Method, apparatus, and sensor for efficient direction of arrival estimation using low rank approximation

Abstract

Novel and useful direction of arrival (DOA) estimation systems and methods with complexity of order O(N log N), which allows for DOA related calibration. In one embodiment, the architecture includes several fast Fourier transform (FFT) machines operating in parallel with coefficient multiplication before and after the FFT operation. These pre- and post-coefficients are computed using singular value decomposition using an optimal low-rank approximation of the distortion matrix. The values of the pre- and post-calibration coefficients before and after the FFT operation for each rank are computed from the singular value decomposition of the distortion matrix C = B / F, where B is the digital beamforming (DBF) matrix and F is the ideal FFT matrix. Methods of obtaining the beamforming matrix B are also disclosed. The architecture implements K x N x log2N operations, which are significantly smaller than the N 2 operations required for full matrix multiplication for relatively low ranks. Circuitry and methods for computing the calibration coefficients are disclosed, as well as a simple proof that this approximates the general beamforming matrix.

Description

Technical Field

[0001] The topics disclosed herein relate to the fields of imaging radar, sonar, ultrasound and other sensors for performing range measurements via FMCW signals and / or angle measurements via digital beamforming and array processing, and more specifically to an efficient mechanism for estimating direction of arrival (DOA) using low-rank approximation (LRA) techniques. Background Technology

[0002] In recent years, many industries, such as the automotive industry and delivery, have been moving towards autonomous solutions. These autonomous platforms are required to operate within their environments and interact with both stationary and moving objects. For this purpose, these systems need sensor suites that allow them to reliably and effectively sense their surroundings. For example, in order for an autonomous car to plan its route through areas with other vehicles on the road, the trajectory planner must have a 3D map of the environment with indications of moving objects.

[0003] Severe weather and poor visibility (due to fog, smoke, sand, storms, etc.) can also degrade visual sensors. They are also limited in estimating radial velocity. Light detection and ranging (LIDAR) devices are used to measure the distance to a target by illuminating it with a laser. However, these devices are expensive, have moving parts, and have a very limited range. Radar is an enhancement technology, not a replacement technology.

[0004] Due to the inherent limitations of visual sensors in terms of range, accuracy, and reliability issues related to optical (e.g., laser) technologies, the optimal solution for generating this 3D map is via radar technology. This presents a new set of requirements that modern radar cannot meet.

[0005] Generally, the larger the aperture of a receiving antenna, the more radiation it receives, resulting in higher sensitivity, or equivalently, a narrower main lobe. Therefore, the receiving antenna can receive weaker signals and provide a relatively accurate indication of their direction.

[0006] On the other hand, vehicle-mounted radars, including vehicle imaging radars, typically have smaller apertures. Although the signals transmitted by vehicle-mounted radars are relatively weak due to power consumption and conditioning limitations, they may require lower sensitivity depending on the link budget, as the range is relatively short and the signals reflected from targets are relatively strong. However, vehicle-mounted radars do not need to detect point targets, such as aircraft detected by missiles, but do require highly accurate images to provide environmental information, which are used as input to one or more tracking and post-processing algorithms and / or Simultaneous Localization and Mapping (SLAM) algorithms (which detect the location of nearby obstacles such as other cars or pedestrians to generate a list of objects from the raw radar detection). Narrow lobes with high accuracy will be able to provide sharper outlines of the target image. The lobe width is determined solely by the equivalent aperture, which is normalized to the wavelength of the transmitted radar signal, and not by the number of receiving antenna elements within the aperture, which affects sensitivity (i.e., the ability to detect weak reflected signals) as well as ambiguity resolution and sidelobe level.

[0007] Another key performance parameter of imaging radar is the sidelobe level of the antenna array. Sidelobes reflected from strong targets can obscure weak targets or lead to false detections. For example, a large object (such as a wall) located in the direction of the sidelobe will cause reflections from the wall to appear in the main lobe. This will obscure reflections from obstacles (such as pedestrians) or create phantom obstacles that could cause vehicles to stop.

[0008] Therefore, minimizing sidelobes is crucial in automotive imaging radar. Additionally, there is a need for a compact radar switch array antenna with high azimuth and elevation resolution and accuracy, providing an increased effective aperture while using a small number of transmit (TX) and receive (RX) elements that meet cost, space, power, and reliability requirements.

[0009] Recently, radar has begun to be used in the automotive industry. High-end vehicles already have radar that provides drivers with parking assistance and lane departure warnings. Currently, interest in autonomous vehicles is growing, and some believe it will be a major driving force for the automotive industry in the coming years.

[0010] Autonomous vehicles offer a new perspective on the application of radar technology in motor vehicles. Motor vehicle radar will be able to play an active role in vehicle control, rather than merely assisting the driver. Therefore, they are likely to become key sensors in the autonomous control systems of vehicles.

[0011] Radar is preferred over other alternatives such as sonar or LiDAR because it is less affected by weather conditions and can be manufactured to a very small size, reducing the impact of the deployed sensor on the vehicle's aerodynamics and appearance. Frequency-modulated continuous wave (FMCW) radar is one type of radar that offers several advantages over other radars. For example, it ensures that the range and speed information of surrounding objects can be detected simultaneously. This information is crucial for providing safe and collision-free operation for the control systems of autonomous vehicles.

[0012] For short-range detection, such as in motor vehicle radar, FMCW radar is commonly used. Several benefits of FMCW radar in motor vehicle applications include: (1) FMCW modulation is relatively easy to generate, and the low-bandwidth processing provides large bandwidth, high average power, high accuracy, low cost, and allows for very good range resolution and the use of Doppler frequency shift to determine speed; (2) FMCW radar can operate with good performance over short ranges; (3) FMCW sensors can be manufactured very small with a single RF transmission source with an oscillator that is also used to downconvert the received signal; and (4) the moderate output power of solid-state components is sufficient because the transmission is continuous.

[0013] Radar systems installed in automobiles should be able to provide the information required by the control system in real time. A baseband processing system is needed to provide sufficient computing power to meet the real-time system requirements. The processing system performs digital signal processing on the received signals to extract useful information, such as the range and speed of surrounding objects.

[0014] Currently, vehicles (especially automobiles) are increasingly equipped with technologies designed to assist drivers in critical situations. In addition to cameras and ultrasonic sensors, automakers are turning to radar as the cost of related technologies decreases. The appeal of radar lies in its ability to provide rapid and clear measurements of the speed and distance of multiple objects in any weather conditions. Correlated radar signals are frequency-modulated and can be analyzed using a spectrum analyzer. In this way, radar component developers can automatically detect, measure, and display signals in the time and frequency domains, even up to frequencies of 500 GHz.

[0015] The use of radar in autonomous vehicles is now attracting significant interest, and such radar is expected to become increasingly prevalent in the future. Millimeter-wave automotive radar is suitable for collision prevention and autonomous driving. Compared to ultrasonic radar and lidar, millimeter-wave frequencies (from 77 to 81 GHz) are less susceptible to rain, fog, snow, and other weather conditions, as well as dust and noise. These automotive radar systems typically include a high-frequency radar transmitter that sends radar signals along a known direction. The transmitter is capable of sending radar signals in continuous or pulsed modes. These systems also include a receiver connected to a suitable antenna system that receives echoes or reflections from the transmitted radar signals. Each such reflection or echo represents an object illuminated by the transmitted radar signal.

[0016] Advanced Driver Assistance Systems (ADAS) are systems developed to automate, adapt, and enhance vehicle systems for safer and better driving. Safety features are designed to avoid collisions and accidents by providing warnings to the driver of potential problems, or by implementing protective measures and taking over vehicle control. Adaptive features can automate lighting, provide adaptive cruise control, automate braking, integrate GPS / traffic warnings, connect to smartphones, warn the driver of other vehicles or hazards, keep the driver in the correct lane, or show blind spots.

