ANGLE-RESOLUTION RADAR SENSOR

Description

[0001] The invention relates to an angle-resolving radar sensor with an antenna array having a number N > 2 of antenna elements offset from one another in a scanning direction and whose aperture A in units of wavelength λ is greater than (N-1) / 2, a digital beamforming device, and an angle estimating device designed for angle estimation on the basis of the signal of the beamforming device. State of the art

[0002] Radar sensors are used in motor vehicles, for example, to measure the distances, relative speeds, and azimuth angles of vehicles or other objects detected in front of the vehicle. The individual antenna elements are then arranged at a distance from one another on a horizontal plane, so that different azimuth angles of the detected objects lead to differences in the path lengths that the radar signals have to travel from the object to the respective antenna element. These path length differences lead to corresponding differences in the phase of the signals received by the antenna elements and evaluated in the associated evaluation channels. By comparing the (complex) amplitudes received in the various channels with corresponding amplitudes in an antenna pattern, the angle of incidence of the radar signal and thus the azimuth angle of the detected object can be determined.

[0003] Elevation angles of objects can also be measured in a similar way. The scanning direction in which the antenna elements are offset from each other is then not horizontal, but vertical.

[0004] To achieve high angular resolution, the antenna aperture should be as large as possible. However, if the distances between neighboring antenna elements are too large, ambiguities in the angle measurement can arise, since the same phase relationships between the received signals are obtained for run-length differences that differ by integer multiples of the wavelength λ. An unambiguous angle measurement can be achieved, for example, with a ULA (Uniform Linear Array) structure, in which the antenna elements are arranged at intervals of λ / 2.

[0005] Using a procedure known as digital beamforming, the main sensitivity direction of the antenna array can be modified, thus effectively directing the radar sensor's main receiving lobe in a specific direction. However, side lobes with a certain sensitivity also occur on either side of the main receiving lobe. In digital beamforming, the complex amplitudes received by each individual antenna element are weighted with an angle-dependent complex phase factor that corresponds to the path length difference of the radar radiation for a given angle. If several beams with different directional angles are formed in this way, a gain function is obtained for each beam. This function specifies the antenna gain for each angle that would result if the detected object were located at that angle.Ideally, for an object located at a certain angle, one should measure a signal strength in each beam that corresponds to the theoretical antenna gain for that beam, and only at exactly one angle, namely the angle at which the object is actually located, should the measured amplitudes show the correct relationship to each other.

[0006] In practice, however, the measured signals are more or less noisy, so the detection angle can only be estimated by finding the angle for which the amplitudes measured in the various beams best correlate with the theoretical values. This correlation can be expressed, for example, by a so-called DML (Deterministic Maximum Likelihood) function, and the angle estimation then consists of finding the maximum of the DML function.

[0007] If the aperture of the antenna array is increased to achieve higher angular resolution while still maintaining unambiguous measurement accuracy, the number of antenna elements must be increased. However, this also increases the number of required analysis channels, thus increasing the required computing power and thus the hardware costs.

[0008] WO2013 / 056880 A1 discloses a radar sensor that uses thinned antenna arrays, in which the spacing is increased for at least some of the pairs of adjacent antenna elements, thus achieving a larger aperture for a given number of evaluation channels. The unambiguousness of the angle measurement is then restored by alternating measurements with different combinations of antenna elements, thereby filling the gaps in the antenna array.

[0009] A radar sensor of the type mentioned above is known from DE 10 2016 108 756 A1. Disclosure of the invention

[0010] Claim 1 defines the invention. The object of the invention is to create an angle-resolving radar sensor that enables high-resolution, unambiguous angle measurements with reduced computational effort for signal evaluation.

[0011] This object is achieved according to the invention in that the center distances between the adjacent antenna elements are different from one another, but do not deviate by more than 25% from the value A / (N-1).

