Self-evolving reconfigurable intelligent surface and control method based on self-supervised neural differential mapping

Abstract

The present application discloses a self-evolving reconfigurable intelligent surface and control method based on self-supervised neural differential mapping, which includes: a self-supervised neural differential mapping coding module, which is used to output the corresponding coding matrix in real time according to the requirements of the target beam based on the coding strategy of the preset self-supervised neural differential mapping coding algorithm, and autonomously evolve and continuously transfer the learning of the coding strategy in different application scenarios; a reconfigurable intelligent surface, which is used to use the coding matrix to perform the corresponding target electromagnetic function among multiple electromagnetic functions; an artificial intelligence chip, which is used to run the preset self-supervised neural differential mapping coding algorithm, autonomously learn the coding strategy and optimize the coding matrix; a phase control circuit board, which is used to adjust the bias voltage of the reconfigurable intelligent surface according to the coding matrix to achieve the target electromagnetic function. Thus, the problem that the related technology cannot provide accurate digital coding for complex electromagnetic functions is solved.

Description

Technical Field

[0001] The present application relates to the field of wireless communication technology, and in particular to a self-evolving reconfigurable intelligent surface and a control method based on self-supervised neural differential mapping. Background Art

[0002] Reconfigurable smart surface is one of the key technologies of the next generation mobile communication technology 6G. Reconfigurable smart surface is an artificial electromagnetic surface, which is a sub-wavelength artificial structure with integrated active devices arranged periodically or quasi-periodically on a two-dimensional plane. It can control the electromagnetic characteristics such as phase, amplitude and polarization of electromagnetic waves through the state of digital coding units, realize diversified electromagnetic functions, and build a new paradigm of intelligent and programmable wireless communication.

[0003] Efficient and accurate coding methods for reconfigurable smart surfaces are a long-standing challenge. In related technologies, nonlinear optimization methods, such as genetic algorithms and particle swarm algorithms, are mainly used to encode reconfigurable smart surfaces to achieve complex electromagnetic functions. In addition, in practical applications, the digital coding of reconfigurable smart surfaces is generally pre-calculated offline according to application requirements, and stored in a field programmable gate array (FPGA). The digital coding is output online in real time, and the bias voltage of the control unit is controlled by the phase control circuit board to achieve the target electromagnetic function.

[0004] However, related technologies, especially nonlinear optimization methods, have high computational complexity and long calculation time, and cannot provide real-time response for digital encoding of reconfigurable smart surfaces in complex electromagnetic functional application scenarios. Summary of the invention

[0005] The present application provides a self-evolving reconfigurable intelligent surface and control method based on self-supervised neural differential mapping to solve the problem that related technologies cannot provide accurate digital encoding for complex electromagnetic functions.

[0006] The first aspect of the present application provides a self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping, including: a self-supervised neural differential mapping coding module, which is used to output a corresponding coding matrix in real time according to the requirements of the target beam based on the coding strategy of a preset self-supervised neural differential mapping coding algorithm, and autonomously evolve and continuously transfer and learn the coding strategy in different application scenarios; a reconfigurable intelligent surface, which is used to use the coding matrix to perform a corresponding target electromagnetic function among multiple electromagnetic functions; an artificial intelligence chip, which is used to run a preset self-supervised neural differential mapping coding algorithm, autonomously learn the coding strategy and optimize the coding matrix; and a phase control circuit board, which is used to adjust the bias voltage of the reconfigurable intelligent surface according to the coding matrix to achieve the target electromagnetic function.

[0007] Optionally, in one embodiment of the present application, the relationship between the reconfigurable smart surface beam pattern and the unit phase shift is expressed as:

[0008]

[0009] Among them, reconfigurable smart surfaces have units, , matrix element , and They are The phase shift and reflection coefficient of each unit, .

[0010] Optionally, in one embodiment of the present application, the self-supervised neural differential mapping encoding module includes: a determination unit, used to determine the upper and lower bounds of the beam corresponding to the target electromagnetic function; an establishment unit, used to establish a first phase shift matrix set and a second phase shift matrix set that compete with each other according to the upper and lower bounds of the beam; an encoding unit, used to build a neural differential mapping algorithm, and construct an objective function to train the neural differential mapping algorithm to output an optimal phase shift matrix, which is quantized into the encoding matrix.

[0011] Optionally, in one embodiment of the present application, the objective function is the error between the beam pattern and the upper and lower boundaries of the target beam, wherein the expression of the objective function is:

[0012]

[0013] in, are the weights of different parts of the objective function, and is the domain defining the upper boundary of the target beam and its side lobes, is the domain defining the lower boundary of the target beam.

