Wireless imaging based on metasurface

Abstract

According to implementations of the disclosure, a metasurface-based wireless imaging solution is provided. In the solution, a metasurface is arranged in a signal propagation path from a transmitter to a receiver via a target object, that is, a wireless signal for imaging sensing is enabled to interact with the metasurface. An image of the target object is generated using the received signal. In addition, in the implementation of the disclosure, a design of end-to-end optimization for a wireless imaging system is proposed. In implementations of the present disclosure, powerful wavefront shaping of the metasurface is used for wireless imaging. In this way, a miniaturized wireless imaging solution can be achieved without the need for mobile device or object movement and with reduced dependence on the number of transceivers.

Description

WIRELESS IMAGING BASED ON METASURFACEBACKGROUND[OOOlJWireless imaging uses a wireless signal reflected by a target object to reconstruct images of its shape. Acoustic imaging is a kind of wireless image. It complements widely used cameras since sound waves work under different lighting conditions and can penetrate through certain materials with high energy efficiency. Unlike radio frequency (RF)-based imaging techniques, acoustic imaging can be easily deployed by using low-cost microphones and speakers on commercial devices. Such wide availability of speakers and microphones makes acoustic imaging attractive for many applications, including gesture recognition, activity detection, and so on.SUMMARY

[0002] According to implementations of the present disclosure, a solution of wireless imaging is provided. In the solution, a sensing signal to a metasurface and a target object is transmitted by a transmitter; a target response signal from the target object is received by a receiver; a target image of the target object is generated based on the target response signal, and a signal measurement matrix describes an effect of a propagation path from transmitting the sensing signal to receiving the target response signal on a wireless signal. In implementations of the present disclosure, the metasurface is used for imaging. In this way. high quality imaging may be performed on the target object without movement of the transmitter, the receiver, and the target object, thereby saving costs. A miniaturized wireless imaging solution can be achieved without device or target movement, and dependence on the number of transceivers can be reduced.

[0003] According to implementations of the present disclosure, a solution for optimizing wireless imaging is provided. In the solution, a signal measurement matrix indicating an effect of the propagation path on the sensing signal is established according to a sensing signal and a propagation path of the sensing signal; a relationship between a response signal and each of the sensing signal, a metasurface, and a reference object is determined through the signal measurement matrix, and the metasurface and the reference object are located in a propagation path from the sensing signal to the response signal; imaging of the reference object is simulated through the established relationship; and the signal measurement matrix is updated based on the simulated result of the imaging until a predetermined condition is satisfied. In this way, the simulated result of the imaging is used to optimize the signal measurement matrix, that is, to optimize the imaging system, end-to-end. For example, the optimized signal measurement matrix may be used to determine an optimized metasurface design, an optimized transceiver beam forming characteristic, and so on, which may improve imaging quality7.

[0004] This section is provided to introduce a selection of the object in a simplified form that isfurther described below in the Detailed Description. This section is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to limit the scope of the claimed subject matter.BRIEF DESCRIPTION OF THE DRAWINGS[0005JFIG. 1 illustrates a block diagram of an example environment in which various implementations of the present disclosure can be implemented;[0006JFIG. 2 illustrates a flow diagram of a process for optimization of wireless imaging in accordance with some implementations of the present disclosure;[0007JFIG. 3 illustrates a schematic diagram of channel modeling for imaging in accordance with some implementations of the present disclosure;[0008JFIG. 4A illustrates a schematic diagram of an optimization framework for reconstructing an image in accordance with some implementations of the present disclosure;[0009JFIG. 4B illustrates a schematic diagram of a structure of an alternate update layer in accordance with some implementations of the present disclosure;[OO1OJFIG. 5 illustrates a flow diagram of a process for wireless imaging in accordance with some implementations of the present disclosure;[OO11JFIG. 6 illustrates a schematic of adaptive image resolution for different distances in accordance with some implementations of the present disclosure; and[0012JFIG 7 illustrates a schematic block diagram of an electronic device capable of implementing various implementations of the present disclosure.DETAILED DESCRIPTION

[0013] The disclosure will now be discussed with reference to a number of example implementations. It should be understood that these implementations are discussed only to enable those of ordinary skill in the art to better understand and thus implement the disclosure, and not to imply any limitation on the scope of the disclosure.

[0014] As used herein, the term "including / comprising" and variations thereof is to be read as meaning an open-ended term that "including / comprising but not limited to". The term "based on" is to be read as "based at least in part on. " The terms "one implementation" and "an implementation" are to be read as "at least one implementation. "The term "another implementation" is to be read as "at least one other implementation". The terms "first", "second", and the like may refer to different or same object. Other explicit and implicit definitions may also be included below.

[0015] It is noted that the headings of any section / subsection provided herein are not limiting. Various implementations are described throughout herein, and any type of implementation can be included under any section / subsection. Furthermore, the implementations described in anysection / subsection may be combined in any manner with any other implementations described in the same section / subsection and / or different sections / subsections.

[0016] Herein, unless explicitly stated otherwise, "in response to A", performing a step does not mean that the step is performed immediately after "A", but may include one or more intermediate steps.

[0017] As used herein, a group of elements, an element group, or the like may include zero, one, or more such elements. The group of elements may be ordered or disordered. For example, "a group of divisions" may include zero, one, or more divisions. As used herein, an element sequence or similar expression may include one or more such elements, and the elements in the sequence are ordered.

[0018] As used herein, the term "model" may leam associations between respective inputs and outputs from training data, such that after training is complete, a corresponding output may be generated for a given input. The generation of the model may be based on the machine learning technology. Depth learning (DL) is a machine learning algorithm, which processes inputs and provides respective outputs by using a multi-layer processing unit. A neural network model is an example of a model based on deep learning. Herein, "model" may also be referred to as "machine learning model", "learning model", "machine learning network", or "learning network", and these terms are used interchangeably herein.

[0019] As briefly described above, acoustic imaging has many advantages. To accurately reconstruct images, conventional acoustic imaging systems require large arrays of transceivers (for example, 40 speakers and 40 microphones) to ensure adequate spatial sampling and signal- to-noise ratio (SNR), which results in dedicated, expensive, and complex hardware designs. In contrast, existing commercial Internet of Things (loT) devices typically have only a small number of transceivers to limit cost and energy consumption. For example, some loT devices only have under 10 microphones and speakers. While these latest loT devices already have the largest number of transceivers in the market, it is still far from what is necessary to produce high-quali ty images.

[0020] In order to address the challenge of limited transceivers, a technical solution made an early stride by introducing synthetic aperture radar (SAR). However, its reliance on mechanical movement spanning tens of centimeters makes it challenging to deploy in static devices (for example, smart speakers). Another technical solution employed a three dimensional (3D)-printed metamaterial stencil to enhance spatial perception by dividing signals from the speakers into multiple replicas. Such a solution uses a rather basic image reconstruction algorithm and stencil design and requires multiple frames through robot motion to get clear images. Therefore, realizing acoustic imaging on low-cost loT devices without device movement remains a challenge. Theissues described by wireless imaging technology are illustrated with acoustic imaging as an example. It should be appreciated that other types of wireless imaging (for example, millimeter waves) also face similar problems.

[0021] To this end. according to the implementations of the present disclosure, a metasurfacebased wireless imaging solution is proposed. In the solution, a metasurface is arranged in a signal propagation path from a transmitter to a receiver via a target object, that is, a wireless signal for imaging sensing is enabled to interact w ith the metasurface. In the implementations of the present disclosure, powerful wavefront shaping of the metasurface is used for wireless imaging. On the one hand, an object to be imaged does not need to move; on the other hand, the dependence on the number of transceivers is reduced. In this w ay, a miniaturized wireless imaging solution can be achieved without device or target movement.

