KR20190124488A - Method of signal subspace based DoA estimation for automotive radar system - Google Patents

Abstract

Provided is a method capable of estimating an angle of arrival even when a FullRank condition is not satisfied in a frequency domain in a technique for estimating an angle of arrival based on a subspace. The reason for estimating the angle of arrival in the frequency domain without estimating the angle of arrival in a time domain is that the angle can be separated only for a desired target because the peak appears at a frequency bin corresponding to a distance of the target in the frequency domain. According to the present invention, using a signal eigenvector without using a noise eigenvector, the angle of arrival based on a subspace is estimated using a signal eigenvector corresponding to an eigenvalue instead of zero root resulting from eigenvalue decomposition with only one bin in the frequency domain.

Description

Method and apparatus for estimating angle of arrival based on signal subspace in radar systems {Method of signal subspace based DoA estimation for automotive radar system}

The present invention provides a technique for reducing the amount of computation in the spectral estimation method of the subspace based arrival angle estimation method for estimating the dirrection of arrival (DoA) of a plurality of targets in a radar system having a plurality of receiving antennas. It is about.

The radar sensor transmits a signal (radar waveform) designed to detect a target, receives a signal reflected from the target, and performs signal processing to detect a distance, speed, and angle of the target. Radar waveforms are classified into pulse radars, continuous waveform (CW) radars, and frequency modulated CW (FMCW) radars.

이고, f_B,up은 업처프에서 검출된 표적에 해당하는 비트 주파수이고, f_B,dn은 다운처프에서 검출된 표적에 해당하는 비트 주파수이다.1 illustrates a method of detecting a distance and a speed by using an FMCW radar waveform as an example. (a) is a triangle shaped frequency modulated CW (FMCW) waveform composed of up-chirp and down-chirp for detecting distance and speed, and (b) is distance and speed using the same. Calculation formula. here

And f _{B, up} is a bit frequency corresponding to the target detected at upchirp, and f _{B, dn} is a bit frequency corresponding to the target detected at downchirp.

However, by transmitting the radar waveform of FIG. 1 and processing the received signal, the distance and velocity of the target can be measured, but the angle must be measured using other methods. For example, two or more reception antennas are used to measure an angle, and a direction of arrival estimation (DoA) is used for angle detection. The angle of arrival estimation steers all the array antennas to find the angle of incidence with the largest output signal.

2 shows a receiving antenna array structure for estimating angle of arrival. The receiving antennas 10-0, 10-1, 10-2,..., And 10-7 for estimating the angle of arrival θ of the signal are arranged at intervals of d. The angle of arrival estimation algorithm 12 performs conventional beamforming, capon, and MUSIC according to the weights w0 to w7 applied to the outputs of the array antennas 10-0, 10-1, 10-2, ..., 10-7. (multiple signal classification), ESPRIT, and ML. Conventional beamforming technique and capon technique have low angular resolution and ML technique has a large amount of calculation, so MUSIC, ESPRIT, and Root-MUSIC techniques, which are subspace-based high resolution techniques, are mainly used.

The core of the subspace-based approach angle estimating technique is that the inner product of the noise eigenvector and the steering vector of the signal is close to zero because the steering vectors corresponding to the incident signals are orthogonal to each other. However, Ergodic means that the ensemble mean, which is the average of random variables obtained by fixing time, is the same as the time average obtained by fixing the sample and averaged over time, but not Ergodic in real environment).

The ESPRIT and Root-MUSIC methods in the subspace method use the subarray instead of the direct spectral estimation method to find the highest signal strength by directing the steering vector containing the antenna phase difference for each angle. Find. However, the above scheme is applicable only when the receiving array antennas are designed at equal intervals.

Among the subspace techniques, MUSIC is applicable to all common array antennas instead of using spectral estimation. In order to estimate the angle of arrival by applying the MUSIC algorithm, the covariance matrix of the received signal must be obtained.

Here, the covariance R must satisfy the Full Rank condition, and the number of full ranks means the number of reception antennas. If the eigenvalue of the covariance matrix is not 0, but the number of ranks is 0 except for the rank number, it is known that the subspace-based arrival angle estimation that distinguishes the signal eigenvector from the noise eigenvector cannot be used. .

The problem to be solved by the present invention is to enable the angle of arrival estimation even when the full rank condition is not satisfied in the frequency domain in the signal subspace based angle of arrival estimation technique in the radar system.

