METHOD FOR CALIBRINGING A MIMO RADAR SENSOR FOR MOTOR VEHICLES

Description

[0001] The invention relates to a method for phase calibration of a MIMO radar sensor with an array of several transmit and receive antenna elements that are offset from each other in at least one direction, and with several high-frequency components that are each assigned to a part of the array, wherein the phase calibration includes a correction of phase differences of the high-frequency components. State of the art

[0002] A MIMO radar sensor with multiple high-frequency components is known from US 2016 / 146931 A1.

[0003] In driver assistance systems for motor vehicles, such as adaptive cruise control or collision warning systems, radar sensors are frequently used to detect the traffic environment. Besides distance and relative speed, the azimuth angle of the detected objects is usually also important, as it allows for lane assignment when locating vehicles ahead. The elevation angle of the detected objects can also be significant, as it provides information about the relevance of the target, for example, whether the target can be driven over or under, or whether it represents a potentially collision-prone obstacle.

[0004] The azimuth and elevation angles of targets can be determined from the amplitudes and / or phase differences of the transmitting and / or receiving antennas of the antenna array. To improve the accuracy and discrimination of the angle estimation, radar sensors operating on the MIMO (Multiple Input Multiple Output) principle can be used. Unlike conventional SIMO (Single Input Multiple Output) radar sensors, which use one transmitting antenna and multiple receiving antennas, MIMO uses multiple transmitting and multiple receiving antenna elements. To separate the signals from the transmitting antenna elements at the receiving antenna elements, the transmitted signals must be uncorrelated (orthogonal). This can be achieved through time-division multiplexing, frequency-division multiplexing, or code-division multiplexing.

[0005] In angle estimation, the received signals are compared with a previously measured angle-dependent antenna pattern. If only a single target is being located (or multiple targets that are clearly distinguishable based on their distance and relative velocity), the estimated angle is the position of best agreement between the received signal and the antenna pattern. For the general case of multi-target estimation, special estimation algorithms are known that provide estimates for the detection angles of all targets involved.

[0006] It has been standard practice to measure the antenna patterns for each individual sensor at the factory before commissioning. However, when the radar sensor is installed in a vehicle, for example behind a bumper or a raised structure such as a brand emblem, distortions of the antenna pattern can occur, leading to systematic errors in angle estimation. This is particularly true for the transmit antenna patterns of MIMO radar sensors.

[0007] From DE 2014 208 899 A1 an alternative method is known which enables calibration or recalibration of an antenna pattern of a MIMO radar sensor with N tx transmit antenna elements and N rx receive antenna elements even after commissioning of the radar sensor.

[0008] This process includes the following steps: Before commissioning the radar sensor: Storing an antenna pattern that assigns a respective control vector to each of several angles θ. a (θ) assigns, which is composed of a transmit control vector a tx (θ) and a receive control vector a rx (θ), after commissioning: Perform a radar measurement to locate an object, check whether the located object is a single target or multiple targets, if it is a single target: perform a SIMO measurement with each of the transmitting antenna elements, estimate the angle θ of the object based on the measurement results, calculate a first, from the components of the transmitting control vector a tx (θ) dependent comparison quantity for each transmitting antenna element, calculation of a second comparison quantity dependent on the results of the SIMO measurements for each transmitting antenna element, and correction of the transmitting control vector atx (θ) based on a known relationship between the first and second comparison quantities for each transmitting antenna element.

[0009] This method has the advantage that interference that only arises during the installation of the radar sensor in the vehicle and therefore cannot be detected during factory calibration can be compensated for subsequently. For example, if a single object is detected at a specific azimuth angle θ while driving the vehicle in which the radar sensor is installed, the transmit antenna pattern for this specific azimuth angle can be recalibrated using the method described above. Since individual objects are detected at different azimuth angles θ over time during vehicle operation, a corrected (calibrated) transmit antenna pattern for the entire azimuth angle range is gradually obtained. The calibration phase can then be completed.Alternatively, it is also possible to continue the recalibration continuously or to repeat it at certain intervals in order to take into account age-related changes in the antenna pattern.

[0010] The method described above only calibrates the transmitting portion of the antenna pattern. However, the publication also describes an analogous method in which the receiving portion of the antenna pattern is calibrated using multiple MISO (Multiple Input Single Output) measurements.

