An intelligent beam modulation system based on a programmable transmissive metasurface

Abstract

The present invention discloses an intelligent beam modulation system based on a programmable transmissive metasurface, which is applied to the field of network communication technology. It includes a programmable transmissive metasurface and a modulation network connected in sequence. The incident electromagnetic wave forms a far-field scattering after passing through the programmable transmissive metasurface, and the far-field scattering data is used as input data to be input into the modulation network, and the modulation network outputs a modulation scheme to control the programmable transmissive metasurface. The present invention designs an adjustable unit with high light transmittance to construct a two-dimensional adjustable metasurface. Taking the far-field scattering data as the input of the network, the network makes predictions and adjustments, and outputs the programming strategy of the adjustable unit of the metasurface, which has good prediction and optimization capabilities for different types of input data, can realize intelligent modulation of the incident signal, enable the signal to bypass obstacles and reach the user side seamlessly, thereby improving the overall quality of communication signals.

Description

Technical Field

[0001] The present invention relates to the field of network communication technologies, and more particularly to an intelligent beam modulation system based on a programmable transmissive metasurface. Background Art

[0002] With the continuous development of network communication, the Internet has entered the era of industrial Internet from the era of mobile Internet. Various new technologies such as AICDE (Artificial Intelligence, Internet of Things, Cloud Computing, Big Data, Edge Computing), combined with enhanced mobile bandwidth, large-scale Internet of Things, and extremely reliable real-time communication in three major 5G scenarios, provide a new engine for the digital transformation of industries. However, while 5G brings more convenience to people's lives, it also poses higher requirements for network quality. In particular, high-frequency signals have a large path loss, a small transmission radius, and are greatly affected by obstacles, weather, and environmental absorption.

[0003] Although 5G has applied many new technologies in aspects such as the air interface and network architecture to bring a brand-new Internet experience to users, in densely populated areas, such as areas with numerous office buildings, various complex obstacles will be encountered during the signal transmission process, and at this time, the network quality received by users will pose challenges. In existing technical solutions, when operators monitor through various indicators that the network coverage in a certain area cannot meet user needs, in order not to affect user experience, they will optimize the coverage of wireless base stations, increase the number of base stations or the transmission power of base stations in the relevant area, and continuously optimize the network to meet user needs. However, although this method can also solve the above problems, it may also affect users in another area. In addition, if it is necessary to increase the base station transmission power or the number of base stations, it means greater investment, and the operating cost of the operator will also increase accordingly. Therefore, how to provide an intelligent beam modulation system based on a programmable transmissive metasurface is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0004] In view of this, the present invention provides an intelligent beam modulation system based on a programmable transmissive metasurface, which uses the metasurface placed outside the building as a dynamic coordinator to intelligently modulate the incident signal, so that the signal can bypass obstacles and reach the user side seamlessly, thereby improving the overall quality of communication signals.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] An intelligent beam modulation system based on a programmable transmissive metasurface, comprising a programmable transmissive metasurface and a modulation network. The incident electromagnetic wave forms a far-field scattering after passing through the programmable transmissive metasurface. The far-field scattering data is used as input data and input into the modulation network, and the modulation network outputs a modulation scheme to control the programmable transmissive metasurface.

[0007] Optionally, the programmable transmissive metasurface is provided with x tunable units with different capacitance sequences arranged non-periodically along the horizontal direction, and y tunable units with different capacitance sequences arranged non-periodically along the vertical direction.

[0008] Optionally, the tunable unit includes two dielectric layers and one PP layer. The relative dielectric constants of the two dielectric layers are 4.5 and 4.4 respectively, and two PIN diodes are embedded at the top of the tunable unit.

[0009] Optionally, the modulation network includes two convolutional layers and two fully connected layers. The activation function is sigmoid. During the training process of the modulation network, the adam method is selected for gradient descent, the gradient threshold is set to 20, and the learning rate is 0.005.

[0010] Optionally, before training the modulation network, the input far-field scattering data is normalized, the output binary phase is replaced with 0 and 1, and 15,000 groups of data are generated by combining MATLAB and CST. 80% is used as the training set, and 20% is used as the validation set for training the modulation network.

[0011] Optionally, the loss function of the modulation network is:

[0012]

[0013] In the formula, n represents the number of samples, represents the predicted value, y i and represents the true value, i ∈ [1, n]. When the predicted value is exactly the same as the true value, MAE is equal to 0, indicating that the model is perfect. The greater the error, the greater the MAE value.