In order to solve the above problem, in the present invention, when the signal eigenvector is used without using the noise eigenvector, the signal covariance matrix can be obtained even in the frequency domain even if the signal is not full rank. We present a solution to the problem of lowering the resolution of frequency components during the process.

The present invention proposes a method and apparatus capable of estimating the angle of arrival even when the full rank condition is not satisfied in the frequency domain in the subspace based method of the angle of arrival. The reason for estimating the angle of arrival in the frequency domain without estimating the angle of arrival in the time domain is that the angle can be separated only for the desired target because the peak appears at the frequency bin corresponding to the distance of the target in the frequency domain. .

In more detail, the concept of the present invention can be obtained by performing fast Fourier transform (FFT) on a single digitally converted chirp signal to obtain data of frequency characteristics (magnitude, phase) components for each reception channel.

After finding the location of the FFT bin corresponding to the target detected through the CFAR process, a signal vector is generated by combining the data of the index corresponding to the target to each receiving channel. At this time, a covariance matrix (M × M) is generated using a signal vector (1 × receiving channel number M) corresponding to the FFT bin of the target. Eigenvalue decomposition is performed on the covariance matrix to find the eigenvalues and eigenvectors. The spatial spectrum is calculated to detect the angle of the target.

The configuration of the present invention described above will be more clearly understood by the drawings and embodiments described later.

Since the present invention takes fewer frequency bins than the number of antennas, it is possible to reduce the number of FFT points, which can be expected to be faster than the general scheme. Since the present invention is capable of fast operation, its utility will be increased, particularly when utilized for automobiles.

1 is an illustration of a distance and speed detection method using the FMCW radar waveform
2 is a diagram illustrating an arrangement of a receiving antenna for estimating angle of arrival
3 is a waveform diagram for explaining the case where there is another peak near the peak
4 is a diagram illustrating a radar signal processing configuration according to an embodiment of the present invention.
5 is a graph showing the results of generating simulated received signals in MATLAB

Advantages and features of the present invention, and methods for achieving them will be apparent with reference to the embodiments described in detail in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but will be implemented in various forms, and only the present embodiments are intended to complete the disclosure of the present invention, and those skilled in the art to which the present invention pertains. It is provided to fully inform the scope of the invention, and the invention is defined by the description of the claims.

On the other hand, the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. In this specification, the singular also includes the plural unless specifically stated otherwise in the phrase. As used herein, “comprises” or “comprising” means the presence of one or more other components, steps, operations and / or elements other than the components, steps, operations and / or elements mentioned or Does not exclude additional

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. First of all, in adding reference numerals to the elements of each drawing, the same elements are denoted by the same reference numerals as much as possible even though they are shown in other drawings, and in describing the present invention, If the description can obscure the subject matter of the present invention, the detailed description is omitted.

According to the method and apparatus for estimating angle of arrival even when the full rank condition is not satisfied in the frequency domain in the subspace-based angle of arrival estimating method proposed by the present invention, the present invention arrives in the frequency domain without estimating the angle of arrival in the time domain. Estimate the angle. The reason is that since a peak appears at a frequency bin corresponding to a distance of a target in the frequency domain, the angle can be separated only for a desired target.

Fast Fourier Transform (FFT) is used to convert to the frequency domain. As the number of points in the FFT increases, the interval between frequency samples decreases, thereby increasing the position accuracy of the target.

If the covariance matrix satisfies the Full Rank condition, the resolution of the frequency component is lowered. In general, when using a single concubine, a signal matrix is constructed using the data near the peak to perform screening. For example, if there are 4 receiving channels, the frequency resolution is 64 point FFT (256/4 = 64) while generating a signal matrix using at least 4 FFT bin data near the peak (Full Rank condition, which will be described later). To get a lower effect. If there are several targets in adjacent areas as shown in FIG. In order to prevent the above performance deterioration, the present invention performs subspace-based arrival angle estimation using one bin.

The present invention describes a subspace-based approach angle estimation method when full rank is not satisfied.

Equation 2 shows a general MUSIC algorithm. This algorithm finds the angle of incidence with the largest signal in the inverse of the length of the steering vector orthogonal to the noise subspace.

는

(입사각)에 대한 스티어링 벡터를 의미하며

은 잡음 고유벡터를 의미한다.In equation 2

Is

Means a steering vector for (incident)

Means noise eigenvector.

In the case of n ranks (n <M (number of antennas)) rather than full ranks, the M–n eigenvalues are found in the middle root, but the eigenvectors are orthogonal vectors, not zero vectors. However, it is impossible to know whether this vector is a noise eigenvector or a signal eigenvector, so MUSIC cannot be obtained.