[0012] Since the aperture is larger than (N-1) / 2, ambiguities can in principle arise in the angle estimation. In the DML function for a ULA, this is reflected in the fact that the function has several maxima of equal height at different angles. Because the antenna elements are not arranged at exactly equal distances, but rather the distances vary slightly from pair to pair, all maxima in the DML function except for one are reduced in height, so that the function once again has a clear absolute maximum and thus a clear angle estimate is possible. However, the deviation of the antenna distances from the value A / (N-1), which would correspond to the distance for a ULA, is limited such that the same techniques can still be used for digital beamforming as for a ULA and the side lobes are sufficiently attenuated.In particular, digital beamforming can be carried out in a particularly efficient manner using a fast Fourier transform (FFT).

[0013] The radar sensor according to the invention thus has a quasi-regular array instead of a ULA, in which the deviations from a perfect ULA are just large enough to allow a clear angle estimate, given the expected noise level. The "predetermined amount" by which the antenna element spacing may deviate from the ULA value A / (N-1) is chosen such that, on the one hand, sufficient robustness against signal noise is achieved, while, on the other hand, the angle-dependent gain functions are not excessively distorted compared to a ULA.

[0014] Advantageous embodiments and further developments of the invention are specified in the subclaims.

[0015] In one embodiment, the deviations of the center distances between the antenna elements from the value A / (N-1) are less than 15%.

[0016] The deviations in the distances between the different pairs of neighboring antenna elements can themselves vary regularly, for example, according to a linear function, a quadratic function, a higher-degree polynomial, or, for example, a sine function. The number N of antenna elements in the array is preferably a power of two, for example, N = 8 or N = 16, which enables efficient digital beamforming using FFT.

[0017] In the following, an embodiment example is explained in more detail using the drawing.

[0018] They show: Fig. 1 is a block diagram of a radar sensor according to the invention with a quasi-regular antenna array; Fig. 2 is a representation of a regular antenna array (ULA) of a conventional radar sensor; Fig. 3 is a gain distribution function for different beams formed by digital beamforming with the antenna array according to Fig. 2 were formed, Fig. 4 a simplified diagram analogous to Fig. 3 , in which, for reasons of clarity, only the gain functions for two beams are shown; Fig. 5 a DML function for the antenna array according to Fig. 2 ; Fig. 6 an example of a ULA with enlarged aperture; Fig. 7 a diagram of two gain functions, analogous to Fig. 4 , but for the array after Fig. 6 ; Fig. 8 a DML function for the array according to Fig. 6 ; Fig. 9 a diagram of two profit functions, analogous to Figuren 4 and 7 , however, for the array according to the invention Fig. 1 ; and Fig. 10 a DML function for the array according to Fig. 1 .

[0019] In Fig. 1 A radar sensor with a quasi-regular antenna array 10 with eight antenna elements 12 is shown in diagram form. Each antenna element 12 is formed by a column of eight serially fed antenna patches 14 extending in the vertical direction z. The column-shaped antenna elements 12 are arranged in a row in a scanning direction y, which in a radar sensor for motor vehicles corresponds to the horizontal transverse direction of the vehicle, so that the radar sensor has an angular resolution capability in azimuth. The distances between the antenna elements 12 as well as the distances between the individual antenna patches 14 of each column are given in units of the wavelength λ of the radar radiation. The distances between the pairs of adjacent antenna elements 12 are in Fig. 1 also given quantitatively and each have approximately the value 2, but with deviations of less than 7% from the mean value of 2.00.

[0020] The width of the antenna array 10 in the scanning direction y is approximately 14 λ, so that the array has an aperture of A = 14 in azimuth. In general, the mean value (2.00 in this example) of the spacing between the antenna elements 12 is equal to A / (N-1), where N is the number of antenna elements 12 in the array.

[0021] In the example shown, the distances between the pairs of antenna elements 12 increase linearly from 1.87 to 2.13.