[0014] Optionally, in one embodiment of the present application, the reconfigurable smart surface is an artificial electromagnetic surface in which active devices are integrated into artificial units.

[0015] The second aspect of the present application provides a control method for a self-evolving reconfigurable intelligent surface based on a self-supervised neural differential mapping, comprising the following steps: based on a coding strategy of the preset self-supervised neural differential mapping coding algorithm, outputting a corresponding coding matrix in real time according to the requirements of a target beam, wherein the coding strategy performs autonomous evolution and continuous transfer learning according to different application scenarios;

[0016] The bias voltage of the reconfigurable smart surface is adjusted according to the coding matrix to utilize the coding matrix to execute the corresponding target electromagnetic function among the multiple electromagnetic functions, and the preset self-supervised neural differential mapping coding algorithm is run to autonomously learn the coding strategy and optimize the coding matrix.

[0017] The third aspect of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement a control method for a self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping as described in the above embodiment.

[0018] The fourth aspect of the present application provides a computer-readable storage medium, which stores a computer program. When the program is executed by a processor, it implements the control method of the self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping as described above.

[0019] Therefore, the embodiments of the present application have the following beneficial effects:

[0020] The embodiment of the present application can be based on the coding strategy of the preset self-supervised neural differential mapping coding algorithm, output the corresponding coding matrix in real time according to the requirements of the target beam, autonomously evolve and continuously transfer the learning of coding strategies in different application scenarios, and use the coding matrix to execute the corresponding target electromagnetic functions in a variety of electromagnetic functions, and then run the preset self-supervised neural differential mapping coding algorithm, autonomously learn the coding strategy and optimize the coding matrix, and adjust the bias voltage of the reconfigurable smart surface according to the coding matrix to achieve the target electromagnetic function, thereby integrating the self-supervised neural differential mapping coding algorithm into the reconfigurable smart surface through the artificial intelligence chip to achieve efficient and accurate digital coding. Thus, the problem that the related technology cannot provide accurate digital coding for complex electromagnetic functions is solved.

[0021] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0023] Figure 1 is an example diagram of a self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping according to an embodiment of the present application;

[0024] Figure 2 A schematic diagram of execution logic of a self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping according to an embodiment of the present application;

[0025] Figure 3 A schematic diagram of the actual architecture of a self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping according to an embodiment of the present application;

[0026] Figure 4 A schematic diagram of a beam upper boundary provided according to an embodiment of the present application;

[0027] Figure 5 A schematic diagram of a beam lower boundary provided according to an embodiment of the present application;

[0028] Figure 6 A one-dimensional schematic diagram of upper and lower boundaries of a beam provided according to an embodiment of the present application;

[0029] Figure 7 A schematic diagram of a process of quantizing a continuous phase shift distribution output by a self-supervised neural differential mapping algorithm into a digital encoding according to an embodiment of the present application;

[0030] Figure 8 A first schematic diagram of a self-supervised neural differential mapping algorithm encoding a 1-bit reconfigurable smart surface to implement dual-beam scanning according to an embodiment of the present application;

[0031] Fig. 9 A second schematic diagram of a self-supervised neural differential mapping algorithm encoding a 1-bit reconfigurable smart surface to implement dual-beam scanning according to an embodiment of the present application;

[0032] Fig.10 A flow chart of a control method for a self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping provided according to an embodiment of the present application;

[0033] Fig.11 A schematic diagram of the structure of an electronic device provided for an application embodiment.

[0034] Explanation of the accompanying drawings: Self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping-10; self-supervised neural differential mapping encoding module-100, reconfigurable intelligent surface-200, artificial intelligence chip-300, phase control circuit board-400; memory-1101, processor-1102, communication interface-1103. DETAILED DESCRIPTION

[0035] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0036] The following describes the self-evolving reconfigurable intelligent surface and control method based on self-supervised neural differential mapping of the embodiment of the present application with reference to the accompanying drawings. In response to the problems mentioned in the above background technology, the present application provides a self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping, which includes: a self-supervised neural differential mapping coding module, which is used to output the corresponding coding matrix in real time according to the requirements of the target beam based on the coding strategy of the preset self-supervised neural differential mapping coding algorithm, and autonomously evolve and continuously transfer the learning of the coding strategy in different application scenarios; a reconfigurable intelligent surface, which is used to use the coding matrix to perform the corresponding target electromagnetic function in a variety of electromagnetic functions; an artificial intelligence chip, which is used to run the preset self-supervised neural differential mapping coding algorithm, autonomously learn the coding strategy and optimize the coding matrix; a phase control circuit board, which is used to adjust the bias voltage of the reconfigurable intelligent surface according to the coding matrix to achieve the target electromagnetic function, so as to integrate the self-supervised neural differential mapping coding algorithm into the reconfigurable intelligent surface through the artificial intelligence chip, and achieve efficient and accurate digital coding. Thus, the problem that the related technology cannot provide accurate digital coding for complex electromagnetic functions is solved.

