Radar system

Abstract

The invention relates to a radar system, comprising a transmitting and receiving device (20), which is designed to transmit a transmission signal, which is divided into temporally repeating frames (26), each frame (26) containing at least one sequence of frequency ramps, and comprising a digital evaluation device (22), which is designed to evaluate the frequency ramps of the transmission signal. The invention relates to a device (22) for detecting a Doppler frequency, comprising a digital evaluation device (22) which is configured for Fourier transform of a received signal frame by frame at least in a distance dimension, characterized in that the evaluation device (22) is further configured for parameterized spectrum estimation (36) of a spectrum obtained by the Fourier transform in order to determine the Doppler frequency.

Description

TECHNICAL FIELD

[0001] The present application relates to a radar system having a transmitting and receiving device which is configured for transmitting a transmission signal which is divided into temporally repeating frames, wherein each frame comprises a sequence of at least one frequency ramp, and having a digital evaluation device which is configured for carrying out a Fourier transformation in at least one range dimension frame-wise on the received signal.

[0002] In particular, the present application relates to a radar system for use in a driver assistance system or in an autonomous driving system of a motor vehicle for detecting the traffic environment. BACKGROUND

[0003] In such known radar sensors, for example in LCS-FMCW systems (linear chirp sequence frequency modulated continuous wave), the transmission signal has a frame structure. The transmission signal is transmitted over the duration of a frame. After the transmitted frame, the transmitter pauses until the next frame is transmitted. The time duration between the start points of two consecutive frames is referred to as the cycle time. The ratio of the frame duration to the cycle time is referred to as the duty cycle. Radar sensors according to the prior art have frame durations of several milliseconds to several tens of milliseconds and duty cycles of less than 50% (for example 20%).

[0004] The signal received again by the radar sensor after reflection at a radar target is mixed with a part of the transmission signal. Due to the ramp-shaped frequency modulation, there is a frequency difference between the received signal and the transmission signal which is proportional to the propagation time of the signal to the radar target and back, so that as a result of the mixing a beat signal is obtained whose frequency well approximates the distance of the radar target. If the radar target has a non-zero radial velocity relative to the radar sensor, a Doppler shift also arises due to the Doppler effect, but at sufficiently large ramp steepness this Doppler shift is negligible in the distance measurement. However, across several transmitted frequency ramps, the Doppler shift leads to a phase step which makes it possible to measure the relative velocity. The minimum velocity difference Av at which the radar echoes of two targets can still be distinguished as separate targets is referred to as the Doppler resolution capability. This Doppler resolution capability is inversely proportional to the coherent integration time, i.e. the duration over which the consecutive frequency ramps are evaluated. Usually, the coherent integration time is equal to the frame duration. In order to improve the Doppler resolution capability, it is therefore desirable for the frame duration to be as long as possible. However, the frame duration cannot be increased arbitrarily, because as the frame duration increases, the latency time which elapses before the evaluation result is available in the driver assistance system and allows a driving reaction also increases. Furthermore, as the frame duration increases, the thermal load on the radar sensor also increases.

[0005] In principle, the coherent integration time can be increased by continuing the evaluation over several frames. Since the frames are separated by pauses, the total observation time is so long that the speed of the radar target can no longer be regarded as constant over this time and migration effects occur which have to be compensated by appropriate transformations, for example, Keystone transformations. High storage requirements arise since a large number of frequency ramps have to be stored for the measurement data over the duration of the evaluation. Furthermore, great delays arise in this system since the evaluation can only be started after all frames to be transformed have been received. SUMMARY

[0006] It is the task of the present application to improve the Doppler resolution without increasing the delay and the storage requirements.

[0007] This task is solved according to the application in that the evaluation device is further configured to perform a parametric spectrum estimation on the spectrum obtained by the Fourier transformation to determine the Doppler frequencies.