[0017] Many forms of ADAS are available: some features are built into the vehicle or are available as additional packages. Furthermore, aftermarket solutions are also available. ADAS relies on input from multiple data sources, including vehicle imaging, LiDAR, radar, image processing, computer vision, and vehicle-to-everything (V2X) connectivity. Additional input from sources outside the primary vehicle platform, such as other vehicles (referred to as vehicle-to-vehicle (V2V)) or vehicle-to-infrastructure systems (e.g., mobile phones or Wi-Fi data networks), is also possible.

[0018] Advanced driver assistance systems are currently one of the fastest-growing segments in automotive electronics. With the steady increase in adoption of industry quality standards in vehicle safety systems ISO 26262, technology-specific standards have been developed, such as IEEE 2020 for image sensor quality and communication protocols such as the Vehicle Information API.

[0019] In recent years, many industries, such as the automotive industry and express delivery, have been moving towards autonomous solutions. These autonomous platforms operate within their environments, interacting with both stationary and moving objects. For this purpose, these systems require sensor suites that allow them to reliably and effectively sense their surroundings. For example, in order for an autonomous vehicle to plan its route through areas with other vehicles on the road, the trajectory planner must have a 3D map of the environment with indications of moving objects.

[0020] Severe weather and poor visibility (such as fog, smoke, sand, heavy rain, or blizzards) can also degrade visual sensors. They are also limited in their ability to estimate radial velocity. Light detection and ranging (LIDAR) devices are used to measure the distance to a target by illuminating it with a laser. However, these are expensive because most have moving parts and very limited range. Therefore, vehicle radar is considered an enhancement rather than a replacement technology.

[0021] In the automotive field, radar sensors are key components of comfort and safety features, such as adaptive cruise control (ACC) or collision mitigation systems (CMS). With the increasing number of vehicle radar sensors operating close to each other simultaneously, radar sensors may receive signals from other sensors. Receiving external signals (interference) can lead to problems such as phantom targets or reduced signal-to-noise ratio. This type of vehicle interference scenario... Figure 1 The image shows direct interference from several surrounding vehicles.

[0022] A well-known way to reduce the number of antenna elements in an array is by using a MIMO technique called a 'virtual array', in which separable (e.g., orthogonal) waveforms are transmitted from different antennas (usually simultaneously), and a larger effective array is generated with the help of digital processing. The shape of this 'virtual array' is a special convolution of the positions of the transmitting and receiving antennas.

[0023] It is also known that, with the aid of bandpass sampling, the de-ramped signal can be sampled at a lower A / D frequency while preserving the target's range information within a range that matches the designed bandpass filter.

[0024] Achieving high resolution simultaneously in angle, range, and Doppler dimension is a significant challenge, especially given the linear increase in hardware complexity.

[0025] Furthermore, direction-of-arrival (DOA) estimation is a critical component of any radar system. For imaging radar, this is typically performed digitally and is often referred to as digital beamforming (DBF). Existing techniques falling into this category include linear operations that can be implemented as matrix multiplication. Other nonlinear methods are often referred to as super-resolution techniques. However, these techniques require significantly higher computational power and are difficult to implement in low-cost consumer sensors.

[0026] However, a problem with existing DOA estimation techniques is the typically high computational load. Common solutions to this problem are (1) using additional computational power, (2) reducing the frame rate, or (3) reducing the number of range-Doppler bins, or any combination thereof. Therefore, it is desirable to have a radar system that performs DOA estimation without compromising the above parameters and exhibits a relatively low computational load. Furthermore, the radar should achieve comparable sidelobe levels (SLLs) compared to existing full DBF calculations. Summary of the Invention

[0027] This invention provides a direction of arrival (DOA) estimation system and method with a complexity of O(n log n). N log N This allows for DOA-related calibration. In one embodiment, the architecture includes several Fast Fourier Transform (FFT) machines that operate in parallel with coefficient multiplication before and after the FFT operation. These pre- and post-coefficients are computed using the optimal low-rank approximation of the distortion matrix via singular value decomposition. C = B / F The singular value decomposition is used to compute the values ​​of the pre-calibration coefficients and post-calibration coefficients before and after the FFT operation for each rank, where B It is a digital beamforming (DBF) matrix, and F It is an ideal FFT matrix. A method for obtaining the beamforming matrix is ​​also disclosed. B The method.

[0028] This architecture implementation K N log2 N The operation, which is significantly less than the full matrix multiplication required for relatively low rank, is performed on matrices with relatively low rank. N 2 Operation, among which K It is an approximate rank, and N It is the length of the Fast Fourier Transform. A circuit and method for calculating the calibration coefficients are also disclosed, along with a simple proof that this approximates a universal beamforming matrix.

[0029] Therefore, according to the present invention, a method for estimating the direction of arrival (DOA) of a signal used in a radar system is provided, comprising: receiving input data; multiplying the input data element-wise by a plurality of pre-coefficient sets to generate a first plurality of results; performing a plurality of fast Fourier transform operations on the plurality of first results to generate a second plurality of results; multiplying the second plurality of results element-wise by the plurality of post-coefficient sets to generate a third plurality of results; and summing the third plurality of results to generate an approximate DOA estimate.

[0030] According to the present invention, a method for estimating the direction of arrival (DOA) of a signal used in a radar system is also provided, comprising: receiving input data x Input data x Each element multiplied by the preceding coefficient V k of k A set to produce k First result diag ( V k ) x ; Regarding the above k The first result is executed. k Fast Fourier Transform operation F To generate k A second result F diag ( V k ) x ; will the k The second result is multiplied element-wise by the postfix coefficient. U k of k A set to produce k A third result U k ( F diag ( V k ) x ); for the above k Summing the third results to produce an approximate DOA estimate y ; and the value therein k It is the rank of the approximation.

[0031] According to the present invention, an apparatus for estimating the direction of arrival (DOA) of a signal for use in a radar system is also provided, comprising an operable means for receiving a receiving antenna array response. xA radar signal processing circuit, operable to multiply each element of the antenna array response by a pre-equalizing coefficient. V k of k A set to produce k First result diag ( V k ) x ; Regarding the above k The first result is executed. k Fast Fourier Transform operation F To generate k A second result F diag ( V k ) x ; will the k The second result is multiplied element-wise by the postfix coefficient. U k of k A set to produce k A third result U k ( F diag ( V k ) x ); for the above k Summing the third results to produce an approximate DOA estimate y ; and the value therein k It is the rank of the approximation.

[0032] According to the present invention, a vehicle radar sensor is also provided, comprising a printed circuit board (PCB) assembly including a plurality of transmitting antennas formed on one side of the PCB assembly, a plurality of receiving antennas formed on the opposite side of the PCB assembly, and a transceiver coupled to the plurality of transmitting antennas and the plurality of receiving antennas. The transceiver is operable to generate and supply transmitting signals to the one or more transmitting antennas, and to receive signals of waves reflected back to the one or more receiving antennas. Radar signal processing circuitry is coupled to the transceiver and is operable to receive input data. x ; the input data x Each element multiplied by the preceding coefficient V k of k A set to produce kFirst result diag ( V k ) x ; Regarding the above k Perform k Fast Fourier Transform operations on the first result. F To generate k A second result F diag ( V k ) x ; will the k The second result is multiplied element-wise by the postfix coefficient. U k of k A set to produce k A third result U k ( F diag ( V k ) x ); for the above k Summing the third results to produce an approximate DOA estimate y ; and the value therein k It is the rank of the approximation. Attached Figure Description

[0033] The present invention is further explained in detail in the following exemplary embodiments with reference to the accompanying drawings, wherein the same or similar elements may be indicated in part by the same or similar reference numerals, and the features of the various exemplary embodiments are combinable. The invention is described herein by way of example only with reference to the accompanying drawings, wherein: Figure 1 It is an illustration of an example street scene, which includes several vehicles equipped with vehicle radar sensor units. Figure 2 This is a diagram illustrating an example radar system that includes multiple receivers and transmitters; Figure 3 This is a diagram illustrating an example radar transceiver constructed according to the present invention; Figure 4 This is a high-level block diagram illustrating an example MIMO FMCW radar according to the present invention; Figure 5 This is a block diagram illustrating an example digital radar processor (DRP) IC constructed according to the present invention; Figure 6This is a high-level block diagram illustrating an example direction of arrival (DOA) estimation using the low-rank approximation (LRA) technique; Figure 7 This is a diagram illustrating an example method for DOA estimation based on the low-rank approximation; and Figure 8 This is a diagram illustrating an example method for calculating the pre-coefficient and post-coefficient. Detailed Implementation

[0034] In the following detailed description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, those skilled in the art will understand that the invention can be practiced without these specific details. In other instances, well-known methods, procedures, and components have not been described in detail so as not to obscure the invention.