[0011] Recently, there has been increasing interest in radar sensors that utilize two or more essentially identical high-frequency components for generating radar signals and receiving and pre-processing the radar echoes. These components can be used individually in low-power radar sensors, such as those found in driver assistance systems, or they can be combined to create a higher-performance radar sensor, particularly one with improved angular resolution. However, in the latter case, precise synchronization of the various high-frequency components is essential to avoid errors caused by phase differences in the receiving and / or transmitting sections of the different components.

[0012] Known solutions include, for example, the use of a central oscillator or master / slave architectures, each in conjunction with precise alignment of the signal paths. However, these solutions are relatively complex.

[0013] Furthermore, radar sensors with multiple high-frequency components present the problem that these components must necessarily be positioned at a certain spatial distance from one another. Consequently, they interact differently with their respective installation environments and / or, for example, due to heat generation within the radar sensor, can have different temperatures. Temperature changes during operation of the radar sensor can therefore lead to phase differences due to the thermal behavior of the electronic components involved, which impair the accuracy of the calibration. These factors are difficult to predict during a single factory calibration of the radar sensor and are therefore hard to control.

[0014] The object of the invention is to provide a method that allows the multiple high-frequency components of a radar sensor to be recalibrated "online", i.e., during the operation of the radar sensor.

[0015] This problem is solved according to the invention with the features specified in the independent claims.

[0016] The core idea of ​​the solution is to divide the array of transmit and receive antennas into transmit and receive subarrays such that each subarray is assigned to exactly one of the RF components, and at least two receive subarrays are offset from each other in one direction and aligned with each other in the perpendicular direction. The antenna pattern calibration procedures described earlier are then applied analogously at the subarray level, with each subarray being treated as a single antenna element. For subarrays belonging to different RF components, the calibration automatically corrects the phase errors caused by the components' asynchronicity.

[0017] Advantageous further developments and embodiments of the invention are specified in the dependent claims.

[0018] The method can be applied to both azimuth and elevation angle estimation. If two or more subarrays belonging to different RF components are horizontally offset, the corresponding RF components can be calibrated using azimuth angle estimation. Conversely, if two or more subarrays belonging to different RF components are vertically offset, the corresponding RF components can be calibrated using elevation angle estimation.

[0019] The invention also relates to a radar sensor for motor vehicles in which one of the methods described above is implemented.

[0020] The following section explains an exemplary embodiment in more detail with reference to the drawing.

[0021] They show: Fig. 1 a schematic diagram of a radar sensor to which the invention is applicable; Fig. 2 a diagram of a MIMO radar sensor with two transmitting antenna elements and four receiving antenna elements; Fig. 3 a diagram of the radar sensor according to Fig. 2 , after installation in a motor vehicle and during the location of an object; Fig. 4 a flowchart of a method for calibrating the radar sensor after Fig. 1 .

[0022] The in Fig. 1 The radar sensor 8 shown comprises an array 10 of transmitting and receiving antennas as well as four identical high-frequency modules HF1-HF4, which are arranged on a common circuit board 12. The transmitting and receiving antennas are formed by antenna elements 14 (patches) arranged in vertical columns.

[0023] In the example shown, array 10 is divided into four domains D1-D4, each of which is assigned one of the high-frequency modules HF1-HF4. Each high-frequency module provides a transmit signal for the transmit antennas of its domain and, as is known per se and therefore not shown here, contains a number of mixers in which the signals received by a receiving antenna are mixed with a portion of the transmit signal and thus down-converted to an intermediate frequency band, so that they are then digitized and further processed in an evaluation and control circuit 16 (in this case outside the circuit board 12).

[0024] Each of the domains D1-D4 contains a number of subarrays of transmit and receive antennas. In the example shown, domain D1 contains two transmit subarrays TX1, TX3 and one receive subarray RS1, domain D2 contains two transmit subarrays TX2, TX4 and one receive subarray RS2, domain D3 contains two transmit subarrays TX5, TX7 and one receive subarray RS3, and domain D4 contains two transmit subarrays TX6, TX8 and one receive subarray RS4. In the example shown, each of the transmitting subarrays TX1-TX8 consists of one column or several parallel vertically (in the z-direction) extending columns of antenna elements 14. Each of the receiving subarrays RS1-RS4, on the other hand, consists of four parallel columns or sub-subarrays RX1-RX4, RX5-RX8, RX9-RX12 and RX13-RX16 respectively, arranged in two parallel horizontally (in the y-direction) extending rows, with uniform spacing between them within each row.

