Target testing based on uncalibrated raw radar data

Abstract

This document describes techniques and systems for target testing based on uncalibrated raw radar data. A channel response (e.g., raw radar data) is obtained from a sensor having a multiple-input multiple-output (MIMO) antenna array. A channel response based on target detection is arranged in a MIMO measurement matrix. Primary singular values of the observation matrix are determined, and the number of primary singular values is compared to the number of targets in the FOV of the sensor, whether from a priori information or estimated after calibration. If the number of main singular values differs from the number of targets (whether from a priori information or estimated after calibration), further analysis may be used to determine the cause of the difference. In this manner, a potentially damaged sensor or degradation caused by reflection of the covering may be identified, sensor status independent of calibration may be monitored, and calibration may be improved.

Description

Background Art

[0001] Multiple-input multiple-output (MIMO) radar systems are a common approach for many radar applications. MIMO radar systems can provide better angular discrimination of targets than non-MIMO systems by forming a large synthetic array of virtual antenna channels. However, due to different factors, such as reflections from damaged physical channels or coverings (e.g., radomes, dashboards), or other factors, the original antenna ray response may be degraded. This degradation may lead to further loss of performance after system calibration. Summary of the invention

[0002] This document describes techniques and systems for target testing based on uncalibrated raw radar data. In one example, a method, the method includes: receiving uncalibrated raw channel responses from a sensor including an uncalibrated raw MIMO antenna array. The method further includes: generating a MIMO measurement matrix based on the uncalibrated raw channel responses. The method further includes: extracting singular values ​​from the MIMO measurement matrix (or extracting eigenvalues ​​from a covariance matrix based on the MIMO measurement matrix). The method further includes: determining the number of dominant singular values ​​based on the singular values. The method further includes: evaluating whether the channel response is valid based on the number of dominant singular values. The method further includes: in response to the channel response being valid, passing the sensor for use in a real-world application. The method further includes: in response to the channel response being invalid, determining a mitigation that results in the verification of the channel response so that the sensor can be used in a real-world application.

[0003] These and other described techniques may be performed by hardware or a combination of hardware and software executed thereon. For example, a computer readable storage medium (CRM) may have instructions stored thereon, and when the instructions are executed, the processor is configured to perform the described techniques. The system may include means for performing the described techniques. The processor or processor unit may be part of a system configured to perform the methods and techniques described herein.

[0004] This summary introduces simplified concepts related to target testing based on uncalibrated raw radar data, which are further described in the detailed description and drawings. This summary is not intended to identify essential features of the claimed subject matter, nor is it intended to be used to determine the scope of the claimed subject matter. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] Details of target testing based on uncalibrated raw radar data are described in this document with reference to the accompanying drawings, which may use the same numerals to refer to similar features and components, and hyphen numbers to designate variations of these similar features and components. The drawings are organized as follows: Figure 1 An example environment is shown in which target testing based on uncalibrated raw radar data of a MIMO radar system can be applied according to the techniques of this disclosure; Figure 2 illustrates a scenario in which target testing based on uncalibrated raw radar data may be applicable to a MIMO radar system having an overlay and integrated into a vehicle according to the techniques of the present disclosure; Figure 3 shows an example vehicle in which target testing based on uncalibrated raw radar data may be applied in accordance with the techniques of this disclosure; Figure 4 The relationship between physical channels and virtual channels for target testing of a MIMO radar system based on uncalibrated raw radar data according to the techniques of this disclosure is shown; Figure 5 shows two graphs representing single target results and multiple target results from target testing of a MIMO radar system based on uncalibrated raw radar data in accordance with techniques of the present disclosure; and Figure 6 An example method for target testing a MIMO radar system based on uncalibrated raw radar data in accordance with the techniques of this disclosure is shown. DETAILED DESCRIPTION Overview

[0006] MIMO radar systems are often integrated into vehicles and used in conjunction with other automotive systems to perform navigation or safety tasks. MIMO radar systems are often superior to other radar systems in detecting and tracking multiple objects. MIMO radar systems can have better angular resolution than other systems due to their ability to form a large synthetic array of virtual channels.

[0007] These virtual channels may be degraded by different factors, such as mutual coupling between channels or length differences between the transmission lines associated with each physical channel. In cases such as these, calibration is often used to correct the raw antenna array response (e.g., channel response) to match the ideal array response. However, other factors may also degrade the raw antenna ray response. Radomes, dashboards, or other coverings may introduce reflections of the sensor's transmitted signals. These reflections may mix with reflections from targets in the sensor's field of view (FOV), resulting in performance degradation, even after the sensor has been calibrated.