[0037] Specifically, Figure 1 A schematic diagram of a self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping provided in an embodiment of the present application.

[0038] like Figure 1 As shown, the self-evolving reconfigurable smart surface 10 based on self-supervised neural differential mapping includes: a self-supervised neural differential mapping encoding module 100, a reconfigurable smart surface 200, an artificial intelligence chip 300 and a phase control circuit board 400.

[0039] Among them, the self-supervised neural differential mapping coding module 100 is used to output the corresponding coding matrix in real time according to the requirements of the target beam based on the coding strategy of the preset self-supervised neural differential mapping coding algorithm, and autonomously evolve and continuously transfer the learning of coding strategies in different application scenarios.

[0040] The reconfigurable smart surface 200 is used to execute a corresponding target electromagnetic function among a plurality of electromagnetic functions using a coding matrix.

[0041] It should be noted that the above-mentioned self-supervised neural differential mapping coding algorithm regards the encoding of intelligent reflective surfaces as a search problem for a solution that satisfies two competing constraints, and trains a binary neural network with a self-supervised learning method to learn the iterative mapping update rules based on the projection differential of the constraint set, thereby integrating the self-supervised neural differential mapping coding algorithm into the reconfigurable intelligent surface through an artificial intelligence chip to achieve efficient and accurate digital coding.

[0042] Optionally, in one embodiment of the present application, the relationship between the beam pattern of the reconfigurable smart surface 200 and the unit phase shift is expressed as:

[0043] ,

[0044] Among them, reconfigurable smart surfaces have units, , matrix element , and They are The phase shift and reflection coefficient of each unit, .

[0045] Specifically, the relationship between the beam pattern of the reconfigurable smart surface and the unit phase shift can be expressed as:

[0046]

[0047] Among them, reconfigurable smart surfaces have units, ,and , and Both indicate size The matrix of No. The elements are , The Matrix elements , and The The matrix elements are and ,and and Respectively represent The phase shift and reflection coefficient of each unit, , , IDFT stands for Inverse Discrete Fourier Transform, and the beam direction is determined by the elevation angle and azimuth Characterization;

[0048] In an embodiment of the present application, it is necessary to calculate the phase shift matrix of the reconfigurable smart surface to form the corresponding target beam pattern. Therefore, the intersection of two competing sets can be solved by utilizing a self-supervised neural differential mapping coding algorithm to provide accurate digital coding for complex electromagnetic functions.

[0049] Optionally, in one embodiment of the present application, the self-supervised neural differential mapping encoding module 100 includes: a determining unit, a establishing unit and an encoding unit.

[0050] The determination unit is used to determine the upper and lower bounds of the beam corresponding to the target electromagnetic function.

[0051] The establishing unit is used to establish a first phase shift matrix set and a second phase shift matrix set competing with each other according to the upper and lower bounds of the beam.

[0052] The encoding unit is used to build the neural differential mapping algorithm and construct an objective function to train the neural differential mapping algorithm to output the optimal phase shift matrix and quantize it into a coding matrix.

[0053] Specifically, the process of the self-supervised neural differential mapping encoding algorithm is as follows:

[0054] 1. Determine the upper and lower bounds of the beam to achieve the target electromagnetic function and ;

[0055] 2. Create two competing sets of phase shift matrices:

[0056] 1) A set of all possible phase shift matrices:

[0057]

[0058] in , is the set of all bounded functions;

[0059] 2) A set of all phase shift matrices that can form the target beam:

[0060]

[0061] in and are the upper and lower boundaries of the target beam, respectively.

[0062] 3. Build a neural differential mapping algorithm. The execution logic of the algorithm is as follows Figure 2 As shown in the figure, BNN is the abbreviation of Binary Neural Network. Its iterative update formula is:

[0063] in, and Represent the binary neural network and its weight parameters respectively to achieve the set and Search for the intersection of .