[0022] According to implementations of the present disclosure, a solution for optimizing wireless imaging is also provided. In the solution, a signal measurement matrix (also referred to as a measurement matrix for short) is established according to a sensing signal and a propagation path of the sensing signal, and the signal measurement matrix indicates an effect of the propagation path on the sensing signal. A relationship between a response signal and each of the sensing signal, a metasurface, and a reference object is determined through the signal measurement matrix. Imaging of the reference object is simulated through the established relationship, and the signal measurement matrix is updated based on the simulated result of the imaging until a predetermined condition is satisfied. In this way. the simulated result of the imaging is utilized to optimize the signal measurement matrix, that is, to optimize the imaging system, end-to-end. For example, the optimized signal measurement matrix may be used to determine an optimized metasurface design, an optimized transceiver beam forming characteristic, and so on, which may improve imaging quality.Example Environment

[0023] FIG. 1 shows a schematic diagram of an example environment 100 in which implementations of the present disclosure can be implemented. As shown in FIG. 1. the environment 100 comprises a transmitter 110, a receiver 120, a metasurface 130, and a target object 140. For example, in the case of acoustic imaging, the transmitter 110 may comprise a speaker array, and the receiver 120 may comprise a microphone array. The transmitter 110 may transmit a sensing signal to the target object 140, and the receiver 120 may receive a response signal from the target object 140. The metasurface 130 is arranged in the signal propagation path, which contributes to the propagation of the radio signal.

[0024] The metasurface 130 may be arranged at any suitable location in the signal propagation path. Although a metasurface is shown, multiple metasurfaces may be included inimplementations of the present disclosure, or a metasurface may include different sections that may be separated in space.

[0025] In some implementations, as shown in FIG. 1, the sensing signal transmitted by the transmitter 110 is incident on the target object 140 through the metasurface 130 and then reaches the receiver 120 through the metasurface 130. In some implementations, the sensing signal transmitted by the transmitter 110 may be incident on the target object 140 through a first metasurface and then reach the receiver 120 through a second metasurface, or may be incident on the target object 140 through a first portion of the metasurface and then reach the receiver 120 through a second portion of the metasurface. The first metasurface and the second metasurface may be different metasurfaces or the same metasurface. The first portion and the second portion may be a same portion or different portions.

[0026] Other types of arrangements are possible. In some implementations, the sensing signal transmitted by the transmitter 110 is incident on the target object 140 through the metasurface and then reaches the receiver 120. In some implementations, the sensing signal transmitted by the transmitter 110 is incident on the target object 140, reaches the receiver 120 through the metasurface.

[0027] It should be appreciated that the components and arrangements illustrated in FIG. 1 are merely an example and that the computing system suitable for implementing the implementations described herein may include one or more different components, other components, and / or different arrangements.

[0028] It should be understood that the structure and function of the various elements in the environment 100 are described for exemplary purposes only, and are not intended to imply any limitation on the scope of the disclosure.

[0029] To better understand the example implementations of the present disclosure, the basic principles of wireless imaging are first described by taking acoustic imaging as an example. In particular, the fundamental idea of acoustic imaging with compressive sensing and the concept of the acoustic metasurface are first introduced.

[0030] Consider the signal xt(t) from N pixels of the target arrive at M microphones at a distance dnm. The signal received by each microphone can be derived as follow s:(1)where ym(t) is the sound pressure at the m'hmicrophone at time t, c is the acoustic signal propagation speed, and xn(t) is the reflected signal from the nthtarget at time t.

[0031] The above relationship may also be captured using the following matrix form:y = Ax + e (2)^ ^ where A stands for a M N measurement matrix, defined as: A„m= - a , y is an Mx 1 vector nm representing the signal from all microphones, x is the target image and reshaped to an N x 1 vector, and e represents Additive White Gaussian Noise (AW GN). % is a greyscale vector, which denotes the fraction of signal that is reflected by the target at each position. Due to the limited number of transceivers available on the low-cost loT devices, acoustic imaging is typically an underconstrained inference problem, which may have an infinite number of solutions.

[0032] Compressive sensing can be used to reconstruct the image % if % or some transformation of x (for example, discrete cosine transformation (DCT)) is sparse. In this case, the image can be reconstructed by solving the following optimization problem: arg (3 ’)where II x H leverages the sparsity prior and II Ax — y ||2< e enforces the accuracy of the reconstructed image. Iterative algorithms may be used to solve this problem. In practice, the reconstruction error depends on the measurement matrix, noise, and the number of unknow ns (that is, pixels).

[0033] The measurement matrix in the imaging task is determined by the channel, as shown in Eq. (1). The channel is dictated by two key parameters: frequency f and distance d,mbetween each target position and receiver. If the measurement matrix A satisfies the Restricted Isometry Property (RIP), then compressive sensing can be applied to accurately reconstruct the image.

[0034] However, in practice, it can be difficult to verify if a matrix satisfies the RIP property. Therefore, the rank of the measurement matrix is used as an approximation. A larger rank suggests more linearly independent constraints for solving the optimization problem, and is preferred.

[0035] In the implementations of the present disclosure, the metasurface is used for wireless imaging. The metasurface enhances the degree of freedom in both spatial and frequency domains to control the measurement matrix. Using the metasurface may significantly increase the rank under the same number of transceivers. Some example implementations of the present disclosure will be described in more detail below with reference to the accompanying figures.Example Optimization Flow for H ireless Imaging[0036JFIG. 2 illustrates a flow diagram of a process 200 for optimizing wireless imaging in accordance with some implementations of the present disclosure. The process 200 may be implemented at any suitable electronic device.

[0037] At block 210, a signal measurement matrix is established according to a sensing signal and a propagation path of the sensing signal. The signal measurement matrix indicates an effect of thepropagation path on the sensing signal. In other words, the signal measurement matrix describes a channel from the sensing signal to the response signal. For example, the measurement matrix signal measurement matrix may be considered as a signal measurement matrix that describes a signal propagation path from the transmitter to the receiver.

[0038] At block 220. a relationship between a response signal and each of the sensing signal, a metasurface, and a reference object is determined through the signal measurement matrix. The metasurface and the reference object are located in a propagation path from the sensing signal to the response signal. The reference object may be used to represent a target object in an actual imaging scenario, in the optimization process.

[0039] The channel modeling may be viewed as performed at block 210 and block 220. FIG. 3 shows a schematic diagram 300 of channel modeling for imaging in accordance with some implementations of the present disclosure. As shown in FIG. 3, Ht mrepresents a channel from a speaker array 310 (as an example of the transmitter 110) to the metasurface 130, wherej) represents the channel from the z-th speaker to the / -th metasurface cell. Similarly, Hm rrepresents the channel from the metamaterial 130 to a microphone array 320 (as an example of the receiver 120). Suppose the reference object is within a given 3D imaging area O 330. and define a channel matrix Hm o, where Hm o(j, k) denotes the channel from the y-th metasurface cell to the / c-th grid in the 3D area. Note that if the k-th grid does not contain the reference object, there is no reflection from this grid. Similarly, define Ho m. where Ho m(k, y) denotes the channel from the / c-th grid in the 3D area to the y-th metasurface cell. Then, let w denote the beamforming of the speakers for a specific frequency. A propagation path of the sensing signal transmitted from the speaker array 310 to the response signal received by the microphone array 320 may be represented by the channels described above.

[0040] Based on the above definitions, the received signals after going through combining beamforming of speakers, metasurface manipulation, and microphones become as follows:where M denotes the manipulation of each metasurface cells on acoustic signals (for example, signal attenuation and phase delay) and • denotes dot product. A dot product is used between the metasurface and incoming signal because each metasurface cell manipulates multipath signals coming through the cell in the same way regardless of which path the signal comes from. Similarly, a dot product may be used to capture the interaction between the incoming signal and the object. If configurations of the metasurface are fixed, the target obj ect may be inferred based on the value of all the other terms. All the channel matrices, includingHm n, Hn m, and Hm rare known based on the transceiver, the metasurface 130. and the 3D imaging area 330. Specifically,Hi = a(di j)e~i2n— . where dt is the distance from the source i to the destination c is the propagation speed of acoustic signals, and a(dij is the amount of signal attenuation at the distance d j. Moreover. Eq. 4 can be further simplified as follows:where Am(M, w) represents the measurement matrix for a specific frequency. To improve imaging perfonnance, the channel noise may be suppressed and more constraints may be accumulated in Eq. 5. In an example, a relationship between the response signal and each of the sensing signal, the metasurface, and the reference object is constructed based on Eq. (5).