[0008] The term "parametric spectrum estimation" refers to a class of known algorithmic methods by which a data sequence can be characterized by specific parameters. If the data sequence represents a superposition of periodic signals with different frequencies, it is in particular possible to identify the frequency components in the spectrum by means of a high-resolution parametric spectrum estimation. According to the application, a high-resolution parametric spectrum estimation is used to determine the Doppler frequencies in a sequence of consecutive frequency ramps. The advantage compared to non-parametric methods is mainly that the achievable resolution is generally not limited by the observation duration, so that it is possible to achieve a better resolution of the evaluation results with the same measurement values and the same observation duration. Here, the spectrum estimation can be performed over several frames without the need for a data transformation of the measurement data.

[0009] Advantageous configurations and extensions of the application are described below.

[0010] Examples of methods suitable for high-resolution spectrum estimation are:

[0011] Multiple Signal Classification (MUSIC)

[0012] Estimation of Signal Parameters by Rotational Invariance Techniques (ESPRIT)

[0013] Min-Norm method

[0014] Higher-order Yule-Walker and singular value decomposition

[0015] Nonlinear least squares method

[0016] In one implementation, as in conventional methods, a two-dimensional digital Fourier transform is first performed frame-by-frame in both the range and Doppler dimensions, thereby obtaining a range-Doppler matrix for each frame. This matrix is ​​divided into range and velocity (Doppler) cells, each cell representing the complex amplitude of the received signal for the corresponding range and velocity combination. At this stage, detection can be performed to classify radar targets based on range and (albeit at lower resolution) velocity. According to the invention, in a subsequent step, parameterized spectral estimation is performed on the corresponding cells of the range-Doppler matrices for multiple consecutive frames to achieve higher resolution in the Doppler dimension.

[0017] In an alternative implementation, a one-dimensional Fourier transform is performed frame-by-frame only in the distance dimension, i.e., a Fourier transform is performed on each individual frequency ramp. As a result, a set of distance vectors is then obtained for each of the multiple frames. Spectral estimation is then performed based on the corresponding components of the distance vectors, and this can also be extended across multiple frames.

[0018] In radar sensors with multiple transmit antennas and / or multiple receive antennas, such as in MIMO radar (multiple-input multiple-output), the present invention can be combined with known angle estimation methods, such as those described in DE102014212280A1, DE102014212284A1, and DE102017200317A1. The parameterized estimation can then be extended across multiple transmit and receive channels.

[0019] To facilitate the resolution of ambiguities in the Doppler dimension, it is desirable to vary the time interval of continuous frequency ramps within a frame. Alternatively or additionally, the time interval can also vary from frame to frame. Attached Figure Description

[0020] Embodiments of the present invention will now be described in detail with reference to the accompanying drawings. The drawings show:

[0021] Figure 1 A simplified block diagram of the radar system according to the present invention is shown;

[0022] Figure 2 Shown in accordance with Figure 1 A block diagram of the measurement method implemented in the radar system;

[0023] Figure 3 A block diagram illustrating alternative measurement methods; and

[0024] Figure 4 and Figure 5 Showing the illustration based on Figure 2 and Figure 3 The flowchart shows the basic steps of the measurement method. Detailed Implementation

[0025] exist Figure 1 The diagram illustrates a simplified block diagram of an FMCW radar sensor 10, mounted, for example, at the front of a motor vehicle, for measuring the distance *d* and relative speed *v* of objects 12 and 14, such as a vehicle traveling ahead. The radar sensor 10 has a voltage-controlled oscillator 16 that provides a frequency-modulated transmit signal to a transmit and receive unit 20 via a mixer 18, the signal being transmitted from the transmit and receive unit towards the objects 12 and 14. Signals reflected at the objects are received by the transmit and receive unit 20 and mixed with a portion of the transmit signal in the mixer 18. In this way, a baseband signal *b* is obtained, which is further analyzed and processed in a digital analysis and processing unit 22. The analysis and processing unit 22 includes a control section 24 that controls the function of the oscillator 16. Within the radar measurement, the frequency of the transmit signal provided by the oscillator is modulated using a sequence of rising or falling ramps.