[0035] Among the benefits and improvements already disclosed, other objects and advantages of the invention will become apparent from the following description taken in conjunction with the accompanying drawings. Detailed embodiments of the invention are disclosed herein. However, it is to be understood that the disclosed embodiments are merely illustrative of the invention as it can be implemented in various forms. Furthermore, each example given in conjunction with the various embodiments of the invention is intended to be illustrative and not restrictive.

[0036] The subject matter considered to be the present invention is specifically pointed out and clearly claimed in the concluding section of the specification. However, when compared with the appended... Figure 1 When reading this invention, the present invention can be best understood with reference to the following detailed description, regarding its organization and operation methods, as well as its purpose, features, and advantages.

[0037] The accompanying drawings form part of this specification and include illustrative embodiments of the invention, illustrating various objects and features of the invention. Furthermore, the drawings are not necessarily drawn to scale, and some features may be exaggerated to show detail of specific components. Additionally, any measurements, specifications, etc., shown in the drawings are intended to be illustrative rather than restrictive. Therefore, the specific structural and functional details disclosed herein should not be construed as limiting, but merely as a representative basis for teaching those skilled in the art to use the invention in various ways. Furthermore, reference numerals may be repeated in the drawings where deemed appropriate to indicate corresponding or similar elements.

[0038] Because most of the embodiments illustrated in this invention can be implemented using electronic components and circuits known to those skilled in the art, the details will not be interpreted to any extent deemed necessary in order to understand and comprehend the basic concepts of this invention and to avoid obscuring or diverting the teachings of this invention.

[0039] Any reference to a method in this specification should be appropriately adapted to a system capable of performing that method. Similarly, any reference to a system in this specification should be appropriately adapted to methods that can be executed by that system.

[0040] Throughout this specification and claims, unless the context clearly specifies otherwise, the following terms shall have the meaning explicitly associated herein. Although possible, the phrases “in one embodiment,” “in an exemplary embodiment,” and “in some embodiments” as used herein do not necessarily refer to the same(s) embodiments(s). Furthermore, although possible, the phrases “in another embodiment,” “in an alternative embodiment,” and “in some other embodiments” as used herein do not necessarily refer to different embodiments. Therefore, as described below, various embodiments of the invention can be readily combined without departing from the scope or spirit of the invention.

[0041] Additionally, as used herein, unless the context clearly specifies otherwise, the term "or" is an inclusive "or" operator and is equivalent to the term "and / or". The term "based on" is not exclusive and allows for basing on additional factors not described, unless the context clearly specifies otherwise. Furthermore, throughout the specification, the meanings of "a", "an", and "the" include plural references. The meaning of "in" includes both "in" and "on".

[0042] Frequency-modulated continuous wave (FMCW) radar is a type of radar that uses frequency modulation. The operating theory of FMCW radar is to transmit a continuous wave with a constantly increasing (or decreasing) frequency. This wave is called a chirp. The transmitted wave, reflected by an object, is received by a receiver.

[0043] Considering the use of radar in automotive applications, vehicle manufacturers currently have access to four frequency bands with varying bandwidths at 24 GHz and 77 GHz. While the 24 GHz ISM band offers a maximum bandwidth of 250 MHz, the 76-81 GHz ultra-wideband (UWB) band provides up to 5 GHz. A band with bandwidth up to 4 GHz lies between the 77 and 81 GHz frequencies. It is currently being used in many applications. Note that other allocated frequencies for this application include 122 GHz and 244 GHz, with bandwidths of only 1 GHz. Since signal bandwidth determines range resolution, having sufficient bandwidth is crucial in radar applications.

[0044] Conventional digital beamforming (FMCW) radars are characterized by very high resolution in the radial, angular, and Doppler dimensions. Imaging radars are based on the well-known phased array technology, which uses a uniformly linearly distributed array (ULA). It is well known that the far-field beam pattern of a linear array architecture is obtained using a Fourier transform. Range measurements are obtained by performing a Fourier transform on the de-sloped signal, which is generated by multiplying the transmitted signal by the conjugate of the received signal. The radar range resolution is determined by the radar's RF bandwidth and is equal to the speed of light c divided by twice the RF bandwidth. Doppler processing is performed by performing a Fourier transform in the slow time dimension, and its resolution is limited by the coherent processing interval (CPI), i.e., the total transmission time used for Doppler processing.

[0045] When using radar signals in motor vehicle applications, it is desirable to determine the speed and distance of multiple objects simultaneously within a single measurement cycle. Ordinary pulse radar cannot easily handle this task because only the distance can be determined based on the time offset between the transmitted and received signals within the cycle. If speed also needs to be determined, frequency-modulated signals are used, such as linear frequency-modulated continuous wave (FMCW) signals. Pulse Doppler radar can also directly measure the Doppler offset. The frequency offset between the transmitted and received signals is also known as the beat frequency. The beat frequency has a Doppler frequency component f. D and delay component f T The Doppler component contains information about velocity, while the delay component contains information about range. With both range and velocity unknown, two beat frequency measurements are required to determine the desired parameters. A second signal with a linearly modified frequency is immediately incorporated into the measurement after the first signal.

[0046] Two parameters can be determined within a single measurement cycle using FM chirp sequences. Since a single chirp is very short compared to the entire measurement cycle, each beat frequency is primarily composed of a delay component f. T This allows the range to be determined directly after each chirp. Determining the phase shift between several consecutive chirs within the sequence allows the use of Fourier transform to determine the Doppler frequency, thus enabling the calculation of the vehicle's speed. Note that the speed resolution increases with the measurement period length.

[0047] Multiple-input multiple-output (MIMO) radar is a type of radar that uses multiple TX and RX antennas to transmit and receive signals. Each transmit antenna in the array independently irradiates a waveform signal that is different from the signal irradiated from the other antenna. Alternatively, the signals can be the same but transmitted at non-overlapping times. The reflected signals belonging to each transmitter antenna can be easily separated in the receiver antenna because (1) orthogonal waveforms are used for transmission, or (2) because they are received at non-overlapping times. A virtual array is created that contains information from each transmit antenna to each receive antenna. Thus, if we have M transmit antennas and N receive antennas, then by using only M+N physical antennas, we will have M·N independent transmit and receive antenna pairs in the virtual array. This characteristic of MIMO radar systems leads to several advantages, such as improved spatial resolution, increased antenna aperture, and potentially higher sensitivity for detecting slowly moving objects.

[0048] As stated above, signals transmitted from different TX antennas are orthogonal. Orthogonality of the transmitted waveforms can be achieved using time division multiplexing (TDM), frequency division multiplexing, or spatial coding. In the examples and descriptions presented herein, TDM is used, which allows only a single transmitter to transmit at a time.

