Radar modulation method with high range resolution under low signal processing load

Abstract

A method for a radar system for detecting the surroundings according to the invention is proposed, the system having a transmitting device for radiating a sequence of transmission signals which contain at least approximately identical single signals, wherein the sequence of individual transmission signals is cyclically repeated, the method being characterized in that the frequency position of the single signals is changed at least approximately linearly over the sequence of single signals, if necessary in addition to a varying and at least approximately mean-free component, and in that here the slope of the linear frequency position change over the individual transmission signals varies at least sometimes over the sequence, in order to improve the radial distance measurement accuracy and / or the relative speed measurement accuracy and / or to be more stable in terms of interference with other radar systems, in particular.

Description

Technical Field

[0001] This invention relates to radar methods and radar systems for driver assistance systems in motor vehicles. According to the invention, the radar system exhibits high range resolution with a low signal processing load. Background Technology

[0002] More and more motor vehicles are equipped with driver assistance systems. These systems use sensor systems to detect the surrounding environment and, based on the identified traffic conditions, initiate automatic vehicle responses and / or provide instructions to the driver, especially issuing warnings. Here, comfort functions and safety functions are distinguished.

[0003] Currently, FSRA (Full Speed ​​Range Adaptive Cruise Control) plays a crucial role as a comfort feature. It adjusts the vehicle's inherent speed to the driver's preset desired speed whenever traffic conditions permit, otherwise automatically adapting the inherent speed to the traffic situation. Furthermore, with the expansion towards at least partially automated lane-changing capabilities, the importance of lane-change assist systems is continuously increasing.

[0004] In addition to improving comfort, safety features are receiving increasing attention, with shortening braking or stopping distances in emergency situations playing a crucial role. The corresponding driver assistance functions range from automatic pre-charging of the brakes to reduce braking wait time to autonomous emergency braking.

[0005] For the aforementioned types of driver assistance systems, radar sensors are now primarily used. They operate reliably even in adverse weather conditions and, in addition to measuring distance to objects, can directly measure the radial relative velocity of objects via the Doppler effect. The transmission frequencies used here are 24 GHz and 77 GHz.

[0006] The aforementioned functions require a fairly high sensor operating range, along with high distance measurement accuracy, distance measurement resolution, and distance measurement separation performance. Therefore, high distance resolution and distance separation performance are also important, as they at least partially compensate for the insufficient angular resolution and angular separation performance of vehicle radar sensors (due to their small size). However, simultaneously achieving high operating range and distance resolution typically requires a high digital signal processing load, which is difficult to achieve because the corresponding signal processors currently used in vehicles are either limited in scope or prohibitively expensive.

[0007] The following methods, proposed in DE 10 2013 200 404 A1 and WO 2018 / 086783 A1, should allow for high range and high distance resolution under moderate digital signal processing loads. However, these methods have failed to achieve the goal of high distance resolution, especially at higher radial relative velocities, and also suffer from reduced sensitivity, i.e., range. Summary of the Invention

[0008] The objective of this invention is to achieve both high effective range and high distance resolution for relatively moving objects using a motor vehicle radar sensor, with a moderate digital signal processing load.

[0009] This task is, in principle, solved using a radar method or radar system according to any one of claims 1 to 12. Here, according to the invention, how to design radar modulation and signal evaluation so that high measurement accuracy and high measurement resolution can be achieved for the distance and relative velocity of the object.

[0010] The advantage of this invention is that it allows the use of signal processors currently available and advantageous for motor vehicle applications to implement sensors with both high range and high distance resolution, enabling the implementation of next-generation, demanding driver assistance systems. Furthermore, the simpler signal processors offer the advantage of consuming less power.

[0011] In the method of a radar system for detecting the surrounding environment according to the invention, the radar system has a transmitting device for radiating transmitted signals, the transmitted signals comprising a sequence of at least approximately identical single signals. Here, the sequences of individual transmitted signals are cyclically repeated. Where necessary, except for varying and at least approximately meanless components, the frequency position of the single signal changes at least approximately linearly on the sequence of single signals, and here, the slope of the linear frequency position change on each transmitted signal varies at least sometimes or partially with the sequence, particularly to improve the accuracy of radial distance measurement and / or relative velocity measurement and / or to be more stable in terms of interference with other radar systems.

[0012] According to a preferred embodiment of the invention, in a sequence of single signals, the frequency position of a single signal (characterized particularly by its center frequency) and its time interval (which, if necessary, excludes varying and at least approximately mean-less components, respectively) can change at least approximately linearly. Here, the relative change in time interval is at least approximately twice the relative change in frequency position, wherein the signs of these changes are opposite.

[0013] Suitable, the frequency position, time interval, and / or phase position of a single signal can be superimposed with random or pseudo-random components.

[0014] The frequency of a single signal is preferably linearly modulated, and the slope of the frequency modulation is at least approximately the same for all single signals, where each transmitted signal refers to the frequency ramp.

[0015] According to the advantageous design of the method, for the K frequency ramps numbered k = 0, ..., K-1 below, I digital received values ​​numbered i = 0, ..., I-1 below can be obtained for multiple receiving channels. Subsequently, a two-dimensional discrete Fourier transform can be performed on each of the I·K received values. If necessary, this two-dimensional discrete Fourier transform can be performed incompletely and preferably by means of a one-dimensional fast Fourier transform. Here, the dimension generated by the exponential dimension i of the received value after the transform can be called the distance gate j = 0, ..., J-1, and the dimension generated by the frequency ramp dimension can be called the Doppler gate l = 0, ..., L-1.

[0016] Furthermore, the linear variation in frequency position and the time interval between each frequency ramp can cause sharp power peaks in the received signal of the transmitted signal reflected on the object after the two-dimensional discrete Fourier transform, when the object moves toward or away from the radar system, i.e., has a relative radial motion component.