[0025] The transmit subarrays TX1-TX8 form four pairs that are offset from each other and from the receive subarrays RS1-RS4 in the z-direction. The transmit subarrays of each pair are at the same height in the z-direction and have a distance from each other in the y-direction that is greater than the combined widths (in the y-direction) of the transmit subarrays RS1 and RS2 (or RS3 and RS4).

[0026] The multi-column transmit subarrays enable high-resolution azimuth angle estimation. Based on the MIMO principle, measurements can be performed with different combinations of transmit and receive subarrays, for example, by alternately transmitting with the transmit subarrays TX1 and TX2 and evaluating the received signals from all eight antenna columns of the receive subarrays RS1 and RS2. Since the phase relationships between the received signals depend on the relative position of the transmit and receive subarrays in the y-direction, this results in a virtual array that is more than twice as wide as the two receive subarrays RS1 and RS2 combined.

[0027] The offset of the transmit and receive subarrays in the z-direction also allows for an angle estimation in elevation according to the same principle.

[0028] However, a prerequisite for successful angle estimations is that the antenna diagrams, which indicate the phase relationships between the received signals depending on the object's location angle, are correctly calibrated.

[0029] However, a complication arises with the radar sensor shown here because, for example, the subarrays in domains D1 and D2 are fed by two different high-frequency components, HF1 and HF2. Therefore, the correct calibration of the antenna pattern also depends on a possible phase offset between the transmit signals of the two high-frequency components, HF1 and HF2. Since this phase offset can change during operation of the radar sensor, for example due to temperature fluctuations in the high-frequency components, a single factory calibration of the radar sensor is insufficient.

[0030] The subject of the invention described here is therefore primarily a method that allows the antenna diagrams to be recalibrated during the operation of the radar sensor in such a way that the phase offsets between the various high-frequency components HF1-HF4 are also correctly taken into account.

[0031] For easier understanding, however, the calibration procedure for a highly simplified antenna array, which is shown in Fig. 2 shown.

[0032] In this example, the antenna array comprises only two transmitting antenna elements 14T and four receiving antenna elements 14R. The transmitting antenna elements 14T are powered by a high-frequency module HF with an attached control and evaluation unit and emit radar signals that are reflected by an object 18 and received by each of the receiving antenna elements 14R. The received signals are evaluated separately in the control and evaluation unit.

[0033] For the sake of clarity only, here (as also in Fig. 1 A bistatic antenna system has been shown in which the transmitting antenna elements 14T are different from the receiving antenna elements 14R. In practice, a monostatic antenna concept can also be used, in which the same antenna elements are used for both transmitting and receiving.

[0034] In this example, the receiving antenna elements 14R are arranged at equal intervals along a straight line (ULA; Uniform Linear Array). The same applies to the transmitting antenna elements 14T, although the transmitting and receiving antenna elements do not necessarily have to be arranged on the same straight line.

[0035] In the example shown, the radar sensor is operated using time-division multiplexing. This means that at any given time, only one of the N tx (= 2) transmitting antenna elements 14T is active. The activity phases of the individual antenna elements alternate cyclically. Fig. 2 This symbolically represents the case in which only the lower of the two transmitting antenna elements 14T is active.

[0036] Alternatively, the radar sensor could also be operated using frequency multiplexing. In this case, all transmitting antenna elements 14T would be active simultaneously, but would operate at slightly different frequencies, so that the signals from the different transmitting antenna elements could be separated again at the receiver.

[0037] Another possibility would be the code division multiplexing method. In this method, a specific code is modulated onto the signal transmitted by each of the 14T transmitting antenna elements, and the signals are separated from each other on the receiving side by code-selective filtering.

[0038] For illustrative purposes, the time-division multiplexing method will be considered below. In a simple signal model, it can be assumed that object 18 is a point-like scattering center at which the signal emitted by the active transmitting antenna element 14T is scattered as a spherical wave, which then reaches the various receiving antenna elements 14R. Fig. 1 However, the distance between the radar sensor and object 18 is unrealistically small. In practice, this distance is so large that the dimensions of the radar sensor 8 are negligibly small compared to the object distance. More realistic conditions are shown Fig. 3 , where the object 18 is located at a greater distance in front of the front of a motor vehicle 20, on which the radar sensor 8 is arranged. The radar waves arriving at the location of the radar sensor can then be considered, to a good approximation, as plane waves that reach all receiving antenna elements 14R at practically the same angle of incidence, the (azimuth) angle θ of the object 18.