[0049] The radar of this invention reduces complexity, cost, and power consumption by implementing time-division multiplexing MIMO FMCW radar relative to full MIMO FMCW. The time-division multiplexing method of vehicle MIMO imaging radar offers significant cost and power benefits compared to full MIMO radar. Full MIMO radar simultaneously transmits several separable signals from multiple transmit array elements. These signals typically need to be separated at each receive channel using matched filter banks. In this case, the entire virtual array is simultaneously filled.

[0050] Using Time Division Multiplexing (MIMO), only one transmit (TX) array element is transmitting at a time. The transmit side is greatly simplified, and each receive (RX) channel does not require a matched filter bank. The virtual array will be gradually filled during the time it takes to transmit from all the TX elements in the array.

[0051] The diagram illustrates a high-level block diagram of an example radar system comprising multiple receivers and transmitters, as shown below. Figure 2As shown. The radar system (generally referred to as 280) includes: a digital radar processor (DRP) / signal processor 282 for performing signal processing functions including direction-of-arrival (DOA) estimation utilizing the low-rank approximation (LRA) mechanism of the present invention; a plurality of N transmitter devices TX1 to TXN 284, each coupled to a transmit antenna 288; and a plurality of M receiver devices RX1 to RXM 286, each coupled to a receive antenna 290. TX data lines 292 connect the DRP to the transmitter devices, RX lines 294 connect the receiver devices to the DRP, and control signals 296 are provided by the DRP to each of the transmitter devices 284 and receiver devices 286, respectively. Note that N and M can be any positive integer greater than 1.

[0052] The diagram illustrates an example radar transceiver constructed according to the present invention. Figure 3 As shown in the diagram. The radar transceiver (typically referred to as 80) includes a transmitter 82, a receiver 84, and a controller 83. The transmitter 82 includes a nonlinear frequency hopping sequencer 88, an FMCW chirp generator 90, a local oscillator (LO) 94, a mixer 92, a power amplifier (PA) 96, and an antenna 98.

[0053] The receiver 84 includes an antenna 100, an RF front end 101, a mixer 102, an IF block 103, an ADC 104, a fast time range processing 106, a slow time processing (Doppler and fine range) 108, and azimuth and elevation processing 110.

[0054] In operation, a nonlinear frequency hopping sequencer 88 generates a nonlinear initiation frequency hopping sequence. The initiation frequency for each chirp is input to an FMCW chirp generator 90, which generates a chirp waveform at a specific initiation frequency. The chirp is up-converted via a mixer 92 to an appropriate band (e.g., an 80 GHz band) according to LO 94. The up-converted RF signal is amplified via a PA 96 and output to an antenna 98, which, in the case of a MIMO radar, may include an antenna array.

[0055] On the receiving side, the echo signal arriving at antenna 100 is input to RF front-end block 101. In a MIMO radar, receiving antenna 100 includes an antenna array. The signal from the RF front-end circuitry is mixed with the transmitted signal via mixer 102 to generate a beat frequency that is input to IF filter block 103. The output of the IF block is converted to digital via ADC 104 and input to fast-time processing block 106 to generate coarse range data. Slow-time processing block 108 is used to generate fine range data and Doppler velocity data. Azimuth and elevation data are then calculated via azimuth / elevation processing block 110. 4D image data 112 is input to downstream image processing and detection. Note that in one embodiment, azimuth / elevation processing block 110 utilizes the low-rank approximation (LRA) mechanism of the present invention to implement direction of arrival (DOA) estimation.

[0056] A high-level block diagram of an example MIMO FMCW radar according to the present invention is illustrated as follows: Figure 4 As shown. The radar transceiver sensor (typically referred to as 40) includes multiple transmit circuits 66, multiple receive circuits 58, a ramp or chirp generator 60 including a local oscillator (LO) 61, a nonlinear frequency hopping sequencer 62, an optional TX element sequencer 75 (dashed line), and a digital radar processor (DRP) / signal processing block 44, which in one embodiment includes a block 45 that implements direction-of-arrival (DOA) estimation using the low-rank approximation (LRA) mechanism of the present invention. In operation, the radar transceiver sensor typically communicates with and can be controlled by a host 42. Each transmit block includes a power amplifier 70 and an antenna 72. The transmitter receives the transmit signal output from the chirp generator 60, which is fed to the PA in each transmit block. The optional TX element sequencer (dashed line) generates multiple enable signals 64 that control the sequence of transmit elements. It is to be understood that DOA estimation can be implemented in radar systems with or without TX element sequencing and with or without MIMO operation. Furthermore, DOA estimation is not limited to implementations in MIMO FMCW radars, but can also be implemented using other types of radar systems.

[0057] Each receiver block includes an antenna 58, a low-noise amplifier (LNA) 50, a mixer 52, an intermediate frequency (IF) block 54, and an analog-to-digital converter (ADC) 56. Signal processing block 44 may include any suitable electronic device capable of processing, receiving, or transmitting data or instructions. For example, a processing unit may include one or more of the following: a microprocessor, a central processing unit (CPU), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a digital signal processor (DSP), a graphics processing unit (GPU), or a combination of such devices. As described herein, the term "processor" is intended to cover a single processing unit, multiple processors, processors, multiple processing units, or other appropriately configured computing elements.

[0058] For example, a processor may include one or more general-purpose CPU cores and optionally one or more dedicated cores (such as DSP cores, floating-point cores, gate arrays, etc.). One or more general-purpose cores execute general-purpose opcodes, while dedicated cores perform functions specific to their intent.

[0059] Attached or embedded memory includes dynamic random access memory (DRAM) or extended data output (EDO) memory, or other types of memory such as ROM, static RAM, flash memory, non-volatile static random access memory (NVSRAM), removable memory, bubble memory, or any combination thereof. Memory stores electronic data that can be used by the device. For example, memory can store electrical data or content such as, for example, radar-related data, audio and video files, documents and applications, device settings and user preferences, timing and control signals, or data from various modules, data structures, or databases. Memory can be configured as any type of memory.

[0060] The transmitted and received signals are mixed (i.e., multiplied) to generate a signal to be processed by the signal processing unit 44. The multiplication process generates two signals: one signal whose phase is equal to the difference between the multiplied signals, and the other signal whose phase is equal to the sum of their phases. The sum signal is filtered out, and the difference signal is processed by the signal processing unit. The signal processing unit performs all necessary processing on the received digital signals and also controls the transmitted signals. Several functions performed by the signal processing unit include determining the coarse range, velocity (i.e., Doppler), fine range, elevation angle, azimuth angle, performing interference detection, mitigation and avoidance, and performing Simultaneous Localization and Mapping (SLAM), etc.

[0061] A block diagram of an example digital radar processor IC of the present invention is shown as follows. Figure 5As shown. The radar processor IC (typically referred to as 390) includes several chip service functions 392 (including temperature sensor circuitry 396, watchdog timer 398, power-on reset (POR) circuitry 400, etc.), a PLL system 394 including power domain circuitry 402, a radar processing unit (RPU) 404 including a parallel FFT engine 406, a data analyzer circuitry 408, a direct memory access (DMA) circuitry 410 and a DOA estimation / LRA mechanism 411, a CPU block 412 including a TX / RX control block 414, a security core block 418 and L1 and L2 cache memory circuitry 424 and a DOA estimation / LRA mechanism 425, a memory system 426, and an interface (I / F) circuitry 428. In one embodiment, the RPU is configured to implement direction of arrival (DOA) estimation in the RPU or in the CPU, or in part in both, using the low-rank approximation (LRA) mechanism of the present invention.

[0062] The TX / RX control circuit 414 may include a setup time control, inter-interference, detection, mitigation, and avoidance block 416 for eliminating frequency source setup time. The safety core block 418 includes a system watchdog timer circuit system 420 and an RFBIST circuit, adapted to perform continuous testing of RF components in the radar system. The I / F circuitry includes interfaces for radar output data 430, TX control 432, RX control 434, external memory 436, and the RF clock 438.