[0026] exist Figure 2 The diagram shows the frequency f of the transmitted signal as a function of time t in more detail. The transmitted signal is divided into frames 26, which are separated from each other by pauses 28. Each individual frame has a duration T. r The time interval between two consecutive frames is the periodic time T. z And it is equal to the sum of the frame duration and the duration of pause 28.

[0027] Within each frame, the transmitted signal consists of a series of linear frequency ramps 30. In the illustrated example, the frequency ramps 30 within the frame are equidistant and have the same steepness and the same center frequency. In other implementations, parameters such as steepness, center frequency, and phase can vary with the ramps. Similarly, the spacing between the ramps can also vary. Furthermore, multiple nested frequency ramps can be transmitted within each frame, these ramps differing from each other in at least one parameter (DE 10 2014212 280A1).

[0028] In the analysis and processing device 22, the received signal corresponding to the continuous ramp 30 is subjected to a two-dimensional Fourier transform 32 frame by frame. Here, the signal of each individual ramp 30 is Fourier-transformed in the distance dimension. Thereby, a spectrum is obtained, in which each located radar target is characterized by a peak at a specific frequency, and the peak frequency position indicates the distance of the target. The second dimension of the two-dimensional Fourier transform is the so-called Doppler dimension. Here, a Fourier integral is formed on the signals received at corresponding times in different frequency ramps 30 of the sequence. As a result, a distance-Doppler matrix 34 is obtained for each frame 26, where each cell represents a specific combination of distance and velocity. The size of the cell corresponds to the resolution, which is mainly determined by the steepness of the ramp 30 in the distance dimension f B and mainly determined by the coherent integration time in the Doppler dimension f D , where this time is equal to the frame duration T R . Once the distance-Doppler matrix 34 of a single cell is obtained, preliminary target detection can be performed on this matrix. In this way, in a known manner, the cells in the matrix whose squared amplitude of the received baseband signal b is higher than a suitable threshold are identified. In other words, the peaks corresponding to the located radar targets are searched for in the two-dimensional spectrum.

[0029] Once the distance-Doppler matrices 34 of a certain number of consecutive frames 26 are obtained, a high-resolution parametric spectrum estimation 36 is performed in a further analysis and processing stage. Figuratively speaking, the distance-Doppler matrices 34 obtained for different frames 26 are stacked together, and the spectrum estimation is performed separately for the columns ( R ) of the matrix cells extending in the third (frame) dimension f . For each combination of distance and velocity, the complex amplitudes recorded in the respective cells are then formed into a data sequence for spectrum estimation. If target detection has been performed previously, the spectrum estimation can be restricted to those cells where radar targets have been located. This significantly reduces the computational workload.

[0030] The spectrum estimation for the columns of the selected matrix cells can be performed, for example, by eigenvalue decomposition of the autocorrelation matrix. This method itself is known and is only outlined roughly here.

[0031] If Q is the number of frames 26 selected for estimation and K < Q is the number of frequency components to be estimated, then the entries (Eintrag) in the selected matrix cells form a data sequence:

[0032]

[0033] x k (q) = a k exp(j(w kq+f k ))

[0034] Where q (=0,...,Q-1) is the index of the computation matrix element, x k (q) is the complex amplitude of the k-th frequency component, e(q) is the noise component, and a k These are the parameters to be estimated, indicating the amplitude of the frequency components, where j is the imaginary unit and w... k and j k It refers to the frequency and phase of the frequency component.

[0035] Using a Q x K matrix A, y(q) can be written as:

[0036] y(q) = Ax(q) + e(q)

[0037] Based on matrix A, and assuming no correlated noise, the autocorrelation matrix R can be calculated. yy .

[0038] R yy =AP A*+s I

[0039] Where P is a diagonal matrix with K diagonal elements that indicate the power portion of the signal, A* is the conjugate transpose of A, s is the square root of the noise power, and I is the identity matrix.

[0040] Then, regarding the autocorrelation matrix R... yy Eigenvalue decomposition is performed, resulting in a diagonal matrix containing Q eigenvalues. The first K eigenvalues ​​are greater than the noise power s. 2 The remaining QK eigenvalues ​​are equal to the noise power. The relevant Q-dimensional eigenvectors are K signal space vectors s1,...,s... K and QK noise space vectors g1,...g Q-K .