[0008] This document describes techniques and systems for target testing based on uncalibrated raw radar data. Channel responses (e.g., raw radar data) are obtained from a sensor with a multiple-input multiple-output (MIMO) antenna array. The channel responses based on target detection are arranged in a MIMO measurement matrix. Dominant singular values ​​of the observation matrix are determined, and the number of dominant singular values ​​is compared to the number of targets in the sensor's FOV (whether from a priori information or estimated after calibration). If the number of dominant singular values ​​is different from the number of targets (whether from a priori information or estimated after calibration), further analysis can be used to determine the cause of the difference. In this way, potentially damaged sensors or degradation caused by reflections from coverings can be identified, sensor status independent of calibration can be monitored, and calibration can be improved. Example Environment

[0009] Figure 1 An example environment 100 is shown in which target testing based on uncalibrated raw radar data of a MIMO radar system 104 may be applied in accordance with the techniques of this disclosure. Environment 100 includes a test chamber 102 for target testing of MIMO radar system 104. Likewise, other environments may be used as a substitute for test chamber 102. Any environment for testing MIMO radar system 104 may be used. Additionally, MIMO radar system 104 may be tested as a standalone system or as a system installed in a vehicle.

[0010] MIMO radar system 104 may be mounted to any platform. For example, a manufacturer may integrate features of MIMO radar system 104 (such as one or more radar sensors) into a vehicle traveling on the road (e.g., Figure 2 The vehicle 204 shown in the figure may be viewed from the side mirror, roof, bumper, or any other interior or exterior location (where the FOV includes the road and any objects moving or stationary near the road) of the vehicle 204. The vehicle may represent different types of motor vehicles (e.g., cars, trucks, motorcycles, buses, tractors, semi-trailers), non-motor vehicles (e.g., bicycles), rail vehicles (e.g., trains), water vehicles (e.g., boats), aircraft (e.g., airplanes), and spacecraft (e.g., satellites), etc.

[0011] MIMO radar system 104 includes hardware that may include a MIMO antenna array system 106, a processor 108, and a computer readable storage medium (CRM) 110. MIMO radar system 104 transmits and receives electromagnetic signals via MIMO antenna array system 106. MIMO antenna array system 106 includes a transmit component and a receive component. The transmit component may include one or more transmit antennas (e.g., transmit channels), one or more transmitters, and circuits that couple the transmit antennas to the transmitters. Similarly, the receive component may include one or more receive antennas (e.g., receive channels), one or more receivers, and circuits that couple the receive antennas to the receivers. Reference Figure 3 The MIMO antenna array system 106 is described in more detail.

[0012] Processor 108 may be a plurality of separate or a single microprocessors, one or more systems on a chip of a computing device, a controller, or a control unit. Processor 108 executes computer executable instructions stored in CRM 110. CRM 110 may store machine readable instructions that, when executed by processor 108 or other logic of MIMO radar system 104, cause processor 108 or other logic to transmit and receive electromagnetic energy to detect and identify objects, such as one or more targets 112 (e.g., actual targets 112, known targets 112). MIMO radar system 104 may include a combination of hardware, software, and / or firmware for detecting objects, such as for automotive systems.

[0013] At least one target 112 may be present in the test chamber 102. Although this document primarily describes techniques and systems related to a single target 112, multiple targets 112 may be used for similar results. The MIMO radar system 104 transmits an electromagnetic beam (e.g., a beam vector) and receives a reflected beam reflected by the target 112 as a channel response 114. Each virtual transmit-receive channel of the MIMO radar system 104 has an associated uncalibrated raw channel response 114. For example, the channel response 114 is a plurality of channel responses, and the number of channel responses is equal to the number of virtual transmit-receive channels of the MIMO radar system 104.

[0014] In addition to beams reflected by the target 112 , the channel response 114 may also include reflections from a covering 202 adjacent to the MIMO antenna array system 106 . Figure 2 Scenario 200 is shown in which target testing based on uncalibrated raw radar data may be applied to MIMO radar system 104 having overlay 202 and integrated into vehicle 204 in accordance with techniques of this disclosure.