[0022] The eight antenna elements 12 are connected via respective signal lines 16 to an evaluation circuit 18, in which the received signals are evaluated in separate reception channels. For example, the radar sensor shown here can be an FMCW (Frequency Modulated Continuous Wave) radar. Each evaluation channel then contains a mixer in which the signal received by the antenna element is mixed with a portion of the transmitted radar signal, resulting in an intermediate frequency signal whose frequency depends on the propagation time of the radar signal from the radar sensor to the object and back, and on the relative speed of the object. The intermediate frequency signals are digitized in the evaluation circuit 18 and each recorded over a specific sampling period in which the frequency of the transmitted signal is ramp-modulated.Based on the frequencies of the intermediate frequency signals, which are obtained on several modulation ramps with different ramp gradients, the distances and relative speeds of the located objects can then be determined in a known manner.

[0023] The signal lines 16 connecting the antenna elements to the evaluation circuit 18 are designed to be all the same length, so that the phase relationships between the signals are not distorted on their way to the evaluation circuit 18. By comparing the amplitudes and phases (i.e., the complex amplitudes) of the signals received in the eight receiving channels, the angle (azimuth angle) can be determined for each object located at a specific distance and relative speed. This angle indicates the direction from the radar sensor to the object. For this purpose, the signals received in the eight receiving channels are subjected to digital beamforming in a beamformer 20, for example, using fast Fourier transformation (FFT). The results of the beamforming are transmitted to an angle estimator 22, where the azimuth angle ϕ of the located object is determined using a maximum likelihood estimation.

[0024] To explain the operation of the invention, a fully filled regular antenna array 24 (ULA) will first be considered, as used in conventional radar sensors and as shown in diagram form in Fig. 2 The spacing between the adjacent antenna elements 12 in this array is uniformly λ / 2, so the uniqueness condition is met. However, this array—with eight receive channels—only has an aperture of A = 3.5, so the angular resolution is significantly limited.

[0025] Fig. 3 is a diagram in which the antenna array 24 is Fig. 2 The antenna gain of the beamforming device 20 is plotted as a function of the azimuth angle ϕ. Specifically, the diagram shows the graphs of normalized gain functions 28 - 40 for ten receive beams whose sensitivity maxima are at 0°, + / -15°, + / -30°, + / -45°, and beyond + / -60°. Each beam has a main lobe with maximum gain and a plurality of side lobes attenuated by approximately 13 dB. This diagram is based on FFT beamforming with a rectangular window.

[0026] For the sake of clarity, Fig. 4 The graphs of gain functions 30 and 36 are shown again in isolation. The graph of gain function 30 is shown in bold lines, while the graph of gain function 36 is shown in dashed lines. It can be seen that gain function 30, with the maximum at 0°, has symmetric side lobes, while the gain function for the beam with the sensitivity maximum at -45° is asymmetric.

[0027] In digital beamforming, a weighted sum is formed from the complex amplitudes of the signals received in the eight antenna elements 12, with complex weighting factors that reflect the runlength difference from antenna element to antenna element. Since these runlength differences depend on the azimuth angle ϕ, a different set of weighting factors is obtained for each beam (with a sensitivity maximum at a certain azimuth angle ϕ). If an object is located at a given azimuth angle ϕ, a signal whose intensity is proportional to the gain function 28 is obtained in the beam with the sensitivity maximum at 0°, whereas for the same object in the beam with the sensitivity maximum at -45° a signal is obtained that is proportional to the gain function 36. Accordingly, a value is obtained for each of the other beams, which is given by the corresponding gain function.

[0028] From the different amplitude values obtained for the various beams after beamforming, the azimuth angle at which the detected object is actually located can be deduced. To do this, the angle at which the measured values best correlate with the values given by the gain functions is sought.

[0029] As an example, Fig. 5 a DML function 42 (Deterministic Maximum Likelihood Function), also known as an angular spectrum, for a target located at an azimuth angle ϕ = 0°. The DML function 42 specifies, for each azimuth angle ϕ, the correlation between the measured values and the gain functions for the various beams. The function is normalized so that the maximum has the value 1. It can be seen that the DML function 42 has only a single, clearly defined maximum at the angle ϕ = 0°, at which the located object is located. If, instead, a signal were received from an object located at an angle ϕ = 20°, the DML function would be shifted so that its maximum would be at 20°.