[0064] It can be understood that since the weights of the above neural network are binarized, the storage occupancy and computational complexity of the model are low, it is hardware-friendly, and suitable for integration into reconfigurable smart surfaces.

[0065] Optionally, in one embodiment of the present application, the objective function is the error between the beam pattern and the upper and lower boundaries of the target beam, wherein the expression of the objective function is:

[0066]

[0067] in, are the weights of different parts of the objective function, and is the domain defining the upper boundary of the target beam and its side lobes, is the domain defining the lower boundary of the target beam.

[0068] Specifically, the objective function of the neural differential mapping algorithm is the error between the beam pattern and the upper and lower boundaries of the target beam:

[0069]

[0070] in, are the weights of different parts of the objective function, and is the domain defining the upper boundary of the target beam and its side lobes, is the domain defining the lower boundary of the target beam.

[0071] The phase shift matrix finally output by the neural differential mapping algorithm is a continuous phase, and according to the requirements of the reconfigurable smart surface, it needs to be quantized into a 1-bit or 2-bit digital coding matrix as needed.

[0072] Optionally, in one embodiment of the present application, the reconfigurable smart surface 200 is an artificial electromagnetic surface in which active devices are integrated into artificial units.

[0073] Specifically, the reconfigurable smart surface is an artificial electromagnetic surface composed of periodically or quasi-periodically arranged artificial units. The artificial units integrate active devices that can exhibit different electromagnetic properties under different voltage biases. By controlling the state of the artificial units, a variety of different electromagnetic functions can be achieved, and each function corresponds to a set of digital codes.

[0074] The artificial intelligence chip 300 is used to run a preset self-supervised neural differential mapping coding algorithm, autonomously learn coding strategies and optimize coding matrices.

[0075] It should be noted that an artificial intelligence chip is an integrated circuit chip that has been specially designed for accelerating artificial intelligence algorithms, that is, a chip that can run artificial intelligence algorithms, including general-purpose chips, semi-customized chips, customized chips, and brain-like chips. The embodiments of the present application use artificial intelligence chips to run a self-supervised neural differential mapping algorithm to manipulate electromagnetic waves to achieve a variety of electromagnetic functions, such as Figure 3 shown.

[0076] The phase control circuit board 400 is used to adjust the bias voltage of the reconfigurable smart surface according to the coding matrix to achieve the target electromagnetic function.

[0077] In addition, the phase control circuit board is used to adjust the phase state of each unit in the reconfigurable smart surface. The phase control circuit board can independently control the offset voltage of the reconfigurable smart surface unit according to the coding matrix to achieve the target electromagnetic function.

[0078] A self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping of the present application is described in detail below through specific embodiments.

[0079] The embodiment of the present application adopts a 1-bit reconfigurable smart surface, where 1 bit means that the surface unit has two different states, corresponding to the two values ​​of "0" and "1" in the coding matrix. The 1-bit reconfigurable smart surface has an operating frequency of 9.37GHz, an aperture size of 24 wavelengths, 2304 units, and is made based on Taconic TLX-8 and FR4 materials. The diode (PIN diode) model used is Skyworks SMP1340.

[0080] Based on a 1-bit reconfigurable smart surface, a self-supervised neural differential mapping algorithm is trained to learn a dual-beam scanning coding strategy, assuming that two beams (the beam directions are determined by The scanning ranges of , , 3dB beamwidth is 3 degrees and sidelobe level is -30dB.

[0081] The training process of the self-supervised neural differential mapping encoding algorithm is as follows:

[0082] 1. Based on the beam parameters of dual-beam scanning, generate the upper and lower bounds of the beam to achieve the target electromagnetic function and

[0083] 2. Create two competing sets of phase shift matrices:

[0084] 1) A set of all possible phase shift matrices:

[0085]

[0086] in, , is the set of all bounded functions;

[0087] 2) A set of all phase shift matrices that can form the target beam:

[0088]

[0089] in, and are the upper and lower boundaries of the target beam, respectively.

[0090] 3. Build a neural differential mapping algorithm. The iterative update formula of the algorithm is:

[0091]

[0092] in, and Represent the binary neural network and its weight parameters respectively, is the number of iteration steps. In the embodiment of the present application, the binary neural network adopts an XNOR network structure (XNOR-Net). As shown in the figure, the input and weight of the network are both binary, and the XNOR operation and the binary count (bitcount) operation are used to approximate the convolution operation. Assume that the input of the XNOR network is , the weight is , then the convolution operation can be expressed as:

[0093]

[0094] in, is the traditional convolution operation. is a convolution operation approximated by XOR and binary counting operations, is a two-dimensional filter of the XOR network, whose elements are .

