Method for performing radar angle estimation - Patents.com

Abstract

1. A method for performing angle estimation based on signals transmitted from a radar sensor having angular resolution in at least one dimension and received after reflection from an object, wherein the radar sensor comprises a MIMO-enabled antenna array, and a cross-path model is used to estimate the radar target's localization angle, the cross-path model also models reflections of the transmitted and / or received signals from reflective surfaces, wherein the angle estimation is performed based on a reduced cross-path model that combines a model for a SIMO mode (48) and a model for a MISO mode (50) of the radar sensor.

Description

[Technical Field]

[0001] The present invention relates to a method for angle estimation based on signals transmitted from a radar sensor having angular resolution in at least one dimension and received after reflection from an object, wherein the radar sensor comprises a MIMO-enabled antenna array, and wherein a cross-path model is used to estimate the radar target's localization angle, the cross-path model also models reflection of the transmitted and / or received signals from reflective surfaces.

[0002] In particular, the present invention relates to a method for radar sensors used in driver assistance systems of motor vehicles for environmental detection. [Background technology]

[0003] In driver assistance systems, in addition to the distance and relative velocity of the object being located, the azimuth and elevation angles are also important. This is because lane assignment can be performed based on this angle information and information about the target's relevance (over / oncoming / under traffic) can be obtained. The azimuth and elevation angles of a target can be determined from the amplitude and / or phase differences of the transmitting and / or receiving antennas of an antenna array. To improve the accuracy and separation capability of angle estimation, radar sensors often use the MIMO (multiple-input, multiple-output) principle. In this case, multiple transmitting and receiving antennas are used, in contrast to traditional SIMO (single-input, multiple-output) radars, which use only one transmitting antenna and multiple receiving antennas. During angle estimation, the received signal is compared with a previously measured angle-dependent antenna diagram. If there is only one target within a (d,v) cell (distance / velocity cell), the estimated angle is obtained as the position that best matches the received signal with the antenna diagram.

[0004] M. Wintermantel, "Radar system comprising overlapping transmitter and receiver antennas," and U.S. Patent No. 8,436,763 B2, describe a MIMO radar sensor that uses code division multiplexing and MIMO principles with two transmit antennas to improve azimuth angle estimation. The two transmit antennas are located at the extreme left and right of the overall array to achieve the largest possible virtual aperture. Multipath propagation due to reflections from barriers, road surfaces, and other surfaces creates four distinct signal propagation paths: the signal propagates directly from the radar sensor to the object without being reflected, then propagates back to the radar sensor; the signal propagates directly from the radar sensor to the object but is reflected on its way back to the radar sensor; the signal is reflected on its way to the object and continues directly to the radar sensor without being reflected on its way back; and finally, the signal is reflected on its way to the object, and the radar echo is reflected again by the same surface on its way back. The signal model that represents this scenario is called the cross-path model.

[0005] If multipath propagation is ignored in the signal model, MIMO angle estimation will provide erroneous estimates with angle errors of several degrees, which can lead to undesirable system behavior such as sidetrack interference and loss of target objects.

[0006] Multipath propagation also precludes the use of virtual array models for MIMO beamforming (both transmit and receive beamforming). As a remedy, F. Engels, P. Heidenreich, M. Wintermantel, L. Stacker, M. AI Kadi, and A. M. Zoubir, "Automotive Radars Signal Processing: Research Directions and Practical Challenges," IEEE Journal of Selected Topics in Signal Processing, doi:10.1109 / JSTSP.2021.3063666, proposes beamforming using separate grids for transmit angle (direction of departure, or DOD) and receive angle (direction of arrival, or DOA).

[0007] Furthermore, methods have been proposed for MIMO radar sensors in which a full multipath signal model is applied, which allows for robust angle estimation but requires significant computational effort and therefore high computing power. Summary of the Invention

[0008] It is an object of the present invention to provide a method that allows accurate and reliable angle estimation in the case of multipath propagation while reducing the amount of computation. This object is solved according to the invention in that the angle estimation is based on a reduced cross-path model that combines a model for the SIMO mode and a model for the MISO mode of the radar sensor.