[0015] The MIMO antenna array system 106 of the MIMO radar system 104 may transmit a signal 206 and receive a signal (not shown) from one or more objects (e.g., target(s) 112). Additionally, the covering 202 (e.g., radome, dashboard, water / ice accumulated on the radome / dashboard surface) may cause one or more reflections 208. The reflections 208 may be mixed with the signal reflected from the object and included in the received channel response of the MIMO antenna array system 106. The reflections 208 may cause the MIMO radar system 104 to detect a false positive, i.e., to sense (e.g., detect) an additional target that is not actually in the FOV of the MIMO radar system 104. The reflections 208 may cause calibration performance degradation or overfitting associated with the MIMO radar system 104. The techniques and systems described herein may identify and help find mitigation techniques to overcome degradation of the MIMO radar system due to reflections 208 or other problems that may be associated with the MIMO radar system 104. Example System

[0016] Figure 3 An example vehicle 204 is shown in which target testing based on uncalibrated raw radar data can be applied according to the techniques of the present disclosure. Vehicle 204 includes controller 302 and MIMO radar system 104. MIMO radar system 104 can be tested using the techniques described herein. Controller 302 and MIMO radar system 104 communicate via link 304. Link 304 can be a wired link or a wireless link, and in some cases includes a communication bus. Controller 302 performs operations based on information received from MIMO radar system 104 via link 304 (such as data output from MIMO radar system 104, including information indicating one or more objects identified and tracked in FOV).

[0017] Controller 302 includes processor 306 and CRM 308, which stores instructions for car module 310. Processor 306 can execute car module 310 to perform driving functions or other operations of vehicle 204, which can include using output from MIMO radar system 104 to help determine driving decisions. For example, car module 310 can provide automatic cruise control and monitor the output of MIMO radar system 104, which indicates the presence of objects in the FOV, for example, to reduce speed and prevent collisions with vehicles or pedestrians. When data obtained from MIMO radar system 104 indicates that one or more objects are crossing in front of vehicle 204, car module 310 can provide an alert or perform a specific maneuver.

[0018] Figure 3Also shown is a cloud system 312 including at least one processor 314 and CRM 316. Cloud system 312 may be any computer system external to vehicle 204 and / or MIMO radar system 104 and may be used to perform testing based on the techniques described herein. Cloud system may communicate with vehicle 204 and / or MIMO radar system 104 via cloud link 320, which may be a wired link or a wireless link (e.g., via the “cloud”).

[0019] CRM 316 stores test module 318. Test module 318 includes a set of instructions that configure processor 314 to perform target testing on MIMO radar system 104 based on uncalibrated raw radar data. Test module 318 receives raw channel responses from MIMO radar system 104 (e.g., before significant processing of the channel responses). Test module 318 can organize the channel responses into uncalibrated raw MIMO measurement matrices (e.g., observation matrices, raw antenna response matrices). The test module extracts singular values ​​from the MIMO measurement matrix using singular value decomposition or other methods. The main singular values ​​can then be determined from the singular values ​​using different processes. Alternatively, the eigenvalues ​​of the covariance matrix derived from the MIMO measurement matrix can be calculated and used to replace the main singular values. The number of targets in the FOV of the MIMO radar system (for some applications, such as in a chamber, the number is known as a priori information, and for some applications, the number can be estimated after calibration) should be equal to the number of main singular values ​​derived from the MIMO measurement matrix. If they are not equal, further analysis can be performed to determine if the channel response includes reflections from the covering, or if the MIMO radar system 104 is damaged. This test can also be used independently of calibration to monitor radar status or blockage, find true raw single target measurements for online calibration, and evaluate the quality of raw radar measurements and find the root cause of poor calibration (e.g., near-field chamber calibration and misalignment analysis, etc.). For example, if the number of major singular values ​​based on the calibrated raw channel response in this example is different from the number of targets detected by the calibrated sensor, online calibration can be applied to the sensor to improve performance.

[0020] Although illustrated in this manner, test module 318 may alternatively be stored in CRM 110 and / or CRM 308. Furthermore, test module 318 may be executed on any MIMO radar system 104 that is integrated into vehicle 204 or that is not yet installed on a platform. Test module 318 may also perform testing based on a single target or multiple targets in the FOV of MIMO radar system 104.

[0021] Figure 4The relationship 400 between physical channels 402 and virtual channels 404 for target testing of a MIMO radar system based on raw uncalibrated radar data according to the techniques of this disclosure is shown. The physical channels 402 include one or more physical transmit (TX) channels 406 and one or more physical receive (RX) channels 408. In this example, there are four physical transmit channels (Tx1 to Tx4) 406 and four physical receive channels (Rx1 to Rx4); however, the number of transmit channels 406 and the number of receive channels 408 can be any number. In addition, the physical transmit channels 406 and the physical receive channels 408 can be equal numbers or different numbers.