[0095] 4. Construct an objective function to train the neural differential mapping algorithm. The objective function is the error between the beam pattern and the upper and lower boundaries of the target beam:

[0096]

[0097] in, are the weights of different parts of the objective function, and is the domain defining the upper boundary of the target beam and its side lobes, is the domain of the lower boundary of the target beam, Figure 4 is a schematic diagram of the upper boundary of the beam, where the light and dark areas are , the dark area is . Figure 5 is a schematic diagram of the lower boundary of the beam, where the light-colored area is In addition, the one-dimensional diagram of the upper and lower boundaries of the beam is as follows Figure 6 As shown;

[0098] 5. Output the optimal phase shift matrix and quantize it into a digital coding matrix according to the requirements of the reconfigurable smart surface, such as Figure 7As shown in the figure, BNN stands for self-supervised neural differential mapping encoding algorithm.

[0099] Understandably, Figure 8 , Fig. 9 Two examples of dual-beam scanning are given and compared with the encoding results of the genetic algorithm, which fully demonstrates the effectiveness and superiority of the self-supervised neural differential mapping encoding. In the figure, GA is the abbreviation of Genetic Algorithm, which represents the genetic algorithm, and RaP is the abbreviation of Radiation Pattern, which represents the beam pattern.

[0100] According to the self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping proposed in the embodiment of the present application, based on the coding strategy of the preset self-supervised neural differential mapping coding algorithm, the corresponding coding matrix is ​​output in real time according to the requirements of the target beam, and the coding strategy in different application scenarios is continuously transferred and learned by autonomous evolution, and the coding matrix is ​​used to execute the corresponding target electromagnetic function among multiple electromagnetic functions, and then the preset self-supervised neural differential mapping coding algorithm is run, the coding strategy and the coding matrix are autonomously learned, and then the bias voltage of the reconfigurable intelligent surface is adjusted according to the coding matrix to achieve the target electromagnetic function. The embodiment of the present application can embed the self-supervised deep learning algorithm into the artificial intelligence chip to drive the reconfigurable intelligent surface, realize real-time and accurate coding of multiple electromagnetic functions, and give the reconfigurable intelligent surface the ability of self-evolution, and can learn the coding strategies of different electromagnetic functions in a self-supervised manner.

[0101] Secondly, if Fig.10 As shown, the control method of the self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping includes the following steps:

[0102] In step S1001, based on the coding strategy of the preset self-supervised neural differential mapping coding algorithm, the corresponding coding matrix is ​​output in real time according to the requirements of the target beam, wherein the coding strategy is autonomously evolved and continuously transferred according to different application scenarios.

[0103] In step S1002, the bias voltage of the reconfigurable smart surface is adjusted according to the coding matrix to utilize the coding matrix to execute the corresponding target electromagnetic function among the multiple electromagnetic functions, and a preset self-supervised neural differential mapping coding algorithm is run to autonomously learn the coding strategy and optimize the coding matrix.

[0104] It should be noted that the aforementioned explanation of the embodiment of the self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping is also applicable to the control method of the self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping in this embodiment, which will not be repeated here.

[0105] According to the control method of the self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping proposed in the embodiment of the present application, the upper and lower bounds of the target beam are determined according to the target electromagnetic function, and the phase shift matrix is ​​obtained by running the self-supervised neural differential mapping coding algorithm based on the artificial intelligence chip. The phase shift matrix is ​​quantified into digital code according to the requirements of the reconfigurable intelligent surface, and the bias voltage of the reconfigurable intelligent surface unit is independently adjusted through the phase shift control circuit board to form the target beam and realize the target electromagnetic function, thereby improving the efficiency and accuracy of the digital coding.

[0106] Next, a control method for a self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping proposed in an embodiment of the present application is described with reference to the accompanying drawings.

[0107] Fig.11 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include:

[0108] A memory 1101 , a processor 1102 , and a computer program stored in the memory 1101 and executable on the processor 1102 .

[0109] When the processor 1102 executes the program, the control method of the self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping provided in the above embodiment is implemented.

[0110] Furthermore, the electronic device further comprises:

[0111] The communication interface 1103 is used for communication between the memory 1101 and the processor 1102 .

[0112] The memory 1101 is used to store computer programs that can be executed on the processor 1102 .

