A signal transmission method, device, apparatus, server and storage medium

Abstract

The application discloses a signal transmission method and device, equipment, server and storage medium, relates to the wireless communication technical field, is applied to the neural network in the wireless communication system, the neural network includes the preset beamforming structure obtained by deep learning, and the neural network is used for determining the beamforming matrix according to the received uplink channel information;The method comprises the following steps: receiving the request information from the electronic equipment, wherein the request information is transmitted through the uplink channel;Determine the downlink channel state information, the reflection phase shift and the transmitting power based on the uplink channel;In response to the request information, the feedback information is generated based on the preset beamforming structure, the downlink channel state information and the reflection phase shift;According to the transmitting power, the feedback information is sent to the electronic equipment.In reducing the complexity of the generated beamforming algorithm, the transmission rate of wireless signal can be effectively improved.

Description

Technical Field

[0001] This application relates to wireless communication technology, and more particularly to a signal transmission method, apparatus, device, server, and storage medium. Background Technology

[0002] Reconfigurable Intelligent Surface (RIS) is a wireless communication technology. Specifically, RIS can improve signal propagation in wireless communication by intelligently controlling the reflection, refraction, and scattering properties of electromagnetic waves. A RIS typically consists of a large number of programmable reflective units, each of which can be individually configured. This allows for intelligent reconfiguration of the propagation channel by adjusting the phase and amplitude of each transmitting unit. For example, by configuring a group of passive electrical components in the RIS to adjust the reflection of the incident signal, the communication system's speed, spectral efficiency, and coverage can be improved. It is worth noting that the passive radio components used in RIS have low power consumption and the ability to improve communication system performance.

[0003] Generally, during wireless signal transmission, when a base station (BS) combines reflection phase shift and beamforming (BD) with beamforming to generate the transmitted signal, a problem arises in optimizing the RIS phase shift. Traditional techniques to address this issue include alternating optimization of reflection and RIS phase shifts, using block coordinate descent (BCD) methods, or employing semidefinite relaxation (SDR) methods. However, the high complexity of these algorithms slows down the convergence speed of wireless communication systems, thus limiting the application of wireless signal transmission in 5G or 6G systems, particularly for services with higher transmission rate and efficiency requirements. Summary of the Invention

[0004] This application provides a signal transmission method that can effectively improve the transmission rate of wireless signals while reducing the complexity of beamforming generation algorithms.

[0005] To achieve the above technical objectives, the embodiments of this application adopt the following technical solutions:

[0006] Firstly, this application provides a signal transmission method applicable to a wireless communication system. The wireless communication system includes a neural network with a preset beamforming structure obtained through deep learning. This neural network is used to determine a beamforming matrix based on received uplink channel information, so as to generate a transmission signal according to the beamforming matrix. Specifically, the signal transmission method involves: receiving a request message from an electronic device, wherein the request message is transmitted via an uplink channel; determining downlink channel state information, reflection phase shift, and transmit power based on the uplink channel information; and further, in response to the request message, generating feedback information based on the preset beamforming structure, downlink channel state information, and reflection phase shift, and sending the feedback information to the electronic device according to the transmit power.

[0007] Understandably, in the process of wireless signal transmission, Channel State Information (CSI) is crucial for beamforming in multi-user, multi-antenna systems. Neural networks are learned in a data-driven manner, trained with a large amount of sample data to obtain a neural network that satisfies data transmission requirements. The neural network has a pre-set beamforming structure and can predict downlink channel state information based on uplink channel information. In other words, the wireless system provided in this application uses a combination of data-driven and model-driven approaches. This approach can satisfy the base station's transmit power requirements when transmitting beam signals and ensure that the transmitted signal conforms to the phase constraints of the Rectified Isochronous Ratio (RIS).

[0008] In conjunction with the first aspect, in one possible implementation, the neural network includes data-driven and model-driven components. The data-driven component receives request information for uplink channel transmission and determines downlink channel state information, reflection phase shift, and transmit power based on the uplink channel state information. The model-driven component includes a preset beamforming structure.

[0009] Generally speaking, using a dual-driven neural system in the beamforming process of wireless communication can effectively solve the problems of beamforming and reflection phase shift in multi-user multiple-input single-output (MISO) systems. Joint optimization can effectively improve the transmission rate of wireless signals while reducing the complexity of the beamforming generation algorithm.