[0022] The number of virtual (transmit-receive) channels 404 is determined by multiplying the number of physical transmit channels 406 by the number of physical receive channels 408. For example, Figure 4 Sixteen virtual channels 404 (Tx1Rx1 through Tx4Rx4) are shown, which are the product of four physical transmit channels 406 and four physical receive channels 408.

[0023] The channel response of each virtual channel 404 measures the round trip of electromagnetic energy transmitted and received by the associated physical transmit and receive channels of the virtual channel. Therefore, the channel response of each virtual channel 404 includes a transmit component and a receive component for mutual coupling, which is equal to the mutual coupling between the associated physical transmit channel 406 and the physical receive channel 408. For example, the mutual coupling of the virtual channel Tx2Rx2 is affected by the physical adjacent channels of Tx2 and Rx2, while the virtual channel Tx3Rx2 is affected by the physical adjacent channels of Tx3 and Rx2. The receive components of the mutual coupling of the two virtual channels are the same (the mutual coupling components generated by Rx2), but the transmit components of the mutual coupling of the two virtual channels are different. Tx2Rx2 has a transmit component associated with Tx2, and Tx3Rx2 has a transmit component associated with Tx3. Furthermore, adjacent virtual channels 404 (eg, Tx2Rx2 and Tx3Rx2) may have a greater combined mutual coupling than virtual channels further away (eg, Tx1Rx1 and Tx4Rx4).

[0024] For a MIMO radar system with N transmit channels and M receive channels, assuming that α(θ) is the ideal composite array response for a single target at angle θ, Among them, a t (θ) and a r (θ) is the ideal steering vector for Tx and Rx. Vec(:) is a vectorized operation. The matrix A(θ) is a rank one matrix because it is the product of two vectors, as shown in Equation 2: A(θ)=a t (θ)·a r (θ) T (2) The raw antenna response (e.g., uncalibrated raw channel response) x(θ) can be rewritten as an NxM matrix X(θ): The raw antenna response data matrix X(θ) (e.g., the MIMO measurement matrix) is usually of full rank. In order for X(θ) to have a rank of 1, the mutual coupling between the Tx antenna and the Rx antenna needs to be negligible, and there is only mutual coupling between the Tx antenna itself and between the Rx antenna itself. This condition is usually applicable to frequency modulated continuous wave radar systems, where transmit and receive isolation is an important requirement in antenna and system design. Otherwise, the lack of isolation may lead to an increased noise floor and reduced sensitivity.

[0025] Ideally, the original measurement matrix of a single target can be approximated as a rank 1 matrix and written as: Among them, C t is a transmit calibration matrix with full rank, and is a receive calibration matrix with full rank.

[0026] X(θ) can be written as: X(θ)=a t (θ)·a r (θ) T ·s+M+n (5) Where s is the singular value, M is the mutual coupling between the TX and RX antennas, and n is the measurement noise. Assuming M and n are negligible, for singular value decomposition, and assuming X(θ) comes from a single target measurement, then ideally: Therefore, in order for X(θ) to be rank 1, there should be one dominant singular value. It should be noted that for multiple targets at different angles, there should be a dominant singular value (e.g., or eigenvalue) corresponding to each target. For example, for the original measurements with multiple targets at different angles: Y=X(θ 1 )+X(θ 2 )+…X(θ N ) (7) Ignore the mutual coupling between TX and RX antennas: Y=a t (θ 1 )·a r (θ 1 )T ·s 1 +a t (θ 2 )·a r (θ 2 ) T ·s 2 +…a t (θ N )-a r (θ N ) T ·s N (8) Then Equation 9 can be derived: It can be seen from Equation 9 that the rank of Y is greater than 1, and for measurements from N targets at different angles, there are N dominant singular values.

[0027] There are several different types of analysis that can be performed on the singular values ​​to determine if they are dominant. Some methods compare the ratio of the singular value to the sum of all singular values ​​(e.g., a weighted average): Or the ratio of one singular value to the next: These ratios can be further compared to a threshold amount to determine whether the singular value is a dominant singular value. The correlation between the original beam vector (e.g., measurement vector) and the beam vector reconstructed from the dominant singular vector can be analyzed. Or other statistical models can be used to find the dominant singular values, such as Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Minimum Description Length (MDL).

