Method, apparatus and radar system

Abstract

The invention relates to a method, an apparatus and a radar system. A method is presented that includes obtaining an IQ representation of radar data indicative of a received radar signal, determining an approximation of the IQ representation, the approximation having a spiral configuration, and determining a phase of the received radar signal based on the approximation.

Description

Technical Field

[0001] The present disclosure relates to phase determination based on IQ data. Examples relate to methods, apparatuses, and radar systems. Background Art

[0002] Conventional phase determination techniques may not be able to accurately compensate for the spiral distortion of IQ data in radar measurements. This may lead to incorrect measurement results. Therefore, improved phase determination may be required. Summary of the Invention

[0003] The subject matter of the independent claims meets this requirement.

[0004] Some aspects of the present disclosure relate to a method that includes obtaining an IQ representation of radar data indicative of a received radar signal, determining an approximation of the IQ representation that has a spiral configuration, and determining the phase of the received radar signal based on the approximation.

[0005] Some aspects of the present disclosure relate to an apparatus that includes processing circuitry configured to obtain an IQ representation of radar data indicative of a received radar signal, determine an approximation of the IQ representation that has a spiral configuration, and determine the phase of the received radar signal based on the approximation. Brief Description of the Drawings

[0006] Some examples of apparatuses and / or methods will be described hereinafter only by way of example and with reference to the drawings, in which:

[0007] Figure 1 Examples of apparatuses are illustrated;

[0008] Figures 2a - 2c An I-Q diagram illustrating an example of an IQ representation of a received radar signal;

[0009] Figure 3a and Figure 3b An I-Q diagram illustrating an example of a center estimate of an IQ representation of a received radar signal;

[0010] Figure 4 An I-Q diagram illustrating an example of a logarithmic spiral function fitted to an IQ representation of a received radar signal;

[0011] Figure 5 An I-Q diagram illustrating an example of a general spiral function fitted to an IQ representation of a received radar signal;

[0012] Figure 6 An I-Q diagram illustrating an example of a cluster of IQ representations of a received radar signal generated by manifold learning;

[0013] Figure 7An I-Q diagram showing an example of a segment of a circle fitted to a corresponding part of the IQ representation of a received radar signal;

[0014] Figure 8 An I-Q diagram showing an example of a polynomial function fitted to the IQ representation of a received radar signal;

[0015] Figure 9 An example of a radar system is shown; and

[0016] Figure 10 An example of a method is shown. DETAILED DESCRIPTION

[0017] Some examples will now be described in more detail with reference to the accompanying drawings. However, other possible examples are not limited to the features of these embodiments of the detailed description. Other examples may include modifications of the features as well as equivalents and alternatives of the features. In addition, the terms used herein to describe certain examples should not limit other possible examples.

[0018] Throughout the description of the drawings, the same or similar reference numerals refer to the same or similar elements and / or features, which may be the same or implemented in a modified form while providing the same or similar functions. For clarity, the thickness of lines, layers, and / or regions in the figures may also be exaggerated.

[0019] Unless otherwise explicitly specified in individual cases, when two elements A and B are combined using "or", this should be understood to discuss all possible combinations, namely only A, only B, and A and B. As an alternative wording for the same combination, "at least one of A and B" or "A and / or B" may be used. This also applies to combinations of more than two elements.

[0020] If a singular form such as "a", "an", "the", and "said" is used and only a single element is not explicitly or implicitly defined as mandatory, other examples may also use several elements to achieve the same function. If a function is described below as being achieved using multiple elements, other examples may use a single element or a single processing entity to achieve the same function. Further understanding, the terms "include", "including", "comprise", and / or "comprising", when used, describe the presence of the specified features, wholes, steps, operations, processes, elements, components, and / or groups thereof, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, processes, elements, components, and / or groups thereof.

[0021] Figure 1FIG. illustrates an example of apparatus 100. Apparatus 100 includes processing circuitry 110 and optional interface circuitry 120. In the presence of interface circuitry 120, interface circuitry 120 may be communicatively coupled (e.g., via a wired or wireless connection) to processing circuitry 110, e.g., for data exchange between interface circuitry 120 and processing circuitry 110.

[0022] Interface circuitry 120 may be any device or component for transmitting or exchanging data. For example, interface circuitry 120 may be a set of electronic components, circuits, and / or subsystems for interaction between different interface-connected entities (such as devices, systems, or components). It may include voltage level shifters, buffers, amplifiers, filters, converters, multiplexers, demultiplexers, and / or various other electronic elements.

[0023] Processing circuitry 110 may be, for example, a single dedicated processor, a single shared processor, or multiple individual processors (some or all of which may be shared), digital signal processor (DSP) hardware, an application specific integrated circuit (ASIC), a microcontroller, or a field programmable gate array (FPGA). Processing circuitry 110 may optionally be coupled to, for example, read only memory (ROM), random access memory (RAM), and / or non-volatile memory for storing software.

[0024] Apparatus 100 will be considered in the context of a radar sensor. For example, apparatus 100 may be partially or fully integrated into a radar sensor. Alternatively, apparatus 100 may be external to the radar sensor. This will be explained below with reference to a radar system and Figure 9 will be explained.

[0025] The radar sensor can be any type of radar sensor, such as a Doppler radar sensor, a continuous wave radar sensor, etc. In some examples, the radar sensor can be a frequency modulated continuous wave (FMCW) radar sensor. The radar sensor is configured to transmit a radar signal and receive the reflection (echo) of the above radar signal, thereby generating a received radar signal. The radar sensor can obtain radar data, for example, by IQ or complex resampling of the received radar signal (e.g., baseband signal). Alternatively, the radar sensor can obtain real-valued radar data and perform a fast Fourier transform (FFT) or discrete Fourier transform (DFT) in the fast time of the radar data, thereby obtaining a complex IQ representation of the range spectrum. For example, for Doppler or CW (continuous wave) radar, complex raw data can be obtained, and the radar data can directly present a spiral shape. However, for FMCW, the raw data is not necessarily complex, that is, the received radar signal can be sampled without a Q channel. In this case, the IQ representation can be generated by applying a complex DFT or FFT to the range bins of the radar data. Optionally, the received radar signal can be downconverted to an intermediate frequency, and then the downconverted signal can be sampled.