[0009] In single-input, multiple-output (SIMO) mode, a signal is transmitted from a single transmit antenna, and the received signal is evaluated by multiple receive channels according to a cross-path model. An angular spectrum is calculated in at least one dimension based on phase and amplitude relationships according to a cross-path model. In multiple-input, single-output (MISO) mode, signals are transmitted from multiple transmit antennas using time, frequency, or code division multiplexing, and only a single receive channel is evaluated for the associated radar echo. This results in an additional angular spectrum, also according to the cross-path model, but with a different information content than the angular spectrum obtained in SIMO. The information available in the two spectra is then combined. In this way, accurate and reliable angle estimation similar to that achieved by the MIMO method is achieved, but the computational complexity is significantly reduced by being limited to a single transmit antenna in SIMO mode and a single receive antenna in MISO mode.

[0010] Advantageous embodiments and further developments of the invention emerge from the dependent claims. In the framework of maximum likelihood angle estimation (azimuth and / or elevation), the angular spectra obtained in SIMO and MISO modes can be non-coherently added to generate a one- or two-dimensional sum spectrum.

[0011] Exemplary embodiments are explained in more detail below on the basis of the drawings. [Brief explanation of the drawings]

[0012] [Figure 1] FIG. 1 is a schematic diagram of the analog portion of a radar system. [Figure 2] FIG. 1 is a diagram of an antenna array of a radar system. [Figure 3] FIG. 1 is a diagram illustrating a scenario with multipath propagation. [Figure 4] FIG. 1 is a block diagram of a method according to the present invention; [Figure 5]FIG. 10 is a diagram showing an example of an angular spectrum obtained in SIMO mode. [Figure 6] FIG. 6 shows the angular spectrum obtained in MISO mode for the same environmental configuration as in FIG. 5. [Figure 7] FIG. 7 shows a sum spectrum generated from the spectra according to FIGS. 5 and 6. [Figure 8] FIG. 10 shows comparative spectra calculated for the same environmental configuration using a full MIMO model. DETAILED DESCRIPTION OF THE INVENTION

[0013] 1 to 4, an exemplary embodiment of a radar system in which the method according to the invention is implemented will be described. FIG. 1 shows a simplified schematic diagram of the analog portion of a radar system.

[0014] A frequency modulation device 10 controls an RF oscillator 12, which generates a series of signals in the form of frequency ramps for multiple transmit antennas 14. Each of the multiple transmit channels is provided with an amplifier 16, which either blocks or amplifies the signal and forwards it to an associated antenna. The oscillator 12 and amplifier 16 are controlled by a multiplexing device 18, for example, according to a time and frequency division multiplexing scheme, so that each transmit antenna 14 transmits a frequency modulated signal in a specific frequency subband within a specific time slot.

[0015] The transmitted signals reflected by the object 24 are received by a number of receive antennas 26 and, in each receive channel, are mixed with a portion of the signal of the RF oscillator 12 by a mixer 28, bringing them into the low frequency range. A / D conversion is then carried out in the usual way by an A / D converter 30. The digitized signals are then further processed in a digital evaluation stage 32.

[0016] 2, the transmit antennas 14 form a transmit array 34, and the receive antennas 26 form a receive array 36. In the example shown, both arrays are two-dimensional, so in principle MIMO angle measurements are possible in both azimuth and elevation.

[0017] In the receive array 36, the receive antennas 26 are equidistantly arranged in the angular resolution direction y, e.g., the azimuth direction. Here, the distance between the individual receive antennas is so large that a large aperture and therefore high angular resolution can be achieved using a small number of antennas. However, here, the distance between the antennas is greater than half the wavelength of the radar radiation, and therefore the Nyquist uniqueness criterion is not met.

[0018] In the example shown here, the receive antennas 26 are also equidistant in elevation (angular resolution direction z), and the distance between the antennas is also very large in this direction, resulting in non-unique undersampling.

