Methods, systems and computer-readable media for phase-conjugated radar target emulation for testing double-sideband (DSB) frequency modulation continuous wave (FMCW) radar systems

Abstract

A phase-conjugated target emulation method for testing a DSB-FMCW radar system comprises selecting a representative transponder from an array of frequency-shift transponders to transmit an emulated target radar signal to a radar DUT. The representative transponder is located at a mirror image location with respect to a desired DoA of a signal from an emulated radar target. The method further comprises receiving a radar signal from the radar DUT and generating an output signal to be transmitted to the radar DUT by the representative transponder by reversing the roles of the normal sideband component and the ghost sideband component of the output signal, such that the ghost sideband component appears to the radar DUT as a desired emulator and the normal sideband component appears to the radar DUT as an undesired ghost sideband component. The method further comprises transmitting the output signal to the radar DUT.

Description

Description

[0001] The subject matter described herein relates to radar target emulation. More specifically, the subject matter described herein relates to a phase-conjugate radar target emulation for testing DSB-FMCW radars.

[0002] FMCW (frequency modulated continuous wave) radars transmit signals called chirps, which are reflected by radar targets and detected by the FMCW radar receivers. A chirp signal is a signal whose frequency increases or decreases linearly with time. The reflected signal detected by the target is a chirp signal with a linearly increasing or decreasing frequency, but delayed in time from the transmitted signal. The FMCW radar receiver mixes the received signal with the transmitted signal to generate an intermediate frequency signal whose instantaneous frequency is equal to the difference between the instantaneous frequencies of the transmitted and received chirp signals. The phase of the intermediate frequency signal is equal to the phase difference between the transmitted and received chirp signals.For a single object in front of the radar, the intermediate frequency signal has a constant frequency, which can be used to determine the distance or pitch of the object in front of the radar. The frequency or pitch of the intermediate frequency signal is related to the distance or pitch of the object in front of the radar by the following equation: Fτ=S2d / c where F τ is the intermediate frequency detected at the radar receiver, S is the chirp slope, d is the range or pitch of the target in front of the radar, and c is the speed of light in vacuum. Thus, using Equation 1, the range of the target in front of the radar can be estimated or determined from the tone of the intermediate frequency signal and the chirp slope.

[0003] FMCW radars are used in applications such as automotive and other radars to determine the distances and directions of radar targets. Systems that use FMCW radars include assisted driving, autonomous driving, and safety features such as automatic braking. One possible way to test FMCW radars in automotive applications would be to mount the radars on vehicles and measure the radars' accuracy in detecting real radar targets while the vehicle and targets are moving. However, such testing is impractical and even dangerous, given the variety of targets and distances encountered by a vehicle radar, and the fact that testing vehicle radars against real targets could result in human injury and / or property damage.

[0004] Given the impracticality of adequately testing vehicle radars against real targets, radar target and scene emulation or mimicking systems have been developed. A radar target and scene emulator designed to test FMCW radars comprises an array of frequency-shifting transponders. The array is placed in front of an FMCW radar under test and typically includes radar-absorbing materials to prevent reflections. The transponders receive radar signals from the radar DUT (device under test) and transmit radar signals back to the radar DUT to mimic a radar at a desired distance and direction from the radar DUT. Mimicking a radar target in a specific direction means that a frequency-shifting transponder is selected to receive the radar signal from the radar DUT and transmit the signal back to the DUT.To mimic the desired distance, the received radar signal is frequency modulated to cause a shift in the intermediate frequency detected by the radar DUT.

[0005] As described in more detail below, a problem with radar scene emulators that test radars with double-sideband receivers is the presence of an unwanted ghost image at the n=1 sideband of the detected signal, where the ghost image appears closer to the radar than the desired image (referred to herein as the desired emulant), which is located at the n=1 sideband of the received signal. The ghost image cannot be completely eliminated by in-phase and quadrature compensation during the frequency shifting process performed on the transponder. Consequently, the unwanted signal could cause the radar to detect the target at a range closer to the radar than the desired emulant. In automotive radar applications, such asautomatic braking systems, detecting the target at a distance closer than the actual target is undesirable, as this may cause the brakes to operate excessively or prematurely.

[0006] Given these and other difficulties, there is a need for improved methods, systems, and computer-readable media for radar target and scene emulation for testing DSB-FMCW radars.

[0007] This object is achieved by a method according to claim 1 and a system according to claim 10.

[0008] A method for phase-conjugate target emulation for testing a double-sideband (DSB) frequency modulation continuous wave (FMCW) radar device comprises providing a radar scene emulator having an array of frequency-shift transponders for emulating physical radar targets. The method further comprises selecting, as a proxy transponder for transmitting a radar signal of an emulated target to a radar DUT, one of the frequency-shift transponders from the array of frequency-shift transponders at a mirror image location with respect to a desired direction of arrival of a signal from an emulated radar target. The method further comprises receiving a radar signal from the radar DUT at the proxy transponder.The method further comprises generating an output signal to be transmitted by the proxy transponder to the radar DUT by swapping roles of a normal sideband component and a ghost sideband component of the output signal such that the ghost sideband component appears to the radar DUT as a desired emulant and the normal sideband component appears to the radar DUT as an undesired ghost sideband component. The method further comprises transmitting the output signal to the radar DUT.

[0009] According to another aspect of the subject matter described herein, providing the radar scene emulator with the array of frequency shift transponders comprises providing a two-dimensional array of frequency shift transponders, and selecting the proxy transponder at the mirror image location with respect to the desired direction of arrival comprises selecting, as the proxy transponder, one of the frequency shift transponders located at a mirror image azimuth and elevation location with respect to the desired direction of arrival.

[0010] According to another aspect of the subject matter described herein, swapping the roles of the normal and ghost sideband components comprises swapping modulation channels used to generate the output signal such that the modulation channels are opposite to those used by the radar DUT to demodulate the output signal.

[0011] According to another aspect of the subject matter described herein, generating the output signal comprises causing the radar DUT to detect an intermediate frequency as follows: fIF=2Sru¨bersch¨ssig,neu / c where f IF is the intermediate frequency, S is a chirp slope of the radar signal from the radar DUT, c is the speed of light in vacuum, and r überschüssig,neu is given by: excess,new=rEmulant+deingestellt where r überschüssig,neu is a distance to which the ghost component is shifted, r Emulant a removal of the desired emulant and d eingestellt is an effective distance from the proxy transponder to the radar DUT, including a distance that takes into account a processing latency of the proxy transponder.