[0063] Note that, depending on the specific implementation, the digital radar processor circuitry 390 may be implemented on a single silicon chip or across several integrated circuits. Similarly, depending on the specific implementation, the transmitter and receiver circuitry may be implemented on a single IC or across several ICs.

[0064] In one embodiment, the DRP 390 is used in a vehicle radar-based FMCW MIMO system. Such systems require multiple transmitter and receiver channels to achieve the desired range, azimuth, elevation, and speed. A higher number of channels results in better resolution performance. Depending on the implementation, multiple transmit channels may be combined into a single chip, and multiple receive channels may be combined into a single chip. The system may include multiple TX and RX chips. Each TX and RX chip can operate as part of a larger system adapted to achieve maximum system performance. In one embodiment, the system also includes at least one control channel. The control channel is operable to configure the TX and RX devices.

[0065] This invention provides a compact radar switch array antenna with high azimuth and elevation resolution and accuracy, and increased effective aperture, while using a small number of TX and RX elements. This invention also provides a compact radar antenna array with high azimuth and elevation resolution and accuracy, and increased effective aperture, while reducing unwanted sidelobes.

[0066] One embodiment of the present invention relates to a method for increasing the effective aperture of a radar switch / MIMO antenna array using a small number of transmit and receive array elements. An array of physical radar receive / transmit elements is arranged in at least two opposing RX rows and at least two opposing TX columns, such that each row includes a plurality of receive elements evenly spaced apart from each other, and each column includes a plurality of transmit elements evenly spaced apart from each other, the array forming a rectangular physical aperture.

[0067] Used as a switch array, the first TX element from one column is activated to transmit a radar pulse during a predetermined time slot. The reflection of the first transmission is received by all RX elements, thereby virtually replicating two opposing RX rows near the origin determined by the location of the first TX element within the rectangular physical aperture.

[0068] During different time slots, this process is repeated for all remaining TX elements, thereby virtually replicating two opposing RX lines near the origin, determined by the location of each activated TX element within the rectangular physical aperture. During each time slot, reflections of the transmission from each TX element are received by all RX elements. In this way, a rectangular virtual aperture with a size twice that of the rectangular physical aperture is achieved using the two replicated opposing RX lines. This virtual aperture determines the radar beamwidth and sidelobes.

[0069] Note that the above replication method works equally well in MIMO or hybrid switch / MIMO designs, where some signals are transmitted simultaneously by multiple TX array elements using orthogonal waveforms that are later separated in the receiver.

[0070] Direction of Arrival (DOA) Estimation

[0071] Note that the DOA estimation / LRA mechanism of this invention is applicable to a variety of radar types and is not intended to be limited to the example radar systems disclosed herein. For example, LRA beamforming is applicable to radars comprising a uniform linear array (ULA) where all antenna sensors are located on a single line and the distance between any two adjacent sensors is the same. MIMO FMCW radar is presented herein, for example only to illustrate the principle of the DOA estimation mechanism of this invention.

[0072] Digital beamforming (DBF) is a well-known technique for determining the direction of arrival (DOA) of a target. An array antenna with multiple antenna elements is used to receive reflected waves from a target. The target's direction is determined by applying a DOA estimation method, such as the well-known beamforming method.

[0073] In direction-of-arrival estimation methods using, for example, array antennas, beamforming methods scan the main lobe of the array antenna in many directions and determine the direction of arrival as the direction of maximum output power. Note that the width of the main lobe determines the angular resolution. Therefore, if it is desirable to improve the resolution so that the directions of multiple targets can be determined, the aperture length of the array is preferably increased by increasing the number of antenna elements. The same applies to minimum norm methods for determining the direction of arrival from the eigenvalues ​​and eigenvectors of the correlation matrix of the received signal from the array, and their extended algorithms (such as Multi-Signal Classification (MUSIC) and Estimation of Signal Parameters via Rotation Invariance Techniques (ESPRIT)). Considering these techniques, since the degree of correlation matrix (i.e., the number of antenna elements) determines the number of targets that can be detected, the number of antenna elements is preferably increased so that the directions of many targets can be determined.

[0074] Typically, during radar calibration, the beamforming matrix is ​​calculated, and during DOA estimation, this matrix is ​​multiplied by the array response vector. The DBF matrix is ​​composed of the calibration vector associated with the DOA. b 1,…, b The expression is composed of N, where N is the number of DOA angles to be estimated (i.e., the number of angles to be scanned in azimuth and / or elevation). b i ∈ C P ,and P This refers to the number of antenna elements. Therefore, the DBF matrix... B It can be written as follows

[0075] make x For the array response, such as in a specific Doppler chamber, the DOA estimate can then be expressed as follows:

[0076] in And | y i | 2 Indicates direction i Received energy, This represents data on a virtual array row or column that has typically (but not necessarily) undergone range and Doppler processing, and PIndicates the length of the rows or columns of the virtual array. Note that... y It is the result of beamforming, which is a measure of how much energy is reflected from each direction. y Each element in the array corresponds to a different direction (i.e., an angle). It can be considered as input data. x The spectrum.

[0077] For the special case of an ideal uniform linear array (ULA), the calibration vector is the corresponding steering vector. The matched filter. Therefore, The steering vector here has the following form

[0078] in d λ is the distance between the antenna elements, and λ is the signal wavelength. In this particular case, the standard Fast Fourier Transform (FFT) operation can be used for DOA estimation as a matched filter for the specific angle to be scanned, as provided in equation (4) below.

[0079]

[0080] in N P ,in N This indicates the length of the FFT, and P Indicates the number of receiving antenna elements. Value N Usually more P The largest and smallest powers of 2. The result of this FFT operation is the well-known sinc response for each direction in the scan. Note that in practice, a window function is typically used before the FFT operation to reduce the sidelobe level (SLL) of the sinc response.

[0081] With DBF matrix B Compared to multiplication, one advantage of using FFT is reduced computational complexity. General matrix multiplication requires... N 2 This operation only requires N log n operations, while FFT computation only requires N log n operations. N This operation is for antennas with a relatively high number of components (e.g., ...). P For high-resolution radars with a range of Doppler cells (≈100), the difference in computational complexity is significant and poses a major challenge for real-time implementations. However, a problem arises: low-complexity FFTs are only effective for ideally uniform arrays without any loss of performance.

[0082] Therefore, for an ideal uniform linear antenna array, a relatively simple FFT operation can be used, with Nlog for DOA estimation. N This operation is called a secondary operation. However, in the real world, when the antenna patterns of different components are not the same due to manufacturing tolerances and other influences, different complex correction vectors are used for each DOA to correct for antenna defects as much as possible; that is, using a digital beamforming (DBF) matrix. This operation requires... N 2 This operation (i.e., matrix multiplication), and in the case of high-resolution radar, where N There could be more than 100 elements, requiring this operation for each range of Doppler (twice in the case of azimuth and elevation). This fact makes standard calculations infeasible for such radar systems.

[0083] Low-rank approximation (LRA) of DBF DOA estimation

[0084] Solutions to the above problems include applying additional computational power, reducing the frame rate, and / or reducing the number of range Doppler bins. Alternatively, a low-rank approximation-based DOA mechanism can be used, which attempts to address these problems while requiring relatively low computational load and not compromising other parameters. Furthermore, an LRA-based DOA mechanism achieves SLL comparable to full DBF computation.