[0026] IQ sampling (i.e., in-phase and quadrature phase sampling) is a demodulation technique that is used to capture the received radar signal as a complex signal to preserve phase information. The received radar signal is represented as two separate radar data streams: one radar data stream for the in-phase (I) component and one radar data stream for the quadrature phase (Q) component. IQ sampling involves sampling (digitizing) the in-phase (I) and quadrature phase (Q) components of the received radar signal, for example, by two synchronous analog-to-digital converters (ADCs). The I and Q components can be sampled with a specific phase shift between each other, such as a 90° phase shift. For example, one path of demodulation can be performed at the original phase position to obtain the I data, and a second path can be performed at a reference frequency with a phase shift of 90° to obtain the Q data. Each data point of the I data is associated with the corresponding Q data, thereby generating IQ data.

[0027] The processing circuitry 110 is configured to obtain an IQ representation 130 of the radar data. The IQ representation 130 of the radar data can be the above IQ data in its original form or data derived therefrom. For example, in the latter case, the IQ representation 130 can indicate complex (e.g., micro-Doppler) data at the target range bin derived from the IQ data. For example, the range or range bin can be derived from the raw data using Fourier transform and binning.

[0028] The interface circuit device 120 can receive, for example, an IQ representation 130 from a radar sensor. Alternatively, the processing circuit device 110 can derive the IQ representation 130 from the radar data, for example, by correlating the data points of the I data with the data points of the Q data based on their temporal relationship, or by deriving distances or distance intervals from the raw data. The IQ representation 130 represents data points in the IQ plane, i.e., having coordinates described by the I-axis and the Q-axis. The IQ representation 130 can include a plurality of (complex) data points or samples of the radar data or data derived therefrom, each sample indicating the respective I and Q values of the radar received signal at a certain point in time.

[0029] For an ideal continuous sine wave radar signal, the data points of the IQ representation 130 would be arranged along a circle around the origin of the IQ coordinate system. This is because the IQ representation 130 indicates the phasor of the received radar signal, i.e., projects the time-dependent radar signal onto the time-independent IQ coordinate system.

[0030] However, the IQ representation 130 may be affected by IQ imbalance (e.g., for Doppler or CW radar sensors). IQ imbalance refers to the imperfection or difference between the I-component and the Q-component of a complex radar signal, i.e., the I and Q components are misaligned. IQ imbalance may be caused by hardware defects of the antennas of the radar sensor receiving the radar signal, misalignment between the I-channel and the Q-channel, or other non-idealities. These IQ imbalances distort the radar data and thus disrupt signal extraction.

[0031] IQ imbalance can generally be classified into amplitude imbalance, phase imbalance, and offset, which shift the data points relative to the origin of the IQ coordinate system, compress the IQ data points into an ellipse, introduce asymmetry, or rotate the ellipse relative to its center (center point).

[0032] In the case of an FMCW radar sensor, distortions may occur due to the radar scenario, e.g., in the case of a static target in the same distance interval of the radar data of the FMCW radar sensor. These may mainly lead to an offset of the IQ representation.

[0033] In the current situation, an IQ representation of 130 may exhibit more complex distortions: the IQ representation of 130 may have or approximately have a spiral shape. That is, the data points of the IQ representation of 130 are arranged or approximately arranged along a spiral structure. The spiral shape can be a coil, a helix, or a winding curve extending outward or inward relative to a center point (or axis). The IQ representation of 130 may thus exhibit discontinuities, that is, between the starting point (e.g., the innermost point or inner end) and the ending point (e.g., the outermost point or outer end) of the winding of the spiral shape. Therefore, since the IQ representation of 130 has a non-closed shape, it cannot be adequately represented by a circle or an ellipse. For example, the starting point of the winding may be farther from the ending point than other data points on the winding. For example, a spiral shape may occur if the measured (received) radar signal does not wind like a corkscrew but increases / decreases its amplitude over time and reverses its direction at a certain point in time. That is, the power of the received radar signal may decrease and increase respectively as the distance between the antenna and the target increases and decreases.

[0034] An example of the IQ representation of 130 of radar data is as Figures 2a - 2c shown. Figures 2a - 2c The I-Q diagram 200 illustrates an example of the IQ representation 210 (here the raw IQ data) of three corresponding receive channels (RX 1, RX 2, and RX 3) of a radar sensor. Each IQ representation 210 is illustrated by a curve that connects the data points of the IQ representation 210 to each other in their chronological order. Figures 2a - 2c The IQ representation 210 in [diagram] has an approximately spiral shape, where one spiral arm extends over one complete winding. The winding can be formed by the spacing or angular interval between different consecutive parts of the spiral. According to the spiral model, the winding can be defined differently. For example, from the perspective of the spiral center, consecutive data points extending an angle of 360° may correspond to one winding.

[0035] The curve is arranged along the spiral shape and goes back and forth several times along the spiral arm. There are partial differences in the adjacent "circles" of the curve depicted by the spiral arm in terms of the extension along the spiral arm. Figures 2a - 2c The spiral shape of the example of the IQ representation 210 in [diagram] is elongated, that is, the radius of the spiral arm does not increase continuously from the starting point of the spiral but evolves in an elliptical shape. Figures 2a - 2c The IQ representations 210 in [diagram] are different from each other in their orientations. For example, Figure 2c the IQ representations 210 in [diagram] rotate relative to Figure 2a and Figure 2b the IQ representations 210 in [diagram] in terms of elongation, starting and ending points, unfolding between curve rounds, etc. In addition to Figures 2a - 2cIn other examples than the ones shown, the IQ representation can have any other spiral shape, e.g., having more windings (turns) or only a partially wound portion, more spiral arms, another orientation, etc.

[0036] For (e.g., very or ultra) near - field measurements, such complex distortions can occur due to (e.g., non - linear) near - field effects and / or high - amplitude variations for short displacements relative to the target. A near - field measurement can be a measurement of a radar sensor in its near - field, e.g., a measurement where the distance between the transmitting antenna and / or receiving antenna of the radar sensor and an object in the radar sensor's field of view is one wavelength or less of the radiation (of the radar signal). Thus, the radar sensor can be configured to transmit a radar signal into its near - field and receive the reflection of the above - mentioned radar signal from the near - field, thereby generating a received radar signal. For example, radar data can indicate such near - field measurements.