[0017] The linear variation of the frequency position on the frequency ramp can be advantageously considered by determining the radial relative velocity of the object, where the position of the power peak after the two-dimensional discrete Fourier transform is corrected primarily by a component linearly related to the range gate dimension j in the Doppler gate dimension l. Here, the linearity factor is derived by the quotient of the variation in frequency position on the frequency ramp and the variation in frequency over the reception time range during each frequency ramp. The position of the power peak is preferably determined by interpolation, thereby typically producing non-integer values ​​for the range gate dimension j and / or the Doppler gate dimension l. This design of the invention is readily applicable to all general radar systems in which the frequency positions of the various frequency ramps vary.

[0018] Suitablely, the linear variation of the frequency position of each frequency ramp can be considered. This is achieved by performing a one-dimensional discrete Fourier transform on the I received values ​​for each frequency ramp k = 0, ..., K-1. In the corresponding case, the phase of the value generated in the range gate dimension j is corrected by a phase component proportional to the product 2π·j·k / K. The scaling factor is primarily derived from the quotient of the change in frequency position on the frequency ramp and the frequency variation within the receiving time range during each frequency ramp. This correction can then be achieved by multiplying the phase component by a complex pointer of length 1 and the corresponding phase. This design of the present invention can be explicitly applied to all general radar systems in which the frequency positions of each frequency ramp vary.

[0019] A sequence of K individual transmitted signals can be cyclically repeated, wherein the slope of the linear frequency position change on each transmitted signal changes at least sometimes (i.e. at least in one or more sequences) with respect to the sequence, especially to improve the accuracy of radial distance measurement and / or relative velocity measurement and / or to be more stable in the event of interference with other radar systems.

[0020] Each transmitted signal preferably represents a frequency ramp, wherein two periods with opposite slopes, i.e., a phase difference factor of -1, are used for precise radial distance measurement and / or relative velocity measurement of the object. Here, the sum and difference of the positions of the object's power peaks after the two-dimensional discrete Fourier transform, generated within the two periods, are used primarily in the Doppler gate dimension, not in the range gate dimension.

[0021] Suitablely, the sequence of K individual transmitted signals can be repeated cyclically, wherein, especially in order to resolve ambiguities in determining radial relative velocities and / or to have greater stability in relation to interference with other radar systems, the average time interval varies at least sometimes with the sequence, i.e., varies at least in one sequence.

[0022] Multiple receiving channels can preferably be implemented using multiple transmitting antennas and / or receiving antennas. In addition, besides performing two-dimensional discrete Fourier transforms on I·K received values ​​respectively, digital beamforming can also be set on the receiving channels or to generate the receiving channels.

[0023] Furthermore, the present invention also claims a radar system for detecting the surrounding environment, the radar system comprising a transmitting device for radiating transmitted signals, the transmitted signals comprising a sequence of at least approximately identical single signals. The radar system is characterized in that it operates using the method according to the present invention. Attached Figure Description

[0024] Figure 1 An exemplary implementation of a radar system is shown.

[0025] Figure 2 This illustrates the frequency of a transmitted signal with a constant frequency position, representing a so-called frequency ramp, according to the prior art.

[0026] Figure 3 The absolute value spectra of three objects after two-dimensional discrete Fourier transforms are shown, and the results are based on... Figure 2 The frequency change process,

[0027] Figure 4 The frequency of the transmitted signal with a linearly varying frequency position is shown.

[0028] Figure 5The absolute value spectra of three objects after two-dimensional discrete Fourier transforms are shown, and the results are based on... Figure 4 The frequency change process, in which the spacing of the frequency ramps is constant.

[0029] Figure 6 The absolute value spectrum is shown with the spacing of the frequency ramps selected according to WO 2018 / 086783 A1.

[0030] Figure 7 The absolute value spectrum is shown with the spacing of the frequency ramps selected according to the present invention. Detailed Implementation

[0031] Considering Figure 1 An exemplary implementation of a radar system is shown in the figure. The radar system has a transmitting antenna TX0 for radiating transmitted signals and M = 4 receiving antennas RX0 to RX3 for receiving transmitted signals reflected from objects; the antennas are implemented as patch antennas in planar technology on a planar circuit board 1.1, wherein the circuit board is oriented with respect to the horizontal and vertical directions within the vehicle, as shown. All antennas (transmitting and receiving antennas) have the same beam characteristics at elevation and azimuth angles, respectively. The four receiving antennas (and thus their phase centers, i.e., radiation centers) are each respectively separated by the same lateral, i.e., horizontal distance d = λ / 2 = 1.96 mm, where λ = c / 76.5 GHz = 3.92 mm is the average wavelength of the radiated signal in the 76 to 77 GHz frequency band used, and c = 3 * 10 8 meters per second is the speed of light.

[0032] The transmitted signal radiated on the transmitting antenna is obtained from a high-frequency oscillator 1.2 in the range of 76 to 77 GHz, the frequency of which can be controlled by the voltage v. 控制 The control voltage is generated in control device 1.7, which includes, for example, a phase-locked loop or a digital-to-analog converter, and controls the oscillator in such a way that the frequency change process corresponds to the desired frequency modulation.

[0033] The signals received by the four receiving antennas are mixed in parallel with the signal from the oscillator 1.2 in the real-value mixer 1.3 to reach the low-frequency range. The received signal then passes through a bandpass filter 1.4 with the shown transfer function, an amplifier 1.5, and an analog-to-digital converter 1.6. The received signal is then further processed in the digital signal processing unit 1.8.

[0034] In order to measure the distance to an object, such as Figure 2 As shown, the high-frequency oscillator and the frequency f of the transmitted signal are... TX Very rapid linear change (in T) ch =In 51.2 microseconds, change Bch =150MHz, where the center frequency is f c =76.5GHz); This refers to the frequency ramp (often also known as the "linear frequency modulation signal"). The frequency ramp at a fixed grid T Dc The frequency repeats periodically over 70 microseconds; there are a total of K = 256 frequency ramps, all of which have the same frequency variation process, i.e., the same frequency slope and the same frequency position (i.e., the same starting frequency and center frequency). In recent years, this modulation method has become increasingly popular in radars used to detect the surrounding environment of motor vehicles.

