Method for radar angle estimation

Abstract

The invention relates to a method for angle estimation on the basis of transmitted signals of a radar sensor, which is angle-resolved in at least one dimension and which has a multiple-input multiple-output antenna array, and signals received after reflection on an object, said radar sensor being characterized in that, in order to estimate the positioning angle of a radar target, the radar target is positioned on the basis of the radar target. A cross-path model is used which also models the reflection of the transmitted and / or received signal on the reflection surface, characterized in that the angle estimation is carried out on the basis of the simplified cross-path model, a model for a single-input multiple-output mode (48) of the radar sensor is combined with a model for a multiple-input single-output mode (50).

Description

TECHNICAL FIELD

[0001] The present invention relates to a method for angle estimation based on transmitted signals and received signals after reflection on an object of a radar sensor which is angle-resolving in at least one dimension and which has a multiple-input multiple-output (MIMO) capable antenna array, wherein for estimating the orientation angle of a radar target a cross-path model is used which also models the reflection of the transmitted and / or received signals on a reflecting surface.

[0002] In particular, the invention relates to a method for a radar sensor which is used in a driver assistance system of a motor vehicle for environmental detection. BACKGROUND

[0003] In driver assistance systems, in addition to the distance and the relative speed of a located object, the azimuth angle and the elevation angle are also important, since on the basis of this angle information lane assignment can be performed and statements about the relevance of a target (driving over / driving towards / driving from below) can be made. The azimuth angle and the elevation angle of a target can be derived from the amplitude and / or phase differences of the transmitting and / or receiving antennas of the antenna array. In order to improve the accuracy and the separation capability of the angle estimation, the multiple-input multiple-output (MIMO) principle is generally used for radar sensors. Here, in contrast to a typical single input multiple output (SIMO) radar (using only one transmitting antenna and multiple receiving antennas), multiple transmitting antennas and multiple receiving antennas are used. In the angle estimation, the received signals are compared with a pre-measured angle-dependent antenna pattern. For the case that only one target is in one (d, v) cell (distance / velocity cell), the estimated angle is derived as the position of the best agreement between the received signals and the antenna pattern.

[0004] In M. Wintermantel, Radar System with Overlapping Transmit and Receive Antennas, and in US 8436763 B2, a multiple-input-multiple-output (MIMO) radar sensor is described which uses the multiple-input-multiple-output (MIMO) principle with code division multiplexing and two transmit antennas in order to improve the azimuth angle estimation. Here, the two transmit antennas are arranged at the left and right edge of the entire array in order to thus obtain the largest possible virtual aperture. In the case of multipath propagation due to reflections at, for example, a guardrail or a roadway surface, four different signal propagation paths occur, namely: a path along which the signal propagates directly from the radar sensor to the object and back to the radar sensor without reflection; a path along which the signal propagates from the radar sensor to the object on the direct path, but is reflected on the way back to the radar sensor; a path along which the signal is reflected on the way to the object and travels directly to the radar sensor without reflection on the way back; and finally, a path along which the signal is reflected on the way to the object and the radar echo is reflected again on the same surface on the way back. The signal model which depicts this scenario is referred to as the cross-path model.

[0005] If the multipath propagation is neglected in the signal model, the multiple-input-multiple-output (MIMO) angle estimation provides false estimates with errors of several degrees. This can then lead to undesired system behavior, such as side lobe interference or loss of target objects.

[0006] The multipath propagation also leads to the fact that the use of the virtual array model is not allowed in multiple-input-multiple-output (MIMO) beamforming (beamforming of transmit and receive beams). In F. Engels, P. Heidenreich, M. Wintermantel, L. Stacker, M. Al Kadi, and A. M. Zoubir, Automotive Radar Signal Processing: Research Directions and Practical Challenges, published in IEEE Signal Processing Magazine, doi: 10.1109 / JSTSP.2021.3063666, as a remedy, a beamforming with a grid separate for the transmit angle (DOD) and the receive angle (DOA) is proposed.

[0007] Furthermore, a method for a multiple-input-multiple-output (MIMO) radar sensor has been proposed in which the complete multipath signal model is used. While this allows a robust angle estimation, a large amount of computing effort is required and thus a high computing power is required. SUMMARY

[0008] The task of the present application is to specify a method which enables an accurate and reliable angle estimation with reduced computing effort in the case of multipath propagation.

[0009] According to the application, this task is solved by angle estimation based on a simplified cross-path model, combining a model for single input multiple output (SIMO) mode of the radar sensor with a model for multiple input single output (MISO) mode.