[0019] In this example, the transmit antennas 14 of the transmit array 42 are also spaced equidistant in azimuth and elevation, but at distances selected to allow for unique angular measurements, but in this regard the aperture is significantly smaller than that in the receive array 44, and therefore the angular resolution is lower.

[0020] The equidistant arrangement of the antenna elements (in azimuth and elevation) facilitates the evaluation of the data, for example by allowing the use of a fast Fourier transform (FFT) to calculate the two-dimensional angular spectrum.

[0021] In the evaluation stage 32, a two-dimensional spectrum in the dimensions of range and relative velocity is first calculated by Fourier transform in a known manner. Based on this spectrum, individual objects can then be identified and their range and relative velocity determined. In a single-target scenario, i.e., when there is only one object in each range / velocity cell, a known MIMO model, briefly described below, can be used to estimate the azimuth and elevation angles for each object. For simplicity, only the azimuth angle estimation is considered here. In this case, only the first row of transmit antennas 14 in the transmit array 34 and the first row of receive antennas 26 in the receive array 36 need to be used.

[0022]

number

[0023] represents a four-component vector whose components (xn,1, xn,2, xn,3, xn,4) indicate the complex amplitudes of the signals transmitted by the nth transmit antenna 14 and received by the four receive antennas 26. If d is the distance between the antenna elements, l is the wavelength of the radar radiation, and s=xn,1 is the complex amplitude of the signal received by the first receive antenna (e.g., the right-most receive antenna in FIG. 2), then the difference in path length between the signals arriving at the various receive antennas 14 gives rise to the following relationship:

[0024]

number

[0025] Although the vectors are written here as row vectors, they should be considered as column vectors, so the superscript "T" denotes transpose. rx is called the receive control vector, which represents the geometric and wave propagation characteristics of the receive array.

[0026] Correspondingly, for the transmit array 34, a control vector σ represents the difference in path length of the optical paths from the transmit antennas to the object 24.

[0027]

number

[0028] can also be defined. For the entire MIMO antenna array, the control vector

[0029]

number

[0030] where the symbol * denotes the Kronecker product. The received signal is a vector with Ntx and Nrx components (in this example, the number of transmit antennas Ntx is 8 and the number of receive antennas Nrx is 12, so 96 components).

[0031]

number

[0032] Forming

[0033]

number

[0034] holds true. Knowing the control vector a(q), the angle q of the object and the received signal

[0035]

number

[0036] By establishing a (unique under appropriate conditions) relationship between θ and θ, the azimuth angle q of the object can be derived from the amplitude and phase relationship of the received signal. However, in reality, the received signal contains more or less noise, so the azimuth angle cannot be calculated accurately and can only be estimated using, for example, deterministic maximum likelihood estimation.

[0037] Generalizing this principle to multi-target estimation, a single angle q can be expressed as a vector whose components represent the angles of the various targets:

[0038]

number

[0039] and the control vector

[0040]

number

[0041] is converted into the control matrix A, and the following relationship holds:

[0042]

number

[0043] Thus, for two targets with localization angles q1 and q2:

[0044]

number

[0045] A scenario characterized by multipath propagation is illustrated schematically in Figure 3. A signal transmitted by radar sensor 8 may propagate not only along a direct path 38 to object 24, but also along an indirect path 40 where it first travels to a reflective surface 42, such as a guardrail, and is then deflected toward object 24. Similarly, a signal 44 reflected from object 24 may propagate not only along a direct path 44 to radar sensor 8, but also along an indirect path 46 where the signal is similarly reflected off surface 42, resulting in the appearance of a mirrored object 24' to the radar sensor.

[0046] For multipath propagation, the complete MIMO signal model is a cross-path model having the following form:

[0047]

number

[0048] Due to the cross paths, i.e., the reciprocity of the path combinations 38, 46 and 40, 44, the last two terms can be collapsed into a single path, but still, the signal vector

[0049]

number

[0050] The computational effort to calculate is significantly larger than in the single-path propagation case. However, this computational complexity can be reduced by using the method shown in the flowchart of FIG. 4. In SIMO mode 48, only signals transmitted by only one of the transmit antennas 14 and received by the multiple receive antennas 26 are evaluated. In parallel, in MISO mode 50, only signals transmitted by the multiple transmit antennas 14 and received by only a single receive antenna 26 are evaluated. In the case of frequency or code division multiplexing, it is possible to distinguish which signal originates from which transmit antenna based on the signal's frequency or coding. In the case of time division multiplexing, multiplexing can be stopped in SIMO mode 48, thereby simultaneously achieving the desired shortening of the measurement cycle.