[0085] In practice, when ULA is used (i.e., not necessarily densely), even when considering damaged arrays, the DBF matrix is ​​similar to an FFT matrix with additional constant phase gain calibration. Therefore, it can be written as

[0086] in B It is a DBF matrix. F is an FFT or DFT matrix, and w is the constant phase gain calibration and window. Note that the DBF matrix B can be derived using the techniques described below or using any other known method. Also note that equation (5) represents an FFT matrix with distortion to illustrate the type of estimator that the mechanism of this invention attempts to determine. Calculation B The actual method is not crucial to this invention, because regardless of the calculation... B (If it is sufficiently similar to an FFT matrix), this mechanism is approximated as... B Matrix F is relative to the beamforming matrix. B Operation generated N The spatial frequency (DOA) corresponding to each direction is calculated. Without loss of generality, equation (5) can be expressed as follows:

[0087] in I It is the identity matrix, and ∈ corresponds to the distortion of the array with respect to the ideal ULA.

[0088] The diagram illustrates a high-level block diagram of an example low-rank approximation for DBF DOA estimation, as shown below. Figure 6 As shown in the diagram, the LRA circuit (typically referenced as 120) includes multiple pre-multipliers 122, an FFT computation block 124, a post-multiplier 126, and an adder 128. A diagram illustrating an example method for DOA estimation based on the low-rank approximation is shown below. Figure 7 As shown in the image.

[0089] The LRA approximation method of this invention utilizes B Similar to an FFT matrix F This is a fact. In one embodiment, it uses parallel processing. K 124 FFT machines, of which K It is an approximate rank. The antenna array response before each FFT operation... x 121 is multiplied by the multiplier 122 by what is called the pre-coefficient. V k The set of coefficients (step 150). Then FFT on the multiplication result X V Each multiplication result X in V The process is executed (step 152). Similarly, after each FFT operation, each FFT module is multiplied by a factor called the post-factor via multiplier 126. U k Different sets of coefficients (step 154). FFT operation K Result U F (X) V The sums are then combined via adder 128 (step 156) to generate the output of the approximate DOA estimation method. y 129 (Step 158), where ' The ' operator represents the well-known Hadamard product, or element-wise multiplication.

[0090] therefore, Figure 6 The diagram illustrates matrix multiplication of DBF matrices. K Rank approximation. This approximation assumes that even with a non-ideal antenna array, it approximates the FFT matrix, but with some deviation. In one embodiment, from the distortion matrix... C = B / FThe singular value decomposition (i.e., element-wise or Hadamard division) is used to compute the values ​​of the calibration coefficients before and after the FFT operation for each rank, where B It is a DBF matrix, and F It is an ideal FFT matrix. This architecture is implemented. K N log2 N This operation, for low-rank cases, is significantly less than the full matrix multiplication required. N 2 This operation is performed in practice. It was found that a rank of 4 is sufficient to compensate for 3D phase center misalignment and weak leakage between the antenna channel. Therefore, the four largest values ​​are used, and the remaining values ​​are set to zero. This is the optimal low-rank approximation of the distortion matrix C.

[0091] Note that, for K = P The approximation is perfect, and any desired DBF matrix can be implemented. However, this is computationally even more intensive than the matrix multiplication described above. In one embodiment, one can choose... K ≤ P At the same time, it still achieves a highly accurate approximation. For an ideal ULA antenna, K =1 is sufficient and is in fact equivalent to a single FFT operation of multiplying the input and output by a constant vector. In the example embodiment, the value 4 is chosen for... K This yielded a satisfactory result (i.e., four largest singular values). This is as follows: Figure 6 As shown, the example circuit uses four pre-multipliers, four FFT computation blocks, and four post-multipliers. It should be understood that the LRA mechanism of this invention can be implemented using any desired rank depending on the specific application.

[0092] It is important to note that in high-resolution radar, the amount of data to be processed is typically enormous. In one embodiment, a matched filter (i.e., maximum likelihood (ML)) is used to estimate the DOA. Other techniques are generally more computationally intensive. In the case of a uniform linear array (ULA), the matched filter becomes an FFT, particularly for ULAs. Note that due to the computational efficiency of the FFT, using any other alternative method would likely require significantly more computation.

[0093] For example, considering a radar with 256 range bins, 1024 Doppler bins would generate a total of 256 range bins. 1024 = 262,144 range Doppler binaries N =128 azimuth angle cells and M=32 elevation angle cells. In the ideal ULA case, ML spatial processing (i.e., azimuth and elevation angles) can be performed via FFT, with a complexity of O(n). N x M xlog2( N x M )≈50e3. However, for non-ULA arrays, full matrix multiplication is required, resulting in a complexity of O(n^2). N x M x( N + M The value is approximately 655e3, which adds a factor greater than 13. Considering that spatial processing is performed for each range of Doppler cells, full matrix multiplication requires approximately 158 billion computations per CPI. However, such a high number of computations is impractical in low-cost consumer radar sensors.

[0094] Existing radars on the market typically have relatively small array sizes, such as 3x4, 6x8, and 12x16. For radars with small arrays, estimating the processing load of DOA is feasible and can even be performed in software. However, even with an array as small as 12x16, the number of computations (i.e., multiplications) becomes prohibitively high, requiring a hardware solution. Therefore, using... N 2 Existing techniques for computational complexity are impractical for radars with larger array sizes (such as 48x48). The LRA method described in this paper (with a complexity of O(n)) is... K N log2 N It is closer to the 'pure' or 'full' FFT DBF mechanism in terms of efficiency, but the resulting performance is close enough to the full matrix multiplication method.

[0095] To overcome these problems, in one embodiment, the DOA estimation mechanism of the present invention uses a maximum likelihood estimation (MLE) matched filter, which is non-ideal. The MLE matched filter provides an index of the amount of energy from a particular direction. b 'Generated as shown in equation (1), specifically for a particular azimuth or elevation angle, such as 25 degrees. These can be considered as FFT coefficients, which are complex numbers corresponding to certain Fourier frequencies. These coefficients actually measure spatial frequencies. With respect to the radar antenna, spatial frequencies infer the direction in the phased array. This is a method for calculating the amount of energy received from a 25-degree direction. This can be applied to all desired DOA angles to be scanned.' X 1… X P(That is, a linear combination of objectives) is repeated. This standard model is used for phased arrays and is referred to as the steering vector. The matched filter is precisely the complex conjugate.

[0096] Each vector 'b' corresponds to 1. ... N In each direction. As previously shown, using a large number of virtual antennas (e.g., 128) results in an excessively large number of calculations. Radar resolution is related to aperture size. Radars with larger spacing between elements have larger apertures, which means narrower beamwidths and thus better resolution. However, regardless of aperture size, the number of DOA calculations required is related to the number of elements.

[0097] It is important to note that even though the radar does not transmit simultaneously from all elements like existing phased array radars, the mathematical process is the same. For example, signals transmitted simultaneously from multiple antennas are combined in the air. In DBF, signals received by multiple receiving antenna elements are digitally summed. Mathematically, this represents the same thing. Nevertheless, it is important to note that the techniques described in this invention are applicable to any DBF setup, whether it is TD-MIMO, simultaneous transmission MIMO (such as OFDM), or even a full ULA with a receiver having a single transmitter.

[0098] In one embodiment, the receiving antenna elements are organized as a ULA. For each direction, there is a different set of coefficients. In a standard beamforming scheme, utilizing an ideal ULA, this matrix becomes the matrix used for a specific { k The discrete Fourier transform (DFT) matrix of the given set of angles is calculated. The set of angles is chosen, and the given set of frequencies is computed very efficiently. In the FFT, symmetries are expected to be utilized, such as even and odd, positive and negative, etc. Matched filters are computed for the specific set of frequencies. DOA estimates are typically calculated for the specific set of frequencies. Therefore, in the case of ULA and a selected set of DOAs, standard FFT operations can be used.

[0099] In one embodiment, the data is multiplied by a calibration window vector, and then the DFT is performed as a matrix multiplication. This forms the system's output before SLAM.