[0010] In single input multiple output (SIMO) mode, a signal is transmitted from a single transmit antenna and the signals received by a plurality of receive antennas are analyzed in a plurality of receive channels in order to calculate an angle spectrum in at least one dimension from the phase and amplitude relationships on the basis of a cross-path model. In contrast, in multiple input single output (MISO) mode, a signal is transmitted from a plurality of transmit antennas in a time-, frequency- or code-division multiplexed manner and, for the analysis of the associated radar echoes, only the signals in a single receive channel are analyzed. In this way, a further angle spectrum is again obtained on the basis of a cross-path model, but which differs in its information content from the angle spectrum obtained in single input multiple output (SIMO) mode. The information available in both spectra is then combined with one another. In this way, an angle estimation is achieved which is similarly precise and reliable as in the multiple input multiple output (MIMO) method, however, in which, due to the restriction to a single transmit antenna in single input multiple output (SIMO) mode and due to the restriction to a single receive antenna in multiple input single output (MISO) mode, the computational effort is significantly reduced.

[0011] Advantageous embodiments and extensions of the application result from the dependent claims.

[0012] In the context of maximum likelihood angle estimation (in azimuth and / or elevation), the angle spectra obtained in single input multiple output (SIMO) mode and multiple input single output (MISO) mode can be non-coherently added in order to form a one- or two-dimensional sum spectrum. BRIEF DESCRIPTION OF DRAWINGS

[0013] Embodiments are explained in more detail below with reference to the drawings. The drawings show:

[0014] Figure 1 Schematic diagram of the analog part of a radar system;

[0015] Figure 2 Diagram of an antenna array of a radar system;

[0016] Figure 3 Diagram for illustrating a scenario with multipath propagation;

[0017] Figure 4 Block diagram of a method according to the application;

[0018] Figure 5 An example of an angular spectrum obtained in single-input multiple-output (SIMO) mode;

[0019] Figure 6 An angular spectrum obtained in multiple-input single-output (MISO) mode for the same environmental configuration as Figure 5

[0020] Figure 7 A sum spectrum formed by the spectra according to Figure 5 and Figure 6 ; and

[0021] Figure 8 A comparison spectrum calculated using a full multiple-input multiple-output (MIMO) model for the same environmental configuration. DETAILED DESCRIPTION

[0022] An embodiment of a radar system to implement the method according to the application is set out in accordance with Figures 1 to 4

[0023] Figure 1 The structure of the analog part of the radar system is shown schematically and in simplified form.

[0024] The frequency modulation device 10 controls a high-frequency oscillator 12 which generates a sequence of signals in the form of a frequency ramp for a plurality of transmitting antennas 14. In each of the transmitting channels an amplifier 16 is arranged which either blocks the signal or forwards it, amplified, to the associated antenna. The oscillator 12 and the amplifiers 16 are manipulated by a multiplexing device 18, for example in accordance with a time- and frequency-division multiplexing scheme, in such a way that each of the transmitting antennas 14 transmits a frequency-modulated signal in a specific frequency subband in a specific time slot.

[0025] The signals transmitted and reflected on the object 24 are received by a plurality of receiving antennas 26 and, in each receiving channel, mixed by a mixer 28 with a part of the signal of the high-frequency oscillator 12 and brought to the low-frequency range. The analog-digital (A / D) conversion then takes place in the usual way by means of an A / D converter 30. The digitized signals are then further processed in a digital analysis processing stage 32.

[0026] As is shown in Figure 2 , the transmitting antennas 14 form a transmitting array 34 and the receiving antennas 26 form a receiving array 36. In the example shown, both arrays are two-dimensional, so that, in principle, a multiple-input multiple-output (MIMO) angular measurement can be made not only in azimuth but also in elevation.

[0027] ​​In the receiving array 36, the receiving antennas 26 are arranged in a uniform spacing in the angularly resolving direction y, for example in the azimuth direction. The spacing between the individual receiving antennas is so great here that a large aperture and a correspondingly high angular resolution can be achieved using a small number of antennas. However, the antenna-to-antenna spacing is greater than half the wavelength of the radar radiation here, so that the Nyquist unambiguity criterion is not met.

[0028] In the example shown here, the receiving antennas 26 are also arranged in a uniform spacing in the elevation direction, in the angularly resolving direction z, and the antenna spacing is so great in this direction too that non- unambiguous undersampling takes place.

[0029] In this example, the transmitting antennas 14 of the transmitting array 42 are also arranged in a uniform spacing in the azimuth and elevation directions, but the spacings are chosen so that unambiguous angle measurement is possible. However, for this, the aperture is significantly smaller than in the receiving array 44, so that the angular resolution is lower.