[0051] The above cross-path model for all four paths is

[0052]

number

[0053] As for MISO,

[0054]

number

[0055] This becomes: This results in a significant reduction in model complexity and computational load. In the DML estimation stage 52, a DML estimation function q is calculated for each combination of the position identification angles θ1 and θ2. 2 The complex amplitudes obtained in the SIMO and MISO modes are calculated using (θ1, θ2) to generate a spectrum, respectively, which is then used to generate a sum spectrum 54, and the global maximum of this sum spectrum is found for each target.

[0056] As an example, FIG. 5 shows the DML estimation function q for the spectrum 56 obtained in SIMO mode 48. 2 (θ1,θ2) is shown. The function value is q 2 The local maxima of the function are indicated by hatching.

[0057] Here, the initial scenario (ground truth) is given by θ1 = -10°, θ2 = 10°, and the same amplitude for the four paths. The distance between the four receiving channels is [0, 1.5, 3, 4.5]λ, and the distance between the three transmitting channels is [0, 1, 2]λ.

[0058] For the same scenario, Figure 6 shows the DML estimate q for the spectrum 58 obtained in MISO mode 50. 2 (θ1, θ2). FIG. 7 shows the estimation function for the associated total spectrum 52 .

[0059] It can be clearly seen that the SIMO spectrum 48 is ambiguous, with multiple maxima having the same height as the global maximum. In contrast, the MISO spectrum 50 allows for angle determination with only very low accuracy, since the global maximum line is very broad due to the very small MISO aperture. In contrast, the sum spectrum 52 allows for a unique and accurate angle determination.

[0060] For comparison, Figure 8 shows spectrum 60 based on a full cross-path model for MIMO, the calculation of which requires much higher computational effort. It can be seen that the effectiveness of spectrum 52, which is much easier to calculate, comes very close to that of the full model (spectrum 60).

Claims

1. 1. A method for performing angle estimation based on signals transmitted from a radar sensor (8) having angular resolution in at least one dimension and received after reflection from an object (24), the radar sensor (8) comprising a MIMO-enabled antenna array (34, 36), wherein a cross-path model is used to estimate a radar target's localization angle, the cross-path model also models reflections of the transmitted and / or received signals from reflecting surfaces (42), the method characterized in that the angle estimation is performed based on a reduced cross-path model that combines a model for a SIMO mode (48) and a model for a MISO mode (50) of the radar sensor (8).

2. 2. The method of claim 1, wherein separate DML estimation functions are calculated for the spectrum obtained in the SIMO mode (48) and the spectrum obtained in the MISO mode (50), and then a total spectrum (52) is generated by arithmetic averaging the two DML estimation functions, and the global maximum of the total spectrum (52) is found.

3. 1. A radar sensor (8) comprising a transmitting / receiving device with a MIMO-capable antenna array (34, 36) and a digital evaluation stage (32), characterized in that the evaluation stage (32) performs the method according to claim 1 or 2.

4. 4. The radar sensor of claim 3, comprising a transmit array and a receive array, wherein a first of the transmit array and receive array is designed for unique angle measurements and the other of the transmit array and receive array is designed for ambiguous angle measurements and has a larger aperture than the first array.

5. 5. The radar sensor of claim 4, wherein all of the transmit antennas (14) of the transmit array (34) are equidistantly spaced in at least one dimension.

6. 6. A radar sensor according to claim 4 or 5, wherein all of the receive antennas (26) of the receive array (36) are equidistantly spaced in at least one dimension.

7. 7. The radar sensor of claim 3, wherein at least one of the transmit array (34) and the receive array (36) is a two-dimensional array.

Images