[0012] According to another aspect of the subject matter described herein, the radar DUT comprises a time division multiplexed (TDM) multiple input multiple output (MIMO) radar device, the method further comprising applying phase corrections to output signals sent to the radar DUT to account for different distances from the agent transponder to different antennas of the TDM MIMO radar device.

[0013] According to another aspect of the subject matter described herein, applying the phase corrections comprises determining the phase corrections using the following equation: ϕk1=±(360° / λ)∗(dkE−dkP−d1E+d1P)mod 360° where ϕ k1 is a phase correction to be applied to a signal output from the representative transponder to a k-th transmit antenna of the radar DUT, λ is a transmit wavelength of the TDM-MIMO radar, d kPis the physical distance from the k-th transmitting antenna of the radar DUT to the representative transponder, d kE is a distance from the k-th transmitting antenna of the radar DUT to a desired emulation mirror image position on the radar scene emulator, d 1P is a distance from the k-th transmitting antenna to the representative transponder and d 1E is a distance from the k-th transmit antenna to the desired emulation mirror position and mod is the modulo function.

[0014] According to another aspect of the subject matter described herein, the radar DUT comprises a code division multiple access (CDM) multiple input multiple output (MIMO) radar, the method further comprising applying phase corrections to output signals sent to the CDM-MIMO radar to account for different distances from the agent transponder to different antennas of the CDM-MIMO radar.

[0015] According to another aspect of the subject matter described herein, applying the phase corrections comprises determining the phase corrections using the following equation: ϕ10=±[ang(AEχ1)−ang(APχ1)−ang(AEχ0)+ang(APχ0)] where ϕ 10 is the phase correction to be applied to an active code χ1, ang() is a function that returns an angle in degrees of a complex argument, A Pχ and A Eχ are given by: APχ=∑k=1NTxakPχ AEχ=∑k=1NTxakEχ where a kPχ is an incident complex amplitude from a k-th transmitting antenna of the radar DUT to a position of the representative transponder during a phase encoding χ and a kEχ is an incident complex amplitude from the k-th transmit antenna of the radar DUT at the position of the desired emulant during phase encoding χ.

[0016] According to another aspect of the subject matter described herein, the method for radar target emulation includes applying a gain correction to the output signal.

[0017] According to another aspect of the subject matter described herein, the method for radar target emulation includes calculating the gain correction using the following equation: Gj=dB(AEχi)−dB(APχj) where G j is the gain correction to be applied to the signal modulated using a j-th code χj , dB() is a function that outputs a resulting field strength in decibels, A Eχj is an amplitude of a signal at a position of the desired emulant and A Pχj is an amplitude of a signal detected at the agent transponder.

[0018] According to another aspect of the subject matter described herein, a system for phase-conjugate target emulation is provided for testing a double-sideband (DSB) frequency modulation continuous wave (FMCW) radar. The system includes a radar scene emulator having at least one processor, a memory, and an array of frequency-shift transponders for emulating physical radar targets. The system further includes a transponder selector, implemented by the at least one processor, for selecting, as a proxy transponder for transmitting an emulated target radar signal to a radar DUT, one of the frequency-shift transponders from the array of frequency-shift transponders at a mirror image location relative to a desired direction of arrival of a signal from an emulated radar target.The proxy transponder is configured to receive a radar signal from the radar DUT, generate an output signal to be transmitted to the radar DUT by swapping roles of a normal sideband component and a ghost sideband component of the output signal such that the ghost sideband component appears to the radar DUT as a desired emulant and the normal sideband component appears to the radar DUT as an undesired ghost sideband component, and transmit the output signal to the radar DUT.

[0019] According to another aspect of the subject matter described herein, the array of frequency shift transponders comprises providing a two-dimensional array of frequency shift transponders, and selecting the representative at the mirror image location with respect to the desired direction of arrival comprises selecting, as a representative transponder, one of the frequency shift transponders located at a mirror image azimuth and elevation location with respect to the desired direction of arrival.

[0020] According to another aspect of the subject matter described herein, the proxy transponder is configured to swap the roles of normal and ghost sideband components by swapping modulation channels used to generate the output signal such that the modulation channels are opposite to those used by the radar DUT to demodulate the output signal.

[0021] According to another aspect of the subject matter described herein, the agent transponder is configured to generate the output signal that causes the radar DUT to detect an intermediate frequency as follows: fIF=2Sru¨bersch¨ssig,neu / c where f IF is the intermediate frequency, S is a chirp slope of the radar signal from the radar DUT, c is the speed of light in vacuum, and r überschüssig,neu is given by: excess,new=rEmulant+deingestellt where r überschüssig,neu is a distance to which the ghost component is shifted, r Emulant a removal of the desired emulant and d eingestellt is an effective distance from the proxy transponder to the radar DUT, including a distance that takes into account a processing latency of the proxy transponder.

[0022] According to another aspect of the subject matter described herein, the radar DUT comprises a time-division multiplexed (TDM) multiple-input multiple-output (MIMO) radar device, wherein the proxy transponder is configured to apply phase corrections to output signals sent to the radar DUT to account for different distances from the proxy transponder to different antennas of the TDM MIMO radar device.

[0023] According to another aspect of the subject matter described herein, the agent transponder determines the phase corrections using the following equation: ϕk1=±(360° / λ)∗(dkE−dkP−d1E+d1P)mod 360° where ϕ k1 is a phase correction to be applied to a signal output from the agent transponder to a k-th transmit antenna of the radar DUT, λ is a transmit wavelength of the TDM-MIMO radar, d kPis a distance from the k-th transmitting antenna of the radar DUT to the representative transponder, d kE is a distance from the k-th transmitting antenna of the radar DUT to a desired emulation mirror image position on the radar scene emulator, d 1P is a distance from the k-th transmitting antenna to the representative transponder and d 1E is a distance from the k-th transmit antenna to the desired emulation mirror position and mod is the modulo function.

[0024] According to another aspect of the subject matter described herein, the radar DUT comprises a code division multiple access (CDM) multiple input multiple output (MIMO) radar, wherein the proxy transponder is configured to apply phase corrections to output signals transmitted to the CDM MIMO radar to account for different distances from the proxy transponder to different antennas of the CDM MIMO radar.