[0010] In conjunction with the first aspect, another possible implementation includes a neural network comprising a first channel state network, a second channel state network, a phase network, and a power network; the first channel state network outputs predicted downlink direct channel state information based on the uplink channel; the second channel state network outputs predicted downlink concatenated channel state information based on the uplink channel; the phase network outputs reconfigurable smart surface RIS phase shift and base station reflection phase shift based on the uplink channel; and the power network outputs transmit power based on the uplink channel power information.

[0011] In conjunction with the first aspect, another possible implementation includes a wireless communication system comprising a base station, a reconfigurable smart surface RIS, and the electronic device.

[0012] The method further includes: a base station receiving the request information from an electronic device, wherein the request information is transmitted via the uplink channel. The base station determines downlink channel state information, reflection phase shift, and transmit power based on the uplink channel; the base station generates response information based on the request information, downlink channel state information, and base station reflection phase shift; the base station sends reflection control information and feedback information to the RIS according to the transmit power; the RIS adjusts the reflection element in response to the reflection control information and sends feedback information to the electronic device.

[0013] In conjunction with the first aspect, another possible implementation of the method may further include: the base station determining equivalent downlink channel information based on request information, downlink channel state information, and base station reflection phase shift; and recovering the beamforming matrix based on the equivalent downlink channel information and transmit power.

[0014] Among them, the beamforming matrix recovery can help the base station recover beamforming. This recovery method is directly calculated and can effectively improve the beamforming generation efficiency.

[0015] Secondly, embodiments of this application also provide a signal transmission device, including a receiving module, a determining module, a generating module, and a transmitting module.

[0016] The receiving module is used to receive request information from an electronic device, wherein the request information is transmitted via an uplink channel. The determining module is used to determine channel state information, reflection phase shift, and transmit power based on the uplink channel. The generating module is used to generate feedback information in response to the request information, based on a preset beamforming structure, downlink channel state information, and reflection phase shift. The transmitting module is used to transmit the feedback information to the electronic device according to the transmit power.

[0017] In conjunction with the second aspect, in one possible implementation, the device further includes a beamforming recovery module.

[0018] The beamforming recovery module is used to determine the equivalent downlink channel information based on the request information, downlink channel state information, and base station reflection phase shift; and to recover the beamforming matrix based on the equivalent downlink channel information and transmit power.

[0019] Thirdly, embodiments of this application also provide a signal transmission device, including: a processor and a memory for storing processor-executable instructions. The processor is configured to execute the instructions to implement the methods described in the first aspect and any of the possible implementations.

[0020] Fourthly, embodiments of this application also provide a server, including: a processor; and a memory for storing processor-executable instructions.

[0021] The processor is configured to execute the instructions to implement the methods described in the first aspect and any possible implementation thereof.

[0022] Fifthly, embodiments of this application also provide a computer-readable storage medium storing computer instructions that, when executed on an electronic device, implement the methods described in the first aspect and any possible implementation.

[0023] It is understood that the beneficial effects achieved by the apparatus of the second aspect and any possible implementation thereof, the device of the third aspect, the server of the fourth aspect, and the computer-readable storage medium of the fifth aspect can be referred to as the beneficial effects of the first aspect and any possible design thereof, which will not be repeated here. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the wireless communication system structure provided in an embodiment of this application;

[0025] Figure 2 This is a flowchart of the signal transmission method provided in an embodiment of this application;

[0026] Figure 3 This is a schematic diagram of a neural network module structure provided in an embodiment of this application;

[0027] Figure 4 This is a schematic diagram of the structure of a wireless signal transmission device provided in an embodiment of this application. Detailed Implementation

[0028] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this embodiment, unless otherwise stated, "a plurality of" means two or more.

[0029] The methods provided in the embodiments of this application will be described in detail below with reference to the accompanying drawings. Those skilled in the art will understand that the structural diagrams of the electronic devices, base stations, or RIS shown in the embodiments of this application are not intended to limit the structure of the devices. In practical applications, the electronic devices, base stations, or RIS may include more or fewer components or modules, or certain components may be combined, or the components may be rearranged.