[0037] If the signal amplitude significantly decreases when the target displacement is half of the wavelength (λ) of the radiation of the radar sensor, a spiral shape may occur. This may be the case if the target is close to the radar sensor relative to the signal's frequency. Near - field measurements can additionally enhance the effect of the spiral - shaped IQ representation. Moreover, a low - reflectivity target or object may cause or amplify the spiral shape in the radar data. In this case, the radar data can indicate near - field measurements of a moving object (e.g., an object moving with an amplitude of (about) half of the radiation wavelength or less) and / or a low - reflectivity object. The spiral shape can be observed in the data points collected for a target moving along λ / 2 because this is the displacement of one winding or one (unit) circle in an ideal measurement (far - field). For shorter displacements, the winding will not be fully covered; larger displacements may result in more windings of the spiral. Alternatively, the radar data can indicate any other measurement that results in the spiral shape of the IQ representation 130.

[0038] The processing circuitry 110 is configured to determine an approximation 140 of the IQ representation 130. The approximation 140 has a spiral configuration (characteristic, i.e., typically a spiral shape). That is, the processing circuitry 110 can determine the approximation 140 such that it has the above - mentioned spiral configuration. The processing circuitry 110 can approximate the IQ representation 130 in the IQ dimension (plane), thereby obtaining the approximation 140. Thus, the approximation 140 can include data points in I and Q values or I and Q coordinates. The processing circuitry 110 can determine an approximation 140 of the overall (e.g., frame) of the IQ representation 130. In some examples, the approximation 140 can exclude the approximation of the received radar signal itself, or the approximation of the baseband signal, or the approximation of the phase values of the received radar signal.

[0039] Approximation 140 can be a simplified, estimated, idealized, fitted, or modified version of IQ representation 130, which retains the basic characteristics of IQ representation 130 in the form of a spiral configuration. Processing circuitry 110 can determine approximation 140 to minimize or reduce the distance between approximation 140 or its data points and the data points of IQ representation 130. The spiral configuration of approximation 140 can mean that approximation 140 has or indicates a spiral shape (e.g., in a mathematical sense), approximately represents a spiral shape, or otherwise utilizes the characteristics of the spiral shape of IQ representation 130: approximation 140 can be based on, for example, a mathematical model or function of a spiral shape.

[0040] Processing circuitry 110 is also configured to determine a phase 150 of the received radar signal based on approximation 140. Thus, phase 150 can be derived from approximation 140. Phase 150 can be an angular phase or a phase angle of the received radar signal. Phase 150 can indicate or quantify the angular displacement or phase difference between two points of approximation 140 in the IQ plane and can indicate the corresponding phase of IQ representation 130. Thus, phase 150 can be related to additional radar processing, such as object detection, motion detection, etc. Processing circuitry 110 can use any phase analysis technique to determine phase 150. Examples of how to extract phase 150 will be described further below.

[0041] Due to the distorted signal, directly deriving phase 150 from IQ representation 130 itself may be inaccurate or result in a large amount of computational work. Conventional calibration methods may not be able to balance the spiral distortion. In contrast, device 100 can compensate for the distortion by generating approximation 140 while considering the spiral shape of IQ representation 130, rather than the conventional complex signal representation that is circular or elliptical in the target distance range.

[0042] Examples of how to determine approximation 140 are at least one of fitting a spiral function to IQ representation 130, manifold learning, matching segments of a circle and / or an ellipse to corresponding parts of IQ representation 130, and polynomial approximation. These examples will be described in more detail below:

[0043] For example, approximation 140 can be based on a mathematical function (spiral function) that describes or reconstructs a spiral shape. Then, approximation 140 can be determined by fitting the spiral function to IQ representation 130. For example, the spiral shape can be mathematically described using a parametric equation that defines the points of the spiral as a function of one or more parameters (spiral function). Certain assumptions can be made about the type of spiral shape exhibited by IQ representation 130 when selecting the spiral function. For example, the spiral shape can be approximated using an Archimedean spiral, a Fermat spiral, a hyperbolic spiral, a logarithmic spiral, etc.

[0044] For example, for a logarithmic spiral (which is also an equiangular spiral), it is assumed that as the angle increases, the distance from the center point of the spiral increases exponentially. For the logarithmic spiral that serves as the basis of the spiral function, Equation 1 can be used as an example of the spiral function:

[0045] r(φ) = ae kφ Equation 1,

[0046] where r is the radius of the spiral related to the angle φ, a is a parameter used to determine the initial distance from the center point, and k is a parameter that affects the spiral expansion rate. The Cartesian format of Equation 1 can also be used as the spiral function.

[0047] As an alternative to a specific spiral function, a general spiral function can be used. That is, the spiral shape can be modeled by the cosine and sine functions of its I and Q coordinates respectively. The cosine and sine functions can be functions of the angle of the spiral. One or more parameters can be used to create the prefactors of the cosine and / or sine functions. Additionally or alternatively, one or more parameters can be used within the argument of the cosine or sine function. The one or more parameters used for the sine and cosine functions can correspond to each other or be different from each other. In the latter case, approximation 140 can have greater flexibility to create more complex and diverse patterns of the spiral shape. For example, different functions for the I and Q coordinates and / or different combinations of parameters for the I and Q coordinates can allow independent control of the I and Q coordinates. The number of parameters used can be adjusted according to the desired approximation accuracy and available computing resources.

[0048] An example of such a general spiral function is described by Equation 2:

[0049]

[0050] where s is the complex-valued point (I and Q value point) of the spiral on the IQ plane, p 1 to p 12 are the parameters of the spiral function, φ is the angle relative to the center point of the spiral, the first term on the right side of Equation 2 from p 1 to p 6 refers to the I value, and the second term on the right side of Equation 2 from p 7 to p 12 refers to the Q value. For the I and Q dimensions, the parameters p 1 to p 12 can independently control the growth rate of the spiral, the scaling of the oscillating radius of the spiral, and the offset of the spiral.

[0051] Then, the above fitting of the spiral function to the IQ representation 130 can be performed based on any fitting method. For a parametric spiral function, fitting the spiral function to the IQ representation 130 can include finding the parameters that fit the resulting spiral to the IQ representation 130. Parameter determination can be solved using an optimization method to reduce the error between the resulting spiral and the IQ representation 130. For example, a regression method, a trained machine learning model, etc. can be used as the fitting method.

[0052] In some examples, the approximation 140 is determined using the least squares (LS) method. The LS method can be a specific type of regression method that is used to estimate the parameters of a regression model (such as the spiral function described above) by reducing (or minimizing) the sum of the squared differences between the observed values (IQ representation 130) and the predicted values (approximation 140). Using the LS method can be beneficial for efficiently achieving robust results. The fitting method (such as the LS method) can similarly be applied to other examples of the approximation 140, as further explained below.