[0025] According to another aspect of the subject matter described herein, the agent transponder is configured to determine the phase corrections using the following equation: ϕ10=±[ang(AEχ1)−ang(APχ1)−ang(AEχ0)+ang(APχ0)] where ϕ 10 is the phase correction to be applied to an active code χ1, ang() is a function that returns an angle in degrees of a complex argument, A Pχ and A Eχ are given by: APχ=∑k=1NTxakPχ AEχ=∑k=1NTxakEχ where a kPχ is an incident complex amplitude from a k-th transmitting antenna of the radar DUT to a position of the representative transponder during a phase encoding χ and a kEχ is an incident complex amplitude from the k-th transmit antenna of the radar DUT at the position of the desired emulant during phase encoding χ.

[0026] According to another aspect of the subject matter described herein, the agent transponder is configured to apply a gain correction to the output signal.

[0027] According to another aspect of the subject matter described herein, the agent transponder is configured to calculate the gain correction using the following equation: Gj=dB(AEχi)−dB(APχj) where G j is the gain correction to be applied to the signal modulated using a j-th code χ j , dB() is a function that outputs a resulting field strength in decibels, A Eχj is an amplitude of a signal at a position of the desired emulant and A PXj is an amplitude of a signal detected at the agent transponder.

[0028] The subject matter described herein may be implemented in software, in combination with hardware and / or firmware, with a preference toward hardware or firmware given the processing speeds required for real-time target emulation for testing an FMCW radar device. For example, the subject matter described herein may be implemented in software executed by a processor. In one exemplary implementation, the subject matter described herein may be implemented using a non-transitory computer-readable medium having stored thereon computer-executable instructions that, when executed by the processor of a computer, direct the computer to perform steps.Exemplary computer-readable media suitable for implementing the subject matter described herein include non-transitory computer-readable media, such as disk storage devices, on-chip memory devices, programmable logic devices, and application-specific integrated circuits. Additionally, a computer-readable medium implementing the subject matter described herein could be located on a single device or computing platform or distributed across multiple devices or computing platforms.

[0029] Preferred embodiments of the present invention are explained in more detail below with reference to the accompanying drawings. They show: Fig. 1 is a plan view of an emulated radar scene illustrating the range and arrival direction of a desired emulator radar signal and a ghost radar signal resulting from imperfect frequency shift manipulation (or spoofing) by a frequency shift transponder of a radar scene emulator; Fig. 2 is a plan view of an emulated radar scene illustrating a range and direction of arrival of a desired emulator radar signal and a ghost radar signal, wherein the roles of the desired emulator and the ghost element are reversed, and the emulated radar signal is transmitted from a frequency shift transponder located at a mirror image location with respect to a desired direction of arrival; Fig. 3 a Mercator-style projection of proxy directions and desired incidence directions relative to the boresight of the radar DUT; Fig. 4 is a top view of an emulated radar scene where the radar DUT is a MIMO radar with three transmit antennas; Fig. 5 is a diagram illustrating a radar scene emulator and a radar DUT; Fig. 6 is a block diagram illustrating a radar scene emulator; and Fig. 7 is a flowchart illustrating an exemplary method for phase-conjugate target emulation for testing a DSB-FMCW radar.

[0030] Automotive radars of today and the near future are either 3D or 4D, with the first three reported dimensions being range, velocity, and azimuth, and the fourth dimension being elevation. The term "direction of arrival" (DoA) is often used to denote azimuth (and elevation for 4D radars). The vast majority of these radars are so-called fast FMCWs, where "fast" means that a series of similar (or somewhat similar) chirps are transmitted at repetition rates on the order of 1–100 kHz, and the Doppler or velocity information is gathered from the phase progression of the received chirps after reflection by the target(s).

[0031] Furthermore, almost all modern fast FMCW radars use MIMO (multiple-in, multiple-out) technology because it offers a larger effective aperture than MISO (multiple-in, single-out) and therefore higher DoA resolution. The two most popular versions of MIMO are TDM-MIMO and CDM-MIMO, where TDM stands for time-division multiplexing and CDM for code-division multiplexing.

[0032] As mentioned above, FMCW radar target emulation, and in particular radar scene emulation (RSE), can be implemented using frequency-shift (f-shift) transponders [Ref. 1, 2]. Due to the radar chirp, an intermediate frequency (IF) shift on the order of sub-kHz to several tens of MHz mimics a real-time delay. The exact IF frequency chosen is proportional to the chirp slope (usually specified in MHz / µs) and to the excess range. Here, "excess" means that the actual range is smaller than the set RSE distance d. eingestellt is emulated, where: deingestellt=dDT+cτspa¨t / 2 where d DT is the physical distance between the radar DUT and the transponder, c is the speed of light and τ spät is the transponder latency, ie τ spätis the time delay required to receive the radar signal, process it electronically, and then send it back. A typical value for d DT could be 75 cm and a typical cτ spät / 2 could be 5 cm, so a typical d eingestellt would be ~0.8 m.

[0033] If the radar uses a double-sideband (DSB) receiver, an undesirable ghost type called an "image," "phase-conjugate," or "n = -1" ghost element may appear in the signal output from the transponder. The desired emulator and the ghost signals are in Fig. 1. In particular, Fig. 1 is a plan view of an emulated radar scene illustrating a range and a direction of arrival of a desired emulator radar signal and a ghost radar signal resulting from imperfect frequency shift manipulation by a frequency shift transponder of a radar scene emulator. In Fig. 1, the desired emulant radar signal is the n=1 sideband signal and the ghost signal is the n=-1 sideband signal.

[0034] The term "n = -1" is used to refer to the ghost signal, as this ghost element arises electronically from an unintended opposite sideband shift stimulus in the transponder's output spectrum. The correct "n = 1" sideband creates a fictitious excess distance: excess,new=rEmulant+deingestellt

[0035] In the Fig. 1 example, the eingestellt equal to 0.8 m, and r Emulantis 40.8 m. Thus, from equation 3, the fictitious excess distance, r überschüssig resulting from the n=1 sideband signal output from the transponder is 0.8 m.

[0036] In contrast, the n = -1 stimulus produces an opposite excess = -r über-schüssig When added to the set distance, the result is: r*=2dset−rEmulant

[0037] Here, r* is the fictitious distance resulting from the n=-1 signal output from the transponder. Fig. 1, r* = 2(0.8)-40.8 = -39.2 m. For the vast majority of real targets, r Emulant > 2d eingestellt , so that r* is negative.