However, the present invention shows that when the signal eigenvector is used without using the noise eigenvector, it can be obtained even if it is not a full rank in the frequency domain (see Equation 3 (subspace method using the signal eigenvector)).

Create a covariance matrix using only one bin where the peaks appear in the frequency domain, and one eigenvalue resulting from eigen decomposition is the eigenvalue generated from the peak with the signal. Can be. (In case of ESPRIT and Root-MUSIC, which finds the root, we need to know eigenvalues and eigenvectors as many as the number of signals.

Since it is a method of estimating each incident angle, the value corresponding to the local maxima is determined as the angle of the target.

4 is a diagram illustrating a radar signal processing process according to an embodiment of the present invention. 4 is a methodological explanatory diagram of the present invention in terms of signal processing flow, and is also an apparatus explanatory diagram of the present invention as a radar receiving apparatus composed of signal processing functions. 4 shows a radar signal processing method or a radar signal receiving apparatus applicable to a vehicle radar system.

The basic concept of the present invention as seen from FIG. 4 is that DoA estimation is performed using only one Chirp. In order to meet the full rank condition with only a single chirp, the data of [(1 Chirp) × (M Receive Channels) × (M or more FFT Bins)] of M FFT Bins corresponding to 1 Chirp is used. DoA estimation should be performed. However, the resolution of the target information (distance, velocity) obtained in addition to the angle is lowered.

In the present invention, even if one FFT bin corresponding to a target or fewer FFT bins is used, DoA estimation based on signal subspace can be performed.

4 shows an embodiment in which signal processing is performed using only one bin corresponding to a target, and the same applies to signal processing with fewer FFT bin data than M (number of receiving channels).

Figure 4 is an example of the configuration of the radar receiver in terms of the device of the present invention. By performing fast Fourier transform (FFT) 16 on a digital signal of a digitally converted chirp, data of frequency characteristic (magnitude, phase) component for each reception channel can be obtained (21).

The following '22, 24, 26, 28, 30 'step (or functional unit) represents a process of performing a subspace-based direction detection to find the angle of the target. After finding the position of the FFT bin corresponding to the target detected through the CFAR process 20 (22, Selecting FFT Bin corresponding to targets), combine the data of the index corresponding to the target in each receiving channel (signal vector) ) At this time, a covariance matrix M × M is generated using a signal vector (1 × number of reception channels M) corresponding to the FFT bin of the target. In the example of FIG. 4, when four reception channels are present, the size of the signal vector X is 4 × 1. As such, a covariance matrix may be generated using the signal vector (24), and the covariance matrix R may be defined as X ^H X.

Eigenvalue decomposition is performed on the covariance matrix to find the eigenvalues and eigenvectors. In the case of FIG. 4, in the case of four reception channels, an eigen value having three middle roots (that is, zero) and an eigenvalue corresponding to a positive real number are generated. Among them, the positive real eigenvalue is determined as the signal eigenvalue (26), the eigenvector corresponding to the signal eigenvalue is defined as the signal eigenvector (28), and the spatial spectrum is calculated using Equation 3 presented above. The angle of the target is detected (30). Computing the received power component by multiplying the matrix by the weight of each incident angle signal eigenvector is called spatial spectral calculation.After calculating the power for each incident angle, the angle corresponding to the local maxima is determined as the target angle (30). ).

Figure 5 shows the performance results when generating a simulated reception signal in MATLAB. There are two received signals and shows the estimation result using only two bins. Since the distances are different, it is actually the result of performing the angle of arrival twice with one bin.

Simulation condition

As mentioned above, the configuration of the present invention has been described in detail through preferred embodiments of the present invention, but a person having ordinary knowledge in the art to which the present invention pertains disclosed herein without changing the technical spirit or essential features of the present invention. It will be appreciated that the present invention may be embodied in other specific forms than. The above described embodiments are to be understood in all respects as illustrative and not restrictive. The protection scope of the present invention is defined by the following claims rather than the above description, and all changes or modifications derived from the claims and their equivalents should be construed as being included in the technical scope of the present invention. .

Claims (1)

In Signal Subspace Based Arrival Angle Estimation in Radar Systems,
Instead of using the noise eigenvectors, the signal eigenvectors are used to generate subspace-based arrival angles using signal eigenvectors corresponding to eigenvalues, not zero roots, which are obtained by performing eigenvalue decomposition with only one bin in the frequency domain. A method for estimating angle of arrival based on signal subspace in a radar system, comprising estimating.

Images