[0030] like Figure 1 The image shows data transmission between devices in a wireless communication system. Figure 1 As shown, the wireless communication system includes a base station (BS), an IRS, and multiple electronic devices (such as...). Figure 1 (U1, U2 and U3).

[0031] In one wireless signal data transmission, an electronic device sends a data request message to the Base Station (BS). The BS responds by generating a corresponding response message and sends it back to the electronic device. In another wireless signal transmission, the electronic device (e.g., U1) cannot directly send a data request message to the BS. U1 then sends a data request message to the IRS, which forwards it to the BS. The BS receives the data request message from the electronic device through the IRS, generates a corresponding response message, and sends it back to the IRS, which then forwards the response message to the electronic device.

[0032] Specifically, information transmission from electronic devices to the IRS, from the IRS to the base station, and from electronic devices to the BS all occur via the uplink channel. Information transmission from the base station to electronic devices, from the IRS to electronic devices, and from the base station to the IRS all occur via the downlink channel.

[0033] Specifically, when beamforming (BS) is combined with beamforming (RIS) to generate transmitted signals, a problem arises involving reflection phase shift and the issue of jointly optimizing RIS phase shift using beamforming (BD). When addressing this technical problem using deep learning, the beamforming problem in base stations was studied from a data-driven perspective. One solution proposed using a multi-layer fully connected neural network (FCNN) to learn the phase shift in a single-electronic device wireless communication system, aiming to reduce the complexity of the RIS algorithm. It is worth noting that these neural networks typically employ a data-driven learning approach, requiring a large number of data samples for training. This reduces the convergence speed of the neural network training and decreases the accuracy of the neural network output. Furthermore, to address this drawback in the neural network training process, one approach is to incorporate classical models from wireless communication theory into the neural network to reduce the output of the neural network. In other words, a model-driven learning method is used. This reduces the number of training samples required.

[0034] It is worth noting that beamforming optimization in wireless communication systems is typically performed using a Multiple-Input-Single-Output (MISO) system. However, in such systems, beamforming optimization and phase shift optimization in RIS-assisted MISO systems are coupled. This means that the phase shift error is greater when beamforming is primarily performed in a RIS-assisted MISO system. Therefore, in wireless communication systems, relying solely on this model-driven approach cannot meet the diverse transmission requirements of wireless communication systems.

[0035] In addition, there is an easily overlooked problem in wireless communication system transmission: the reciprocity between the uplink and downlink channels may not exist.

[0036] Reciprocity refers to the fact that in a communication system, if the uplink and downlink transmit information on different time slots under the same frequency resources, then within a relatively short period of time (the coherence time of channel propagation), the channel fading experienced by the transmitted signals of the uplink and downlink can be considered to be the same.

[0037] Channel State Information (CSI) is crucial for beamforming in multi-user, multi-antenna systems. Based on CSI, key information such as channel quality and environmental conditions can be determined in wireless communication. In some implementations, research on wireless communication systems combining Base Station (BS) and Resonant Radio Interference (RIS) assumes a good CSI at the BS, allowing the BS to estimate the downlink CSI based on the uplink CSI. However, this method relies on the reciprocity of uplink and downlink. In practice, the presence of analog front-end circuitry in the radio transmitter unit causes differences between the uplink and downlink CSI, making it difficult to estimate the downlink CSI from the uplink. In other words, in real-world applications, uplink and downlink channels are non-reciprocal.

[0038] Based on this, embodiments of this application provide a signal transmission method applied to a wireless communication system, enabling the system to become a data-driven and model-driven wireless system. For example, the system may include a neural network, in which a preset beamforming structure obtained through deep learning is set. In this way, the wireless communication system can determine the beamforming matrix based on the received uplink channel information. While reducing the complexity of the beamforming generation algorithm, this effectively improves the transmission rate of the wireless signal.

[0039] like Figure 2 The diagram shown is a flowchart of a signal transmission method provided in an embodiment of this application. Figure 2 As shown, the signal transmission method includes steps 202-204.

[0040] Step 201: Receive request information from electronic device.

[0041] The request information is transmitted via the uplink channel.