[0041] Channel noise has significant impact on the image reconstruction error. An effective approach is to leverage multiple microphones at the receiver side, where beamforming of the multiple microphones can be used to harness spatial diversity and suppress channel noise. Let D denote beamforming codebooks of the microphones that consist of multiple weights, and Rmdenotes the received signal at all microphones. R = DRmrepresents combining the received signals across microphones. For each frequency, the codebook size is equal to the number of microphones, since further increasing it does not yield new information due to linear dependence.

[0042] A simple way to obtain more constraints is to use frequency diversity. However, the benefit of frequency diversity is limited by the number of speakers and microphones. Therefore, a metasurface and beamforming of the speakers and microphones are jointly designed across multiple frequencies. To achieve this, the impact of the metasurface at different frequencies, including the phase shift and amplitude attenuation, needs to be firstly derived. The phase offset introduced by the metasurface can be directly calculated from the propagation distance of internal structure of each metasurface cell. To derive the impact on the amplitude, it is observed that the metasurface is designed to achieve close to 100% penetration at 20 KHz (that is, \Mf=2QkHz\ — 1), and the penetration of the metasurface decays at other frequencies. The impact of each of multiple different metasurface cells may be simulated and a lookup table may be generated to record the resulting amplitude.

[0043] In the examples described herein, since the transmit beamforming and the receive beamforming are employed, the signal measurement matrix can be denoted as A(M.W.D).

[0044] The aforementioned formulation is derived in the frequency domain, so that the impact of frequencies on the metasurface and the channel propagation model can be captured. As a result, the signal measurement matrix A(M.W.D) is complex-valued, which affects the image reconstruction process. New M and Rmmay be constructed by stacking the real and imaginary parts from the original M and Rm. and the signal measurement matrix A(M.W.D) is convertedinto a real value. In this way, the rank of the initial complex-valued measurement matrix may be effectively doubled.

[0045] The process 300 is continued to be described. At block 230, imaging of the reference object is simulated through the established relationship. That is, the reconstructed image of the reference object may be simulated according to the received signal R at the receiver.

[0046] With continued reference to the above example, the following describes how to simulate the reconstructed image. When the constraints are insufficient to yield a unique solution, additional information about the object may be leveraged. One commonly used regularization term is sparsity, which indicates the target occupies a small portion of the 3D imaging area. This could be enforced by selecting an appropriate imaging area. Correspondingly, the following optimization problem may be constructed:0 = argwhere II 0 II i is a commonly used Li norm to promote the sparsity in 0 and a captures the relative importance of the sparsity versus fitting error.

[0047] The image reconstruction problem can be solved in a number of ways. In some implementations, an Alternating Direction Method of Multipliers (ADMM) may be employed. It is an iterative method that optimizes one variable at a time in each iteration while fixing the other variables. Based on ADMM, Eq. 6 may be rewritten as follows:where s.t. denotes the condition of the optimization problem.

[0048] The ADMM algorithm is based on the augmented Lagrangian:in addition, sequential minimization of the 0 and z variables is performed by the following dual variable updates: where for termbased on domain knowledge (for example, S(-) =|| z H . p and a are hyperparameters used to adjust the importance of the fidelity term and sparsity term during optimization

[0049] The classic ADMM algorithm may have some problems. First, the hand-picked priors (forexample, the sparsification transform S) and hyperparameters (a and p) may not work well in some scenarios. Additionally, it takes hundreds of iterations to converge, which results in long running time.

[0050] In view of this, in some implementations, a machine learning model, such as a neural network, may be utilized to reconstruct an image based on the received response signal. For example, the simulated signal of the response signal, for example, Rm, may be generated based on the relationship established at block 220. An initial image of the reference object may be reconstructed, using a first model, based on the simulated signal and the signal measurement matrix. The simulated image of the reference object may be generated from the initial image. For example, the first model is used to achieve an initial reconstruction of the image, and the second model may be used to refine the initially reconstructed image.

[0051] The first model and the second model may be implemented using any suitable neural network structure, which is not limited in the implementations of the present disclosure.

[0052] An example implementation is described below with reference to FIG. 4A and FIG. 4B. FIG. 4A illustrates a schematic diagram 400A of an optimization framework for reconstructing an image in accordance with some implementations of the present disclosure. As show n in FIG. 4A, a neural network may be used instead of the classical ADMM algorithm. Such a neural network may be referred to as a physics-informed learning model. The model has two parts: an unrolled ADMM network 410 with learnable layers (as an example of the first model) and a refinement network 420 (as an example of the second model). The unrolled ADMM network 410 may perform the bulk of image reconstruction and include knowledge of the forw ard physical model, while the refinement network 420 may denoise the image and correct model mismatch errors.

[0053] In some implementations, the first model may comprise a plurality of alternate update layers for iteratively reconstructing the initial image with each alternate update layer being configured to alternately update the initial image and a corresponding multiplier.

[0054] For example, an ADMM network 410 implemented with aneural network may unfold each iteration of traditional ADMM into a layer with learnable hyperparameters (also referred to as the alternate update layer). The alternate update layer will be described below with reference to FIG. 4B. FIG. 4B shows a schematic diagram 400B of an alternate update layer. As shown in FIG. 4B, the design of the present disclosure replaces the sparse prior with a CNN, and leams imaging priors from imaging process in a data-driven manner. The sparsity may not strictly hold in real images and it is best to directly learn a prior from real data, and using a prior from real data can enhance the image quality and speed up inference. Therefore, there is an ADMM layer for each iteration, with the hyperparameter p and the learnable CNN A More specifically, the update Eq. in Eq. (9) becomes:zk+1= N(Ok+1, uk') (10) where Eq. (10) does not impose sparsity constraints when updating zk+1. Since the sparsity term II z II t is no longer used, the hyperparameter a is also removed.

[0055] In such implementations, the unrolled ADMM algorithm uses a prior from real data instead of sparsity regularization, thereby enhancing the imaging quality. Each layer in the neural network is associated with hyperparameters and these hyperparameters are learned from real data, whereas the corresponding parameters in ADMM are fixed during the iteration process, which slows down the convergence. For example, the network proposed by the present disclosure takes 8 iterations to converge, while ADMM takes over 200 iterations to converge.

[0056] To enhance the robustness of algorithm against noise and further correct mismatch errors caused by imperfect modeling, the unrolled ADMM network 410 may be concatenated with the refinement network 420 1. For example, the refinement network 420 may consist of a 3-layer encoder concatenated with a 3-layer decoder and output a ID vector, which is then reshaped to the desired imaging dimensions (for example, 10x 10x10 3D images). Such a design effectively improves system robustness since it is well-suited for noise reduction in images, thanks to skip connections and hierarchical processing.

[0057] The simulated signal of the response signal is generally pure and noise-free, while various noises are difficult to avoid in practical imaging scenarios. To this end, in some implementations, noise may be added to the simulated signal before the simulated signal is input to the first model. With continuing reference to FIG. 4A, a noise generator 425 may add noise to the simulated signal / ?mbefore the simulated signal is input to the unrolled ADMM network 410. In this way, the generalization capability of the model can be improved, the prediction capability of the model for new data can be improved, and the sensitivity of the model to a specific noise sample can be reduced.

[0058] At block 240, the signal measurement matrix is updated based on the simulated result of the imaging until a predetermined condition is satisfied. For example, a difference between the simulated imaging and the true imaging of the reference object may be minimized as an optimization goal. The predetermined condition may include, for example, the difference being less than a threshold or the optimization reaching a predetermined number of iterations, and so on.

[0059] As an example, an end-to-end optimization framework may be built to jointly optimize the imaging algorithm and the measurement matrix. Firstly, the measurement matrix may be initialized by minimizing the coherence of the measurement matrix, so that a reasonable measurement matrix can be obtained as a starting point. The measurement matrix optimization can then be concatenated with the image reconstruction and iteratively optimized in an end-to-end manner.