[0038] For an FMCW radar, a physically realistic positive range, when using up-chirps (down-chirps), appears as a downshift (upshift), while a physically unrealistic negative range appears as an upshift (downshift). If the radar uses a single-sideband (SSB) receiver, the receiver automatically filters out the negative range interference because the receiver is designed to accept only the correct shift polarity. However, a DSB radar receiver convolves the two shift cases, meaning that the negative range becomes a positive range again: RImage=−r*=rEmulant−2deset

[0039] In equation 5, r Bild the removal of the unwanted ghost image as detected by the double sideband receiver. In the example in Fig. 1 applies r Bild = 39.2 m, which means that the ghost image is perceived by the radar device as if it were a distance of 2 d eingestellt or 1.6 m closer to the radar than the desired emulant. The folding effect of a DSB radar also causes the ghost image to appear to have a direction of arrival (DoA) that originates from a mirror image direction of the desired emulant. This explains why the n = -1 ghost element is a DoA mirror image of the desired emulant, but appears slightly closer.

[0040] The ghost disappears with perfect I / Q (in-phase / quadrature) balancing during the frequency shifting process, but perfection is never achieved. A 30-40 dB ghost suppression relative to the desired emulator requires considerable effort.

[0041] For many automakers and automotive component manufacturers, especially those interested in testing AEB (Automatic Emergency Braking), even 30-40 dB of suppression is insufficient and unacceptable if the ghost element appears closer to the host vehicle than the emulator itself. The autonomous driving controller could initiate emergency braking when none is requested, resulting in poor fuel mileage, an unpleasant passenger experience, and so on.

[0042] On the other hand, if the mirror-like weak ghost element were slightly further away than the emulator, the same manufacturers (or at least some of them) would accept the situation.

[0043] The subject matter described herein includes a method for keeping the desired emulation target closer to the radar under test than its opposite-sideband ghost partner. It should be noted that an excellent I / Q balance is still sought to attenuate the ghost element. However, given the unavoidable result that the ghost signal is not infinitely suppressed, the method described herein achieves a weak ghost element that is farther away than the emulator.

[0044] Fig. Figure 2 is a diagram illustrating a top view of an emulated radar scene in which the roles of the ghost signal and the desired emulator signal are reversed by swapping the in-phase (I) and quadrature (Q) modulation channels to generate a transponder output signal, with the desired emulator closer to the radar under test than the ghost signal. Additionally, the transponder selected to output the emulated radar signal is located in a mirror image direction from the desired arrival direction, causing the newly selected desired emulator to appear to originate from the desired arrival direction.

[0045] Referring to Fig. 2, we first propose swapping the roles of the normal (n = 1) and phase-conjugate (n = -1) transponder output frequency sidebands. This immediately achieves the change of "closer" vs. "farther," but the emulator's DoA is still mirrored with respect to the desired situation. Therefore, second, we deliberately choose a transponder on the opposite side of the radar scene emulator [Ref. 2]. "Mirror-opposed" means opposite in both the horizontal and vertical directions—see Fig. 3.

[0046] In Fig. 2, the point labeled n=-1 is the location of the desired emulator, due to the role reversal of the normal and phase-conjugate sidebands of the emulated radar signal output from the transponder described above. The point labeled n=1 is the location of the ghost sideband. In Fig. 2 is d eingestellt 0.8 m, r Emulant is 40.8 m and rGeist is 42.4 m, which achieves the goal that the ghost signal is further away from the radar under test than the desired emulator.

[0047] The desired direction of incidence in Fig. 2 is 75°, so the desired emulator appears to originate from 75°, a transponder located at 105° is selected to be the starting point for the emulated radar signal. It should be noted that Fig. 2 shows only the horizontal or azimuth dimension of the emulated radar scene and that the selected transponder would also be in a mirror image vertical (elevation) direction from the desired direction of incidence. The selected transponder in Fig. 2 is called a representative transponder because the transponder acts as a representative for the desired direction of arrival.

[0048] Fig. Figure 3 is a Mercator-type plot of representative vs. desired directions relative to the radar boresight. Typically, the boresight points to the geometric center of the radar scene emulator, e.g., Ref. 2. Fig. 3, the square labeled "desired DoA 1" is located at a desired azimuth and elevation position. The square labeled "proxy transponder 1" is the location of the proxy transponder used to generate the emulated radar signal, so that the desired emulator appears to emanate from the direction of the transponder located at desired DoA 1. Proxy transponder 1 is located in the mirror image azimuth and elevation direction of the transponder located at desired DoA 1. Similarly, the hexagon labeled "desired DoA 2" is located at a different desired azimuth and elevation position. The hexagon labeled "Proxy Transponder 2" is the location of the proxies transponder from which the emulated radar signal is transmitted, so that the desired emulator appears to emanate from the desired DoA 2.The agent transponder 2 is located at a mirror image azimuth and elevation position of the desired DoA 2. It is noted that it is not required that a transponder be located at the desired DoA position, but only that a transponder be located at a mirror image position of the desired DoA.

[0049] For multiple-input, single-output (MISO) radars, the procedures described above are sufficient to achieve phase-conjugate target emulation. In practice, however, we swap the I and Q signals applied to the transponder and set the IF to a higher frequency (absolute value) than normal, specifically: fIF=2Sru¨bersch¨ssig,neu / c, where S is the chirp slope, and excess,new=rEmulant+dset,

[0050] It is noted that the sign of f IFis equal to the sign of the chirp slope, which is suitable for FMCW phase conjugation. For a normal target emulation (no phase conjugation), the sign of f IF opposite to the sign of S.

[0051] In addition, the change in sign of the d eingestellt -terms compared to equation (3). In this new formulation, the n = 1 sideband becomes the ghost element at a distance r Emulant + 2d eingestellt , while the n = -1 sideband to the new desired emulant is at a distance of r Emulant + d eingestellt Incidentally, for some implementations, the target radial velocity relative to the own vehicle must also be reversed in sign; however, some radar scene emulators, such as the Keysight AD1012A, automatically account for this using the SwapIQ software button and the subroutines activated by this button.