[0042] It is understandable that the requested information could be information sent from the electronic device to the base station by the user when operating the electronic device.

[0043] Electronic devices transmit request information to the base station via the uplink channel. In some implementations, when the distance between the electronic device and the base station exceeds a preset distance value, or when there is an obstruction to the wireless transmission between the electronic device and the base station, the electronic device can send request information to the IRS via the uplink channel. The IRS then forwards the received request information to the base station. For example, a user k (which could be electronic device U1 in Figure 1) sends request information to the base station.

[0044] For example, in an IRS-assisted downlink MISO system, a base station (BS) with N antennas serves K single-antenna users. An IRS equipped with an array of N passive reflective elements is deployed to facilitate transmission between the BS and the users. The BS controls the IRS via an IRS controller, which adjusts the phase of the array elements to reflect the incident signal in the desired direction.

[0045] In some implementations, with Figure 1 The wireless communication system shown is an example. This represents the downlink channel from the base station to user k. This represents the downlink channel from the IRS to user K. This indicates the downlink path from the base station to the IRS. This represents the uplink channel from user k to the base station. This represents the uplink channel from user K to the IRS. This represents the uplink path from the IRS to the base station. The signal received at user k can then be represented by the following formula 1:

[0046] Formula (1)

[0047] in, ( ) This represents the cascaded channel between the base station and user k via IRS reflection.

[0048] Furthermore, the downlink channel matrix can be represented as: .set up For beamforming vectors, This is the information transmitted from BS to user k. It is the phase shift vector at the IRS, where For the phase shift of the i-th element, It is additive white Gaussian noise with zero mean and zero variance. Therefore, the signal-to-noise ratio (SINR) at user k can be expressed by Equation 2:

[0049] Formula (2)

[0050] in, This represents the equivalent downlink channel.

[0051] The total speed of all electronic devices can then be expressed by the following formula 3:

[0052] Formula (3)

[0053] in, It's the weight.

[0054] In the signal transmission method provided in this application, the optimization problem of beamforming can be described as maximizing the sum rate of all users while satisfying the constraints of transmit power and continuous phase shift. This can be expressed by the following formula 4:

[0055] Formula (4)

[0056] in, Let C1 be the beamforming matrix at BS. C1 is the transmit power constraint, and P is the maximum transmit power of BS. C2 is the continuous phase shift constraint at RIS. When the downlink CSI is known, the above problem can be solved iteratively using known methods. In this case, if deep learning can be used to optimize the problem, better optimization results can be obtained. This is the dual-drive model provided in the embodiments of this application. In addition, in reality, there may be no channel reciprocity between the uplink and downlink. Therefore, assuming that only uplink channel information is available on BS, channel reciprocity does not hold.

[0057] For example, suppose there exists a function Its input is the uplink CSI, and its output is the optimal beamforming matrix W and phase shift v that satisfy the constraints of problem (P). Then the problem can be expressed by the following formula 5:

[0058] Formula (5)

[0059] in, Uplink channel matrix . This represents the uplink channel from user k to the base station. This represents the uplink channel from user K to the IRS. This indicates the uplink channel from the IRS to the base station.

[0060] In general, functions A closed-form expression for does not exist. Since any continuous-valued function can be approximated by a small error according to the general approximation theorem, problem (P) can be transformed into a trainable problem. Assume the network... There is a trainable set of parameters. Therefore, formula 5 above can be approximately represented by formula 6 as follows:

[0061] Formula (6)

[0062] It should be noted that the data-driven approach in this embodiment is based on a model-based beamforming structure. This data-driven approach can effectively reduce the output dimension of the neural network, reduce the training difficulty of the neural network, thereby improving training efficiency and thus enhancing the performance of the data-driven approach, and improving the data transmission efficiency and performance of wireless communication. It is worth noting that this structure can be applied to multi-antenna MISO systems, but cannot be directly applied to RIS-assisted MISO systems. This is because RIS systems include complex direct and cascaded channels, as well as phase shift vectors. Therefore, in one possible implementation of this application, a fixed phase shift vector can be given in the wireless communication system. This allows for data-driven matching with RIS-assisted MISO systems.