[0035] During each frequency ramp k = 0, ..., K-1, the received signals of each of the M = 4 analog-to-digital converters m = 0, ..., M-1 are sampled (or scanned) I = 256 times at intervals of 200 nanoseconds (i.e., at 5 MHz), wherein the sampling always begins at the same time point relative to the start of the ramp (see...). Figure 2 The resulting digital sample values ​​with exponents i = 0, ..., I⁻¹ are represented by s(i,k,m). Signal sampling is only meaningful within the time range during which the received signal from the object arrives at the distance of interest—that is, after the ramp begins, at least the propagation time corresponding to the maximum distance of interest must be waited (which corresponds to 0.66 microseconds at a maximum distance of interest of 99 meters); it should be noted that, here and thereafter, distance is always understood as radial distance.

[0036] As is known from the prior art and can be readily derived, in the case of a single point object at a distance r, the sampled signal s(i,k,m) represents a sinusoidal oscillation over an exponent i, which can be described, with a very good approximation, as follows:

[0037]

[0038] That is, the oscillation frequency is proportional to the object distance r. Generally, even in the case of radial relative motion between the object and the sensor, a constant distance can be assumed for the frequency of the sinusoidal oscillation with a very good approximation. However, the relative motion with a radial component v varies with the phase position of the sinusoidal oscillation. It has the following effects:

[0039]

[0040] That is, the phase position changes linearly on the frequency ramp k, where the rate of phase change is proportional to the radial relative velocity v of the object. Due to the linearity of the receiver, in the case of multiple and / or extended objects, the sampled signal s(i,k,m) is generated as a linear superposition of sine functions of the shape described above.

[0041] This signal form allows for further processing using a two-dimensional Fourier transform (DFT), which includes a signal window applicable to each received channel m. The DFT is preferably implemented in two stages using two one-dimensional fast Fourier transforms (FFT). After this two-dimensional discrete Fourier transform (DFT), a power peak appears in the resulting spectrum S(j,l,m), the corresponding position of which corresponds to the distance r and relative velocity v of the associated object. See [reference needed]. Figure 3 It shows the absolute spectrum of the receiver channel m, |S(j,l,m) / A(m)|, in dB, for three objects with the same radar cross section, at least approximately the same azimuth angle, and the following distances and relative velocities: [r1 = 29.5 m, v1 = 1.09 m / s], [r2 = 30 m, v2 = 1.09 m / s], and [r3 = 45 m, v3 = 60.4 m / s]. Receiver noise is also superimposed on the object signals, and the receiver noise in the spectrum is significantly lower than the power peak of the object labeled with the object number. Dimensions j = 0, ..., J-1, generated by dimension i (sampled value index), are represented using distance gates, and dimensions l = 0, ..., L-1, generated by dimension k (frequency ramp), are represented using Doppler gates. This is because the location of the power peak in dimension j is primarily determined by object distance, and in dimension l by relative velocity (reflected through the Doppler effect). It is negligible here that the power peak location also has a very small dependence on one of the two physical parameters, distance and relative velocity. It is important to note that velocity cannot be explicitly calculated from the Doppler gates of the power peak, because in the proposed design, only a precise range of 28 m / s is achieved using K = L = 256 Doppler gates. Ambiguity can be addressed, for example, by adjusting the spacing T of the frequency ramps. Dc This is achieved by varying the radar cycle (see also below). Figure 3 The number of distance gates is only J = 100, which is significantly less than the number of sampled values ​​I = 256. The background is that, on the one hand, the sampled values ​​are real values, and therefore their spectrum is symmetric, meaning that no additional information is contained in the upper half of the Discrete Fourier Transform (DFT). On the other hand, according to... Figure 1 The simulated bandpass filter 1.4 has an upper transition range with a frequency bandwidth of 1.09 MHz (corresponding to a range of 56 frequency interpolation points). The modulation bandwidth B used here... ch =150MHz, distance gate width B ch / 150MHz·1m is exactly 1 meter, therefore J=100 distance gates allow a maximum effective range of 99 meters.

[0042] from Figure 3It can be seen that the first two objects with [r1 = 29.5 m, v1 = 1.09 m / s] and [r2 = 30 m, v2 = 1.09 m / s] cannot be separated, but are merged into a single power peak because they have the same relative velocity and only slightly different distances, with a distance difference of 0.5 m, which is only half a distance gate. Typically, to separate two point objects with the same relative velocity, a difference of approximately two distance gates is required. For the distance separation performance of these two objects, a significantly higher modulation bandwidth B is needed. ch It is at least 4 times higher, that is, B ch =600MHz, which results in a distance gate width of B ch / 150MHz·1m=0.25m. With the same maximum sensor range of approximately 99m, each frequency ramp requires more than 4 times the number of sampling values. This requires, on the one hand, a faster analog-to-digital converter, and on the other hand, and more seriously, approximately 4 times the processing power and memory in the digital signal processing device.

[0043] To avoid this situation, one can use, for example, what is known from DE 10 2013 200 404 A1 and... Figure 4 The alternative modulation scheme is shown in the diagram. Compared to the previous one, based on... Figure 2 The modulation scheme under consideration, in which the only variation is in the frequency position, specifically through the starting frequency and the center frequency F. c The frequency position marked (k) is now located on K = 256 frequency ramps, each with a frequency B. s / K, where B s =600MHz linear increase; thus effectively achieving a higher modulation bandwidth, and consequently resulting in significantly better distance separation performance. The frequency ramp spacing is constant, i.e., T is constant. D (k) = 70 microseconds. In this modulation scheme, the two-dimensional discrete Fourier transform (DFT) signal processing remains unchanged. Based on the example above, the absolute value spectrum |S(j,l,m) / A(m)| obtained for the three objects is... Figure 5 As shown in the image. (Based on...) Figure 3 Compared to the original absolute spectrum, the position of the power peak now shifts in the Doppler gate dimension l, but not in the range gate dimension j. This is because the linear increase in frequency position via the frequency ramp increases the number of wave trains in the beam path from the sensor to the corresponding object and the returning beam (wavelength decreases with increasing frequency), which affects the phase position of the received value s(i,k,m) according to equation (1). The component that varies linearly on the frequency ramp k has an effect; this component is superimposed on the component that is also linear in k caused by radial relative motion according to equation (2), so that the two components have essentially the same effect, namely, the shift of the power peak in the Doppler gate dimension. As shown later, the shift in the Doppler gate dimension l caused by the change in frequency position is approximated by the B of the distance gate dimension j of the corresponding object. s / B ch times.