[0053] In some cases, the fitting method can benefit from an optional initialization of the approximation 140. For example, the initialization can reduce the processing time. In the case of the spiral function, the parameters of the selected spiral function can be initialized. For example, the initialization can be based on the specific reference coordinates of one or more points or the reference characteristics of the spiral shape. For example, the initialization can be based on the position of the center (or center point, origin) of the spiral shape of the IQ representation 130 in the IQ plane. In this case, the processing circuitry 110 can estimate the center of the spiral shape and determine the approximation 140 based on the estimated center. The reference length, number of turns, rotation direction, etc. of the spiral can also be initialized. The estimation of the center can additionally or alternatively be used to determine the phase 150, as further explained below. This center can be considered the IQ offset of the IQ representation 130.

[0054] The center can be estimated by any estimation technique. For example, techniques such as the method of moments, maximum likelihood method, radial profile analysis, etc. can be used to analyze the distribution of the data points of the IQ representation 130. An ellipse fitting method or a circle fitting method can achieve high accuracy and low computational complexity. For example, the processing circuitry 110 can estimate the center by applying such an ellipse fitting method or circle fitting method to the IQ representation 130.

[0055] Ellipse fitting and circle fitting are techniques that can be used to estimate the parameters of an ellipse and a circle, respectively, which fit, approximate, or best match the data points of IQ representation 130. Ellipse fitting and circle fitting can involve reducing or minimizing the error between the data points and the equation of the ellipse or the circle. This error can be measured as the sum of the squared differences between the observed data points and the points on the fitted ellipse (least squares). Ellipse fitting can include LS fitting, RANSAC (Random Sample Consensus), or direct ellipse fitting, etc. Circle fitting can include LS fitting, Kasa method, or Taubin method, etc. The result of ellipse fitting or circle fitting is an ellipse or a circle fitted to the data points. Then, the center of the ellipse or the circle can be determined as an estimate of the spiral center.

[0056] For example, LS fitting of an ellipse or a circle can be performed using "fminsearch" based on the optimization of the (spiral) center. fminsearch is a function for unconstrained nonlinear optimization, which is used to find the minimum of a defined objective (cost) function. In this case, the cost function is the LS error. The function of fminsearch can be based on the Nelder-Mead simplex method, which is a direct search optimization algorithm. It can be particularly advantageous for finding the spiral center of IQ representation 130 because the resulting cost function does not necessarily have to be smooth or differentiable. fminsearch can include starting from an initial guess of the spiral center, iteratively exploring the parameter space, for example, by forming a simplex (a geometric shape in an N-dimensional space) around the initial guess, evaluating the objective function at specific points within the simplex, and updating the simplex based on the evaluation. The fminsearch algorithm can continue to iteratively refine the simplex until the convergence criterion is met.

[0057] The accuracy of ellipse fitting or circle fitting can be improved by applying it to a predefined winding part of the spiral shape or a part thereof, that is, the data points of other winding parts (and optionally, some data points of the predefined winding part) can be excluded. For example, the ellipse fitting method or the circle fitting method can be applied to the innermost winding part of the spiral shape or a part thereof. The data points in IQ representation 130 that do not belong to this predefined winding part can be discarded. Therefore, ellipse or circle fitting can be specifically applied to the data points (partial data) belonging to the predefined winding part. This is shown by Figure 3a and Figure 3b shown. Figure 3a and Figure 3b Fig. 300 shows an I-Q diagram of an example of an ellipse 320 fitted to an example of IQ representation 310. IQ representation 310 corresponds to Figure 2a the IQ representation 210 of. The ellipse 320 has a center 325, which can be assumed to be the estimated center of the spiral shape of IQ representation 310. Figure 3aThe ellipse 320 is fitted to the complete data of the IQ representation 310, while Figure 3b the ellipse 320 is only fitted to the partial data 330 of the IQ representation 310. The partial data 330 only includes data points belonging to the innermost wound part (turn) of the spiral arms of the spiral shape. Such an ellipse fitting to the partial data 330 can obtain more accurate results when reconstructing the spiral shape in subsequent approximation techniques.

[0058] Figure 4 An example of a logarithmic spiral function fitted to the Figure 2a IQ representation 210 is shown. Figure 4 The I-Q diagram 400 showing an example of the IQ representation 410 (corresponding to the IQ representation 210) is shown. The I-Q diagram 400 also includes an approximation 420 obtained from the logarithmic spiral function fitted to the IQ representation 410. Figure 4 The fitting of the

[0059] has been initialized with an initial offset 430 assumed to be the spiral center. After performing the fitting, i.e., after at least one iteration of modifying the parameters of the spiral function to reduce the error between the curve 420 and the IQ representation 410, the spiral center is set to the fitting or result offset 440.

[0059] As Figure 4 shown, due to its simplicity, fitting a logarithmic spiral function to the IQ representation 130 can achieve a very low running time. In addition, the estimated spiral can be substantially independent of the initialization, such as the initialization with an IQ offset.

[0060] Figure 5 An example of a general spiral function fitted to the Figure 2a IQ representation 210 is illustrated. Figure 5 The I-Q diagram 500 showing an example of the IQ representation 510 (corresponding to the IQ representation 210) is shown. The I-Q diagram 500 also includes an approximation 520 obtained from the general spiral function fitted to the IQ representation 510. Figure 5 The fitting of the

[0061] is initialized with an initial offset 530 (equal to the fitting offset 540) assumed to be the spiral center.

[0061] As Figure 5 shown, fitting a general spiral function to the IQ representation 130 can improve the accuracy of the approximation 140, i.e., the error is smaller when representing the IQ representation 130 compared to the logarithmic spiral function. However, compared to the logarithmic spiral function, it can have a longer running time and a higher dependence on guessing a good initialization value (e.g., the spiral center).

[0062] In addition to using spiral functions, manifold learning can also be used. For example, the processing circuit device 110 can determine the approximation 140 by applying manifold learning to the IQ representation 130.