[0060] In particular, before the end-to-end optimization, the measurement matrix A(M.W.D) may be initialized by minimizing the coherence of the matrix. The coherence measures the maximum inner product between any two columns of a matrix. Matrices with low coherence tend to be more easier to invert, and may lead to better reconstruction error in compressive sensing. Therefore, this property can be used to increase the diversity of A(M.W.D) so that each measurement provides new information. The coherence-based optimization can be modeled as rnin^^G^ , where Gik— A M, Wi A M, Wk) computes the correlation between the i-th and / c-th rows in the measurement matrix. This optimization problem can be solved using a gradient descent approach and the physical constraints on the tunable parameters can be ignored for simplicity. In such an implementation, the coherence is used as the initialization criteria because it is easier to optimize.

[0061] The Mean Square Error (MSE) is computed between the reconstructed images and the ground-truth images, as demonstrated below:where N is the number of grids in a 3D scene. Then the error is back propagated to update all learnable parameters of the first model and the second model. Since the training objective is to minimize the distance to the ground truth image instead of intermediate indicators, the quality of reconstructed images can be significantly optimized.

[0062] In some implementations, determining, based on the updated signal measurement matrix, at least one of: a beamforming characteristic of a transmitter of the sensing signal, for example, a codebook for a speaker HZ ; a beamforming characteristic of a receiver for receiving the response signal, for example, a microphone D across all frequencies; or a structure of the metasurface, that is, M. As an example, the objective function of end to-end training is the same as Eq. (11) and the training process iteratively performs the following two steps: (1) designing the measurement matrix and (2) updating the physics-informed imaging reconstruction network. Step (1) treats the imaging reconstruction network as known and optimizes M and W, D. Step (2) treats M, W. and D as known and optimizes the imaging reconstruction network, and then iterates until convergence.

[0063] An example implementation for determining the structure of the metasurface is described in detail below. In some implementations, an electrical characteristic or an acoustic characteristic of each metasurface cell of the metasurface is determined based on the updated signal measurement matrix, and the structure of each metasurface cell is determined according to the electrical characteristic or the acoustic characteristic of the metasurface cell. For example, the phase shift, amplitude attenuation, or the like of each metasurface cell may be determined based on the optimized measurement matrix. The structure of each of the metasurface cells may then be determined based on the determined phase shift, and amplitude attenuation.

[0064] In some implementations, the shape and parameter of the metasurface cell may be determined for each location of the metasurface to satisfy desired acoustic or electrical characteristics (such as phase, amplitude, polarization, or the like) at the location. In some implementations, the wireless signal for imaging may be a sound wave. Accordingly, the metasurface may be a sound wave metasurface. In such an implementation, a target cell structure may be selected for each metasurface cell from a plurality of cell structures according to the acoustic characteristic of the metasurface cell. The plurality of cell structures may be predetermined candidate cell structures, and the acoustic characteristic corresponding to each candidate cell structure may be predetermined.

[0065] By way of example, metasurface design involves microscopic design, which determines a structure of the metasurface cell, and macroscopic design, which determines the phase map across the entire metasurface (that is, which type of metasurface cell is placed at each location of the metasurface). For example, the metasurface cell may have 16 types of candidates, which correspond to respective phase offsets (for example, 0, 1 / 16*2TT, 2 / 16*2TT, ..., 15 / 16*2TT).

[0066] In some implementations, the wireless signal for imaging may be a wave of other frequency band, such as a millimeter wave. Accordingly, the metasurface may be a millimeter-wave metasurface. In such implementations, to achieve effective phase manipulation while maintaining high transmissivity over a wide frequency range of 77-81 GHz, the metasurface needs to exhibit electromagnetic resonance over the wide frequency band. Therefore, the millimeter wave metasurface design targets cells that are highly transmissive and capable of achieving phase modulation close to 2 at the target frequency. To this end, the metasurface cell for a millimeter wave is designed such that the transmissivity and a phase modulation range at the target frequency satisfies a predetermined condition. For example, the metasurface cell which has high transmissivity at the target frequency and has a phase modulation range close to 2it may be determined. For example, the predetermined condition may include that the transmissivity at the target frequency is higher than a threshold and a difference between the phase modulation range and 2n is less than a threshold.

[0067] In some implementations, the millimeter wave metasurface cell may have a structure in which each metasurface cell comprises a dielectric layer and at least one metal layer. For example, a dielectric layer is sandwiched between two metal layers. A metal pattern in the metal layers may include a metal ring and a metal patch located inside the ring. For example, the metal pattern may include an outer square ring and an inner rectangular patch. By way of example, the spacing between the cells may be 1.8 mm, which is close to half the wavelength of the design frequency of 79 GHz. The width of the outer ring may be 0. 1 mm and the thickness of the dielectric substrate may be 0.252 mm. Different electrical characteristics, such as phase shift, may be achieved byadjusting the dimension (for example, width and length) of the inner metal patch (for example, a rectangular patch).

[0068] In some implementations, the metasurface comprises one or more metal layers, respective electrical characteristics of the one or more metal layers may be determined based on the updated signal measurement matrix, and respective structures of the one or more metal layers may be determined using a machine learning model, based on a respective electrical characteristic of the one or more metal layers. In some implementations, the machine learning model comprises at least one of: a convolutional neural network, a generative adversarial network, or a diffusion model. These machine learning models may be pretrained using a predefined data set and then used in an inference process to predict a respective structure of one or more metal layers.

[0069] Optimization of wireless imaging is described above, and application of optimized wireless imaging will be described below.

[0070] FIG. 5 illustrates a flow diagram of a process 500 of wireless imaging in accordance with some implementations of the present disclosure. The process 500 may be implemented in any suitable imaging device, such as the imaging device shown in FIG. 1.

[0071] At block 510, a sensing signal is transmitted to the metasurface 130 and the target object 140 by the transmitter 110. At block 520, a target response signal is received from the target obj ect 140 by the receiver 120. For example, the target response signal may be a reflected signal that reaches the metasurface 130 or a transmitted signal that does not reach the metasurface 130.

[0072] In some implementations, the sensing signal is incident onto the target object 140 through the metasurface 130. The metasurface 130 may be positioned between the transmitter 110 and the target object 140 such that the sensing signal transmitted by the transmitter 110 is incident onto the target object 140 through the metasurface 130.

[0073] Altematively, or in addition, in some implementations, the sensing signal is incident onto the target object 140 and then reaches the receiver 120 through the metasurface 130. The transmitter 110 and the receiver 120 may be positioned on either side of the metasurface 130, respectively, such that the sensing signal transmitted by the transmitter 110 is incident onto the target object 140 and then reaches the receiver 120 through the metasurface 130.

[0074] Altematively, or in addition, in some implementations, the sensing signal is incident onto the target object 140 through the metasurface 130 and then reaches the receiver 120 through the metasurface 130. The transmitter 110 and the receiver 120 may be positioned on one side of the metasurface 130, so that the sensing signal transmitted by the transmitter 110 is incident onto the target object 140 through the metasurface 130 and then reaches the receiver 120 through the metasurface 130.

[0075] In some implementations, the sensing signal comprises at least one of a sound wave or anelectromagnetic wave. For example, the wireless imaging may be sound wave imaging or millimeter wave imaging.

[0076] In some implementations, the target image comprises at least one of: a one-dimensional image, a two-dimensional image, or a three-dimensional image.

[0077] In some implementations, a structure of the metasurface may be determined for a first imaging resolution and a first imaging distance. An imaging size of the target object may be determined according to an imaging distance for the target object, based on the first imaging resolution and the first imaging distance.

[0078] In some implementations, a response signal resolution may be determined based on the first imaging resolution and the first imaging distance, and the response signal resolution indicates a distance between separable peaks in the target response signal.

[0079] Due to reduced signal-to-noise ratio (SNR) and reduced angular resolution, the imaging quality degrades with the distance. To better support imaging across a varying distance, in this implementation, adaptive resolution image reconstruction may be adopted. The resolution between two pixels of the target may be scaled as a function of the distance d between the target object and the transceiver.

[0080] Specifically, let Ad denote the resolution of the reflected pulse (also referred to as the response signal) (two closest peaks in the reflected pulse that can be separated). Let r denote the separation between two closest pixels on an image plane that can be separated. The relationship Ad = d2+ r2— d may be obtained according to Pythagorean Theorem, where d denotes the distance (also referred to as the first imaging distance) between the closest pixel on the target plane and the transceiver.