[0044] According to Figure 5 The spectrum shows that the first two objects with [r1 = 29.5 m, v1 = 1.09 m / s] and [r2 = 30 m, v2 = 1.09 m / s] are now separated, forming two independent power peaks. This separation occurs in the Doppler gate dimension because a slightly different distance of 0.5 m results in different frequency position variations, shifted by two Doppler gates (the difference in the Doppler gate dimension is higher than the difference in the distance gate dimension by B). s / B ch That is, four times higher, where the difference is half a distance from the gate.

[0045] However, according to Figure 5 The unfavorable aspect in the spectrum is that the third object with [r3 = 45 m, v3 = 60.4 m / s] no longer has a sharp power peak, but instead exhibits a very strong extension in the Doppler gate dimension. This leads to several drawbacks: first, it reduces the possible detection range (because the level is smaller); second, relative velocity measurements become less accurate (because the power peak is blurred); and third, it is no longer possible to separate another target with the same relative velocity but a slightly different distance (because the blurred power peaks overlap). The larger the absolute value of the relative velocity, the stronger the dissipation of the power peak; in the first two objects, the effect is still not visible because their relative velocities are very small.

[0046] Proposed in WO 2018 / 086783 A1, it has a center frequency F c The spacing T of the frequency ramps k = 0, ..., K-1 of (k) D (k) is no longer constant, but is changed, thus making T D (k)·F c The product of (k) is constant. Then, for the example above, we derive... Figure 6 The absolute value spectrum |S(j,l,m) / A(m)|. And according to... Figure 5 Unlike the spectrum, the power peak dissipation of the third object with high relative velocity [r3 = 45 m, v3 = 60.4 m / s] is now smaller (about halved), but still exists and is unacceptably large.

[0047] Therefore, the method according to the invention is now derived, which prevents the blurring of power peaks even at high relative speeds.

[0048] The relative time t∈[-T] within the frequency ramp k ch / 2,T ch / 2] On the high-frequency oscillator and thus the frequency f of the transmitted signal TX (t,k) is applicable to:

[0049] f TX (t,k)=F c (k)+B ch / T ch ·t (3)

[0050] Among them, the center frequency F of the frequency ramps k = 0, ..., K-1 c (k) is:

[0051] F c (k)=F cc +B s / K·(k-(K-1) / 2), (4)

[0052] Among them, F cc It is all center frequencies F c The average value of (k). The phase of the oscillator signal and the transmitted signal is obtained through integration. Therefore, it is given by the following equation:

[0053]

[0054] The integral constant has no effect here and is therefore omitted.

[0055] For a single point object, the phase of the received signal at the output of the mixer It is generated by the phase difference between the current oscillator signal and the signal reflected back from the object, and its delayed transmission time Δt is:

[0056]

[0057] Among them, s ch B represents the modulation bandwidth of the linear frequency modulated signal. ch The symbols, i.e., +1 represents a rising frequency ramp, and -1 represents a falling frequency ramp. The received signal after the mixer is also called the intermediate frequency (IF) signal. The sampled signal s(i,k,m) of the associated receiving channel m passes through time t∈[-T] ch / 2,Tch The I sample values ​​with exponents i = 0, ..., I-1 are obtained by forming I sample values ​​on [ / 2].

[0058] The transmission time Δt is calculated for an object with a radial relative velocity v:

[0059] Δt=2(r c (k)+vt) / c; (7)

[0060] Where, r c (k) is the distance of the object at the center of the frequency ramp k:

[0061] r c (k)=r+v·T c (k), (8)

[0062] Where r is the average distance across all frequency ramps, and T c (k) is the absolute time of the center of frequency ramp k (the absolute time of the centers of all frequency ramps is defined as 0). Note that the relative velocity is assumed to be constant here, because the entire sequence of K frequency ramps lasts only a very short time, typically ≤20 milliseconds.

[0063] After transformation and omitting negligible terms, the phase of the intermediate frequency signal is obtained from equations (5) to (7) as follows:

[0064]

[0065] The average intermediate frequency signal phase (i.e., at t=0) is derived on the frequency ramp k as follows:

[0066]

[0067] From equation (9), the frequency of the intermediate frequency signal, i.e., the intermediate frequency itself, can be derived:

[0068] f IF (t,k)=|B ch | / T ch ·2(r c (k)+v·t) / c+s ch ·(F c (k)+B ch / T ch ·t)·2v / c。 (11)

[0069] For the average intermediate frequency f of frequency ramp k IF (k) (i.e., at t=0) yields:

[0070] f IF (k)=|B ch | / T ch·2r c (k) / c+s ch ·F c (k)·2v / c; (12)

[0071] The first component reflects the distance-dependent effect of linear frequency modulation, and the second component represents the Doppler effect, i.e., the frequency shift due to relative motion. Here, the second component is usually significantly smaller than the distance-dependent component. In the average across all frequency ramps, the following is derived with the average distance r (see Equation (8)) and the average center frequency F. cc (See the mid-frequency f of equation (4)) IF :

[0072] f IF =|B ch | / T ch ·2r / c+s ch ·F cc ·2v / c. (13)

[0073] If a one-dimensional discrete Fourier transform is formed on the sampled signal s(i,k,m) of the frequency ramp k and the receiving channel m, then at the distance gate j(k) = f IF (k)*T ch The peak power is obtained from equation (12):

[0074] j(k)=|B ch |·2r c (k) / c+s ch ·F c (k)·T ch ·2v / c, (14)