[0113] The memory 1101 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0114] If the memory 1101, the processor 1102 and the communication interface 1103 are implemented independently, the communication interface 1103, the memory 1101 and the processor 1102 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Fig.11 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0115] Optionally, in a specific implementation, if the memory 1101, the processor 1102 and the communication interface 1103 are integrated on a chip, the memory 1101, the processor 1102 and the communication interface 1103 can communicate with each other through an internal interface.

[0116] The processor 1102 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0117] This embodiment also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the control method for the self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping as described above.

[0118] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0119] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0120] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0121] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection with one or N wirings (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways as necessary and then storing it in a computer memory.

[0122] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above embodiment, N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0123] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0124] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0125] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping, characterized in that: include: A self-supervised neural differential mapping coding module is used to output the corresponding coding matrix in real time according to the requirements of the target beam based on the coding strategy of the preset self-supervised neural differential mapping coding algorithm, and autonomously evolve and continuously transfer the coding strategy in different application scenarios; A reconfigurable smart surface, used to perform a corresponding target electromagnetic function among a plurality of electromagnetic functions using the coding matrix; An artificial intelligence chip, used to run a preset self-supervised neural differential mapping coding algorithm, autonomously learn the coding strategy and optimize the coding matrix; as well as A phase control circuit board, used for adjusting the bias voltage of the reconfigurable smart surface according to the coding matrix to achieve the target electromagnetic function; The relationship between the beam pattern and the unit phase shift of the reconfigurable smart surface is expressed as: Among them, reconfigurable smart surfaces have units, ,and , and Both indicate size The matrix of No. The elements are , The Matrix elements , and The The matrix elements are and ,and and Respectively represent The phase shift and reflection coefficient of each unit, , , IDFT stands for Inverse Discrete Fourier Transform, and the beam direction is determined by the elevation angle and azimuth Characterization; The self-supervised neural differential mapping encoding module includes: A determination unit, used to determine the upper and lower bounds of the beam of the target electromagnetic function and ; Establish a unit for determining the upper and lower bounds of the beam and A first phase shift matrix set and a second phase shift matrix set that compete with each other are established, wherein the first phase shift matrix set is a set including all possible phase shift matrices, the second phase shift matrix set is a set including all phase shift matrices that can form a target beam, and the mathematical expression of the first phase shift matrix set is: in , is the set of all bounded functions; The mathematical expression of the second phase shift matrix set is: in is the lower boundary of the target beam, is the upper boundary of the target beam; The encoding unit is used to build a neural differential mapping algorithm and construct an objective function to train the neural differential mapping algorithm to output an optimal phase shift matrix, and quantize it into the encoding matrix according to the requirements of the reconfigurable smart surface, wherein the iterative update formula of the neural differential mapping algorithm is: in, and Represent binary neural networks and their weight parameters respectively; k is the number of iterations; the binary neural network adopts an XOR network, the input and weight of the XOR network are both binary, and the XOR operation and the binary counting operation are used to approximate the convolution operation; the input of the XOR network is , the weight is , then the convolution operation is expressed as: in, It is a traditional convolution operation; It is a convolution operation approximated by XOR and binary counting operations; is a two-dimensional filter of the XOR network, and the elements of the two-dimensional filter are ; The objective function is the error between the beam pattern and the upper and lower boundaries of the target beam, wherein the expression of the objective function is: in, are the weights of different parts of the objective function, and is the domain defining the upper boundary of the target beam and its side lobes, is the domain of the lower boundary of the target beam, represents the lower boundary of the target beam, Indicates the upper boundary of the target beam.

2. The self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping according to claim 1, characterized in that: The reconfigurable smart surface is an artificial electromagnetic surface in which active devices are integrated into artificial units.

3. A control method for a self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping, characterized in that: The self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping as claimed in any one of claims 1 to 2 is characterized in that it includes the following steps: Based on the coding strategy of the preset self-supervised neural differential mapping coding algorithm, the corresponding coding matrix is ​​output in real time according to the requirements of the target beam, wherein the coding strategy is autonomously evolved and continuously transferred and learned according to different application scenarios; The bias voltage of the reconfigurable smart surface is adjusted according to the coding matrix to utilize the coding matrix to execute the corresponding target electromagnetic function among the multiple electromagnetic functions, and the preset self-supervised neural differential mapping coding algorithm is run to autonomously learn the coding strategy and optimize the coding matrix.

4. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the control method of the self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping as claimed in claim 3.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the control method of the self-evolving reconfigurable intelligent surface based on self-supervised neural differential mapping as claimed in claim 3.

Images