[0081] In some implementations, after obtaining the response signal resolution, a second imaging resolution for the target object may be determined based on the response signal resolution and a second imaging distance for the target object, and the sensing signal and the imaging size are determined according to the second imaging resolution. The second imaging resolution may be represented as r = V2Add + Ad2~ jl dd, where d denotes the second imaging distance.

[0082] FIG. 6 illustrates a schematic diagram 600 of adaptive image resolution for different distances in accordance with some implementations of the present disclosure. As shown in FIG. 6, since the location of the target imaging area is known, the appropriate imaging resolution may be derived based on the relative distance of the target imaging area to the transceivers. Note that one resolution can support a range of distances (for example. 1cm resolution for imaging within 15cm, and 3cm resolution for imaging between 15cm and 135cm). In this way, by adjusting the resolution, high imaging quality can be achieved at high noise and at long distances.

[0083] At block 530, a target image of the target object is generated based on the target responsesignal and a signal measurement matrix describing an effect of a propagation path from transmitting the sensing signal to receiving the target response signal on a wireless signal.

[0084] In some implementations, an initial image of the target object may be reconstructed using a first model, based on the target response signal and the signal measurement matrix, and the target image may be generated from the initial image using a second model. With continuing reference to FIG. 4A, the unrolled ADMM network 410 (as an example of the first model) with a learnable layer may reconstruct an initial image of the target object. The refinement network (as an example of the second model) 420 may generate the target image from the initial image.

[0085] In some implementations, the first model comprises a plurality of alternate update layers for iteratively reconstructing the initial image with each alternate update layer being configured to alternately update the initial image and a corresponding multiplier. For example, an example of the first model described above with reference to FIG. 4B.

[0086] In some implementations, an image of the target object may be reconstructed as the target image using a diffusion model, based on the target response signal and the signal measurement matrix. The evolution of the data is simulated by gradually adding noise (forw ard diffusion) and removing noise (back diffusion), and an image corresponding to the target object is generated.

[0087] In some implementations, the first model and the second model may be corrected by: transmitting, by the transmitter 110, a reference sensing signal to the metasurface 130 and a reference object; receiving, by the receiver 120, a reference response signal from the reference object; generating a reconstructed image of the reference object using the first model and the second model based on the reference response signal and the signal measurement matrix; and updating model parameters of the first model and the second model based on a difference between the reconstructed image and a truth image of the reference object. For example, the loss function shown in Eq. (11) may be computed, and the model may be updated by minimizing the loss function. This correction process can be view ed as an on-site update to the first model and the second model, and such on-site update utilizes real environment data.[0088JFIG. 7 show s a schematic block diagram of an electronic device capable of implementing multiple implementations of the present disclosure. It should be understood that the electronic device 700 shown in FIG. 7 is merely an example and should not constitute any limitation on the functionality and scope of the implementations described in this disclosure.

[0089] As shown in FIG. 7, the electronic device 700 comprises an electronic device 700 in the form of a general-purpose computing device. Components of the electronic device 700 may include, but are not limited to, one or more processors or processing devices 710, a memory 720, a storage device 730, one or more communication units 740, one or more input devices 750, and one or more output devices 760.

[0090] In some implementations, the electronic device 700 may be implemented as a device with computing capability, such as a computing device, a computing system, a sen- er, a mainframe and the like.

[0091] The processing device 710 may be a physical or virtual processor and can execute various processing based on the programs stored in the memory 720. In a multi-processor system, a plurality of processing units execute computer-executable instructions in parallel to enhance parallel processing capability' of the electronic device 700. The processing device 710 may include a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor, a controller, and / or a microcontroller.

[0092] The electronic device 700 usually includes various computer storage medium. Such medium may be any available medium accessible by the electronic device 700, including but not limited to, volatile and non-volatile medium, or detachable and non-detachable medium. The memory 720 may be a volatile memory (for example, a register, cache, Random Access Memory (RAM)), non-volatile memory (for example, a Read-Only Memory (ROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), a flash memory), or any combination thereof. The storage device 730 may be any detachable or non-detachable medium and may include computer-readable medium such as a memory, a flash memory' drive, a magnetic disk or any other medium that can be used for storing information and / or data and are accessible by the electronic device 700.

[0093] The electronic device 700 may further include additional detachable / non-detachable, volatile / non-volatile memory' medium. Although not shown in FIG. 7, there may be provided a disk drive for reading from or writing into a detachable and non-volatile disk, and an optical disk drive for reading from and writing into a detachable non-volatile optical disc. In such cases, each drive may be connected to a bus (not shoyvn) via one or more data medium interfaces.

[0094] The communication unit 740 implements communication with another computing device via the communication medium. In addition, the functionalities of components in the electronic device 700 may be implemented by a single computing cluster or a plurality of computing machines that can communicate with each other via communication connections. Thus, the electronic device 700 may operate in a networked environment using a logic connection with one or more other servers, network personal computers (PCs), or further general netw ork nodes.

[0095] The input device 750 may include one or more of a variety of input devices, such as a mouse, keyboard, data import device and the like. The output device 760 may be one or more output devices, such as a display, data export device and the like. By means of the communication unit 740, the electronic device 700 may further communicate with one or more external devices (not shown) such as storage devices and display devices, one or more devices that enable the userto interact with the electronic device 700, or any devices (such as a network card, a modem and the like) that enable the electronic device 700 to communicate with one or more other computing devices, if required. Such communication may be performed via input / output (I / O) interfaces (not shown).

[0096] In some implementations, as an alternative of being integrated on a single device, some or all components of the electronic device 700 may also be arranged in the form of cloud computing architecture. In the cloud computing architecture, the components may be provided remotely and work together to implement the functionalities described in the subject matter described herein. In some implementations, cloud computing provides computing, software, data access and storage service, which will not require end users to be aware of the physical locations or configurations of the systems or hardware provisioning these services. In various implementations, the cloud computing provides the services via a wide area network (such as Internet) using proper protocols. F or example, a cloud computing provider provides applications over the wide area network, which may be accessed through a web browser or any other computing components. The software or components of the cloud computing architecture and corresponding data may be stored in a server at a remote position. The computing resources in the cloud computing environment may be aggregated or distributed at locations of remote data centers. Cloud computing infrastructure may provide the services through a shared data center, though they behave as a single access point for the users. Therefore, the cloud computing infrastructure may be utilized to provide the components and functionalities described herein from a service provider at remote locations. Alternatively, they may be provided from a conventional server or may be installed directly or otherwise on a client device.

[0097] The electronic device 700 may be used to implement resource management in accordance with various implementations of the subject matter described herein. The memory 720 may include one or more modules having one or more program instructions. These modules may be accessed and run by the processing unit 710 to perform functions of various implementations described herein. For example, the memory 720 may include an imaging module 725 for performing imaging according to the one or more implementations above. As shown in FIG. 7, the electronic device 700 may obtain an input for imaging via the input device 750, and provide output for imaging, such as the predicted properties, via the output device 760. In some implementations, the electronic device 700 may further receive the input from other devices (not shown) via the communication unit 740.EXAMPLE IMPLEMENTATIONS

[0098] Some example implementations of the present disclosure are listed below.

[0099] In one aspect, the disclosure provides a method of wireless imaging. The method comprises:transmiting, by a transmiter, a sensing signal to a metasurface and a target object; receiving, by a receiver, a target response signal from the target object; and generating a target image of the target object based on the target response signal and a signal measurement matrix describing an effect of a propagation path from transmiting the sensing signal to receiving the target response signal on a wireless signal.[OlOOJIn some implementations, the sensing signal is incident onto the target object through the metasurface, the sensing signal is incident onto the target object and then reaches the receiver through the metasurface, the sensing signal is incident onto the target object through a first portion of the metasurface and then reaches the receiver through a second portion of the metasurface, or the sensing signal is incident onto the target object through a first metasurface and then reaches the receiver through a second metasurface. The first portion and the second portion are a same portion or different portions. The first metasurface and the second metasurface are a same metasurface or different metasurfaces.[OlOlJIn some implementations, the sensing signal compnses at least one of a sound wave or an electromagnetic wave.