[0014] Optionally, after training is completed, multiple groups of data randomly selected from the test set are used to verify the fitting ability of the modulation network, and the prediction accuracy is calculated:

[0015]

[0016] In the formula, Accuracy is the prediction accuracy, Num total is the total data volume, Num pre is the number of correctly predicted data.

[0017] As can be seen from the above technical solutions, compared with the prior art, the present invention provides an intelligent beam modulation system based on a programmable transmissive metasurface, which has the following beneficial effects: The present invention designs an adjustable unit with high light transmittance, constructs a two-dimensional adjustable metasurface, places an irregular object above the metasurface, uses the far-field scattering data generated in this scenario as the input of the network, uses the predefined scattering field as the objective function, and the network makes predictions and adjustments to output the programming strategy of the metasurface. The network has good prediction and optimization capabilities for different types of input data; the network outputs a series of voltage modulation sequences to change the state of each unit on the metasurface, resulting in changes in the overall scattering field, and finally realizes signal regulation; the intelligent beam modulation system can realize intelligent modulation of the incident signal, enabling the signal to bypass obstacles and reach the user side seamlessly, thereby improving the overall quality of communication signals. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on the provided drawings.

[0019] Figure 1 Schematic diagram of the incident electromagnetic environment of the present invention;

[0020] Figure 2 Schematic diagram of the modulation network structure of the present invention;

[0021] Figure 3 Schematic diagram of the overall structure of the adjustable unit of the present invention;

[0022] Figure 4 Schematic diagram of the bottom structure of the adjustable unit of the present invention;

[0023] Figure 5 Schematic diagram of the top structure of the adjustable unit of the present invention;

[0024] Figure 6 Schematic diagram of the simulated transmission amplitude of the adjustable unit of the present invention;

[0025] Figure 7 Schematic diagram of the transmission phase of the diode of the adjustable unit of the present invention in different states;

[0026] Figure 8 Schematic diagram of the first beam modulation result in the embodiment of the present invention;

[0027] Figure 9 Schematic diagram of the second beam modulation result in the embodiment of the present invention;

[0028] Figure 10 This is the schematic diagram of the third beam modulation result in the embodiments of the present invention;

[0029] Figure 11 This is the schematic diagram of the fourth beam modulation result in the embodiments of the present invention. Detailed implementation manners

[0030] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0031] The embodiments of the present invention disclose an intelligent beam modulation system based on a programmable transmissive metasurface, which includes a programmable transmissive metasurface and a modulation network. The incident electromagnetic wave forms a far-field scattering after passing through the programmable transmissive metasurface, and the far-field scattering data is used as input data to be input into the modulation network, and the modulation network outputs a modulation scheme to control the programmable transmissive metasurface.

[0032] As Figure 1 shown, the incident wave and the programmable transmissive metasurface together constitute an incident electromagnetic environment. In the embodiments of the present invention, there are obstacles behind the programmable transmissive metasurface. After the intelligent modulation of the programmable transmissive metasurface and the modulation network, the signal bypasses the obstacles, improving the overall quality of the communication signal.

[0033] Further, the programmable transmissive metasurface is provided with x tunable units with different capacitance sequences arranged non-periodically along the horizontal direction and y tunable units with different capacitance sequences arranged non-periodically along the vertical direction.

[0034] Furthermore, in the embodiments of the present invention, the number of tunable units arranged in the horizontal direction is 43, the number of tunable units arranged in the vertical direction is 21, and the capacitance sequence of the tunable units in the vertical direction is consistent with that of the first tunable unit in this column.

[0035] Further, as Figure 3 shown, the tunable unit includes two dielectric layers and one PP layer. The relative dielectric constants of the two dielectric layers are 4.5 and 4.4 respectively, and two PIN diodes are embedded at the top of the tunable unit.

[0036] Furthermore, in the embodiments of the present invention, the bottom and top structures of the tunable unit are respectively as Figure 4 and Figure 5As shown, where p = 19 mm, d1 = 6.2 mm, px = 10 mm, py = 6.5 mm, w0 = 4 mm, w1 = 4.6 mm, w2 = 2 mm, w3 = 2.5 mm. The PIN diode uses a MACOM MA4FCP305 flip chip, with a resistance of 1.7 Ω in the on state and a capacitance of 50 fF in the off state. The connection method of "+-+-" is used to ensure that only one diode is in the conductive state each time power is supplied. The transmission amplitude of the cell simulated using CST2023 software, and the results are as Figure 6 shown. The solid line and the dashed line respectively represent different states of the diode. It can be seen that this structure has a high transmittance at the operating frequency (3.5 GHz), and the amplitude is better than -0.8 dB. Figure 7 shows that when the diode is in different operating states, when the current flows in the opposite direction, a transmission phase of just flipped 180° is generated.