[0028] Based on these equations, a MIMO radar system can be tested based on its raw channel response. If the number of dominant singular values ​​is equal to the number of targets in the FOV of the MIMO radar system (whether estimated from a priori information or after calibration), the MIMO radar system passes the test and the channel response is valid. If the number of dominant singular values ​​is not equal to the number of targets in the FOV of the MIMO radar system (whether estimated from a priori information or after calibration), the channel response is not valid, and further analysis can be performed, and mitigation techniques for overcoming the inaccuracies of the MIMO radar system can be sought. Based on the channel response being valid, the sensor can be used in real-world applications (e.g., deployed in the field). Example Implementation

[0029] Figure 5Two graphs representing single target results (graph (a)) and multiple target results (graph (b)) from target testing of a MIMO radar system based on raw uncalibrated radar data are shown in accordance with techniques of the present disclosure. In graph (a), one target represented by curve 502 can be determined. The other curves in graph (a) have amplitudes closer to zero and can be assumed to be noise or other unrelated indications. For example, if the test is performed on a MIMO radar system with a covering, any reflections from the covering may not adversely affect the MIMO radar system.

[0030] Graph (b) shows two main curves, curve 504 and curve 506. Continuing with the same assumptions as graph (a), if the graph is generated based on one actual target in the FOV (e.g., represented by curve 504), then a second perceived target (e.g., represented by curve 506) may prove that the MIMO radar system is degraded due to reflections from an overburden or some other problem associated with the MIMO radar system. However, if there are two known targets in the FOV of the MIMO radar system, then the MIMO radar system will pass the test and the channel responses representing curves 504 and 506 are valid. In this way, a MIMO radar system can be tested based on the raw channel responses to determine the effects of overburden or blockage, identify bad units, or set a baseline for the raw channel responses prior to calibration. Example Method

[0031] Figure 6 An example method 600 for target testing a MIMO radar system based on uncalibrated raw radar data according to the techniques of the present disclosure is shown. The method 600 is shown as a plurality of sets of operations (or actions) being performed, but is not necessarily limited to the order or combination of operations shown herein. In addition, any one or more operations may be repeated, combined, or reorganized to provide other methods.

[0032] At 602, uncalibrated raw channel responses are received by a computing unit from a sensor including a MIMO antenna array. The computing unit may be a test computer, or the computing unit may be part of the sensor system or part of another system communicatively connected to the sensor system. The channel responses may be raw antenna responses including detections made by the sensor.

[0033] At 604, an uncalibrated raw MIMO measurement matrix (e.g., raw antenna response data matrix, X(θ)) is generated based on the channel responses. Each of the elements of the MIMO measurement matrix corresponds to a virtual channel formed by the MIMO antenna array as part of a synthetic array aperture. That is, for N transmit channels and M receive channels, there may be N×M virtual channels, and the MIMO measurement matrix may have a dimension of N×M.

[0034] At 606, singular values ​​are extracted from the MIMO measurement matrix. The singular values ​​may be decomposed based on singular value decomposition.

[0035] At 608, the number of dominant singular values ​​is determined from the set of singular values ​​extracted at 606. Alternatively, the eigenvalues ​​of the covariance matrix derived from the MIMO measurement matrix can be determined and used in place of the dominant singular values ​​of the target test. The dominant singular values ​​can be determined based on different statistical models. For example, a simple statistical model can be used, such as comparing the weighted average of each singular value to a threshold, or comparing one singular value to the next singular value. Another method can analyze the correlation between the original beam vector and the beam vector reconstructed from the dominant singular vector. Similarly, more complex statistical models such as AIC, BIC, or MDL can be used. These statistical models can be analyzed individually or in combination with each other.

[0036] At 610, the number of dominant singular values ​​is evaluated to determine whether the channel response is valid. That is, ideally, the number of dominant singular values ​​should be equal to the number of actual targets in the FOV of the sensor, the number of targets being known as a priori information or estimated after calibration. For a single target test, the number of dominant singular values ​​should be one. If there are multiple targets in the FOV (e.g., actual targets), the number of singular values ​​should be equal to the number of multiple targets, and if they are equal, the channel response is considered valid. If the number of singular values ​​is not equal to the number of targets, the channel response can be determined to be invalid. Some reasons why the number of singular values ​​is not equal to the number of multiple targets is that the channel response may include reflections from a covering on the sensor's MIMO antenna array (e.g., a radome, a dashboard), or the sensor may be damaged.