[0102] In some implementations, the target image comprises at least one of: a one-dimensional image, a two-dimensional image, or a three-dimensional image.

[0103] In some implementations, a structure of the metasurface is determined for a first imaging resolution and a first imaging distance, and the method further comprises: determining, based on the first imaging resolution and the first imaging distance, an imaging size of the target object according to an imaging distance for the target object.

[0104] In some implementations, determining the imaging size of the target object according to the imaging distance of the target object comprises: determining a response signal resolution based on the first imaging resolution and the first imaging distance, the response signal resolution indicating a distance between separable peaks in the target response signal; determining a second imaging resolution for the target object based on the response signal resolution and a second imaging distance for the target object; and determining the sensing signal and the imaging size according to the second imaging resolution.

[0105] In some implementations, generating the target image of the target object comprises: reconstructing, using a first model, an initial image of the target object based on the target response signal and the signal measurement matrix; and generating, using a second model, the target image from the initial image.

[0106] In some implementations, the first model comprises a plurality of alternate update layers for iteratively reconstructing the initial image with each alternate update layer being configured to alternately update the initial image and a corresponding multiplier.

[0107] In some implementations, generating the target image of the target object comprises: reconstructing, using a diffusion model, an image of the target object as the target image based on the target response signal and the signal measurement matrix.

[0108] In some implementations, the method further comprises correcting the first model and the second model by: transmitting, by the transmitter, a reference sensing signal to the metasurface and a reference object; receiving, by the receiver, a reference response signal from the reference object; generating a reconstructed image of the reference object using the first model and the second model based on the reference response signal and the signal measurement matrix; and updating model parameters of the first model and the second model based on a difference between the reconstructed image and a truth image of the reference object.

[0109] In some implementations, a method for optimizing wireless imaging is provided in the present disclosure. The method comprises: establishing, according to a sensing signal and a propagation path of the sensing signal, a signal measurement matrix indicating an effect of the propagation path on the sensing signal; determining, through the signal measurement matrix, a relationship between a response signal and each of the sensing signal, a metasurface, and a reference object, the metasurface and the reference object being located in a propagation path from the sensing signal to the response signal; simulating imaging of the reference object through the established relationship; and updating the signal measurement matrix based on the simulated result of the imaging until a predetermined condition is satisfied.[OllOJIn some implementations, the method further comprises determining, based on the updated signal measurement matrix, at least one of: a beamforming characteristic of a transmitter of the sensing signal, a beamforming characteristic of a receiver for receiving the response signal, or a structure of the metasurface.[OlllJIn some implementations, determining the structure of the metasurface comprises: determining an electrical characteristic or an acoustic characteristic of each metasurface cell of the metasurface based on the updated signal measurement matrix; and determining the structure of each metasurface cell according to the electrical characteristic or the acoustic characteristic of the metasurface cell.

[0112] In some implementations, determining the structure of each metasurface cell according to the electrical characteristic or the acoustic characteristic of the metasurface cell comprises: determined, for each location of the metasurface, the shape and parameter of the metasurface cell, to satisfy desired acoustic or electrical characteristics.

[0113] In some implementations, determining the structure of each metasurface cell according to the electrical characteristic or the acoustic characteristic of the metasurface cell comprises: selecting a target cell structure from a plurality of cell structures according to the acousticcharacteristic of the metasurface cell.

[0114] In some implementations, the metasurface comprises one or more metal layers, and determining the structure of the metasurface comprises: determining respective electrical characteristics of the one or more metal layers based on the updated signal measurement matrix; determining, using a machine learning model, respective structures of the one or more metal layers based on a respective electrical characteristic of the one or more metal layers.

[0115] In some implementations, the machine learning model comprises at least one of: a convolutional neural network, a generative adversarial network, or a diffusion model.

[0116] In some implementations, simulating the imaging of the reference object through the established relationship comprises: generating a simulated signal for the response signal based on the established relationship; reconstructing, using a first model, an initial image of the reference object based on the simulated signal and the signal measurement matrix; and generating, using a second model, a simulated image of the reference object from the initial image.

[0117] In some implementations, the first model comprises a plurality of alternate update layers for iteratively reconstructing the initial image with each alternate update layer being configured to alternately update the initial image and a corresponding multiplier.

[0118] In some implementations, before inputting the simulated signal to the first model, the method further comprises: adding noise to the simulated signal.

[0119] In some implementations, the metasurface comprises a metasurface cell for a millimeter wave, transmissivity and a phase modulation range of the metasurface cell for the millimeter wave at a target frequency satisfying a predetermined condition.

[0120] In some implementations, the metasurface cell for the millimeter wave comprises a dielectric layer and at least one metal layer, and a metal pattern in the at least one metal layer comprises a metal ring and a metal patch located inside the metal ring.

[0121] In another aspect, an electronic device is provided in the present disclosure. The electronic device comprises: a processing unit; a memory coupled to the processing unit and having instructions stored thereon which, when executed by the processing unit, causing the device to perform acts comprising: receiving, by a receiver, a target response signal from the target object; and generating a target image of the target object based on the target response signal and a signal measurement matrix describing an effect of a propagation path from transmitting the sensing signal to receiving the target response signal on a wireless signal.

[0122] In some implementations, the sensing signal is incident onto the target object through the metasurface, the sensing signal is incident onto the target object and then reaches the receiver through the metasurface, the sensing signal is incident onto the target object through the metasurface and then reaches the receiver through the metasurface.

[0123] In some implementations, the sensing signal comprises at least one of a sound wave or an electromagnetic wave.

[0124] In some implementations, the target image comprises at least one of: a one-dimensional image, a two-dimensional image, or a three-dimensional image.

[0125] In some implementations, a structure of the metasurface is determined for a first imaging resolution and a first imaging distance, and the acts comprises: determining, based on the first imaging resolution and the first imaging distance, an imaging size of the target object according to an imaging distance for the target object.

[0126] In some implementations, determining the imaging size of the target object according to the imaging distance of the target object comprises: determining a response signal resolution based on the first imaging resolution and the first imaging distance, the response signal resolution indicating a distance between separable peaks in the target response signal; determining a second imaging resolution for the target object based on the response signal resolution and a second imaging distance for the target object; and determining the sensing signal and the imaging size according to the second imaging resolution.

[0127] In some implementations, generating the target image of the target object comprises: reconstructing, using a first model, an initial image of the target object based on the target response signal and the signal measurement matrix; and generating, using a second model, the target image from the initial image.

[0128] In some implementations, the first model comprises a plurality of alternate update layers for iteratively reconstructing the initial image with each alternate update layer being configured to alternately update the initial image and a corresponding multiplier.

[0129] In some implementations, generating the target image of the target object comprises: reconstructing, using a diffusion model, an image of the target object as the target image based on the target response signal and the signal measurement matrix.

[0130] In some implementations, the acts further comprises correcting the first model and the second model by: transmitting, by the transmitter, a reference sensing signal to the metasurface and a reference object; receiving, by the receiver, a reference response signal from the reference object; generating a reconstructed image of the reference object using the first model and the second model based on the reference response signal and the signal measurement matrix; and updating model parameters of the first model and the second model based on a difference between the reconstructed image and a truth image of the reference object.

[0131] In yet another aspect, an electronic device is provided in the present disclosure. The electronic device comprises: a processing unit; a memory coupled to the processing unit and having instructions stored thereon which, when executed by the processing unit, causing thedevice to perform acts comprising: establishing, according to a sensing signal and a propagation path of the sensing signal, a signal measurement matrix indicating an effect of the propagation path on the sensing signal; determining, through the signal measurement matrix, a relationship between a response signal and each of the sensing signal, a metasurface, and a reference object, the metasurface and the reference object being located in a propagation path from the sensing signal to the response signal; simulating imaging of the reference object through the established relationship; and updating the signal measurement matrix based on the simulated result of the imaging until a predetermined condition is satisfied.

[0132] In some implementations, the acts further comprise determining, based on the updated signal measurement matrix, at least one of: a beamforming characteristic of a transmitter of the sensing signal, a beamforming characteristic of a receiver for receiving the response signal, or a structure of the metasurface.