[0037] Furthermore, as Figure 2 shown, the modulation network includes two convolutional layers and two fully connected layers, the activation function is sigmoid, in the training process of the modulation network, the adam method is selected for gradient descent, the gradient threshold is set to 20, and the learning rate is 0.005, so that the occurrence of overfitting can be prevented during the training process.

[0038] Furthermore, before training the modulation network, the input far-field scattering data is normalized, the output binary phase is replaced with 0 and 1, and 15,000 groups of data are generated by combining MATLAB and CST. 80% is used as the training set, and 20% is used as the validation set for training the modulation network.

[0039] Furthermore, the loss function of the modulation network is:

[0040]

[0041] In the formula, n represents the number of samples, represents the predicted value, y i and represents the true value, i ∈ [1, n]. When the predicted value is exactly the same as the true value, MAE is equal to 0, indicating that the model is perfect. The greater the error, the greater the MAE value.

[0042] Furthermore, after training is completed, multiple groups of data randomly selected from the test set are used to verify the fitting ability of the modulation network, and the prediction accuracy is calculated:

[0043]

[0044] In the formula, Accuracy is the prediction accuracy, Num total is the total data volume, Num pre is the number of correctly predicted data.

[0045] Furthermore, in the embodiment of the present invention, the loss of the validation set is 0.45, and the loss of the training set is 0.42. Four groups of data randomly selected from the test set are used to verify the fitting ability of the network. The average prediction accuracy of these four cases is calculated to be 93%, that is, 40 out of 43 phases are correctly predicted. The phase of each unit of the programmable transmissive metasurface along the horizontal direction is modified according to the four groups of data, and CST2023 is used for full-wave time-domain simulation. The incident direction of the electromagnetic wave is perpendicular incidence, and the near-field and far-field scattering are observed. The results of the four groups of data are respectively as Figure 8 , Figure 9 , Figure 10 and Figure 11 shown. It can be seen that the theoretical curve and the predicted curve have a high degree of consistency. Calculate the correlation coefficient r(E Theory ,E Test ):

[0046]

[0047] C(E Theory ,E Test ) = E[(E Theory -E[E Theory])( E Test -E[E Test )]

[0048] = E[E Theory E Test -2E[E Test E[E Theory +E[E Theory E[E Test

[0049] = E[E Theory E Test -E[E Theory E[E Test

[0050] In the formula, C(E Theory ,E Test ) is the covariance of E Theory and E Test , Var[E Theory is the variance of E Theory , and Var[E Test is the variance of E Test . E is the mathematical expectation. In the embodiment of the present invention, the correlation coefficients of the four cases are all above 95%.

[0051] ​​In the present specification, the various embodiments are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the various embodiments, reference can be made to each other. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple. For the relevant parts, reference can be made to the description in the method section.

[0052] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features disclosed herein.

Claims

1. An intelligent beam modulation system based on a programmable transmissive metasurface, characterized in that, it includes a programmable transmissive metasurface and a modulation network. The incident electromagnetic wave forms far-field scattering through the programmable transmissive metasurface, and the far-field scattering data is used as input data to input into the modulation network, and the modulation network outputs a modulation scheme to control the programmable transmissive metasurface; the programmable transmissive metasurface is provided with x tunable units with different capacitance sequences arranged non-periodically along the horizontal direction and y tunable units with different capacitance sequences arranged non-periodically along the vertical direction; the tunable unit includes two dielectric layers and one PP layer, the relative dielectric constants of the two dielectric layers are 4.5 and 4.4 respectively, and two PIN diodes are embedded at the top of the tunable unit; the modulation network includes two convolutional layers and two fully connected layers, the activation function is sigmoid, in the training process of the modulation network, the adam method is selected for gradient descent, the gradient threshold is set to 20, and the learning rate is 0.005; before the modulation network is trained, the input far-field scattering data is normalized, the output binary phase is replaced with 0 and 1, and 15,000 groups of data are generated by combining MATLAB and CST, 80% of which is used as the training set and 20% is used as the validation set for the modulation network training; the loss function of the modulation network is: Where n represents the number of samples, represents the predicted value, and y i represents the true value, i ∈ [1, n]. When the predicted value is exactly the same as the true value, MAE is equal to 0, indicating that the model is perfect. The larger the error, the larger the MAE value; after the training is completed, multiple groups of data randomly selected from the test set are used to verify the fitting ability of the modulation network, and the prediction accuracy is calculated: Where, Accuracy is the prediction accuracy, Num total is the total data volume, Num pre is the correctly predicted data volume.

Images