[0039] With xn The four-component vector whose components (xn,1 , xn,2 , xn,3 , xn,4 ) represent the complex amplitudes of the signals transmitted by the nth transmitting antenna element 14T and received by the four receiving antenna elements 14R is to be denoted by . Where d is the distance between antenna elements, λ is the wavelength of the radar radiation, and s = xn,1 is the (time-dependent) complex amplitude of the signal received by the first of the receiving antenna elements 14R (for example, the rightmost antenna element in ). Fig. 3 ), so, due to the differences in path length between the signals reaching the various receiving antenna elements 14R, the following relationship applies: xn ¯ θ = s 1 , e − 2 πi d / λ sin θ , e − 2 πi 2 d / λ sin θ , e − 2 πi 3 d / λ sin θ , T = s a _ rx θ

[0040] The superscript symbol "T" denotes transposition, since vectors are written here as row vectors but should be considered as column vectors. The vector arx is referred to as the receive control vector. This control vector specifies the geometric and wave propagation properties of the receiving antenna array under consideration. Such a control vector can be defined not only for ULA antenna arrays, but also more generally for other antenna configurations.

[0041] Accordingly, a control vector can also be defined for the array of (in this example only two) transmitting antenna elements 14T. a define tx, which in this example would essentially specify the path length differences of the optical paths from the transmitting antenna elements to object 18.

[0042] The control vector is obtained for the entire MIMO antenna array. a _ θ = a _ tx θ * a _ rx θ

[0043] The symbol * here represents the Kronecker product. In the example considered here, the following applies: a _ θ = a tx , 1 a rx , 1 , a tx , 1 a rx , 2 , a tx , 1 a rx , 3 , a tx , 1 a rx , 4 , a tx , 2 a rx , 1 , a tx , 2 a rx , 2 , a tx , 2 a rx , 3 , a tx , 2 a rx , 4 T

[0044] The received signals form a vector x with N tx · N rx components (8 components in this example), and the following holds: x _ θ = s a θ

[0045] Knowledge of the control vector a (θ) allows us to establish a (under suitable conditions unique) relationship between the angle θ of the object and the received signals x, and to deduce the azimuth angle θ of the object from the amplitude and phase relationships of the received signals. However, since in practice the received signals will be more or less noisy, the azimuth angle cannot be calculated exactly, but only estimated, for example using a maximum likelihood estimation.

[0046] When this principle is generalized to multi-target estimation, the single angle θ becomes a vector. θ , whose components specify the angles of the different targets, from the control vector aA control matrix A is created, and the following relationship applies: x _ = A θ _

[0047] In principle, for a given radar sensor, the antenna diagram, i.e., the entirety of all control vectors, can be determined. a (θ) for all possible azimuth angles θ, must be measured before the sensor is put into operation. In the Fig. 3 In the example shown, however, the radar sensor 10 is installed behind a bumper 22 of the vehicle, and this bumper 22 also has a specific relief 24, for example, because an emblem of the vehicle brand is embossed or stamped on it. Since the optical density of the bumper differs from the optical density of the air, the refraction of the radar waves at and within the bumper 22 leads to differences in propagation length, which affect the antenna pattern. This effect depends on the exact installation location of the radar sensor relative to the relief 24 and is therefore difficult to compensate for computationally, especially since effects such as dirt on the bumper can further distort the antenna pattern. It is understood that these problems are exacerbated when the antenna array has larger dimensions, as, for example, in Fig. 1 .

[0048] The aim now is to calibrate the antenna pattern after the radar sensor has been installed in the vehicle, so that such distortions of the antenna pattern do not lead to significant errors in angle estimation.

[0049] First, one way to define the transmit control vector will be described. a to recalibrate tx(θ). It is useful to use relative control vectors. a' tx (θ) and a' To define rx (θ): a ′ ¯ tx θ : = a ′ ¯ tx θ / a tx , 1 θ and a ′ ¯ rx θ : = a tx , 1 θ a _ rx θ Because of equation (2), then a _ θ = a ′ ¯ tx θ * a ′ ¯ rx θ since the factor a' cancels out tx (θ). A recalibration of the relative transmit control vector. a' tx (θ) is therefore used for a recalibration of the original transmit control vector. a tx (θ) equivalent.