[0100] Therefore, in general, the expected number of calculations B X But it is usually O( N 2 Order 1) requires too much computation. Conversely, the number of... B X Use the FFT operation to approximate as described above. Note that this is an approximation.B The equation used for calculation is not the FFT matrix, but one that is close enough. The distortion matrix is ​​calculated using... B Calculated B It is calculated using any expected, well-known techniques. Assuming... B Approximates FFT, and the distortion matrix uses B Then the singular value decomposition (SVD) is calculated to determine the prefix and postfix coefficients.

[0101] In one embodiment, the DOA estimation mechanism can be elegantly and efficiently implemented in hardware, wherein X This represents a virtual element array. After range / Doppler processing, the DOA estimate is calculated. For each range, Doppler, and row in the virtual array, azimuth processing is performed, representing 256. 1024 128 = 33.5 million calculations. If the elevation angle also needs to be estimated, assuming the azimuth and elevation angles have the same resolution, an additional 33.5 million calculations are required. Note that different coefficients and different ranks can be used for the elevation angle, because the ranks used for the azimuth and elevation angles do not need to be the same and can be different. The higher the rank, the better the approximation. The rank and other relevant parameters can be dynamically programmed and selected.

[0102] Beamforming matrix B The determination

[0103] In the following section, we disclose a method for using array response matrices. A Inverse measurement of beamforming matrix B The method.

[0104] To recap, the beamforming matrix is ​​defined by the following relationship:

[0105] in And | y n | 2 Indicates direction n ∈[1, N The energy received. After range and Doppler processing, This typically (but not necessarily) represents complex-valued data present in rows or columns of a virtual array, and P Indicates the length of the rows or columns of the virtual array.

[0106] Although usually each direction N It can represent any arbitrary angle of choice, but in Fourier beamforming (or beamforming closely related to Fourier beamforming, such as in the case of this invention),N The possible angle values ​​are determined by the input wavelength. λ Array spacing d and FFT length N FFT Confirmed, as shown below:

[0107] Notice, N FFT > P And it is usually a power of 2 (due to FFT efficiency), while P Determined by the range of virtual rows or columns, and not typically considered a power of 2. We will discuss this with... N FFT = P The angles corresponding to special cases are called natural angles or orthogonal angles, because they correspond to lengths of... P The orthogonal (spatial) frequencies of the DFT.

[0108] The array response matrix is ​​defined by the reciprocal relation of equation (7) as follows:

[0109] in This represents the complex backscattering distribution of the environment (usually, but not necessarily, at the range Doppler chamber). Indicates the angle of arrival θ n The complex backscattering at that point n ∈[1, N ], w It is a constant phase gain calibration and windowing vector, and diag( w ) -1 having on the diagonal w The inverse of the diagonal matrix of elements (and also a diagonal matrix). Note that... It is a property of the environment (i.e., what exists in the environment), and it is different in principle from... y , y It is the environment perceived via beamforming equation (7). We multiply by diag( w And using constant phase gain calibration data. The definition rewrites equation (9) as follows:

[0110] In index notation, the first... p The elements are given below.

[0111] Note that A is P ×N Via.

[0112] If we construct a controlled measurement setup (e.g., in an anechoic chamber, or an outdoor setup with minimal clutter), such that the environment only contains RCS... R A single point target (which we set to one without losing generality), and at an angle q At position 0, then:

[0113] in It is the Kroenecker delta function.

[0114] Then we have:

[0115] Note the symbol. This represents the p-th element of the experiment, where the target is at an angle. q At position 0, that is, using this setting, the data on the virtual array We are given a single column of the array response matrix By repeating the experiment from different angles, we can, in principle, retrieve... A All columns of the matrix.

[0116] Note that the required setup equation (12) is idealized and requires an infinite signal-to-noise ratio (SNR), but for practical purposes, it can be achieved in a typical anechoic chamber within a ~3 dB field of view (FOV) around the line of sight. As we move further away from this FOV, the SNR deteriorates to the point where equation (12) no longer holds. The maximum angle at which equation (12) holds is called the angle at which the SNR is highest. We label the set of angles within this FOV as... .

[0117]

[0118] We define the total number of angles measured within this FOV as Q Therefore, the measured array response A measured yes P × Q Via.

[0119]

[0120] These Q Each measured angle is not constrained to belong to any angle grid. We continue to interpolate the rows of the measured array response to the array length. P Defined angle grid, that is, we use fromQ From one angle Resampling to a set using simple interpolation (16) The size (i.e., the number of angles within the FOV defined on the natural grid) is defined as Note the angle. This corresponds to the null value of the sinc array response. Therefore, after this interpolation, we obtain a dimension of of .

[0121]

[0122] In order to complete arrive P × P The matrix is ​​formed by summing the ideal array response vectors of all array null angles outside the FOV; that is, we add the columns to the matrix. The left and right sides make

[0123] Among them, v p It is the first ideal array response (Fourier matrix) on a grid of natural angles. p OK.

[0124] We assume that, like the beamforming matrix B In that way, A With the inverse Fourier transform matrix Slightly different:

[0125] therefore,

[0126] This completes the inversion, because

[0127] Note that this produced a result. And we want to be in The result is as follows. Therefore, we obtain the result in the Fourier plane via... Perform zero-filling.

[0128]

[0129] Calibration coefficient calculation

[0130] The diagram illustrates an example method for calculating the pre-coefficient and post-coefficient. Figure 8As shown in the diagram. In one embodiment, it is assumed that the DBF matrix is ​​similar to the FFT matrix, and the coefficients in the LRA architecture are computed in an optimal manner. To do this, the residual matrix or distortion matrix is ​​first calculated below. C The calculation (step 130) is as follows:

[0131] The division operation is performed element-wise. .in addition, C and They have the same dimensions. Note that in the case of an ideal ULA, C All elements of it are 1, and it also has rank 1. Then C The singular value decomposition is calculated (step 132) as follows to generate V (Step 134) and U (Step 136), that is, Therefore, for the distortion matrix C The decomposition produces three matrices.

[0132]

[0133] in , ,and It is by C A diagonal matrix composed of the singular values. A diagonal matrix representing singular values. H Indicates transpose. N This indicates the length of the Fourier transform, and P Indicates the number of virtual array elements in the processing direction (e.g., non-virtual or virtual effective array rows or columns). Note that... This includes diagonal matrices containing singular values, similar to the eigenvalues ​​of non-rectangular matrices. Finally, the rank in the LRA mechanism... k The prefix and suffix use the first k Use singular values ​​to calculate

[0134] The result of this operation is the approximate residual matrix given below (step 138).

[0135] in ,and (Step 140). Therefore, also with B They have the same dimensions. C K Having rank KFurthermore, according to the matrix approximation lemma or the Eckart-Young-Mirsky theorem, it is relative to the well-known Frobenius method and the spectral norm approximation residual matrix. C optimal rank K matrix.

[0136] To show that the coefficients of equations (27) and (28) are Figure 6 The coefficients in the architecture and circuitry are compared with existing DBF-based and LRA-based DOA estimation operations. On the one hand, the DBF method obtains DOA estimation results from the antenna array response. x Calculation for scanning direction θ i The following DOA estimates y i As shown below

[0137] Then we use C K approximate C to obtain

[0138] Inserting equation (31) into equation (30) produces the following result

[0139] therefore,

[0140] in' The ' operator represents the well-known Adamant product or element-wise multiplication. Note that equation (33) and Figure 6 The operations performed in the circuit are exactly the same. Specifically, diag ( V )· x It is the product of each element and the calibration pre-factor. F · diag ( V )· x Apply FFT to the results. It is an element-wise multiplication of the post-calibration value after FFT. exist k The sum of the above is k The sum of all ranks, for example, is 4 in the example disclosed herein.