[0075] And from the average of all distance gates, we get from (13):

[0076] j = |B ch |·2r / c+s ch ·F cc ·T ch ·2v / c, (15)

[0077] This typically represents j(k) or a non-integer value of j, meaning the actual maximum value of the power peak lies between two integer distance gates considered in the Discrete Fourier Transform (DFT), and its non-integer position can be determined by interpolation. After the two-dimensional Discrete Fourier Transform (DFT), the power peak lies at the average distance gate j according to equation (15). The variation of the distance gate j(k) on the frequency ramp k according to equation (14) is mainly due to the distance r that varies slightly in the relative velocity. c(k) causes, however, a very small change, because the distance change is small over a short period of time (typically within ≤20 milliseconds) of a total of K frequency ramps, which, after the two-dimensional discrete Fourier transform (DFT), only results in a slight widening of the power peak in the distance gate dimension. The first component in the distance gate j according to equation (15) is generated by the distance r of the object, and the second component is generated by its relative velocity v; the second component is typically much smaller than the first component, thus the distance gate is primarily determined by the distance.

[0078] From the average intermediate frequency signal phase according to equation (10) Using the average distance r according to equation (8) c (k) yields:

[0079]

[0080] The first component in the frequency slope k changes linearly (because the center frequency F) c (k) Linear variation). For the constant spacing of the frequency ramps studied at the beginning of this paper, that is, the time T of the linear variation of the center of the frequency ramp. c In the case of (k), the second component is nonlinear with respect to relative velocity v≠0 because the corresponding linear term T c (k) and F c (k) appears in the product. This nonlinear characteristic, after performing a second one-dimensional discrete Fourier transform (DFT) on the frequency ramp dimension k, does not yield a sharp power peak in the resulting Doppler gate dimension l; originating from T c (k)·F c (k)·s ch The higher the nonlinear component of 2V / C, and consequently the higher the relative speed, the more ambiguous the power peak becomes (e.g., according to...). Figure 5 As can be seen in the examples.

[0081] To avoid ambiguity related to relative velocity, according to the second component of equation (16) It must also be linear in k, that is:

[0082] s ch ·T c (k)·F c (k)·2v / c=(k-(K-1) / 2)·constant

[0083] By according to T c (k) Solve the equation and replace the average ramp frequency F according to equation (4). c (k), ignoring very small terms, yields:

[0084] T c (k)=(k-(K-1) / 2)·T Dc / (1+(k-(K-1) / 2) / K·B s / F cc (17)

[0085] Among them, T Dc = constant · s ch / (2v / c·F cc );

[0086] As can be seen from equation (17), the parameter T Dc It is the average spacing of the frequency ramps (i.e., the average sampling time used to obtain the Doppler gate dimension through the second discrete one-dimensional Fourier transform, which should symbolize T). Dc The "D" in the index). Due to the modulation bandwidth B on the sequence of frequency ramps. s Typically, it is higher than the average transmission frequency F cc Much smaller, therefore the denominator of equation (17) is of the form (1+x), where |x| < 1, thus, for example, up to the second-order series expansion 1 / (1+x) = 1 - x + x 2 -+... can be used as an excellent approximation:

[0087] T c (k)=(k-(K-1) / 2)·T Dc ·(1-(k-(K-1) / 2) / K·B s / F cc )+(k-(K-1) / 2) / K·B s / F cc ) 2 (18)

[0088] The time interval T between two adjacent frequency ramps D (k)=T c (k)-T c (k-1) Using equation (18) and omitting negligible terms, we arrive at:

[0089] T D (k)=T Dc ·(1-2((kK / 2) / K·B s / F cc )+3((kK / 2) / K·B s / F cc ) 2 (19)

[0090] Because the third component contains a very small ratio B in quadratic form. s / Fcc And therefore usually more so than in B s / F cc The second component of the linear equation is much smaller, so it can also be omitted:

[0091] T D (k)=T Dc ·(1-2(kK / 2) / K·B s / F cc (20)

[0092] Therefore, the time interval of the frequency ramps varies at least approximately linearly on the frequency ramp k. According to equation (20), the frequency ramp spacing T... D (k)=T c (k)-T c The relative changes of (k-1) on the frequency ramps k=1, ..., K-1 are:

[0093] (T D (k)-T Dc ) / T Dc = -2(kK / 2) / K·B s / F cc . (twenty one)

[0094] According to equation (4), its center frequency F, which varies linearly on the frequency ramps k = 0, ..., K-1, is obtained. c The relative change of (k):

[0095] (F c (k)-F cc ) / F cc = +(kK / 2 + 1 / 2) / K·B s / F cc . (twenty two)

[0096] As can be seen from the two equations above, the slope of the linear relative change in the center frequency of the frequency ramp is = +B s / F cc And the slope of its time interval = -2B s / F cc That is, the relative change in the time interval is twice the relative change in the frequency position of the frequency ramp, and the signs of these changes are opposite. It should be noted that, when the time interval is precisely determined, for example according to equation (19), the correlation of the relative changes is not entirely accurate, but only approximately given. For the modulation bandwidth B considered above... s =600MHz, average frequency F ccFor the example of 76.5 GHz, looking at the entire sequence of K frequency ramps, the relative change in frequency position is approximately 0.78%, and the relative change in time interval is -1.56%. It should also be mentioned that when the ramp spacing is designed according to WO 2018 / 086783 A1, the relative changes in frequency ramp time interval and frequency position are inversely related and equal in absolute value, meaning they do not differ by a factor of two.

[0097] The choice of time interval for the frequency ramp (i.e., T according to equation (20)) D (k) is generated after the two-dimensional discrete Fourier transform (DFT). Figure 7 The absolute value spectrum shown is |S(j,l,m) / A(m)|. (This is in contrast to the spectrum shown below.) Figure 6 Unlike the spectrum, the power peak of the third object with a high relative velocity [r3 = 45 m, v3 = 60.4 m / s] is now also sharp, indicating that dissipation has been blocked, which also has an effect at a level approximately 2 dB higher. The two objects with the same relative velocity and only slightly different distances [r1 = 29.5 m, v1 = 1.09 m / s] and [r2 = 30 m, v2 = 1.09 m / s] remain unchanged and separated.