[0052] For multiple-input, multiple-output (MIMO) radars, there is an additional consideration: only the returning beams from the transponder to the radar are perceived as phase conjugate, while the outgoing beams from the radar to the transponder are not. Since there are, by definition, multiple transmit antennas in MIMO, there are multiple outgoing beams to each transponder, and the relative delays of these beams remain "normal" and not phase conjugate because the transponder electronics causally process them.

[0053] The result of this mixed phase-conjugate processing of incoming beams and normal processing of outgoing beams is an unintentional emulation analogy of the physical situation known in the world of automotive radar as "DoA not equal to DoD" (DoA ≠ DoD). In this case, the two directions mirror each other. When this situation occurs—and this is hardly a strong effect in the real world—the radar's DoA algorithm becomes confused. Most MIMO radars have more Rx (or receive) channels than Tx (or transmit) channels, and the latter are more densely spaced; thus, the perceived DoA is usually "in-place" because the Rx channels carry a greater determination weight than the Tx channels, but the angular accuracy falls outside the vendor's specifications. Sometimes the perceived DoA can be off by several tens of degrees.

[0054] A popular MIMO antenna geometry in many 4D radars is to use a nearly equal number of Rx and Tx channels, with the Rx antennas handling most of the azimuth sensing and the Tx antennas handling most of the elevation sensing. In this case, inconsistent processing of the outgoing versus returning beams often leads to gross DoA errors in both directions.

[0055] To correctly handle TDM-MIMO radars, our method includes a third ingredient, time slot phase correction. Specifically, since phase control of the applied I and Q IF signals is already present, we adjust the returning signal phases in a manner that depends on which Tx channel is the active chirper and which transponder in the radar scene emulator is used as the proxy direction.

[0056] Any of the radar's Tx antennas can be selected as the phase reference for the time slot phase correction. This is due to an already sufficient template for multiple-input single-output (MISO) radars—after all, the presence of the rest of the N Tx- -1 Tx antennas (N Tx = #Tx channels > 1), which makes the DUT a MIMO radar device. Without loss of generality, Tx1 should be selected as the reference channel.

[0057] For each time slot in which Tx1 is the active transmit channel, there is no phase correction at all. For each time slot in which Tx k is the active channel, where k > 1, the following phase correction is added to both the I and Q signals: ϕk1=±(360° / λ)∗(dkE−dkP−d1E+d1P)mod 360° where λ is the radar transmission wavelength, d kP the physical distance from Tx k to the representative or proxy transponder, d kEthe distance from Tx k to the desired emulation mirror image position on the radar scene emulator, d 1P is the distance from Tx1 to the representative and d 1E is the distance from Tx1 to the mirror image position. The "mod" function is the usual modulo wrapping function for limiting the phase to a 360° span. (One could, of course, also use unwrapped phases, but these consume more bits of the emulator system's digital memory.)

[0058] Fig. Figure 4 shows an example of a MIMO DUT with three transmit antennas. Fig. Figure 4 also shows the distances shown in equation 7 between each transmit antenna and the agent and desired DoA transponders. Fig. The distances shown in Figure 4 are not to scale, they only serve to illustrate the distances between the desired emulator position and the representative transponder position, and the transmitting antennas vary depending on which antenna is currently transmitting.

[0059] The ± sign on the right-hand side of equation (7) is specified as follows: For up-chirps, we use the + sign, while for down-chirps, we use the - sign. The reason for this is as follows: For a normal (no phase conjugation) emulation, a true returning signal from an up-chirp appears down-shifted, and thus a positive distance increment appears as a negative phase increment to the radar's associated receiver. For a phase-conjugated emulation, the opposite occurs; hence the + sign. For a down-chirp, everything is reversed.

[0060] By applying the phase corrections from Equation (7) in the appropriate time slots and at the appropriate transponder locations, we achieve a realization of the correct Tx-to-mirror direction phases relative to the reference. Thus, the processing of the outgoing (Tx to transponder) and returning (transponder to Rx) beams is now fully consistent for MIMO DoA signal processing.

[0061] The AD1012A has a “screen” of transponders symmetrically around the central axis; thus, the filled star in Fig. 4 another transponder. However, this is not generally necessary—that is, in principle, the filled star could be a virtual transponder position (in which case, not an actual transponder device).

[0062] Likewise, the proxy position could be an interpolated (virtual) position between two adjacent transponders in the AD1012A, in which case the filled star representing the desired emulant position in Fig. 4, would mirror interpolate between the two Mercator-mirrored transponder positions of the AD1012A. As with normal target interpolation (no phase conjugation) [Ref. 3], the DoA accuracy of our proposed method is not limited by transponder spacing. Of course, when applied to 4D radars, this applies to both azimuth and elevation.

[0063] Certain parameters necessary to implement the proposed method are known based on knowledge of the radar DUT. The wavelength is known. The transmit channel sequence and chirp slope are often publicly known, e.g., through Federal Communications Commission (FCC) or CTC reports, and are sometimes even provided upon request by the radar vendor. The slope may change over the course of the waveform, but this information is also either publicly available or available upon request. If not, slope information can be measured prior to a scenario deployment. Modern radars often have multiple operating modes that differ in their signal characteristics.All modes can be characterized before deployment; during deployment, detection of which mode is in use must be done in passing not with the chirp timescale, but rather with the much slower frame timescale.

[0064] The Fig. The distances shown in Figure 4 do not vary over time. They depend only on the DUT's Tx antenna geometry, the DUT's mounting within the radar scene emulator chamber, and the scene emulator transponder screen geometry. The first of these can be obtained from radar "trim" photos, in particular, a controlled photograph of the antenna printed circuit board (PCB). Careful mounting is an essential component of a good radar scene emulator system. The emulator screen geometry is certainly known. Calculating the distances in equation (7) is reduced to an application of the Pythagorean theorem. The transmit sequence can also be derived by selectively blocking Tx antennas, e.g., with anechoic foam, and observing which time slots have no signal.

[0065] We now cover the other commonly used MIMO type, called CDM-MIMO (CDM = Code Division Multiplexing). With CDM-MIMO, all radar Tx antennas are active simultaneously, but different chirp time slots have different phase encodings of the Tx channels. Walsh-Hadamard BPM (binary phase encoding) is the oldest CDM technique and is still in use. For example, if N Tx = 2, Tx1 and Tx2 would be in phase in one time slot and 180° out of phase in another time slot. A more recent variation of CDM-MIMO, called DDM-MIMO (DDM = Doppler Division Multiplexing), uses non-binary phasing, i.e., 4-phase or 6-phase, to handle N Tx > 2.