[0141] Any arrangement of components used to achieve the same functionality is effectively “associated” so that the desired functionality is achieved. Therefore, any two components combined in this paper to achieve a specific functionality can be considered “associated” with each other so that the desired functionality is achieved, regardless of the architecture or intermediate components. Similarly, any two such associated components can also be considered “operably connected” or “operably coupled” with each other to achieve the desired functionality.

[0142] Furthermore, those skilled in the art will recognize that the boundaries between the above operations are merely illustrative. Multiple operations may be combined into a single operation, a single operation may be distributed among additional operations, and operations may be performed at least partially overlapping in time. Additionally, alternative embodiments may include multiple instances of a particular operation, and the order of operations may be changed in various other embodiments.

[0143] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used herein, the singular forms “a,” “an,” and “the” are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that, when used in this specification, the terms “comprising” and / or “including” specify the presence of the stated features, integers, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0144] In the claims, any reference numerals placed between parentheses should not be construed as limiting the claims. The use of introductory phrases such as “at least one” and “one or more” in the claims should not be construed as meaning that the introduction of another claim element by the indefinite article “a” or “an” limits any particular claim containing such an introduced claim element to an invention containing only one such element, even if the same claim includes the introductory phrases “one or more” or “at least one” and indefinite articles (such as “a” or “an”). The same applies to the use of definite articles. Unless otherwise stated, terms such as “first” and “second” are used to arbitrarily distinguish the elements described by such terms. Therefore, these terms are not necessarily intended to indicate the time or other priority of such element. The fact that certain measures are recited in mutually different claims does not indicate that a combination of these measures cannot be used advantageously.

[0145] All components or steps in the following claims, along with their corresponding structures, materials, actions, and equivalents of functional elements, are intended to encompass any structure, material, or action for performing a function in combination with other claimed elements of the specific claim. The description of the invention has been presented for illustrative and descriptive purposes, but is not intended to be exhaustive or to limit the invention to the forms disclosed. Since many modifications and alterations will readily occur to those skilled in the art, it is intended that the invention not be limited to the limited number of embodiments described herein. Therefore, it is to be understood that all suitable variations, modifications, and equivalents can be invoked within the spirit and scope of the invention. The embodiments have been chosen and described in order to best explain the principles and practical application of the invention and to enable others of ordinary skill in the art to understand the invention with respect to various embodiments having various modifications suitable for the particular intended use.

Claims

1. A method for estimating the direction of arrival (DOA) of a signal used in a radar system, comprising: The antenna array response vector is used as input data. The input data is multiplied element by element by a set of pre-coefficients to produce a first plurality of results; Multiple Fast Fourier Transform (FFT) operations are performed on the first plurality of results to generate a second plurality of results, wherein the FFT operations are performed in the spatial domain on the channel spacing of the antenna array response; The second plurality of results are multiplied element-wise by the set of postfix coefficients of the number of factors to produce a third plurality of results; The third set of results are summed to produce an approximate DOA estimate; and Wherein, the quantity is the rank of the approximation.

2. The method according to claim 1, further comprising: Different approximate DOA estimates are calculated for each desired azimuth and / or elevation direction angle to be scanned.

3. The method according to claim 1, wherein, The preceding and following coefficients are calculated from the singular value decomposition of the distortion matrix.

4. The method according to claim 3, wherein, The distortion matrix is ​​determined by taking the inverse of the array response.

5. The method according to claim 4, wherein, The array response is determined by performing a set of measurements on a single target at multiple angles in a controlled environment.

6. A method for estimating the direction of arrival (DOA) of a signal used in a radar system, comprising: Receive antenna array response vector as input data x ; The input data x Each element multiplied by the preceding coefficient V k of k A set to produce k First result diag ( V k ) x ; Regarding the k The first result is executed. k Fast Fourier Transform operation F To generate k A second result F diag ( V k ) x ; The k The second result is multiplied element-wise by the postfix coefficient. U k of k A set to produce k A third result U k ( F diag ( V k ) x ); Regarding the k Summing the third results to produce an approximate DOA estimate y ; as well as Wherein, the value k It is the rank of the approximation.

7. The method according to claim 6, further comprising: Different approximate DOA estimates are calculated for each desired azimuth and / or elevation direction angle to be scanned. y .

8. The method according to claim 6, wherein, The approximate rank k The DOA estimation is different for azimuth and elevation angles.

9. The method according to claim 6, wherein, From the distortion matrix The singular value decomposition is used to calculate the preceding coefficients. V k and the post-coefficient U k ,in , , , , yes C The diagonal matrix of singular values, H Indicates transpose. N This indicates the length of the Fast Fourier Transform, and P This indicates the number of virtual array elements in the processing direction.

10. The method according to claim 9, wherein, The distortion matrix is ​​determined by taking the inverse of the array response.

11. The method according to claim 10, wherein, The array response is determined by performing a set of measurements on a single target at multiple angles in a controlled environment.

12. An apparatus for estimating the direction of arrival (DOA) of a signal in a radar system, comprising: Radar signal processing circuitry, which operates to receive the response of the receiving antenna array. x ; The radar signal processing circuit operates as follows: Multiply each element of the antenna array response by a pre-coefficient. V k of k A set to produce k First result diag ( V k ) x ; Regarding the k The first result is executed. k Fast Fourier Transform operation F To generate k A second result F diag ( V k ) x ; The k The second result is multiplied element-wise by the postfix coefficient. U k of k A set to produce k A third result U k ( F diag ( V k ) x ); Regarding the k Summing the third results to produce an approximate DOA estimate y ; as well as Wherein, the value k It is the rank of the approximation.

13. The apparatus according to claim 12, wherein, The radar signal processing circuitry operates to calculate different approximate DOA estimates for each desired azimuth and / or elevation azimuth angle to be scanned. y .

14. The apparatus according to claim 12, wherein, The approximate rank k The DOA estimation is different for azimuth and elevation angles.

15. The apparatus according to claim 12, wherein, From the distortion matrix The singular value decomposition is used to calculate the preceding coefficients. V k and the post-coefficient U k ,in , , , , yes C The diagonal matrix of singular values, H Indicates transpose. N This indicates the length of the Fast Fourier Transform, and P This indicates the number of virtual array elements in the processing direction.

16. The apparatus according to claim 15, wherein, The distortion matrix is ​​determined by taking the inverse of the array response.

17. The apparatus according to claim 16, wherein, The array response is determined by performing a set of measurements on a single target at multiple angles in a controlled environment.

18. A motor vehicle radar sensor, comprising: Printed circuit board (PCB) assemblies, including: Multiple transmitting antennas are fabricated on one side of the PCB assembly; Multiple receiving antennas are fabricated on opposite sides of the PCB assembly; and A transceiver coupled to the plurality of transmitting antennas and the plurality of receiving antennas, the transceiver operating to generate and supply the transmitting signals to the plurality of transmitting antennas, and to receive signals of waves reflected back to the plurality of receiving antennas; Radar signal processing circuitry, which is coupled to the transceiver and operates to: Receive antenna array response vector as input data x ; The input data x Each element multiplied by the preceding coefficient V k of k A set to produce k First result diag ( V k ) x ; Regarding the k Perform k Fast Fourier Transform operations on the first result. F To generate k A second result F diag ( V k ) x ; The k The second result is multiplied element-wise by the postfix coefficient. U k of k A set to produce k A third result U k ( F diag ( V k ) x ); Regarding the k Summing the third results to produce an approximate DOA estimate y ;as well as Wherein, the value k It is the rank of the approximation.

19. The sensor according to claim 18, wherein, From the distortion matrix The singular value decomposition is used to calculate the preceding coefficients. V k and the post-coefficient U k ,in , , , , yes C The diagonal matrix of singular values, H Indicates transpose. N This indicates the length of the Fast Fourier Transform, and P This indicates the number of virtual array elements in the processing direction.

Images