[0050] The quality of an angle estimation is determined by the so-called quality function q( θ ) described. This function is a measure of the probability that the estimates obtained with the estimation algorithm correspond to the actual angular positions of the located objects. For the general case of multi-target estimation, where several targets, indistinguishable based on their distances and relative velocities, are located at different angles θj, the performance function is given by the equation q 2 θ _ = x _ H P A θ x _

[0051] In this equation, x is the vector of signals obtained with the different combinations of transmitting and receiving antenna elements. x H< the Hermitian conjugate vector to x, and PA ( θ ) = A(AH< A) -1< AH< , where A is the control matrix containing the control vectors of all targets. If necessary, the control matrix may also contain multiple control vectors for the same target if, due to reflections from guardrails or the like, multiple optical paths lead from the same target to the radar sensor.

[0052] In the case of a single destination with only one path, this equation simplifies to q 2 θ = a _ H θ x _ 2 / a _ θ 2 x _ 2

[0053] The control vector a (θ) can be measured and normalized before the radar sensor is activated. The signal vector x can also be normalized after each measurement. Therefore, it will be assumed in the following that both the control vector and the signal vector are normalized, which further simplifies the equation to: q 2 θ = a _ H θ x _ 2

[0054] The vector x can be written as x _ = xi ¯ , i = 1 … N tx

[0055] It contains xi The vectors represent the measurement signals transmitted by the i-th transmitting antenna element and received by the N rx receiving antenna elements. For a (single-target) MIMO angle estimation with all 14T transmitting antenna elements, the quality function is then obtained. q 2 θ = a _ H θ x _ 2 = ∑ i a tx , i * θ a _ rx H θ xi ¯ 2

[0056] In this, the sum runs over all N tx transmit antenna elements (summation index i), and a tx,i *(θ) is the complex conjugate of the individual component a tx,i (θ) of the transmit control vector. a tx (θ). The last transformation in equation (13) follows from equation (2).

[0057] If we now define sizes yi (θ) as: y i θ = a _ rx H θ xi ¯ Thus, from equation (13) it follows: q 2 θ = ∑ i a tx , i * θ y i θ 2

[0058] For a single-target SIMO angle estimation performed only with the i-th transmitting antenna element, one obtains (with normalization): q 2 = y i θ 2 / xi ¯ 2 .

[0059] Based on this angle accuracy, it can now be determined whether the angle estimation involves a single-objective or a multi-objective situation. In a multi-objective situation, the accuracy function will have a significantly lower value. Therefore, the criterion used to decide is that the accuracy function for the estimated angle θ lies below a suitably chosen threshold value.

[0060] Alternatively, other methods and criteria can be used to distinguish between a single-goal situation and a multi-goal situation.

[0061] If signal noise is neglected, the following approximates the true angle θ: xi ¯ = s a tx , i θ a _ rx θ where s represents the complex amplitude of the signal emitted by the target and θ is the actual angle of the target.

[0062] Substituting equation (17) into equation (14) yields: y i θ = a _ rx H θ s a tx , i θ a _ rx θ

[0063] There aWhen rx (θ) is normalized, this simplifies to y i θ = s a tx , i θ

[0064] If SIMO measurements are performed successively with each of the transmitting antenna elements 14T, N tx relationships of the type given in equation (19) are obtained. However, it is not yet possible to directly verify whether and how exactly these relationships (19) are satisfied, since the amplitude s is unknown. This problem can be circumvented by dividing the vectors on the left and right sides of equation (19) by one of their components (without loss of generality, by the first component a tx,1 (θ) and y 1 (θ) respectively). This yields: a tx , n θ / a tx , 1 θ = y n θ / y 1 θ

[0065] The quantities on the left-hand side of equation (20) are the components of the relative transmit control vector a tx '(θ). The quantities on the right-hand side are obtained from the measurement results according to equation (14). xi and the known, originally used receiver control vector a rx (θ).

[0066] Recalibration is therefore easily possible by changing the previously used relative transmit control vector. a tx '(θ) is replaced by the vector with the components yn (θ) / y 1 (θ). It is equally easy to determine how much the previously used antenna pattern was distorted by the deviation between the old and the new relative transmit control vector.