[0066] Our principle for handling CDM-MIMO is a generalization of our TDM-MIMO treatment. Again referring to Fig. 4 as an example, we calculate for each phase encoding χ the complex results (sum) of the N Tx (= 3 in Fig. 4) outgoing waves incident on the representative (P) and at the position of the filled star (E): APχ=∑k=1NTxakPχ AEχ=∑k=1NTxakEχ where a kPχ the impact complex amplitude of Tx k on the representative during a phase encoding χ and a kEχ the impact complex amplitude of Tx k at the position of the filled star during phase encoding χ. Following textbook electromagnetism, the a's can be calculated (in units of electric field V / m) from the Tx phasing, the Tx antenna gains, the total transmit power for the given radar operating mode, and the "1 / d" radiation rule for amplitude propagation. 1 / d kP for a kPχ and 1 / d kE for a kEχTypically, the Tx gains and power distribution in CDM-MIMO are all the same, so calculations in equations (8) and (9) mainly depend on the Tx phasing.

[0067] In analogy to the TDM case, where one of the Tx channels is selected as the reference, one of the DCM codes is now selected as the phase reference. Fig. 4 assumes that a ternary (3-phase) DDM MIMO is used. The three mutually orthogonal codes are 0°-0°-0°, 0°-120°-240°, and 0°-240°-120°, with the phases listed in the order applied to Tx1-Tx2-Tx3 (Note: In this example, the number of codes is N). Tx , but more generally the number of codes ≥ N Tx .) These codes are called χ0, χ1, and χ2, respectively. Without loss of generality, χ0 can be chosen as the phase reference code.

[0068] For each time slot where χ0 is the active code, no phase correction of the transponder I and Q signals is required. If χ1 is the active code, we add the following phase correction to I and Q: ϕ10=±[ang(AEχ1)−ang(APχ1)−ang(AEχ0)+ang(APχ0)] where ang is the angle in degrees of the complex argument. As in the TDM case, the ± sign is chosen to be + for upstream chirps and - for downstream chirps. Similarly, when χ2 is the active code, we add the following phase correction to I and Q: ϕ20=±[ang(AEχ2)−ang(APχ2)−ang(AEχ0)+ang(APχ0)]

[0069] This covers the phase correction portion of our CDM-MIMO treatment. Since the radar's transmit far-field radiation pattern is code-dependent in CDM-MIMO, we also need to perform amplitude correction. To accurately handle an RCS (radar cross-section), we cannot select an arbitrary code as the amplitude reference. Instead, we must perform a P-to-E correction for each of the CDM codes. Continue with Fig. 4 as an example, when a chirp time slot uses χj (j = 0, 1 or 2) as the phase code, a gain correction is applied to the proxy transponder: Gj=dB(AEχj)−dB(APχj) where the dB function measures the resulting field strength in decibels (relative to, for example, 1 V / m). Gain correction is implemented by adjusting both the I and Q driver strength and / or adjusting the transponder's output variable attenuation.

[0070] The combination of phase and gain / amplitude correction reproduced the CDM time-dependent superposition of the outgoing wave at a position E, although the transponder processing these waves is located at a position P.

[0071] It should be clear that a general CDM-MIMO, that is not only Fig. 4, is treated by considering all code-dependent phase corrections in analogy to equations (10) and (11) and all code-dependent amplitude corrections according to equation (12). Furthermore, the attentive reader will recognize that TDM-MIMO is only a special case of CDM-MIMO, whereby the "coding" is a Tx channel selection. Referring to Fig. 4 the “codes” are 1-0-0, 0-1-0 and 0-0-1 when selecting Tx1, Tx2 and Tx3 respectively.

[0072] As with TDM-MIMO, parameter knowledge for the CDM-MIMO case is obtained either from published information or pre-deployment measurement. If the code-time sequence is not public or provided by the vendor, one way to derive it from a measurement is to observe the signal strength versus time over a series of chirps at multiple angular positions relative to the DUT. The number of positions should be approximately the number of codes, or preferably slightly larger. The signal strength at each position increases or decreases depending on the code; as the code changes, the strong-signal and weak-signal positions also change. This "poor man's" code-dependent far-field information is used to derive the code-time sequence.

[0073] Fig. Figure 5 is a diagram illustrating a radar scene emulator comprising an array of frequency shift transponders and a radar DUT. Fig. In the example illustrated in Figure 5, the array includes 512 frequency-shifting transponders, where transponders in the same row can be activated to emulate radar targets at a desired elevation position relative to the radar DUT, and transponders in the same column can be activated to emulate radar targets at a desired azimuth position relative to the radar DUT. An example of a radar scene emulator suitable for implementing the subject matter described herein is the aforementioned AD1012A, available from Keysight Technologies, Inc.

[0074] Fig. 6 is a block diagram of a radar scene emulator having a plurality of frequency-shifting transponders suitable for emulating radar targets using phase conjugation, as described herein. Fig. Figure 6 illustrates the components of each frequency-shift transponder for achieving phase-conjugate target emulation. Note that the frequency-shift transponder includes other components, such as transmit and receive antennas and amplifiers for transmitting signals to and receiving signals from the radar DUT. Referring to Fig. 6, a radar scene emulator 100 includes an array of frequency-shift transponders 101. Each frequency-shift transponder 101 includes double-sideband (DSB) mixers 102 and 104. Mixer 102 mixes a 90° phase-shifted version of the incoming signal from the radar DUT with a modulation signal output from a digital-to-analog converter (DAC) 106. Mixer 104 mixes an in-phase version of the incoming radar signal from the radar DUT with a modulation signal output from a DAC 108. The modulation signal output from DACs 106 and 108 is generated by a modulation controller 112 based on a sequence of digital data input to a splitter 110.

[0075] A summer 114 sums the signals output from the mixers 102 and 104 to produce the frequency-shifted output signal that results in the intermediate frequency signal detected by the radar DUT so that the radar DUT perceives the desired emulant as being at the distance given by Equation 6.

[0076] Phase conjugation is achieved in transponder 101 by switching the interleaving of data delivered to splitter 110. The IF (the amount of either upshift or downshift) is generally achieved at <30 MHz. Automotive radar chirps are located in the 76-77 GHz or 77-81 GHz band, so this frequency shift is a very small fraction of the operating frequency. With such a low shift frequency, digital signals are simply sent to DACs 106 and 108, which convert the digital signals into analog IF signals, which appear on the left sides of DSB mixers 102 and 104 in the diagram.