[0037] At 612, if the channel response is valid, the sensor is considered to have passed the target test. The sensor can then be used in a real-world application (e.g., deployed in the field). Other applications include calibration of the sensor, evaluating raw sensor measurements, and online calibration.

[0038] At 614, if the channel response is invalid, the sensor is considered to have failed the target test. A mitigation of the channel response may be determined that can be verified. In examples where it is determined that the invalid channel response is due to degradation associated with reflections from the covering, techniques may be applied to reduce the reflections or otherwise filter them out. In some cases, including where the sensor is determined to be a bad unit, the sensor may be repaired or discarded. In other cases, the invalid channel response may indicate that a poor calibration was applied to the sensor, and the channel response may help find the root cause of the poor calibration and, in some cases, improve calibration performance through online calibration. Additional Examples

[0039] Some additional examples of target testing of MIMO radar systems based on uncalibrated raw radar data are provided below.

[0040] Example 1: A system comprising: at least one processor, the at least one processor being configured to: receive a channel response from a sensor comprising a multiple-input multiple-output (MIMO) antenna array; generate a MIMO measurement matrix based on the channel response; extract singular values ​​from the MIMO measurement matrix; determine a number of dominant singular values ​​based on the singular values; evaluate whether the channel response is valid based on the number of dominant singular values; in response to the channel response being valid, pass the sensor for use in a real-world application; and in response to the channel response being invalid, determine a mitigation resulting in validation of the channel response so that the sensor can be used in a real-world application.

[0041] Example 2: The system of Example 1, wherein the received channel responses include channel responses of a plurality of virtual channels.

[0042] Example 3: The system of any of the previous examples, wherein the elements of the MIMO measurement matrix correspond to respective channel responses.

[0043] Example 4: The system of any of the previous examples, wherein the processor is configured to extract singular values ​​from the MIMO measurement matrix by at least: decomposing the MIMO measurement matrix based on singular value decomposition; or calculating eigenvalues ​​of a covariance matrix derived from the MIMO measurement matrix.

[0044] Example 5: A system of any of the previous examples, wherein the processor is configured to determine the number of dominant singular values ​​by at least: comparing a weighted average of each singular value to a threshold amount; comparing a ratio of a first singular value to a second singular value to a threshold; evaluating a correlation between an original beam vector (measurement vector) associated with a sensor and a beam vector reconstructed from the singular values; analyzing based on an Akaike information criterion model; analyzing based on a Bayesian information criterion model; or analyzing based on a minimum description length model.

[0045] Example 6: The system of any of the previous examples, wherein the number of principal singular values ​​corresponds to the number of perception targets detected by the sensor.

[0046] Example 7: The system of any of the previous examples, wherein the processor is configured to evaluate whether the channel response is valid by at least comparing the number of dominant singular values ​​to the known number of targets in the field of view of the sensor; if the number of dominant singular values ​​is equal to the known number of targets, determining that the channel response is valid; and if the number of dominant singular values ​​is not equal to the known number of targets, determining that the channel response is invalid.

[0047] Example 8: The system of any of the previous examples, wherein mitigation includes reducing effects due to reflections from a covering of the sensor.

[0048] Example 9: The system of any of the previous examples, wherein the processor is configured to: evaluate whether a calibrated raw channel response is valid by comparing at least the number of dominant singular values ​​to an estimated number of targets detected after calibration of the sensor; determine that the channel response is invalid in response to the number of dominant singular values ​​not being equal to the estimated number of targets detected after calibration of the sensor; and perform online calibration of the sensor based on a difference between the number of dominant singular values ​​and the estimated number of targets detected in response to the channel response being invalid.

[0049] Example 10: A method comprising: receiving a channel response from a sensor comprising a multiple-input multiple-output (MIMO) antenna array; generating a MIMO measurement matrix based on the channel response; extracting singular values ​​from the MIMO measurement matrix; determining a number of dominant singular values ​​based on the singular values; evaluating whether the channel response is valid based on the number of dominant singular values; and in response to the channel response being valid, using the sensor for use in a real-world application.

[0050] Example 11: The method of any of the previous examples, wherein the received channel responses include channel responses for a plurality of virtual channels.

[0051] Example 12: The method of any of the previous examples, wherein the elements of the MIMO measurement matrix correspond to respective channel responses.