[0063] Manifold learning is a dimensionality reduction technique for machine learning. It can be based on discovering the underlying structure or geometry of higher-dimensional data in a lower-dimensional space, i.e., it can be based on the assumption that high-dimensional data lies on or near a low-dimensional manifold embedded in the high-dimensional space. The manifold can be assumed to be smooth, continuous, and / or connected, similar to the geometry of a spiral structure (shape). Manifold learning techniques can aim to project the data onto a lower-dimensional space while preserving the basic geometric properties and relationships between data points. Manifold learning can improve the accuracy of approximation 140, especially when there are non-linear relationships between IQ representations 130 among data points.

[0064] Manifold learning algorithms can be configured to find clusters of contiguous data points in IQ representation 130. For example, it can include searching for a one-dimensional data order in IQ representation 130 by surveying data clusters and finding the nearest neighbors. The cluster or its compressed version can be approximation 140.

[0065] Any manifold learning algorithm can be applied, such as Isomap (Isometric Feature Mapping), Locally Linear Embedding (LLE), Multidimensional Scaling (MDS), Autoencoders, etc. Isomap (Isometric Feature Mapping) can be beneficial for capturing the intrinsic structure of the data because it is based on the geodesic distance (shortest path distance) between data points in IQ representation 130. Thus, Isomap is a non-linear dimensionality reduction method that preserves the pairwise geodesic distances (shortest path distances) between data points in the original high-dimensional space of IQ representation 130 while projecting the data onto a low-dimensional space.

[0066] Isomap can include constructing a pairwise distance matrix that captures the Euclidean distances between data points in the high-dimensional space of IQ representation 130. Isomap can also include constructing a neighborhood graph where data points are connected to their nearest neighbors in the high-dimensional space, e.g., connected to their k nearest neighbors, where k is a parameter that determines the level of local versus global information retained in the feature mapping. Then, Isomap can compute the geodesic distances between data points on the neighborhood graph. These geodesic distances represent the shortest path distances along the graph edges and capture the intrinsic structure of the data, even if it is non-linear. Isomap can use Multidimensional Scaling (MDS) on the geodesic distance matrix to find the low-dimensional representation of IQ representation 130 that best preserves these distances. The result of Isomap can be a low-dimensional feature mapping of the data points, where the data points are represented in the reduced space. This low-dimensional representation or the data derived from it can be approximation 140.

[0067] To improve clustering using Isomap manifold learning, phase evaluation can be performed, for example, by polynomial fitting of an IQ representation 130 to the manifold parameters of the manifold learning. The manifold parameters can be at least one of distance metrics (such as Euclidean distance or another metric) used to calculate pairwise distances between data points and multiple dimensions in the low-dimensional representation generated by the manifold learning algorithm. Additionally or alternatively, clustering of Isomap manifold learning can be improved by considering the temporal dependence between data points of the IQ representation 130.

[0068] Figure 6 Illustrates the clustering of an IQ representation 210 generated by Isomap manifold learning Figure 2a as an example. Figure 6 An I-Q diagram 600 showing an example of a cluster 610 of the IQ representation 210. In Figure 6 it, the cluster 610 is represented by gray scale, that is, data points within one cluster have the same gray level.

[0069] Some approximation techniques (such as manifold learning, piecewise fitting, or polynomial approximation) can benefit from optional upstream compression of the IQ representation 130. Compression can reduce the computational workload for determining the approximation 140. In particular, LS fitting can be negatively affected by extended data clusters. As shown in reference Figures 2a - 2c it, the distribution of data points of the IQ representation 130 can exhibit a certain statistical dispersion (variability, scatter, diffusivity), that is, there is a certain deviation between repetitions (rounds) of the path passed by the data points along the spiral arm. Compression can achieve an initial simplification of the IQ representation 130, which can then be used to determine the approximation 140. Compression refers to the process of reducing the size or volume of the IQ representation 130 while retaining relevant geometric information. Any compression technique can be used, for example, eliminating redundancy (that is, only considering data points of one (representative) round or fewer rounds along the spiral arm while discarding data points of other rounds) or encoding the IQ representation 130 (that is, determining representative data points of one round along the spiral arm).

[0070] An example of compression can be based on the estimated temporary phase of the received radar signal. For example, the processing circuitry 110 can estimate the above-mentioned temporary phase based on the IQ representation 130, compress the IQ representation 130 with respect to the temporary phase, and determine the approximation 140 of the compressed IQ representation. For example, the estimated spiral center can be used for compression. Thus, compression can be based on the estimated (temporary) phase relative to the spiral center. The temporary phase can be derived, for example, from an ellipse fitting method or a circle fitting method, as described above.

[0071] In the case of using an ellipse or circle fitting to estimate the (initial) helix center, the resulting ellipse or circle can optionally be offset and then reused for estimating the temporary phase. The phase of the ellipse or circle can be estimated by determining the angular position of points on the ellipse or circle relative to an axis (e.g., the axis of symmetry). The phase can indicate how far the data points have traveled around the ellipse or circle from a reference point, such as a focus of the ellipse or circle or the point where the major and minor axes of the ellipse intersect. After offset removal, the phase value of each data point of the ellipse or circle can be used as the temporary phase.

[0072] Compression can be performed by associating the data points of the IQ representation 130 with the corresponding temporary phase. The data points can be clustered based on the above association. Then, for example, based on statistical processing of the data points of each cluster, at least one compressed data point can be determined for each cluster. For example, the mean or median or a similar value of the data points of the cluster can be determined and set as the new compressed data point, or the data point closest to the mean or median can be selected as the (representative) compressed data point. Alternatively, compression can include selecting any other type of representative data point for the corresponding temporary phase value and omitting the remaining data points. In this way, compression that preserves the original phase of the data points can be performed.

[0073] Alternatively, for example, in the case where no initial helix center is estimated or for less complex compression, compression can be performed by using every z≥2 data points of the IQ representation 130 and ignoring the remaining data points. For example, such compression can be beneficial for Isomap manifold learning because this technique does not necessarily rely on an initial helix center approximation.

[0074] The approximation 140 can alternatively or additionally be determined by piecewise fitting. For example, the processing circuit device 110 can match corresponding segments of multiple circles or ellipses with corresponding parts of the IQ representation 130. These segments can together reproduce the helix configuration (or helix shape) fitted to the IQ representation 130. For example, the (e.g., compressed) data points can be grouped into parts or segments of an approximate helix shape. For example, these parts or segments can be selected by covering a predefined angle along the (approximate) helix shape. Then, ellipse or circle fitting can be applied to each segment of the helix shape. The result of the fitting algorithm can be a segment of a circle or ellipse that fits (matches) the segment of the helix shape. The accuracy and complexity of determining the approximation 140 in the case of piecewise fitting can depend on the number of segments used to reconstruct the helix shape.