[0030] The equidistant arrangement of the antenna elements, in the azimuth and elevation directions, facilitates the analytical processing of the data, since it enables, for example, the use of a fast Fourier transform (FFT) to calculate a two-dimensional angle spectrum.

[0031] In the analytical processing stage 32, a two-dimensional spectrum in the range and relative velocity dimension is first calculated by Fourier transformation in a known manner. On the basis of this spectrum, individual objects can then be identified and their range and relative velocity determined. In the case of a single-target scenario, i.e. when there is one object in each range / velocity cell, a known multiple-input multiple-output (MIMO) model can be used for angle estimation in the azimuth and elevation directions, which will be briefly outlined below. For the sake of simplicity, only angle estimation in the azimuth direction is considered here, in which only the first row of transmitting antennas 14 in the transmitting array 34 and the first row of receiving antennas 26 in the receiving array 36 are used.

[0032] The following should be used xn to denote a four-component vector whose components (xn,1, xn,2, xn,3, xn,4) specify the complex amplitudes of the signals transmitted by the nthtransmitting antenna 14 and received by the four receiving antennas 26. If d is the antenna element-to-antenna element spacing, λ is the wavelength of the radar radiation and s = xn,1is the complex amplitude of the signal received by the first receiving antenna, for example the one placed furthest to the right in Figure 2

[0033] xn(q) = s(1, e -2pi(d / λ)sin(q) ,e -2pi(2d / λ)sin(q) ,e -2pi(3d / λ)sin(q) ) T =s a rx (q)

[0034] The superscript "T" should indicate the transpose, since the vector is written as a row vector but should be treated as a column vector. rx This is called the reception steering vector and describes the geometry and wave propagation properties of the reception array.

[0035] Accordingly, a control vector can also be defined for the transmit array 34 a tx , which accounts for the difference in the length of the optical path from the transmitting antenna to the object 24.

[0036] For the entire multiple-input multiple-output (MIMO) antenna array, the steering vector is obtained

[0037] a (q)= a tx (q)* a rx (q)

[0038] Here, the symbol * represents the Kronecker product.

[0039] The received signal forms a vector with Ntx·Nrx components x (96 components in this example, since the number of transmit antennas Ntx is 8 and the number of receive antennas Nrx is 12), and applies

[0040] x (q)=s a (q)

[0041] Knowing the control vector a(q) allows to establish the angle q between the object and the received signal x The (unique under appropriate conditions) relationship between and is used to infer the object's azimuth q from the amplitude and phase relationship of the received signals. However, since in practice the received signals will be more or less contaminated by noise, the azimuth cannot be calculated exactly but can only be estimated, for example by means of deterministic maximum likelihood estimation.

[0042] If this principle is extended to multi-target estimation, the single angle q becomes a vector q , whose components describe the angles of different targets, controlled by the vector a becomes the steering matrix A, and the applicable relationship

[0043] x =A q

[0044] Therefore, in the case of two targets with positioning angles q1 and q2:

[0045] x = [ a tx (q1)* a rx (q1) a tx (q2)* a rx (q2)]

[0046] In Figure 3 a scenario is outlined that is characterized by multipath propagation. The signal emitted by the radar sensor 8 can propagate not only on the direct path 38 to the object 24, but also on an indirect path 40 that first leads to a reflecting surface 42, e.g. a guard rail, and then deflects to the object 24. Likewise, the signal 44 reflected on the object 24 can propagate not only on the direct path 44 to the radar sensor 8, but also on an indirect path 46 on which the signal likewise reflects on the surface 42, so that a mirror image object 24' is pseudoed to the radar sensor.

[0047] In the case of multipath propagation, the complete multiple-input-multiple-output (MIMO) signal model is the cross-path model, which has the following form:

[0048] x = [ a tx (q1)* a rx (q1) a tx (q2)* a rx (q2) a tx (q2)* a rx (q1) a tx (q1)* a rx (q2)]

[0049] Although the last two terms can be merged into a single path due to the reciprocity of the cross-paths, i.e. the path combinations 38, 46 and 40, 44, the computational effort for calculating the signal vector x is still much greater than in the case of single-path propagation.

[0050] However, by applying Figure 4The method illustrated as flow chart can reduce this effort. In the single-input multiple-output (SIMO) mode 48, only signals are analyzed which are transmitted by a single one of the transmit antennas 14 and received by a plurality of receive antennas 26. In parallel thereto, in the multiple-input single-output (MISO) mode 50, only signals are analyzed which are transmitted by a plurality of transmit antennas 14 and received by only a single receive antenna 26. In the case of frequency- or code-division multiplexing, it can be distinguished which signal comes from which transmit antenna according to the frequency or the coding of the signal. In the case of time-division multiplexing, the multiplexing can be suspended (ausgesetzt) in the single-input multiple-output (SIMO) mode 48, whereby a desired shortening of the measurement period is achieved at the same time.