[0077] During operation, the modulation controller 112 provides an interleaved stream of digital numbers that flow into the splitter 110. In the "normal" case, the modulation controller 112 may interleave the data provided to the splitter 110, with the even time slots representing cos(2pi*f_IF*t) and the odd time slots representing sin(2pi*f_IF*t). In the phase-conjugated case, even slots become sin(2pi*f_IF*t) and odd slots become cos(2pi*f_IF*t). What is considered "normal" depends on the sign of the chirp slope, as described above. This means that for an upchirp, the receiver in the non-emulated, non-phase-conjugated case would normally receive a downshifted version of the transmitted signal. For a down-chirp, the receiver would typically receive an up-shifted version of the transmitted signal.Whatever the "normal" assignment, phase conjugation simply reverses the interleaving, causing the transponder to upshift the frequency of the transmitted radar signal for upchirps and downshift the frequency of the transmitted radar signal for downchirps. The amount of frequency upshift or downshift in the signal output from transponder 101 is equal to F. IF , as defined in equation (6.2). The modulation controller 112 could also apply amplitude and phase correction for the MIMO cases, as described above.

[0078] The radar scene emulator 100 includes at least one processor 116 with an on-chip memory 118. A transponder selector 120 selects the desired proxy transponder to emulate a radar target from the desired direction of arrival. The transponder selector 120 may be implemented using computer-executable instructions stored in the memory 118 and executed by the processor 116.

[0079] Fig. Figure 7 is a flowchart illustrating an exemplary method for phase-conjugate target emulation for testing a DSB-FMCW radar. Referring to Fig. 7, the method includes, at step 700, providing a radar scene emulator having an array of frequency-shifting transponders for emulating physical radar targets. For example, a radar scene emulator may be provided, such as the one described in Fig. 5 and Fig. 6 is shown.

[0080] At step 702, the method includes selecting, as a proxy transponder for transmitting an emulated target radar signal to a radar DUT, one of the frequency-shift transponders from the array of frequency-shift transponders at a mirror image location relative to a desired arrival direction of a signal from an emulated radar target. For example, one of the frequency-shift transponders in the array that is located at a mirror image location of the desired arrival direction of the emulated radar signal may be selected as the proxy transponder. Examples of desired arrival directions and proxy transponder locations are described in the Fig. 2 and Fig. 3 shown.

[0081] At step 704, the method further includes receiving a radar signal from the radar DUT at the agent transponder. For example, the selected agent transponder may receive a chirped radar signal from the radar DUT.

[0082] At step 706, the method further includes generating an output signal to be transmitted by the proxy transponder to the radar DUT by swapping roles of a normal sideband component and a ghost sideband component of the output signal such that the ghost sideband component appears to the radar DUT as a desired emulator and the normal sideband component appears to the radar DUT as an undesired ghost sideband component. For example, the selected proxy transponder could swap the roles of the in-phase and quadrature components of the output signal by switching the I and Q modulation channels used to generate the output signal.If the radar DUT expects the in-phase modulated signal at even time slots and the quadrature modulated signal at odd time slots, the proxy transponder outputs a quadrature modulated radar signal at even time slots and an in-phase modulated signal at odd time slots. If the radar DUT expects the quadrature modulated signal at even time slots and the in-phase modulated signal at odd time slots, the proxy transponder outputs an in-phase modulated radar signal at even time slots and a quadrature modulated signal at odd time slots. Switching of the modulation channels can be performed by inverting the digital bit stream input to the proxy transponders to control the I and Q mixing, as described above with respect to . Fig. 6 was described.

[0083] At step 708, the method further includes transmitting the output signal to the radar DUT. For example, the proxy transponder transmits the signal to the radar DUT. The radar DUT receives the signal, mixes the signal with the transmitted signal, determines the intermediate frequency, and calculates the range of the emulated radar target using Equation 1. Because the modulation role swapping and frequency shifting (see Equation 6) occur, the desired emulator appears closer to the radar DUT than the unwanted ghost signal.

[0084] The subject matter described herein includes at least the following ideas: 1. FMCW radar scene emulator method that shifts a DSB radar clutter ghost, however strong or weak, to a distance greater than a desired emulator distance by phase-conjugate frequency shifting and selecting a frequency shift transponder in a direction mirroring a desired emulator direction. 2. Using the radar scene emulator, the noise ghost is made further away than the desired emulator by frequency shifting a returning signal to the radar so that a detected intermediate frequency f IF = 2Sr überschüssig,neu / c, where S is an FMCW chirp slope and r überschüssig,neu is given by equation (6.2). 3. The frequency shift radar scene emulator can also be extended for TDM MIMO radars to apply time slot phase corrections to the transponder I and Q frequency shift signals, where the time slots are categorized by an active radar Tx channel. 4. The phase corrections are calculated based on selecting an arbitrary reference Tx channel. 5. The phase correction = 0 whenever a reference channel is the active channel. 6. The phase correction is managed by equation (7) when the active channel is not the reference channel. 7. The sign of the phase correction in equation (7) is positive for up-chirps and negative for down-chirps. 8. A controlled photograph of the radar antenna PCB is used to calculate the distances (of outgoing rays) in equation (7). 9. An X-ray of the radar case is used when a PCB photograph is unavailable. (The antennas show up well due to the copper and gold in their metallization.) 10. A DUT transmit sequence is determined by selectively blocking DUT Tx antennas prior to a scenario deployment if the transmit sequence is not publicly known or provided by the vendor. 11. The radar scene emulator can perform phase-conjugate DoA interpolation if the desired direction of arrival lies between two of the frequency-shift transponders. 12. The radar scene emulator can be used to test CDM-MIMO radars, where time slot phase corrections are applied to the transponder I and Q signals based on phase differences of Tx complex results (see equations (10) and (11)) and time slot gain corrections are applied based on a magnitude difference of the complex results (see equation (12)). 13. When testing CDM MIMO radars, gain corrections are applied to the transponder I and Q signals. 14. The gain corrections are applied to the transponder's output variable attenuator. 15. Multiple power acquisition positions are used prior to a scenario deployment to record a “poor man’s” far-field pattern versus a timeslot, and this information is used to derive the CDM code versus a timeslot.