[0098] Now, it is also necessary to determine the location of the power peak of the object in the Doppler gate dimension. For this purpose, the time T of the frequency ramp center, determined above according to equation (17), is... c (k) is inserted into the phase of the intermediate frequency signal according to equation (16). In the middle; using the center frequency F c In the case of equation (4) of (k), omitting the unrelated constant phase components, we obtain:

[0099]

[0100] Wherein, the duration T of the entire frequency ramp sequence s for:

[0101] T s = K·T Dc ; (twenty four)

[0102] It must be emphasized again that, as required and through the appropriate selection of T c As achieved by (k), this represents a linear phase change process on k.

[0103] If we now form a second one-dimensional discrete Fourier transform along the frequency ramp dimension k, then at the Doppler gate...

[0104] The power peak is generated at that point, which can be expressed by equation (23):

[0105] l = s ch ·(B s ·2r / c + F cc ·T s ·2v / c); (25)

[0106] Here, the first component originates from the object's distance r, and the second component originates from its relative velocity v. Unlike the distance gate j, which is controlled solely by the object's size, i.e., its distance, according to equation (15), the Doppler gate considers the relative velocity and distance with a degree of similarity.

[0107] As can be seen by comparing equations (25) and (15) used to generate the Doppler gate l and the range gate j, the influence of distance in the Doppler gate dimension is higher than that in the range gate dimension. s / |B ch This results in a corresponding improvement in distance separation performance.

[0108] The equations (15) and (25) used for the distance gate and the Doppler gate will now be rewritten in such a way that the distance and relative velocity are related to their gate lengths:

[0109] j = r / R Lch +s ch ·v / D Lch (26)

[0110] l = s ch ·r / R Ls +s ch ·v / D Ls (27)

[0111] The distance and Doppler gate length are:

[0112] R Lch =c / (2|B ch |), R Ls =c / (2B) s ), D Lch =c / (2F) cc T ch ), D Ls =c / (2F) cc T s (28)

[0113] In sensor applications, the distance and relative velocity of an object are unknown, but the underlying task is to determine this distance and relative velocity from the location of the power peak after a two-dimensional discrete Fourier transform (DFT). Therefore, two equations (26) and (27) are solved based on the distance r and velocity v; thus, we obtain:

[0114] r = R L ·(jl·T ch / T s (29)

[0115] v=D L ·s ch (lj·B s / B ch (30)

[0116] The revised gate length is

[0117] R L =R Lch / (1-B s / B ch ·T ch / T s ), D L =D Ls / (1-B s / B ch ·T ch / T s (31)

[0118] The distance gate j and Doppler gate l of the object are generally non-integer and can be determined by interpolation from the shape of the power peak in the two-dimensional discrete Fourier transform (DFT), which only provides values ​​at integer gates.

[0119] Furthermore, it must be considered that the Doppler gate l can typically lie within a numerical range larger than the defined range L = K of the Discrete Fourier Transform (DFT); therefore, the Doppler gate can only be determined from the DFT to unknown integer multiples of K. A method to resolve the ambiguity is similar to the method proposed in DE 10 2009 016 480 A1, where the average frequency ramp spacing T is varied with the radar period. Dc That is, in the sequence of K frequency ramps transmitted in the current radar cycle, a different T is used than in the previous sequence. Dc Value. By changing D in equation (27) Ls With the relative velocities roughly the same, the Doppler gate l is produced in the current radar cycle with a different value than in the previous radar cycle, which can resolve ambiguity (the relative velocity can change only slightly with the radar cycle in about 50 milliseconds).

[0120] Based on equation (30) used to determine the relative velocity of the object, the frequency position (in B) of the linearly changing frequency ramp is considered in the following manner. s The effect caused by (≠0 as a characteristic) is that the Doppler gate l of the generated power peak is subtracted from the component jB proportional to its distance from the gate. s / B ch Furthermore, according to equation (31), B s ≠0 will still slightly affect the Doppler gate width D L .

[0121] Alternatively, the effect caused by the linearly varying frequency position can be considered as follows: after performing a one-dimensional discrete Fourier transform on the I received values ​​for each frequency ramp k = 0, ..., K-1, subtract 2π·j·B from each value. s / B ch • k / K corrects the phase of the value of the distance gate dimension j for all j and k (i.e., regardless of whether there is an object there, which is not yet known at that point in time); the correction can be achieved by multiplying the phase by a complex pointer of length 1 and the corresponding phase.

[0122] As described above, in order to determine the distance gate and Doppler gate of an object, the accurate location of the power peak is obtained by interpolation; especially since the power peak has a level not only in one gate but also in at least one adjacent gate due to the signal window used in the Discrete Fourier Transform (DFT), the actual location, which is generally non-integer, can be determined from the shape of the power peak, for example, by parabolic interpolation or by using the known shape of the power peak (derived from the Discrete Fourier Transform of the window function itself). However, this interpolation is not arbitrarily accurate; for example, interpolation errors may occur due to superimposed noise (especially in the case of poor signal-to-noise ratio) or due to extended, i.e., non-point-like objects. This leads to inaccuracies in determining the distance and relative velocity of the object according to equations (29) and (30); for the relative velocity according to equation (30), it is particularly important that the distance gate there is factored by B. s / B ch The factor is considered (in the example above, it is 4). In equation (29) used to determine the distance, almost only the distance gate is considered (the Doppler gate has only a very small weight T). ch / T s And therefore almost only the interpolation error (derived from the distance gate) is considered; however, this error is due to the large gate width R. Lch =c / (2|B ch |) is taken into consideration, i.e., not with the typically significantly smaller door width R Ls =c / (2B) sThis is taken into account; that is, for the accuracy of distance determination, there is no factor from the large modulation width B. s And therefore, it does not benefit from changes in frequency position on the frequency ramp (which so far has only improved distance separation performance for objects with the same relative velocity). Therefore, the inaccuracy, for both distance and relative velocity, stems primarily from errors in the distance gate.