[0075] Figure 7 Illustrates an example of piecewise fitting a circle to Figure 2a the IQ representation 210. Figure 7An I-Q diagram 700 showing an example of an IQ representation 710 (corresponding to the IQ representation 210) is shown. The I-Q diagram 700 also includes an approximation obtained by matching segments of five circles 721 to 725 with corresponding parts of a compressed version of the IQ representation 710. The compressed IQ representation is indicated by specific compressed data points, such as data points 731 - 735. For example, the compressed data point 731 belongs to the first part of the compressed IQ representation, and a segment of the first circle 721 is fitted to this part. The compressed data point 732 belongs to the second part of the compressed IQ representation, and a segment of a different second circle 722 is fitted to this part, and so on.

[0076] Another example of how to determine the approximation 140 is the polynomial method. For example, the processing circuitry 110 can determine the approximation 140 by determining a polynomial approximation of the spiral shape of the IQ representation 130. A (e.g., two-dimensional) polynomial approximation (polynomial fitting or polynomial regression) can be based on the mathematical technique of curve fitting. Determining the polynomial approximation can involve fitting a polynomial equation to the IQ representation 130.

[0077] The general form of the polynomial equation is = a 0 + a 1 x + a 2 x 2 + a 3 x 3 +... + a n x n , where y is the dependent variable and represents the Q value (or I value), and x is the independent variable and represents a linear example vector. Thus, the I and Q values can be individually fitted. a 0 , a 1 , a 2 , ……, a n are the coefficients of the polynomial equation. These coefficients are determined through a fitting process and represent the relationship between the I value and the Q value, which describe the data points of the approximation 140. The goodness of fit can depend on the chosen degree (n) of the polynomial equation. The degree determines the highest power of x in the polynomial equation. For example, a linear fit uses n = 1, a quadratic fit uses n = 2, and so on. The degree can be heuristically selected by validating the fit or based on domain-specific assumptions. Generally, an nth-order polynomial can be used to fit the time series Q(t) and I(t) to correct the phase by evaluating the polynomial fit.

[0078] As described above, any fitting method can be used to fit a polynomial equation to the IQ representation 130 or its compressed version. For example, LS fitting can be used to find the coefficients (a 0 , a 1 , a 2 , ……, a n)。 This process may involve solving a set of equations. Compared with other approximation techniques described in this document, the polynomial method can be both accurate and computationally inexpensive.

[0079] A specific example of the polynomial method may include the following steps: performing an initial IQ offset estimation (center estimation) using ellipse fitting, compressing the IQ representation 130 based on the estimated temporary phase relative to the ellipse center, and fitting a polynomial equation to the compressed IQ representation.

[0080] Figure 8 The polynomial fitting is illustrated in. Figure 8 An I-Q diagram 800 of an example of an IQ representation 810 (corresponding to the IQ representation 210) is shown. The I-Q diagram 800 also includes a compression 820 of the IQ representation 810 and an approximation 830 of the compression 820 obtained by fitting a polynomial equation to the compression 820. The approximation 830 may have a starting data point 840 and an ending data point 850.

[0081] The phase 150 is derived from the approximation 140 determined by any of the above techniques. The phase 150 may correspond to the angle observed from a reference point. For example, the reference point may be the center of the spiral shape approximated by the approximation 140. Thus, as described above, the center estimation technique can be applied to the approximation 140. If the center has been estimated as the initialization of the approximation process, the estimated center can be reused as the reference point for phase determination.

[0082] Phase determination can be directly applied to the approximation 140, or applied to the original IQ representation 130 mapped to the approximation 140. In the latter case, the processing circuitry 110 may optionally map the data points of the IQ representation 130 to the data points of the approximation 140, and determine the phase 150 of the received radar signal based on the mapped data points. The mapping of the data points can associate the original data points with the corresponding approximate data points. The mapping can preferably be performed such that the phase 150 of the original data points remains substantially unchanged.

[0083] For example, LS fitting can be used to find the association between the original data points and the approximate data points. LS fitting can optionally be used in combination with "fmincon". fmincon (function minimization with constraints) is a function for solving non-linear constrained optimization problems. fmincon includes reducing or minimizing the error between the phase of the original data points and the phase of the approximate data points, where the error of LS fitting is used here.

[0084] Another example of a mapping technique can be the estimated temporary phase using ellipse or circle fitting, as described above for the compression technique. Thus, the original data points with a temporary phase that matches or approximately represents the phase of the data points of approximately 140 can be mapped to the above data points of approximately 140. This can be faster than ordinary LS fitting and provide higher performance. Note that the mapping can alternatively be performed by an external circuit device: in this case, device 100 can receive the mapping from the above external circuit device (e.g., via interface circuit device 120).

[0085] Any technique can be used to determine phase 150. For example, this can depend on the specific method chosen to determine approximately 140. Two specific examples of phase determination techniques can be arctangent demodulation and phase linearization. In the former case, processing circuit device 110 can determine phase 150 by applying arctangent (inverse tangent) demodulation to the data points of approximately 140 or the data points mapped to approximately 140. To apply the above arctangent demodulation, for example, the arctangent of the ratio of the approximate Q value (component) to the I value can be determined. Then, phase 150 can be determined based on the arctangent value. For example, the value of phase 150 can correspond to the value of the arctangent. Arctangent demodulation can particularly achieve high precision in phase determination.

[0086] In the case of phase linearization, processing circuit device 110 can determine the phase by linearizing phase 150 with respect to the arc length of approximately 140 or the mapped data points. In other words, the arc length can be determined for approximately 140, for example, for a polynomial approximation of a spiral shape, and the arc length can be linearized based on a predefined relationship between the length and the phase. This relationship can be guessed and optionally corrected, or directly derived from the estimated temporary phase of the ellipse or circle fitting or from the phase fitted (e.g., LS fitting) to the arc length function for linearization.

[0087] Phase evaluation can optionally be performed using the IQ offset removal version of approximately 140 or the mapped data points. As described above, the IQ offset can be estimated by using the center estimation technique.

[0088] Optionally, device 100 can perform further processing on the determined phase 150. For example, processing circuit device 110 can determine the movement of the target based on the determined phase 150. Processing circuit device 110 can determine, for example, an oscillation of λ / 2 or less in the near field of the radar sensor. Processing circuit device 110 can alternatively determine any movement of any amplitude in the near field of the radar sensor. For example, the movement can include or be the movement of a biological blood vessel. In this case, processing circuit device 110 can also determine the blood pressure of the organism based on the determined phase 150 (and / or based on the movement).