[0051] The cross-path model described above for all four paths simplifies for single-input multiple-output (SIMO) to:

[0052] x = [q1(q2)] a rx (q1)* a rx (q2)]

[0053] and for multiple-input single-output (MISO) to:

[0054] x = [q1(q2)] a tx (q1)* a tx (q2)]

[0055] Hereby a significant reduction of the model complexity and the computational load is achieved.

[0056] In the DML estimation stage 52, from the complex amplitudes obtained in the single-input multiple-output (SIMO) mode and the multiple-input single-output (MISO) mode, respectively, a spectrum is formed in such a way that for each combination of the pointing angles θ1, θ2, a DML estimation function q 2 (θ1, θ2) is calculated. From these spectra, a sum spectrum 54 is then formed, and in this sum spectrum, a global maximum is sought for each target.

[0057] As an example, in Figure 5 the DML estimation function q 2 (θ1, θ2) is shown for the spectrum 56 obtained in the single-input multiple-output (SIMO) mode 48. The function values are illustrated by the contour lines for the same values of q 2 . The local maxima of the function are shown in hatching.

[0058] Here, the original scenario (true case) is given by θ1= -10°, θ2= 10° and the same amplitude for the four paths. The spacing of the four receive channels is [0, 1.5, 3, 4.5]λ, and the spacing of the three transmit channels is [0, 1, 2]λ.

[0059] For the same scenario, Figure 6 The DML estimation function q is shown for the spectrum 58 obtained in the multiple-input single-output (MISO) mode 50 2 (θ1,θ2).

[0060] Figure 7 The estimation function is shown for the sum spectrum 52 of the correlation.

[0061] It can be clearly seen that the single-input multiple-output (SIMO) spectrum 48 is ambiguous, here there are multiple maxima of the same height as the global maximum. In contrast, the multiple-input single-output (MISO) spectrum 50 allows an angle determination only with very low accuracy, since the line of the global maximum is very wide, determined by the very small multiple-input single-output (MISO) aperture. In contrast, the sum spectrum 52 allows a unambiguous and precise angle determination.

[0062] For comparison, Figure 8 The spectrum 60 is shown, which is based on the complete cross-path model for multiple-input multiple-output (MIMO) and which requires a much higher effort for the calculation. It can be seen that the plausibility of the much simpler calculated spectrum 52 is very close to the plausibility of the complete model (spectrum 60).

Claims

1. A method for angle estimation based on a signal transmitted by a radar sensor (8) that performs angle resolution in at least one dimension and a signal received after reflection from an object (24), the radar sensor having an antenna array (34, 36) with multiple-input and multiple-output capabilities, wherein: In order to estimate the positioning angle of a radar target, a cross-path model is used, which also models the reflection of the transmitted and / or received signal on a reflecting surface (42), characterized in that the angle estimation is performed based on a simplified cross-path model and the model for the single-input multiple-output mode (48) of the radar sensor (8) is combined with the model for the multiple-input single-output mode (50).

2. The method according to claim 1, wherein separate DML estimation functions are calculated for the spectrum obtained in the single-input multiple-output mode (48) and the spectrum obtained in the multiple-input single-output mode (50), and then a sum spectrum (52) is formed by arithmetic averaging the two DML estimation functions, and the global maximum of the sum spectrum (52) is found.

3. A radar sensor (8) having a transmitting and receiving device with an antenna array (34, 36) with multiple-input and multiple-output capability and a digital evaluation stage (32), characterized in that The method according to claim 1 or 2 is implemented in the evaluation stage (32).

4. The radar sensor according to claim 3 , comprising a transmitting array (34) and a receiving array (36), wherein a first array of the transmitting array and the receiving array (34, 36) is designed for unambiguous angle measurement, and a further array is designed for ambiguous angle measurement and has a larger aperture than the first array. 5 . The radar sensor according to claim 4 , wherein all transmitting antennas ( 14 ) of the transmitting array ( 34 ) are arranged at uniform spacings in at least one dimension. 6 . The radar sensor according to claim 4 , wherein all receiving antennas ( 26 ) of a receiving array ( 36 ) are arranged at uniform spacings in at least one dimension.

7. The radar sensor according to any one of claims 3 to 6, wherein at least one of the transmitting array and the receiving array (34, 36) is a two-dimensional array.

Images