[0085] The disclosure of each of the following references is incorporated herein by reference in its entirety. References 1. “Automotive Forward Looking Sensor Test Station,” Mark E. Russell and Clifford A. Drubin, U.S. Patent No. 6,114,985, 9 / 5 / 2000. 2. “Evolving Your ADAS and AV Tests with Emulation Capability, Keysight, White Paper, June 1, 2022. 3. “Coordinated Mini-Radar Target Simulators for Improved Accuracy and Improved Ghost Cancellation,” Gregory S. Lee, Gregory Douglas Vanwiggeren and Ken A. Nishimura, U.S. Patent Application Publication No. 22 / 0019834, January 20, 2022.

[0086] It should be understood that various details of the subject matter described herein may be changed without departing from the scope of the subject matter described herein. Furthermore, the foregoing description is for illustrative purposes only, and not for limiting purposes, as the subject matter described herein is defined by the claims set forth below. QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited non-patent literature

[0000] Automotive Forward Looking Sensor Test Station“, Mark E. Russell und Clifford A. Drubin, U.S.-Patent No. 6,114,985, 5.9.2000

[0085] „Evolving Your ADAS and AV Tests with Emulation Capability, Keysight, White Paper, 1.6.2022

[0085] Coordinated Mini-Radar Target Simulators for Improved Accuracy and Improved Ghost Cancellation“, Gregory S. Lee, Gregory Douglas Vanwiggeren und Ken A. Nishimura, U.S.-Patentanmeldung Veröffentlichungsnr. 22 / 0019834, 20.1.2022

[0085]

Claims

[1] A method for phase-conjugated target emulation for testing a double-sideband (DSB) frequency modulation continuous wave (FMCW) radar device, the method comprising the following steps: Providing (700) a radar scene emulator having an array of frequency shift transponders for emulating physical radar targets; Selecting (702), as a proxy transponder for transmitting an emulated target radar signal to a radar DUT, one of the frequency shift transponders from the array of frequency shift transponders at a mirror image location with respect to a desired direction of arrival of a signal from an emulated radar target; receiving (704) a radar signal from the radar DUT at the agent transponder; generating (706) an output signal to be transmitted by the proxy transponder to the radar DUT by swapping the roles of a normal sideband component and a ghost sideband component of the output signal such that the ghost sideband component appears to the radar DUT as a desired emulant and the normal sideband component appears to the radar DUT as an undesired ghost sideband component; Send (708) the output signal to the radar DUT. [2] The method of claim 1, wherein providing (700) the radar scene emulator with the array of frequency shift transponders comprises providing a two-dimensional array of frequency shift transponders, and selecting the proxy transponder at the mirror image location with respect to the desired direction of arrival comprises selecting, as the proxy transponder, one of the frequency shift transponders located at a mirror image azimuth and elevation location with respect to the desired direction of arrival. [3] A method according to claim 1 or 2, wherein swapping the roles of normal and ghost sideband components comprises swapping modulation channels used to generate the output signal so that the modulation channels are opposite to those used by the radar DUT to demodulate the output signal. [4] The method of claim 3, wherein generating the output signal (706) comprises causing the radar DUT to detect an intermediate frequency as follows: fIF=2Sru¨bersch¨ssig,neu / c, where f IF is the intermediate frequency, S is a chirp slope of the radar signal from the radar DUT, c is the speed of light in vacuum, and r überschüssig,neu is given by: excess,new=rEmulant+dset, where r überschüssig,neu is a distance to which the ghost component is shifted, r Emulant a removal of the desired emulant and d eingestellt is an effective distance from the proxy transponder to the radar DUT, including a distance that takes into account a processing latency of the proxy transponder. [5] The method of any one of claims 1 to 4, wherein the radar DUT comprises a time division multiplexed (TDM) multiple input multiple output (MIMO) radar, the method further comprising applying phase corrections to output signals sent to the radar DUT to account for different distances from the agent transponder to different antennas of the TDM MIMO radar. [6] The method of claim 5, wherein applying the phase corrections comprises determining the phase corrections using the following equation: ϕk1=±(360° / λ)∗(dkE−dkP−d1E+d1P)mod 360° where ϕ k1 is a phase correction to be applied to a signal output from the agent transponder to a k-th transmit antenna of the radar DUT, λ is a transmit wavelength of the TDM-MIMO radar, d kPis the physical distance from the k-th transmitting antenna of the radar DUT to the representative transponder, d kE is a distance from the k-th transmitting antenna of the radar DUT to a desired emulation mirror image position on the radar scene emulator, d 1P is a distance from the k-th transmitting antenna to the representative transponder and d 1E is a distance from the k-th transmit antenna to the desired emulation mirror position and mod is the modulo function. [7] The method of any one of claims 1 to 6, wherein the radar DUT comprises a code division multiple access (CDM) multiple input multiple output (MIMO) radar, the method further comprising applying phase corrections to output signals sent to the CDM-MIMO radar to account for different distances from the agent transponder to different antennas of the CDM-MIMO radar. [8] The method of claim 7, wherein applying the phase corrections comprises determining the phase corrections using the following equation: ϕ10=±[ang(AEχ1)−ang(APχ1)−ang(AEχ0)+ang(APχ0)] where ϕ 10 is the phase correction to be applied to an active code χ1, ang() is a function that returns an angle in degrees of a complex argument, A Pχ and A Eχ are given by: APχ=∑k=1NTxakPχ AEχ=∑k=1NTxakEχ where a kPχ is an incident complex amplitude from a k-th transmitting antenna of the radar DUT to a position of the representative transponder during a phase encoding χ and a kEχ is an incident complex amplitude from the k-th transmit antenna of the radar DUT at the position of the desired emulant during phase encoding χ. [9] A method according to claim 7 or 8, comprising applying a gain correction to the output signal and calculating the gain correction using the following equation: Gj=dB(AEχj)−dB(APχj) where G j is the gain correction to be applied to the signal modulated using a j-th code χ j , dB() is a function that outputs a resulting field strength in decibels, A Eχj is an amplitude of a signal at a position of the desired emulant and A Pχj is an amplitude of a signal detected at the agent transponder. [10] A phase-conjugate target emulation system for testing a double-sideband (DSB) frequency modulation continuous wave (FMCW) radar device according to the method of any one of claims 1 to 9.

Images