[0123] However, the inaccuracies in determining the distance and relative velocity for the distance gate error can now be avoided by not always targeting the modulation bandwidth B on the frequency ramp sequence. s Instead of using the same sign, the absolute value is changed over multiple radar cycles while remaining constant; that is, for example, alternating +B. s and -B s Thus, every radar cycle, the frequency position linearly increases on the frequency ramp, and linearly decreases in other radar cycles. Consequently, in the Doppler gate l according to equation (27), the sign of the range component changes; if now from two having B s The sum of the Doppler gates of the object acquired by the radar period with different symbols roughly eliminates the range component and yields the relative velocity; conversely, the difference in the Doppler gates yields the opposite. In fact, it must also be considered that, on the one hand, the range will vary slightly with the radar period when the relative velocity v≠0; on the other hand, the average spacing T of the frequency ramp... Dc The range varies over multiple radar cycles. After some intermediate calculation steps and simplifications, the average range r over two cycles is obtained. m and relative velocity v m :

[0124] v m =s ch ·D Ls+- ·(l + +l - ) / 2 / (1-D Ls+- ·t +- / (2|R Ls |)) (32)

[0125] r m =|R Ls |·(s ch ·(l + -l - ) / 2-v m / 2·(1 / D Ls+ -1 / D Ls- (33)

[0126] Among them, l +It is the Doppler gate in the first radar cycle, with a positive modulation bandwidth of +B. s , and l - It is time t +- The Doppler gate in the next radar cycle has a negative modulation bandwidth of -B s "Average" Doppler gate width D Ls+- Consisting of two radar cycles (with different average frequency ramp spacing), and if necessary, different Doppler gate widths D. Ls+ and D Ls- From this, we can conclude that:

[0127] D Ls+- =2 / (1 / D) Ls+ +1 / D Ls- (34)

[0128] Therefore, to determine the range and relative velocity of an object, only the Doppler gate of the object is needed from two radar cycles, instead of the range gate that could lead to significant errors in previous methods. Large modulation bandwidth B s The width of the small door R Ls It is now also related to distance determination, i.e., interpolation error is considered accordingly smaller.

[0129] Accurate distance measurement is important, especially at close range, for functions such as avoiding collisions with obstacles (like guardrails) or other vehicles located to the side of the vehicle. Here, the distance is often less than the large door width R. Lch =c / (2|B ch |), that is, located within the first distance gate (due to bumper reflections and / or the superposition of negative frequency components), where interpolation functionality is generally particularly poor. This is achieved by using the method described above only from two points with opposite B... s The distance is determined by the periodic Doppler gate, so even close distances can be accurately determined.

[0130] In the above implementation scheme, the modulation bandwidth B s The sign of B has been changed over two radar cycles, while the absolute value remains constant. However, it is sufficient in principle to change B over two radar cycles. s The values ​​and / or the slope of the linear frequency position change are determined so that the effect of the distance gate can be eliminated. Then, a weighting factor is also introduced into the sum and difference of the required Doppler gates, i.e., the weights of the Doppler gate values ​​generated in the two periods are different.

[0131] To ensure the radar system is stable against interference from other radar systems, it is preferable, and particularly similar to, the method described in documents WO 2008 / 040341 ​​A1, DE 10 2009 016 480 A1, and EP 2 629 113 B1, to modify the modulation parameters, for example:

[0132] - The average distance of the frequency ramp with the period (as mentioned above, this also allows for a simple solution to velocity ambiguity);

[0133] - Modulation bandwidth B with period s and / or B ch (absolute value and / or sign);

[0134] -By superimposing additional meanless components, typically ranging from a few microseconds to random or pseudo-random components varying on k, the time interval T of the frequency ramp according to equations (19) and (20) D (k); For a relatively moving object, the received phase has a component that changes slightly on the frequency ramp, but it is still very small, so that the resulting effect (reduction in noise and power peak levels) after the Discrete Fourier Transform (DFT) is negligible.

[0135] - By superimposing additional random or pseudo-random meanless components varying on k, the frequency position F of the frequency ramp according to equation (4) is obtained. c (k) (i.e. its center frequency); this change in frequency position can also be achieved by always using the same frequency ramp, but changing the time point at which the received signal sampling values ​​are first obtained; the resulting phase change of the received signal, which is proportional to the distance gate, can be compensated by the corresponding general phase correction after the first one-dimensional discrete Fourier transform (DFT).

[0136] - The phase position of each transmitted signal by an additional phase modulator in the transmitting device, wherein the phase position varies randomly or pseudo-randomly on the frequency ramp, is compensated again on the receiving side, preferably in a digital signal processing device.

[0137] According to Figure 1In the radar system under consideration, there are M = 4 receiving antennas and associated receiving channels m = 0, ..., M-1. After a two-dimensional Discrete Fourier Transform (DFT), digital beamforming is preferably calculated again in each range-Doppler gate (j,l), for example, in the form of a Discrete Fourier Transform (DFT) or a Fast Fourier Transform (FFT); thus, a three-dimensional Fourier Transform is performed. The power peak is then determined in the three-dimensional spectrum. The azimuth of the object is derived from the position of its power peak in the third dimension, which is generated by the dimension m of the receiving channel; according to the above relationship, the range and relative velocity are derived from the other two dimensions. To provide more channels for angle forming, it is preferable to use not only multiple receiving antennas but also multiple transmitting antennas, and to evaluate the signals of all combinations of transmitting and receiving antennas to achieve multiple virtual receiving channels. If all or some of the transmitting and / or receiving antennas are not operating simultaneously, then multiple frequency ramp sequences of the above type, preferably similar, are nested together.