[0067] In the procedure described above, a coherent summation (yi (θ) = ) must first be performed. a H< rx (θ) xi ) are performed and then the ratio yn(θ) / y1(θ) is calculated. In general, however, only a tolerably small error arises if the calculation is simplified by first calculating the ratios xn,k / x1,k and then averaging over all k (i.e., over all receiving antenna elements). Therefore, the following relationships can also be used as an approximation instead of equation (20): a tx , n θ / a tx , 1 θ = 1 / N rx ∑ k x n , k / x 1 , k

[0068] In this, xn,k = ( xn ) k = xj , with j = (k-1)N tx +n, the nth component of the partial vector xn from x.

[0069] The core of the process therefore comprises the following steps: Calculating a first component of the transmit control vector aCalculate a reference value dependent on tx (θ) (e.g. a tx,n (θ) / a tx,1 (θ) ) for each transmitting antenna element (with the index n), calculate a second reference value dependent on the results of the SIMO measurements (e.g. yn (θ) / y 1 (θ) or Σ k (xn,k / x 1,k ) ) for each transmitting antenna element, and correct the transmit control vector. a tx (θ) (or equivalently the relative transmit control vector) a' tx (θ) ) based on a known relationship (equation (20) or (21) ) between the first and second comparison quantities for each transmitting antenna element.

[0070] An analogous method also allows for the calibration or recalibration of the receive antenna pattern, i.e., the receive control vector. a rx .

[0071] Equation (13) is replaced by: q 2 θ = ∑ i a tx , i * θ a _ rx H θ xi ¯ 2 = a _ rx H θ x ′ ¯ θ 2 with x ′ ¯ θ = ∑ i a tx , i * θ xi ¯ and analogous to equation (12): x ′ ¯ = xn ¯ ′ , n = 1 … N rx

[0072] For a single-target MISO angle estimation (Multiple Input Single Output), performed only with the nth receiving antenna element, one obtains (with normalization): q 2 = xn ¯ ′ 2 / ∑ i xn ¯ ′ i 2 − 1 / 2

[0073] Again, the angle accuracy can be used to determine whether the angle estimation involves a single-target situation or a multi-target situation.

[0074] If a single-target situation exists and signal noise is neglected, then the following approximate values ​​apply to the true angle θ: xi ¯ = s a tx , i θ a _ rx θ

[0075] A calculation analogous to equations (18) to (20) then leads to: a rx , n θ / a rx , 1 θ = x ′ n θ / x ′ 1 θ

[0076] The first comparison quantities in this case are therefore the components a rx,n (θ) / a rx,1 (θ) of a relative receive control vector. a"rx (defined analogously to equation (6) ). The second comparison quantities, which now depend on the results of the MISO measurements, are formed by the quantities x' n (θ) / x' 1 (θ) for each receiving antenna element (index n).

[0077] As a good approximation, the comparison quantities and relationships analogous to equation (21) can also be used in this case: a rx , n θ / a rx , 1 θ = 1 / N tx ∑ k x k , n / x k , 1

[0078] The calibration procedures described above can also be applied analogously to radar sensor 8 according to Fig. 1 The calibration of the subarrays within a domain corresponds exactly to the procedures described above, whereby the transmitting subarrays, for example TX1 and TX3, take the place of the transmitting antenna elements 14T, while for the receiving antennas the sub-subarrays, for example RX1-RX4, take the place of the receiving antenna elements 14R.

[0079] However, calibration procedures are also possible that cross the boundaries between the domains. For example, the antenna pattern can be calibrated for the two receive subarrays RS1 and RS2. The receive control vector then has eight components, four for each of the two subarrays. The MISO measurements are then performed for each of the eight sub-subarrays RX1, RX8, for example, using the four transmit subarrays TX1-TX4. The measurements with the antenna combinations TX1, RX1-RX4 and TX3, RX1-RX4 provide a calibration for the sub-subarrays within domain D1. The measurements with the antenna combinations TX2, RX1-RX4 and TX4, RX1-RX4 provide a calibration for the same sub-subarrays, however, these two calibrations will generally not agree, as a phase offset between the high-frequency components HF1 and HF2 may also be noticeable in the measurements with TX2 and TX4.The same applies to the two possible calibrations of the sub-subarrays RX5-RX8 in domain D2. Generally, the phase offset between the two RF components will cause the corrected receive control vectors to differ from each other in either the first four components or the last four components, depending on the domain in which the transmit subarray is located. This deviation can then be used to determine the phase offset between the RF components HF1 and HF2 and, based on this, to perform a phase calibration of the two RF components.