[0052] Example 13: The method of any of the previous examples, wherein extracting singular values ​​from the MIMO measurement matrix comprises: decomposing the MIMO measurement matrix based on singular value decomposition; or calculating eigenvalues ​​of a covariance matrix derived from the MIMO measurement matrix.

[0053] Example 14: The method of any of the previous examples, wherein determining the number of dominant singular values ​​comprises: comparing a weighted average of each singular value to a threshold amount; comparing a ratio of a first singular value to a second singular value to a threshold; evaluating a correlation between an original beam vector associated with a sensor and a beam vector reconstructed from the singular values; performing an analysis based on an Akaike information criterion model; performing an analysis based on a Bayesian information criterion model; or performing an analysis based on a minimum description length model.

[0054] Example 15: The method of any of the previous examples, wherein the number of principal singular values ​​corresponds to the number of perceptual targets detected by the sensor.

[0055] Example 16: The method of any of the previous examples, wherein evaluating whether the channel response is valid comprises comparing the number of dominant singular values ​​to a known number of targets in the field of view of the sensor.

[0056] Example 17: The method of any of the previous examples, further comprising: determining that the channel response is valid if the number of dominant singular values ​​is equal to the known number of targets.

[0057] Example 18: The method of any of the previous examples, further comprising: if the number of dominant singular values ​​is not equal to the known number of targets, determining that the channel response is invalid.

[0058] Example 19: The method of any of the previous examples, wherein, in response to the channel response being invalid, determining a mitigation that results in validation of the channel response such that the sensor can be used in a real world application.

[0059] Example 20: A computer-readable medium comprising instructions that, when executed, cause a processor to: receive a channel response from a sensor comprising a multiple-input multiple-output (MIMO) antenna array; generate a MIMO measurement matrix based on the channel response; extract singular values ​​from the MIMO measurement matrix; determine a number of dominant singular values ​​based on the singular values; evaluate whether the channel response is valid based on the number of dominant singular values; in response to the channel response being valid, pass the sensor for use in a real-world application; and in response to the channel response being invalid, determine a mitigation that results in validation of the channel response so that the sensor can be used in a real-world application. Conclusion

[0060] Although various embodiments of the present disclosure are described in the foregoing description and shown in the accompanying drawings, it should be understood that the present disclosure is not limited thereto, but can be implemented in various ways within the scope of the following claims. From the foregoing description, it will be apparent that various changes can be made without departing from the spirit and scope of the present disclosure as defined by the following claims. Problems associated with raw channel response may occur in other systems. Therefore, although described as a way to test the channel response of a radar system, the techniques described above can be applied to other systems that would benefit from analyzing the raw channel response. In addition, these techniques can also be applied to other systems that utilize MIMO antenna arrays.

[0061] Unless the context clearly dictates otherwise, the use of "or" and grammatically related terms represents unlimited, non-exclusive alternatives. As used herein, a phrase referring to "at least one" of a list of items refers to any combination of those items, including individual members. As an example, "at least one of a, b, or c" is intended to cover: a, b, c, ab, ac, bc, and abc, as well as any combination with multiple identical elements (e.g., aa, aaa, aab, aac, abb, acc, bb, bbb, bbc, cc, and ccc, or any other ordering of a, b, and c).

Claims

1. A system, wherein the system include: at least one processor configured to: receiving an uncalibrated raw channel response from a sensor comprising a multiple-input multiple-output (MIMO) antenna array; generating an uncalibrated raw MIMO measurement matrix based on the uncalibrated raw channel response; extracting singular values ​​from the MIMO measurement matrix; determining a number of dominant singular values ​​based on the singular values; evaluating whether the channel response is valid based on the number of the dominant singular values; responsive to the channel response being valid, via the sensor, for use in a real world application; as well as In response to the channel response being invalid, a mitigation is determined that results in validation of the channel response such that the sensor can be used in a real-world application.

2. The system according to claim 1, It is characterized in that The received channel responses include channel responses of a plurality of virtual channels.

3. The system according to claim 1, It is characterized in that The elements of the MIMO measurement matrix correspond to corresponding channel responses.

4. The system according to claim 1, It is characterized in that The processor is configured to extract singular values ​​from the MIMO measurement matrix by at least: Decomposing the MIMO measurement matrix based on singular value decomposition; or Eigenvalues ​​of a covariance matrix derived from the MIMO measurement matrix are calculated.