[0089] When measuring the pulsation of an artery below the skin surface with a radar sensor directly attached to or not far above the skin surface, such target displacement in the (e.g., very / ultra) near field of the radar sensor (e.g., an FMCW radar sensor) may be highly distorted due to near-field effects. For a target displacement of λ / 2, the device 100 can be more approximated as a spiral shape rather than being conventionally approximated as circular or elliptical in the target distance range. Based on this approximation 140, the device 100 can enable the data points of the IQ representation 130 to be projected onto the unit circle to achieve more accurate displacement measurement.

[0090] Figure 9 An example of a radar system 900 is illustrated. The radar system 900 includes a device 910 (such as device 100) as described herein and a radar sensor 920. The radar sensor 920 is configured to transmit a radar signal and receive the reflection of the above-mentioned radar signal, thereby generating a received radar signal. For example, the radar sensor 920 may include at least one antenna configured to transmit a radar signal.

[0091] Although the device 910 and the radar sensor 920 are depicted as separate blocks in Figure 9 , in other examples, the device 910 may be partially or fully included in the radar sensor 920, so the radar sensor correspondingly includes all or part of the processing circuitry 110 of the device 910.

[0092] In the case where the device 910 is only partially included in the radar sensor 920, the radar system 900 may include distributed processing circuitry that performs the corresponding parts of the processing steps, for example, in the following form: a first processing (sub)circuitry included in the radar sensor 920, and a second processing (sub)circuitry external to the sensor and communicatively coupled to the first processing circuitry through an interface circuitry (e.g., interface circuitry 120) for exchanging data between the first processing circuitry and the second processing circuitry, for example.

[0093] In the case where the device 910 is integrated in the radar sensor 920, the processing circuitry and the radar sensor 920 may be jointly integrated in a single semiconductor chip or in multiple semiconductor chips.

[0094] In the case where the device 910 is not included in the radar sensor 920, the circuitry may take the form of circuitry external to the radar sensor 920 and communicatively coupled thereto through an interface circuitry.

[0095] The radar sensor 920 may optionally be configured to receive reflections from the near field of the radar sensor 920. The radar sensor 920 may be an FMCW radar sensor.

[0096] In combination with the proposed technology or one or more of the above examples, for example, with reference to Figure 1 , more details and aspects of the radar system 900 are explained. The radar system 900 may include one or more additional optional features corresponding to one or more aspects of the proposed technology or one or more of the above examples.

[0097] Figure 10 An example of the method 1000 is illustrated. The method 1000 may be performed by an apparatus (such as apparatus 100) described herein. The method 1000 includes obtaining 1010 an IQ representation of radar data indicative of a received radar signal, determining 1020 an approximation of the IQ representation that has a spiral characteristic configuration, and determining 1030 a phase of the received radar signal based on the approximation.

[0098] More details and aspects of the method 1000 are explained in combination with the proposed technology or one or more of the above examples, for example, with reference to Figure 1 . The method 1000 may include one or more additional optional features corresponding to one or more aspects of the proposed technology or one or more of the above examples.

[0099] Some examples of the proposed technology are given below:

[0100] One example (e.g., example 1) relates to a method that includes obtaining an IQ representation of radar data indicative of a received radar signal, determining an approximation of the IQ representation, the approximation having a spiral configuration, and determining a phase of the received radar signal based on the approximation.

[0101] Another example (e.g., example 2) relates to the previous example (e.g., example 1) or any other example, and further includes that the approximation is determined by determining a polynomial approximation of a spiral shape for the IQ representation.

[0102] Another example (e.g., example 3) relates to the previous example (e.g., one of examples 1 or 2) or any other example, and further includes determining that the approximation includes matching corresponding segments of a plurality of circles or ellipses with corresponding parts of the IQ representation, the segments together reproducing the spiral configuration.

[0103] Another example (e.g., example 4) relates to the previous example (e.g., one of examples 1 to 3) or any other example, and further includes that the approximation is determined by applying manifold learning to the IQ representation.

[0104] Another example (e.g., example 5) relates to the previous example (e.g., example 4) or any other example, and further includes that the manifold learning includes using isometric feature mapping.

[0105] Another example (e.g., Example 6) relates to a previous example (e.g., one of Examples 1 to 5) or any other example, and further includes an approximation determined by fitting a spiral function to the IQ representation.

[0106] Another example (e.g., Example 7) relates to a previous example (e.g., one of Examples 1 to 6) or any other example, and further includes an approximation determined using the least squares method.

[0107] Another example (e.g., Example 8) relates to a previous example (e.g., one of Examples 1 to 7) or any other example, and further includes a phase determined by applying an arctangent demodulation to the approximation.

[0108] Another example (e.g., Example 9) relates to a previous example (e.g., one of Examples 1 to 8) or any other example, and further includes a phase determined by linearizing the phase with respect to the arc length of the approximation.

[0109] Another example (e.g., Example 10) relates to a previous example (e.g., one of Examples 1 to 9) or any other example, and further includes estimating the center of the spiral shape of the IQ representation and determining the approximation based on the estimated center.

[0110] Another example (e.g., Example 11) relates to a previous example (e.g., Example 10) or any other example, and further includes a center estimated by applying an elliptical fitting method or a circular fitting method to the IQ representation.

[0111] Another example (e.g., Example 12) relates to a previous example (e.g., Example 11) or any other example, and further includes an elliptical fitting method or a circular fitting method applied to a predefined winding portion of the spiral shape.

[0112] Another example (e.g., Example 13) relates to a previous example (e.g., one of Examples 1 to 12) or any other example, and further includes estimating the temporary phase of the received radar signal based on the IQ representation, compressing the IQ representation with respect to the temporary phase, and determining the approximation of the compressed IQ representation.

[0113] Another example (e.g., Example 14) relates to a previous example (e.g., one of Examples 1 to 13) or any other example, and further includes mapping the data points of the IQ representation to the data points of the approximation and determining the phase of the received radar signal based on the mapped data points.

[0114] Another example (e.g., Example 15) relates to a previous example (e.g., one of Examples 1 to 14) or any other example, and further includes determining the motion of the target based on the determined phase.