[0138] In summary, the method presented here as an example allows for high-precision and high-separation-performance range measurements by using high modulation bandwidth. This approach does not degrade measurement and detection quality when dealing with relatively moving objects, nor does it require the high computational power of a digital signal processing unit (as is required in conventional methods utilizing high modulation bandwidth). The need for only moderate computational power stems from the fact that the Discrete Fourier Transform can be used as the Fast Fourier Transform (FFT) in its rapid implementation for computation, and also from the fact that the multidimensional Fast Fourier Transform (FFT) has fewer dimensions than conventional methods with high range resolution and high measurement accuracy, since the range measurement is partially shifted to the dimension where relative velocity measurement is also performed. This fully leverages the fact that in radar systems for motor vehicles used for ambient detection, high range separation performance is primarily required for targets with the same radial relative velocity. Corresponding examples of radar systems used for ambient detection of a vehicle include measuring the length and width of traffic jams ahead, stopped vehicles under bridges or beside guardrails, the stationary surrounding environment of the road (guardrails, trees, buildings, etc.), and other vehicles (which typically have numerous reflective points). Therefore, good distance separation performance is also important, because the angle separation performance of radar systems is relatively poor due to the typically large beamwidth (due to the limited size). This may, for example, cause reflections from the left and right guardrails to not be separated and merged, resulting in the measured angle being in its own lane and thus mistakenly identifying a stationary obstacle (e.g., a stopped vehicle).

[0139] It is worth noting that for scenarios involving multiple targets with slightly different relative speeds and distances, this method only partially demonstrates its advantages, because, precisely, the total number of detection gates, i.e., range-Doppler gates, does not increase by increasing the modulation width on the linear change of frequency position on the frequency ramp. However, these scenarios are generally rarely associated with the aforementioned driver assistance functions.

[0140] It should be noted that it will be apparent to those skilled in the art that the ideas and embodiments of the invention illustrated by the above examples can be transferred to general measurement and parameter design, i.e., they can also be used for other numerical values. Therefore, general parameters are also given in the equations and figures.

[0141] Even when the design according to the invention, which uses the time interval between two adjacent frequency ramps according to equation (20), is not used, i.e., a constant interval is used instead, other exemplary inventive designs can still be used.

Claims

1. A method for a radar system for detecting a surrounding, the radar system being provided with a transmitting device for radiating a transmission signal comprising a sequence of individual signals which are at least approximately identical, wherein, The sequence of individual transmit signals is cyclically repeated, wherein the frequency of the individual signals is linearly modulated and the slope of the frequency modulation is at least approximately the same for all individual signals, wherein the individual transmit signals represent frequency ramps, wherein on the sequence of individual signals the frequency position of the individual signals changes at least approximately linearly and wherein here the slope of the linear change of the frequency position over the individual transmit signals changes at least sometimes with the sequence, wherein during the following K frequency ramps numbered with k = 0,..., K-1 the following I digital reception values numbered with i = 0,..., I-1 are obtained respectively and wherein a two-dimensional discrete Fourier transform is carried out on the I-K reception values respectively, wherein the dimension produced after the transform by the reception value index dimension i is called the distance gate dimension j = 0,..., J-1 and the dimension produced by the frequency ramp dimension is called the Doppler gate dimension l = 0,..., L-1, wherein two cycles with slopes of opposite slope, i.e. with a difference factor of -1, are employed for the precise radial distance measurement and / or relative velocity measurement of the object, wherein the sum and the difference of the positions produced in the two cycles within the two-dimensional discrete Fourier transform of the power peak of the object are used only in the Doppler gate dimension and not in the distance gate dimension.

2. The method of claim 1, wherein, On the sequence of individual signals the frequency position of the individual signals and its time interval change at least approximately linearly respectively, apart from varying and at least approximately mean-free components, wherein the relative change amount of the time interval is at least approximately twice the relative change amount of the frequency position and the signs of the changes are opposite.

3. The method of claim 2, wherein, The frequency position of the individual signals is characterized by a center frequency.

4. The method of any one of claims 1 to 3, wherein, A random or pseudo-random component is superimposed on the frequency position, the time interval and / or the phase position of the individual signals.

5. The method of any one of claims 1 to 3, wherein, The linear change of the frequency position of the frequency ramps is taken into account in such a way that, in order to determine the radial relative velocity of the object, the position of the power peak of the object after the two-dimensional discrete Fourier transform is corrected in the Doppler gate dimension l by a component which is linearly dependent on the distance gate dimension j, wherein the linear factor is derived from the quotient of the change of the frequency position over the frequency ramp and the change of the frequency over the reception time range during the individual frequency ramp, and the position of the power peak is determined by interpolation, whereby non-integer values are produced for the distance gate dimension j and / or the Doppler gate dimension l.

6. The method of any one of claims 1 to 3, wherein, The linear change of the frequency position of the individual frequency ramps is taken into account in such a way that, after the one-dimensional discrete Fourier transform of the I reception values of each frequency ramp k = 0,..., K-1, the phase of the values produced in the distance gate dimension j is corrected respectively by a phase component proportional to the product 2π·j·k / K, wherein the proportional factor is derived from the quotient of the change of the frequency position over the frequency ramp and the change of the frequency over the reception time range during the individual frequency ramp.

7. The method of claim 6, wherein, The correction is achieved by multiplication by a complex vector of length 1 with the respective phase.

8. The method of any one of claims 1 to 3, wherein, The sequence of K individual frequency ramps is cyclically repeated and here the slope of the frequency ramps themselves changes at least sometimes with the sequence.

9. The method of any one of claims 1 to 3, wherein, The sequence of K individual transmit signals is cyclically repeated, and the average time interval of the transmit signals is at least sometimes varied with the sequence.

10. The method of any one of claims 1 to 3, wherein, A plurality of receive channels is realized by a plurality of transmit antennas and / or receive antennas, and in addition to the two-dimensional discrete Fourier transform by the I K receive values respectively, there is also a digital beamforming on the receive channels or for generating the receive channels.

11. A radar system for detecting a surrounding environment, the radar system having a transmitting device for radiating a transmitting signal comprising a sequence of at least approximately identical individual signals, wherein, The radar system operates with the method according to any one of the preceding claims 1 to 10.

Images