[0041] Based on the K largest eigenvalues, the estimated signal space and noise space vectors can then be calculated. Then, the K frequency components w K The estimation can be performed, for example, using one of the parameterization methods (such as MUSIC, Min-Norm, or ESPRIT).

[0042] Figure 3A variation of the method is shown in which a one-dimensional Fourier transform 38 is performed in the distance dimension instead of a two-dimensional Fourier transform 32. Thus, a vector 40 is obtained for each ramp 30, whose components are complex amplitudes in different distance cells. High-resolution parameterized spectral estimation 36 is performed here in the Doppler dimension. That is, corresponding components in the sequence of vectors 40 extending across multiple frames 26 form a data sequence, which is estimated to obtain high-resolution data in the Doppler dimension.

[0043] exist Figure 4 The text describes the basic steps of a method, which corresponds to... Figure 3 The process is described below. In step S1, radar signals are transmitted, received, and digitized. In step S2, a one-dimensional discrete Fourier transform (DFT) 38 is performed. In step S3, high-resolution parametric spectral estimation (HPS) 36 is performed in the Doppler dimension. If multiple (possibly nested) frequency ramp sequences 30 are transmitted in each frame 26, then in step S3, data obtained for different sequences are also included in the estimation. This provides a larger sample size, thereby improving the accuracy of the estimation. Accordingly, in the case of MIMO radar, coherent parametric estimation is also performed in step S3 for combinations of multiple transmit and receive channels.

[0044] If multiple high-resolution spectra are obtained in step S3, for example, for multiple MIMO channels, then incoherent integration is performed on these spectra in step S4. Finally, in step S5, target detection with high resolution in the Doppler dimension is performed on the integration result, and ambiguity is resolved using known methods if necessary.

[0045] Figure 5 It shows that according to Figure 2 The flowchart of the method is shown. Step S2 is replaced by step S2', in which a two-dimensional digital Fourier transform is performed, and step S3 is replaced by step S3', in which spectral estimation at the frame dimension and possible estimations for different sequences and MIMO channels are performed.

Claims

1. A radar system having a transmitting and receiving device (20) which is configured for transmitting a transmission signal, which is divided into frames (26) which are repeated in time, wherein, Each frame (26) contains a sequence of at least one frequency ramp, and the radar system has a digital analysis processing device (22) configured for performing a Fourier transform (32, 38) on the received signal frame-wise at least in the range dimension, characterized in that the analysis processing device (22) is further configured for performing a parametric spectrum estimation (36) on the frequency spectrum obtained by the Fourier transform to determine the Doppler frequency.

2. The radar system of claim 1, wherein, The spectrum estimation is performed across data for a plurality of consecutive frames (26).

3. The radar system of claim 2, wherein, The digital Fourier transform (32) is a two-dimensional Fourier transform in the range dimension and the Doppler dimension, and wherein the parametric spectrum estimation (36) is performed in the frame dimension.

4. The radar sensor of claim 2, wherein, The digital Fourier transform (38) is a one-dimensional transform in the range dimension, and wherein the parametric spectrum estimation (36) is performed in the Doppler dimension.

5. The radar system of any one of the preceding claims, wherein, The signal transmitted within each frame (36) comprises a sequence of a plurality of frequency ramps (30) which differ from each other in at least one parameter, and wherein the spectrum estimation is performed based on data obtained for the plurality of sequences.

6. The radar system according to any one of the preceding claims, having a plurality of transmit and receive channels, wherein, The spectrum estimation is performed based on frequency spectra obtained for the plurality of transmit and receive channels. The signal transmitted within each frame (36) comprises a sequence of a plurality of frequency ramps (30) which differ from each other in at least one parameter, and wherein the spectrum estimation is performed based on data obtained for the plurality of sequences. The spectrum estimation is performed based on frequency spectra obtained for the plurality of transmit and receive channels.

Images