[0115] Another example (e.g., Example 16) relates to a previous example (e.g., Example 15) or any other example, and further includes that the movement includes the movement of the blood vessels of a living being, and the method further includes determining the blood pressure of the living being based on the determined phase.

[0116] Another example (e.g., Example 17) relates to a non-transitory machine-readable medium having a program stored thereon, the program having program code for performing the method according to any one of Examples 1 to 16 when the program is executed on a processor or programmable hardware.

[0117] Another example (e.g., Example 18) relates to a program having program code for performing the method according to any one of Examples 1 to 16 when the program is executed on a processor or programmable hardware.

[0118] One example (e.g., Example 19) relates to a device including processing circuitry configured to obtain an IQ representation of radar data indicative of a received radar signal, determine an approximation of the IQ representation, the approximation having a spiral configuration, and determine the phase of the received radar signal based on the approximation.

[0119] One example (e.g., Example 20) relates to a radar system including the device according to Example 19 and a radar sensor configured to transmit a radar signal and receive a reflection of the radar signal, thereby generating a received radar signal.

[0120] Another example (e.g., Example 21) relates to a previous example (e.g., Example 20) or any other example, and further includes that the radar sensor is configured to receive the reflection from the near field of the radar sensor.

[0121] Another example (e.g., Example 22) relates to a previous example (e.g., one of Examples 20 or 21) or any other example, and further includes that the radar sensor is a frequency-modulated continuous-wave radar sensor.

[0122] Aspects and features described with respect to a particular example among the foregoing examples may also be combined with one or more other examples to replace the same or similar features of the other example, or to introduce these features additionally into the other example.

[0123] The example may also be or relate to a (computer) program that includes program code which, when the program is executed on a computer, a processor, or other programmable hardware component, performs one or more of the above methods. Thus, the steps, operations, or processes of the above different methods may also be performed by a programmed computer, processor, or other programmable hardware component. The example may also cover a program storage device, such as a digital data storage medium, which is machine, processor, or computer-readable and encodes and / or contains machine-executable, processor-executable, or computer-executable programs and instructions. For example, the program storage device may include or be a digital storage device, a magnetic storage medium such as disks and tapes, a hard disk drive, or an optically readable digital data storage medium. Other examples may also include a computer, a processor, a control unit, a (field) programmable logic array ((F)PLA), a (field) programmable gate array ((F)PGA), a graphics processing unit (GPU), an application-specific integrated circuit (ASIC), an integrated circuit (IC), or a system-on-chip (SoC) system programmed to perform the steps of the above methods.

[0124] It is further understood that the disclosure of several steps, processes, operations, or functions in the specification or claims should not be construed as implying that these operations necessarily depend on the described order, unless explicitly stated in individual cases or necessary for technical reasons. Thus, the foregoing description does not limit the execution of several steps or functions to a certain order. Additionally, in further examples, a single step, function, process, or operation may include and / or be decomposed into several sub-steps, sub-functions, sub-processes, or sub-operations.

[0125] If some aspects have been described in connection with a device or system, these aspects should also be understood as a description of the corresponding method. For example, the block, apparatus, or functional aspects of a device or system may correspond to the features of the corresponding method, such as method steps. Thus, the aspects described with respect to the method should also be understood as a description of the corresponding block, corresponding element, property, or functional feature of the corresponding device or corresponding system.

[0126] The following claims are hereby incorporated into the detailed description, where each claim may stand alone as a separate example. It should also be noted that although in the claims, the dependent claims relate to a particular combination with one or more other claims, other examples may also include combinations of dependent claims with the subject matter of any other dependent or independent claim. Such combinations are hereby expressly provided, unless it is stated in individual cases that a particular combination is not intended. Additionally, the features of one claim should also be included in any other independent claim, even if that claim is not directly defined as dependent on that other independent claim.

Claims

1. A method (1000), comprising: acquiring (1010) an IQ representation (130) of radar data indicative of a received radar signal; determining (1020) an approximation (140) of the IQ representation (130), the approximation (140) having a spiral configuration; and A phase (150) of the received radar signal is determined (1030) based on the approximation (140).

2. The method (1000) of claim 1, wherein the approximation is determined by determining a polynomial approximation to a spiral shape of the IQ representation.

3. The method (1000) of any one of the preceding claims, wherein determining the approximation (140) comprises matching corresponding segments of a plurality of circles or ellipses with corresponding portions of the IQ representation (130), the segments together reproducing the spiral configuration.

4. The method (1000) according to any one of the preceding claims, wherein the approximation (140) is determined by applying manifold learning to the IQ representation (130).

5. The method (1000) of any one of the preceding claims, wherein the approximation (140) is determined by fitting a spiral function to the IQ representation (130).

6. The method (1000) according to any one of the preceding claims, wherein the phase (150) is determined by applying an inverse tangent demodulation to the approximation (140).

7. The method (1000) of any one of the preceding claims, wherein the phase (150) is determined by linearizing the phase (150) with respect to an arc length of the approximation (140).

8. The method (1000) according to any one of the preceding claims, further comprising: The center of the spiral shape of the IQ representation (130) is estimated, and the approximation (140) is determined based on the estimated center.

9. The method (1000) of claim 8, wherein the center is estimated by applying an ellipse fitting method or a circle fitting method to the IQ representation (130).

10. The method (1000) according to claim 9, wherein the ellipse fitting method or the circle fitting method is applied to a predefined winding portion of the spiral shape.

11. The method (1000) according to any one of the preceding claims, further comprising: estimating a temporary phase of the received radar signal based on the IQ representation (130); compressing the IQ representation relative to the temporary phase (130); and The approximation (140) of the compressed IQ representation (130) is determined.

12. The method (1000) according to any of the preceding claims, further comprising determining a motion of a target based on the determined phase (150).

13. The method (1000) of claim 12, wherein the movement comprises movement of a blood vessel of a living being, the method (1000) further comprising determining a blood pressure of the living being based on the determined phase (150).

14. An apparatus (100) comprising a processing circuit device (110), the processing circuit device being configured to: obtaining an IQ representation of radar data indicative of a received radar signal (130); determining an approximation (140) of the IQ representation (130), the approximation (140) having a spiral configuration; and A phase (150) of the received radar signal is determined based on the approximation (140).

15. A radar system (900), comprising: The device (910) according to claim 14; as well as A radar sensor (920) is configured to transmit a radar signal and receive reflections of the radar signal, thereby generating the received radar signal.

Images