[0080] The same applies analogously to domains D3 and D4 and the calibration of the high-frequency components HF3 and HF4.

[0081] The calibration of the high-frequency components can also be carried out as part of a recalibration of the transmit control sectors using SIMO measurements.

[0082] By recalibrating the antenna diagrams for the elevation angle ϕ, the phase offset between the high-frequency components HF1 and HF3, as well as the phase offset between the high-frequency components HF2 and HF4, can be determined in a similar way, so that ultimately a phase calibration for all four high-frequency components is achieved.

[0083] In Fig. 4 A complete calibration procedure is shown that can be performed while driving the motor vehicle which has the radar sensor 8.

[0084] In step S1, normal localization operations are performed first, i.e., objects in the vicinity of the vehicle are located using radar sensor 8. Multi-target angle estimates are then performed for the located objects based on the antenna patterns (in azimuth and elevation) that were originally measured when the radar sensor was commissioned or recalibrated during previous recalibration procedures.

[0085] In step S2, it is checked whether an object has been located in the current positioning cycle at an azimuth angle θ and / or at an elevation angle ϕ for which no recalibration has yet taken place or the last recalibration was some time ago.

[0086] If this is the case (J), a SIMO measurement and angle estimation for the azimuth or elevation angle is performed in step S3 using one of the transmitting subarrays TX1 - TX4. Otherwise (N), the process returns to step S1, and the loop with steps S1 and S2 is executed until an object is found at an angle requiring recalibration.

[0087] Based on the angular accuracy of the SIMO angle estimation performed in step S3, a decision is made in step S4 as to whether the object located at the angle θ or ϕ is a single object or not.

[0088] If it is not a single object (N), the loop returns to steps S1 and S2. If it is a single object (J), further SIMO measurements are taken in step S5 with all transmit subarrays TX1–TX4. This yields a complete set of the quantities yi (θ) for all transmit subarrays. Based on the resulting comparison quantities, the (relative) transmit control sector a' tx (θ) is then corrected (recalibrated) in step S6.

[0089] Optionally, a further step S7 can be added, in which MISO measurements are performed for the same object using all sub-subarrays of the receive subarray RX1 - RX16. Based on these measurements, the (relative) receive control vector a" rx (θ) is then corrected in step S8.

[0090] If time-division multiplexing is used, the time intervals between individual SIMO or MISO measurements should not be too large to prevent errors caused by any displacement of the object during the time elapsed between measurements from becoming significant. However, it is possible to interlock the measurements and / or combine the measurement results in such a way that the errors caused by the time offset average out. An example of such a method is described in DE 10 2013 209 708 A1.

[0091] An analogous refinement also exists in the frequency division multiplexing method, since here the distance of the target in conjunction with the frequency offset between the transmitting antenna elements can lead to a phase difference between the sizes yi (θ), which may need to be compensated.

Claims

1. Method for phase calibration of a MIMO radar sensor (8) having an array (10) composed of multiple transmitting and receiving antenna elements (14) offset with respect to one another in at least one direction (y, z), and having multiple high-frequency modules (HF1 - HF4), each of which is associated with a part (D1 - D4) of the array (10), the phase calibration including a correction of phase differences of the high-frequency modules, characterized in that the array (10) is subdivided into transmitting subarrays (TX1 - TX8) and receiving subarrays (RS1 - RS4) in such a way that each subarray is associated with exactly one of the high-frequency modules and at least two receiving subarrays (RS1, RS2; RS3, RS4) belonging to different high-frequency modules are offset with respect to one another in the at least one direction (y, z) and are aligned with one another in the direction (z, y) perpendicular thereto, and in that the method comprises at least one calibration routine having the following steps: before start-up of the radar sensor: - storing an antenna pattern which assigns to each of multiple angles θ a respective control vector as(θ) composed of a transmission control vector astx(θ) and a reception control vector asrx(θ), the control vector having at least one component for each subarray, after start-up: - carrying out a radar measurement for locating an object (18), - checking whether the located object is a single target or a multiple target, and, - if it is a single target: - carrying out a MISO measurement with each of the at least two receiving subarrays (RS1, RS2; RS3, RS4), each MISO measurement for a receiving subarray also including a measurement with a transmitting subarray associated with another high-frequency module, - estimating the angle θ of the object on the basis of the measurement results, - calculating a first comparison variable, which is dependent on the components of the reception control vector asrx(θ), for each of the at least two receiving subarrays (RS1, RS2; RS3, RS4), - calculating a second comparison variable, which is dependent on the results of the MISO measurements, for each of the at least two receiving subarrays (RS1, RS2; RS3, RS4), and - correcting the reception control vector asrx(θ) on the basis of a known relationship between the first and second comparison variables for the relevant receiving subarrays (RS1, RS2; RS3, RS4).

