOA estimation value together,
The M virtual reception signals include (a) a transformation matrix representing the relationship between the reception signals received from the ith antenna to the (N-2 + i) th antenna and the reception signal received to the (N-1 + i) (B) multiplying the obtained transformation matrix by a value corresponding to the reception signals received from the (i + 1) -th to the (N-1 + i) th antennas, (A) and (b) of a process of generating a virtual received signal received by a virtual antenna are repeated one by one until the number i reaches M while increasing the number i of signals received by the virtual antenna by one, And generating a received signal DOA using the ULA-based virtual receive signal generation. 12. The apparatus of claim 11, wherein the predetermined number M is at least three times the number N of the actual antennas. 12. The method of claim 11, wherein the signal processor obtains the transform matrix using a linear least squares (LLS) method based on a relationship between received signals. A received signal DOA estimator. 12. The method of claim 11, wherein the signal processing unit comprises: (c) a process of receiving N time-domain radio signals reflected from a target in the forward direction and returning the time domain; (d) And (e) generating a signal used to obtain the transform matrix through a process of converting a signal corresponding to a bit frequency among the received signals of the transformed frequency domain into a signal in a time domain, An apparatus for estimating a received signal DOA using an antenna - based virtual receive signal generation. The method of claim 14, wherein the signal processing unit further comprises: (f) converting the received radio signal of the time domain into a complex signal form using Hilbert transform, (G) a process of converting the complex-valued signal into a signal of a frequency domain for performing a Fourier transform, and (h) a process of calibrating a signal of the transformed frequency domain. Receiving signal DOA estimation apparatus using virtual - based receiving signal generation. The signal processor as set forth in claim 14, wherein, for the processing (e) of converting the signal corresponding to the bit frequency into a signal in the time domain, (i) And (j) performing inverse Fourier transform on the signal corresponding to the extracted bit frequency to convert the signal into a time domain signal. The reception signal DOA estimation using the ULA antenna based virtual reception signal generation Device.

Images