[0030] In order to improve the angular resolution, the ULA 24 is now Fig. 2 The distances between the antenna elements 12 are increased to 2λ, but without increasing the number of antenna elements. This results in a ULA 44 with an aperture of A = 14, as shown in Fig. 6 is shown.

[0031] Fig. 7 shows how this change affects the profit functions 28 - 40. Similar to Fig. 4 are also in Fig. 7 only the graphs for the profit function 28 and 36 are shown. It can be seen that these profit functions each have several maxima of approximately the same height. For example, the profit function 28 has a maximum at 0° (as in Fig. 4 ), but further equally high maxima at + / - 30°. In between there is a larger number of more strongly damped side lobes. The same applies accordingly to the gain function 36 as well as to any other (in fig. 7 (not shown) profit function.

[0032] The DML function 42 for the ULA according to Fig. 6 is in Fig 8 This function also now has several equally high main maxima, so that a clear angle estimate is no longer possible.

[0033] If, on the other hand, the quasi-regular antenna array 10 is used according to Fig. 1 , then unambiguous angle estimates are possible again. The profit functions 28 and 36 for this array are in Fig. 9 shown. Similar to Fig. 7 Each gain function in the angular range from -50° to +50° has three distinct main maxima, but the side lobes are more "shouldered" here, ie the side lobes are raised and moved closer to the flanks of the main maxima.

[0034] Fig. 10 shows the corresponding DML function 42. Since the beam forming in the beam forming device 20 is done in the same way as in the ULAs according to Figuren 2 and 6, with the same weighting factors, but the distances between the adjacent antenna elements 12 are slightly non-uniform, only a main maximum 46 at ϕ = 0° (in the case of a target at 0°) remains at full height, while the flanking maxima 48 at + / -30° are more strongly suppressed. This means that a unique maximum is found in an angle estimation, thus enabling a unique angle estimation.

[0035] Incorrect angle estimates can only occur if the signals are so noisy that the differences between maxima 46 and 48 are blurred, and one of the maxima 48 has the highest value and is incorrectly selected to determine the azimuth angle. The more non-uniform the spacing between the neighboring antenna elements 12, the more strongly the maxima 48 are suppressed, and the more robust the angle estimate is against the signal noise. On the other hand, however, with increasing non-uniformity of the array, the side lobes become increasingly pronounced. However, with a suitable choice of the non-uniformity of the spacing between the antenna elements 12, it is possible to achieve a clear angle estimate despite the typically present signal noise, and due to the enlarged aperture, a higher angular resolution is achieved without the need for additional evaluation channels.

Claims

1. Angle-resolving radar sensor having an antenna array (10), which has a number N > 2 of antenna elements (12) offset with respect to one another in a scanning direction (y) and the aperture A of which in units of the wavelength λ is greater than (N-1) / 2, having a digital beam shaping device (20), and having an angle estimation device (22) designed to estimate an angle on the basis of the signal from the beam shaping device (20), wherein the centre-to-centre distances between the adjacent antenna elements (12) differ from one another, but do not differ by more than 25% of the value A / (N-1).

2. Radar sensor according to Claim 1, in which the distances for the adjacent antenna elements do not differ by more than 15% of the value A / (N-1).

3. Radar sensor according to Claim 1 or 2, in which the distances between the antenna elements (10) vary according to a regular pattern.

4. Radar sensor according to Claim 3, in which the distances between the antenna elements (12) vary according to a polynomial function.

5. Radar sensor according to Claim 4, in which the distances between the antenna elements (12) vary according to a linear function.

6. Radar sensor according to one of the preceding claims, in which N is a power of two.

7. Radar sensor according to one of the preceding claims, in which A > 1, preferably A ≥ 2, units of the wavelength.