[0133] In some implementations, determining the structure of the metasurface comprises: determining an electncal characteristic or an acoustic characteristic of each metasurface cell of the metasurface based on the updated signal measurement matrix; and determining the structure of each metasurface cell according to the electrical characteristic or the acoustic characteristic of the metasurface cell.

[0134] In some implementations, determining the structure of each metasurface cell according to the electrical characteristic or the acoustic characteristic of the metasurface cell comprises: determined, for each location of the metasurface, the shape and parameter of the metasurface cell, to satisfy desired acoustic or electrical characteristics.

[0135] In some implementations, determining the structure of each metasurface cell according to the electrical characteristic or the acoustic characteristic of the metasurface cell comprises: selecting a target cell structure from a plurality of cell structures according to the acoustic characteristic of the metasurface cell.

[0136] In some implementations, the metasurface comprises one or more metal layers, and determining the structure of the metasurface comprises: determining respective electrical characteristics of the one or more metal layers based on the updated signal measurement matrix; determining, using a machine learning model, respective structures of the one or more metal layers based on a respective electrical characteristic of the one or more metal layers.

[0137] In some implementations, the machine learning model comprises at least one of: a convolutional neural network, a generative adversarial network, or a diffusion model.

[0138] In some implementations, simulating the imaging of the reference object through the established relationship comprises: generating a simulated signal for the response signal based on the established relationship; reconstructing, using a first model, an initial image of the referenceobject based on the simulated signal and the signal measurement matrix; and generating, using a second model, a simulated image of the reference object from the initial image.

[0139] In some implementations, the first model comprises a plurality of alternate update layers for iteratively reconstructing the initial image with each alternate update layer being configured to alternately update the initial image and a corresponding multiplier.

[0140] In some implementations, the acts further comprises: before inputting the simulated signal to the first model, adding noise to the simulated signal.

[0141] In some implementations, the metasurface comprises a metasurface cell for a millimeter wave, transmissivity and a phase modulation range of the metasurface cell for the millimeter wave at a target frequency satisfying a predetermined condition.

[0142] In some implementations, the metasurface cell for the millimeter wave comprises a dielectric layer and at least one metal layer, and a metal pattern in the at least one metal layer comprises a metal ring and a metal patch located inside the metal ring.

[0143] In a further aspect, a device for wireless imaging is provided in the present disclosure. The device for wireless imaging comprises: a transmitter configured to transmit a sensing signal to a metasurface and a target object; a receiver configured to receive a target response signal from the target object; and a controller configured to generate a target image of the target object based on the target response signal and a signal measurement matrix describing an effect of a propagation path from transmitting the sensing signal to receiving the target response signal on a wireless signal.

[0144] In some implementations, the sensing signal is incident onto the target object through the metasurface, or the sensing signal is incident onto the target object and then reaches the receiver through the metasurface, or the sensing signal is incident onto the target object through the metasurface and then reaches the receiver through the metasurface.

[0145] In some implementations, the sensing signal comprises at least one of a sound wave or an electromagnetic wave.

[0146] In some implementations, the target image comprises at least one of: a one-dimensional image, a two-dimensional image, or a three-dimensional image.

[0147] In some implementations, a structure of the metasurface is determined for a first imaging resolution and a first imaging distance, and the controller is further configured to determine, based on the first imaging resolution and the first imaging distance, an imaging size of the target object according to an imaging distance for the target object.

[0148] In some implementations, determining the imaging size of the target object according to the imaging distance of the target object comprises: determining a response signal resolution based on the first imaging resolution and the first imaging distance, the response signal resolutionindicating a distance between separable peaks in the target response signal; determining a second imaging resolution for the target object based on the response signal resolution and a second imaging distance for the target object; and determining the sensing signal and the imaging size according to the second imaging resolution.

[0149] In some implementations, generating the target image of the target object comprises: reconstructing, using a first model, an initial image of the target object based on the target response signal and the signal measurement matrix; and generating, using a second model, the target image from the initial image.

[0150] In some implementations, the first model comprises a plurality of alternate update layers for iteratively reconstructing the initial image with each alternate update layer being configured to alternately update the initial image and a corresponding multiplier.

[0151] In some implementations, generating the target image of the target object comprises: reconstructing, using a diffusion model, an image of the target object as the target image based on the target response signal and the signal measurement matrix.

[0152] In some implementations, the transmitter is further configured to: transmit a reference sensing signal to the metasurface and a reference object; the receiver is further configured to receive, a reference response signal from the reference object; the controller is further configured to generate a reconstructed image of the reference object using the first model and the second model based on the reference response signal and the signal measurement matrix; and updating model parameters of the first model and the second model based on a difference between the reconstructed image and a truth image of the reference object.

[0153] In another aspect, a computer program product is provided in the present disclosure. The computer program product tangibly stored in a computer storage medium and comprises computer-executable instructions which, when executed by a device, cause the device to perform acts comprising: receiving, by a receiver, a target response signal from the target object; and generating a target image of the target object based on the target response signal and a signal measurement matrix describing an effect of a propagation path from transmitting the sensing signal to receiving the target response signal on a wireless signal.

[0154] In some implementations, the sensing signal is incident onto the target object through the metasurface, the sensing signal is incident onto the target object and then reaches the receiver through the metasurface, the sensing signal is incident onto the target object through the metasurface and then reaches the receiver through the metasurface.

[0155] In some implementations, the sensing signal comprises at least one of a sound wave or an electromagnetic wave.

[0156] In some implementations, the target image comprises at least one of: a one-dimensionalimage, a two-dimensional image, or a three-dimensional image.

[0157] In some implementations, a structure of the metasurface is determined for a first imaging resolution and a first imaging distance, and the acts comprises: determining, based on the first imaging resolution and the first imaging distance, an imaging size of the target object according to an imaging distance for the target object.

[0158] In some implementations, determining the imaging size of the target object according to the imaging distance of the target object comprises: determining a response signal resolution based on the first imaging resolution and the first imaging distance, the response signal resolution indicating a distance between separable peaks in the target response signal; determining a second imaging resolution for the target object based on the response signal resolution and a second imaging distance for the target object; and determining the sensing signal and the imaging size according to the second imaging resolution.

[0159] In some implementations, generating the target image of the target object comprises: reconstructing, using a first model, an initial image of the target object based on the target response signal and the signal measurement matrix; and generating, using a second model, the target image from the initial image.

[0160] In some implementations, the first model comprises a plurality of alternate update layers for iteratively reconstructing the initial image with each alternate update layer being configured to alternately update the initial image and a corresponding multiplier.

[0161] In some implementations, generating the target image of the target object comprises: reconstructing, using a diffusion model, an image of the target object as the target image based on the target response signal and the signal measurement matrix.

[0162] In some implementations, the acts further comprises correcting the first model and the second model by: transmitting, by the transmitter, a reference sensing signal to the metasurface and a reference object; receiving, by the receiver, a reference response signal from the reference object; generating a reconstructed image of the reference object using the first model and the second model based on the reference response signal and the signal measurement matrix; and updating model parameters of the first model and the second model based on a difference between the reconstructed image and a truth image of the reference object.

[0163] In yet another aspect, a computer program product is provided in the present disclosure. The computer program product is tangibly stored in a computer storage medium and includes computer-executable instructions. The computer-executable instructions, when executed by the device, cause the device to perform actions comprising:

[0164] In some implementations, the acts further comprise determining, based on the updated signal measurement matrix, at least one of: a beamforming characteristic of a transmitter of thesensing signal, a beamforming characteristic of a receiver for receiving the response signal, or a structure of the metasurface.

[0165] In some implementations, determining the structure of the metasurface comprises: determining an electrical characteristic or an acoustic characteristic of each metasurface cell of the metasurface based on the updated signal measurement matrix; and determining the structure of each metasurface cell according to the electrical characteristic or the acoustic characteristic of the metasurface cell.

[0166] In some implementations, determining the structure of each metasurface cell according to the electrical characteristic or the acoustic characteristic of the metasurface cell comprises: determined, for each location of the metasurface, the shape and parameter of the metasurface cell, to satisfy desired acoustic or electrical characteristics.