2. Method according to Claim 1, characterized in that the at least one calibration routine comprises the following further steps, which are carried out after start-up of the radar sensor: - if the located object is a single target: - carrying out a SIMO measurement with each transmitting subarray, each SIMO measurement for a transmitting subarray also including a measurement with a receiving subarray associated with another high-frequency module, - estimating the angle θ of the object on the basis of the measurement results, - calculating a third comparison variable, which is dependent on the components of the transmission control vector astx(θ), for each of the at least two transmitting subarrays (TX1, TX2; TX3, TX4), - calculating a fourth comparison variable, which is dependent on the results of the SIMO measurements, for each of the at least two transmitting subarrays (TX1, TX2; TX3, TX4), and - correcting the transmission control vector astx(θ) on the basis of a known relationship between the third and fourth comparison variables for the relevant transmitting subarrays (TX1, TX2; TX3, TX4).

3. Method for phase calibration of a MIMO radar sensor (8) having an array (10) composed of multiple transmitting and receiving antenna elements (14) offset with respect to one another in at least one direction (y, z), and having multiple high-frequency modules (HF1 - HF4), each of which is associated with a part (D1 - D4) of the array (10), the phase calibration including a correction of phase differences of the high-frequency modules, characterized in that the array is subdivided into transmitting subarrays (TX1 - TX8) and receiving subarrays (RS1 - RS4) in such a way that each subarray is associated with exactly one of the high-frequency modules (HF1 - HF4) and at least two transmitting subarrays (TX1, TX2; TX3, TX4) belonging to different high-frequency modules are offset with respect to one another in the at least one direction (y, z) and are aligned with one another in the direction (z, y) perpendicular thereto, and in that the method comprises at least one calibration routine having the following steps: before start-up of the radar sensor: - storing an antenna pattern which assigns to each of multiple angles θ a respective control vector as(θ) composed of a transmission control vector astx(θ) and a reception control vector asrx(θ), the control vector having at least one component for each subarray, after start-up: - carrying out a radar measurement for locating an object (18), - checking whether the located object is a single target or a multiple target, and, - if it is a single target: - carrying out a SIMO measurement with each transmitting subarray, each SIMO measurement for a transmitting subarray also including a measurement with a receiving subarray associated with another high-frequency module, - estimating the angle θ of the object on the basis of the measurement results, - calculating a first comparison variable, which is dependent on the components of the transmission control vector astx(θ), for each of the at least two transmitting subarrays (TX1, TX2; TX3, TX4), - calculating a second comparison variable, which is dependent on the results of the SIMO measurements, for each of the at least two transmitting subarrays (TX1, TX2; TX3, TX4), and - correcting the transmission control vector astx(θ) on the basis of a known relationship between the first and second comparison variables for the relevant transmitting subarrays (TX1, TX2; TX3, TX4).

4. Method according to one of the preceding claims, for a radar sensor having at least three high-frequency modules (HF1-HF4), in which the array (10) is subdivided into transmitting subarrays (TX1-TX8) and receiving subarrays (RS1-RS4) in such a way that at least two subarrays RS1, RS2 belonging to a first (HF1) and a second (HF2) high-frequency module are arranged horizontally offset from one another and at least one further subarray (TX5-TX8) belonging to a third high-frequency module (HF3, HF4) is arranged vertically offset from the first two subarrays, and in which an azimuth angle estimation is performed for a phase calibration of the first and second high-frequency modules (HF1, HF2) and an elevation angle estimation is performed for a phase calibration of the third high-frequency module (HF3, HF4).

5. MIMO radar sensor (8) having an array (10) composed of multiple transmitting and receiving antenna elements (14) offset with respect to one another in at least one direction (y, z), and having multiple high-frequency modules (HF1-HF4), each of which is associated with a part (D1-D4) of the array, and having an evaluation and control circuit (16) for the high-frequency modules, characterized in that the evaluation and control circuit is designed to carry out a method according to one of Claims 1 to 4.