5. The system according to claim 1, It is characterized in that The processor is configured to determine the number of the principal singular values ​​by at least: comparing the weighted average of each singular value to a threshold amount; comparing a ratio of the first singular value to the second singular value to a threshold; evaluating a correlation between an original beam vector associated with the sensor and a beam vector reconstructed from the singular values; The analysis was conducted based on the Akaike Information Criterion model; Perform analysis based on a Bayesian Information Criterion model; or The analysis was performed based on the minimum description length model.

6. The system of claim 1, It is characterized in that The number of the dominant singular values ​​corresponds to the number of sensory targets detected by the sensor.

7. The system of claim 1, It is characterized in that The processor is configured to evaluate whether the channel response is valid by at least: comparing the number of principal singular values ​​to a known number of objects in a field of view of the sensor; determining that the channel response is valid if the number of the dominant singular values ​​is equal to the known number of the targets; as well as If the number of dominant singular values ​​is not equal to the known number of targets, then the channel response is determined to be invalid.

8. The system of claim 1, It is characterized in that The mitigation includes reducing effects due to reflections from a covering of the sensor.

9. The system of claim 1, It is characterized in that The processor is further configured to: receiving a calibrated raw channel response from the sensor; evaluating whether the calibrated raw channel response is valid by comparing at least the number of the dominant singular values ​​to an estimated number of targets detected after calibration of the sensor; determining that the calibrated raw channel response is invalid in response to the number of dominant singular values ​​not being equal to an estimated number of targets detected after the calibration of the sensor; as well as In response to the calibrated raw channel response being invalid, an online calibration of the sensor is performed based on a difference between the number of dominant singular values ​​and an estimated number of detected targets.

10. A method, the method comprising: include: receiving an uncalibrated raw channel response from a sensor comprising a multiple-input multiple-output (MIMO) antenna array; generating an uncalibrated raw MIMO measurement matrix based on the uncalibrated raw channel response; extracting singular values ​​from the MIMO measurement matrix; determining a number of dominant singular values ​​based on the singular values; evaluating whether the channel response is valid based on the number of the dominant singular values; The channel response is valid through the sensor to be used in real world applications.

11. The method according to claim 10, It is characterized in that The received channel responses include channel responses of a plurality of virtual channels.

12. The method according to claim 10, It is characterized in that The elements of the MIMO measurement matrix correspond to corresponding channel responses.

13. The method according to claim 10, It is characterized in that Extracting singular values ​​from the MIMO measurement matrix includes: Decomposing the MIMO measurement matrix based on singular value decomposition; or Eigenvalues ​​of a covariance matrix derived from the MIMO measurement matrix are calculated.

14. The method according to claim 10, It is characterized in that Determining the number of dominant singular values ​​includes: comparing the weighted average of each singular value to a threshold amount; comparing a ratio of the first singular value to the second singular value to a threshold; evaluating a correlation between an original beam vector associated with the sensor and a beam vector reconstructed from the singular values; The analysis was conducted based on the Akaike Information Criterion model; Perform analysis based on a Bayesian Information Criterion model; or The analysis was performed based on the minimum description length model.

15. The method according to claim 10, It is characterized in that The number of the dominant singular values ​​corresponds to the number of sensory targets detected by the sensor.

16. The method of claim 10, It is characterized in that Evaluating whether the channel response is valid includes: The number of dominant singular values ​​is compared to a known number of objects in a field of view of the sensor.

17. The method of claim 16, further comprising: include: If the number of dominant singular values ​​is equal to the known number of targets, then the channel response is determined to be valid.

18. The method of claim 16, further comprising: include: If the number of dominant singular values ​​is not equal to the known number of targets, then the channel response is determined to be invalid.

19. The method of claim 18, It is characterized in that In response to the channel response being invalid, a mitigation is determined that results in validation of the channel response such that the sensor can be used in a real-world application.

20. A computer readable medium comprising instructions which, when executed, cause a processor to: receiving an uncalibrated raw channel response from a sensor comprising a multiple-input multiple-output (MIMO) antenna array; generating an uncalibrated raw MIMO measurement matrix based on the uncalibrated raw channel response; extracting singular values ​​from the MIMO measurement matrix; determining a number of dominant singular values ​​based on the singular values; evaluating whether the channel response is valid based on the number of the dominant singular values; responsive to the channel response being valid, via the sensor, for use in a real world application; as well as In response to the channel response being invalid, a mitigation is determined that results in validation of the channel response such that the sensor can be used in a real-world application.

Images