[0167] In some implementations, determining the structure of each metasurface cell according to the electrical characteristic or the acoustic characteristic of the metasurface cell comprises: selecting a target cell structure from a plurality of cell structures according to the acoustic characteristic of the metasurface cell.

[0168] In some implementations, the metasurface comprises one or more metal layers, and determining the structure of the metasurface comprises: determining respective electrical characteristics of the one or more metal layers based on the updated signal measurement matrix; determining, using a machine learning model, respective structures of the one or more metal layers based on a respective electrical characteristic of the one or more metal layers.

[0169] In some implementations, the machine learning model comprises at least one of: a convolutional neural network, a generative adversarial network, or a diffusion model.

[0170] In some implementations, simulating the imaging of the reference object through the established relationship comprises: generating a simulated signal for the response signal based on the established relationship; reconstructing, using a first model, an initial image of the reference object based on the simulated signal and the signal measurement matrix; and generating, using a second model, a simulated image of the reference object from the initial image.

[0171] In some implementations, the first model comprises a plurality of alternate update layers for iteratively reconstructing the initial image with each alternate update layer being configured to alternately update the initial image and a corresponding multiplier.

[0172] In some implementations, the acts further comprises: before inputting the simulated signal to the first model, adding noise to the simulated signal.

[0173] In some implementations, the metasurface comprises a metasurface cell for a millimeter wave, transmissivity and a phase modulation range of the metasurface cell for the millimeter wave at a target frequency satisfying a predetermined condition.

[0174] In some implementations, the metasurface cell for the millimeter wave comprises a dielectric layer and at least one metal layer, and a metal pattern in the at least one metal layer comprises a metal ring and a metal patch located inside the metal ring.

[0175] In yet another aspect, the disclosure provides a computer-readable medium having stored thereon computer-executable instructions which, when executed by a device, cause the device to perform one or more example implementations of the methods of the above aspects.

[0176] The functionalities described herein above may be performed, at least in part, by one or more hardware logic components. For example, without limitation, example types of hardware logic components that can be used include Field-programmable Gate Arrays (FPGAs), Application-specific Integrated Circuits (ASICs), Application-specific Standard Products (ASSPs), System-on-a-chip systems (SOCs), Load Programmable Logic Devices (CPLDs), and so on.

[0177] Program code for carrying out the methods of the subject matter described herein may be written in any combination of one or more programming languages. The program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing flowchart such that the program code, when executed by the processor or controller, causes the functions / operations specified in the flow-charts and / or block diagrams to be implemented. The program code may be executed entirely or partly on a machine, executed as a stand-alone software package partly on the machine, partly on a remote machine, or entirely on the remote machine or server.

[0178] In the context of the subject matter described herein, a machine-readable medium may be any tangible medium that may contain or store a program for use by or in connection with an instruction execution system, flowchart, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include but is not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, flowchart, or device, or any suitable combination of the foregoing. More specific examples of the machine-readable storage medium would include an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random-access memory (RAM), a read-only memory (ROM), an erasable programmable readonly memory- (EPROM or Flash memory), an optical fiber, a portable compact disc read-only- memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0179] Further, although the operations are depicted in a particular order, this should not be understood as requiring that such operations are performed in the particular order shown or in sequential order, or that all illustrated operations are performed to achieve the desired results. Incertain circumstances, multitasking and parallel processing may be advantageous. Likewise, while several specific implementation details are contained in the above discussions, these should not be construed as limitations on the scope of the subject matter described herein, but rather as descriptions of features that may be specific to particular implementations. Certain features that are described in the context of separate implementations may also be implemented in combination in a single implementation. Rather, various features described in a single implementation may also be implemented in various implementations separately or in any suitable sub-combination.

[0180] Although the subject matter has been described in language specific to structural features and / or methodological acts, it is to be understood that the subject matter specified in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.

Claims

CLAIMS1. A method of wireless imaging, comprising: transmitting, by a transmitter, a sensing signal to a metasurface and a target object; receiving, by a receiver, a target response signal from the target obj ect; and generating a target image of the target object based on the target response signal and a signal measurement matrix describing an effect of a propagation path from transmitting the sensing signal to receiving the target response signal on a wireless signal.

2. The method of claim 1, wherein the sensing signal is incident onto the target object through the metasurface, or the sensing signal is incident onto the target object and then reaches the receiver through the metasurface, or the sensing signal is incident on the target object through a first portion of the metasurface and then reaches the receiver through a second portion of the metasurface, the first portion and the second portion being a same portion or different portions.

3. The method of claim 1, wherein a structure of the metasurface is determined for a first imaging resolution and a first imaging distance, and the method further comprises: determining, based on the first imaging resolution and the first imaging distance, an imaging size of the target object according to an imaging distance for the target object.

4. The method of claim 3, wherein determining the imaging size of the target object according to the imaging distance of the target object comprises: determining a response signal resolution based on the first imaging resolution and the first imaging distance, the response signal resolution indicating a distance between separable peaks in the target response signal; determining a second imaging resolution for the target object based on the response signal resolution and a second imaging distance for the target object; and determining the sensing signal and the imaging size according to the second imaging resolution.

5. The method of claim 1, wherein generating the target image of the target object comprises: reconstructing, using a first model, an initial image of the target object based on the target response signal and the signal measurement matrix; and generating, using a second model, the target image from the initial image.

6. The method of claim 5, wherein the first model comprises a plurality of alternate update layers for iteratively reconstructing the initial image with each alternate update layer being configured to alternately update the initial image and a corresponding multiplier.

7. The method of claim 1, wherein generating the target image of the target object comprises: reconstructing, using a diffusion model, an image of the target object as the target image based onthe target response signal and the signal measurement matrix.

8. The method of claim 5, further comprising correcting the first model and the second model by: transmitting, by the transmitter, a reference sensing signal to the metasurface and a reference object; receiving, by the receiver, a reference response signal from the reference object; generating a reconstructed image of the reference object using the first model and the second model based on the reference response signal and the signal measurement matrix; and updating model parameters of the first model and the second model based on a difference between the reconstructed image and a truth image of the reference object.

9. A method for optimizing wireless imaging, comprising: establishing, according to a sensing signal and a propagation path of the sensing signal, a signal measurement matrix indicating an effect of the propagation path on the sensing signal; determining, through the signal measurement matrix, a relationship between a response signal and each of the sensing signal, a metasurface, and a reference object, the metasurface and the reference object being located in a propagation path from the sensing signal to the response signal; simulating imaging of the reference object through the established relationship; and updating the signal measurement matrix based on the simulated result of the imaging until a predetermined condition is satisfied.

10. The method of claim 9, further comprising determining, based on the updated signal measurement matrix, at least one of: a beamforming characteristic of a transmitter of the sensing signal, a beamforming characteristic of a receiver for receiving the response signal, or a structure of the metasurface.

11. The method of claim 10, wherein determining the structure of the metasurface comprises: determining an electrical characteristic or an acoustic characteristic of each metasurface cell of the metasurface based on the updated signal measurement matrix; and determining the structure of each metasurface cell according to the electrical characteristic or the acoustic characteristic of the metasurface cell.

12. The method of claim 10, wherein the metasurface comprises one or more metal layers, and determining the structure of the metasurface comprises: determining respective electrical characteristics of the one or more metal layers based on the updated signal measurement matrix; determining, using a machine learning model, respective structures of the one or more metal layers based on a corresponding electrical characteristic of the one or more metal layers.

13. The method of claim 9, wherein simulating the imaging of the reference object through theestablished relationship comprises: generating a simulated signal for the response signal based on the established relationship; reconstructing, using a first model, an initial image of the reference object based on the simulated signal and the signal measurement matrix; and generating, using a second model, a simulated image of the reference object from the initial image.

14. The method of claim 11, wherein the metasurface comprises a metasurface cell for a millimeter wave, transmissivity and a phase modulation range of the metasurface cell for the millimeter wave at a target frequency satisfy ing a predetermined condition.

15. The method of claim 14, wherein the metasurface cell for the millimeter wave comprises a dielectric layer and at least one metal layer, and a metal pattern in the at least one metal layer comprises a metal ring and a metal patch located inside the metal ring.

Images