[0063] Specifically, from a mathematical perspective, the RIS system can be equivalently viewed as a multi-user MISO system. Therefore, the optimal beamforming structure of the RIS-assisted MISO system can be expressed by the following formula 7:

[0064] Formula (7)

[0065] in, It is an identity matrix. and All are satisfied The positive parameter. For the optimal downlink power vector, The auxiliary vector can be used to determine the beamforming direction.

[0066] It should be noted that power vector This is considered a key feature of beamforming solutions. Therefore, compared to directly learning high-dimensional beamforming matrices... Unlike other methods, the data model used in this application's embodiments is capable of learning low-dimensional power features. Furthermore, the beamforming vector can be recovered from the learned power features. In practice, this approach reduces the output dimension of the neural network learning beamforming from 2NK to 2K, thereby improving training efficiency.

[0067] Step 202: Determine the downlink channel state information, reflection phase shift, and transmit power based on the uplink channel.

[0068] The neural network architecture provided in this application includes learning phase shifts. and power characteristics A two-part neural network maximizes the sum rate while adhering to transmit power and phase shift constraints. This neural network uses the uplink CSI... The input is the beamforming matrix at RIS, and the final output is the beamforming matrix at RIS. and phase shift .

[0069] For example, the entire neural network architecture consists of four modules, such as Figure 3 As shown. The neural network includes a channel state network (CSI-Net), a phase network (Phase-Net), and a power network (Power-Net). Figure 3 As shown, the channel state network includes a first channel state network (CSI-Net1) and a second channel state network (CSI-Net2).

[0070] CSI-Net1 and CSI-Net2 are two fully connected neural networks (FCNNs) used to learn the uplink and downlink relationships (i.e., unitary transform relationships) of direct and cascaded channels, respectively. The input to both CSI-Net1 and CSI-Net2 is the uplink CSI. The output of CSI-Net1 is the predicted downlink direct CSI. The output of CSI-Net2 is the prediction of downlink cascaded CSI. .

[0071] In some implementations, supervised learning methods can be used to train the CSI-Net1 and CSI-Net2 networks. Since its input is a complex matrix, it is split into real and imaginary parts to facilitate the training of the neural network.

[0072] Specifically, suppose we have a channel training dataset, with There are two uplink-downlink CSI pairs. The prediction outputs of CSI-Net1 and CSI-Net2 are denoted as follows: and ( ), where the corresponding labels of the real channels are denoted as . and The loss function for CSI-Net1 can be the mean squared error (MSE), as shown in Equation 8 below:

[0073] Formula (8)

[0074] The loss function of CSI-Net2 can be expressed by the following formula 9:

[0075] Formula (9)

[0076] Phase-Net: This module is designed to learn and optimize RIS phase shifts. Specifically, neural networks use uplink CSI. As input, the phase shift is output in a supervised learning manner. Similarly, complex vectors It is divided into real and imaginary parts. The predicted output of Phase-Net is denoted as... And its label is Meanwhile, the MSE-based Phase-Net loss function is expressed as follows (Equation 10):

[0077] Formula (10)

[0078] Power-Net: Power-Net aims to understand the power characteristics of beamforming schemes. This indicates that Power-Net's prediction output is... and ,in and These are the t-th training samples of the downlink and uplink power vectors obtained by the BCD algorithm during power training. Similar to previous networks, this network is trained using supervised learning. The loss function for Power-Net is also given, as shown in Equation 11 below:

[0079] Formula (11)

[0080] Step 203: In response to the request information, generate feedback information based on the preset beamforming structure, downlink channel state information, and reflection phase shift.

[0081] Specifically, the base station transmits information to the downlink channel by recovering the optimal beamforming based on beamforming in a neural network. In some implementations, estimated direct and concatenated channels, as well as estimated phase shifts, are used to obtain the equivalent downlink CSI. Then, the beamforming matrix is ​​recovered using power characteristics and equivalent downlink channel information. .

[0082] It is worth mentioning that this module does not contain any training parameters, but only calculates the beamforming matrix. Furthermore, to maximize the sum and rate, a hybrid training method is used, and the overall loss function is expressed by the following formula 12:

[0083] Formula (12)

[0084] in These are the weights for each loss component. Learning the channel matrix, phase shift vector, and power characteristics is merely an intermediate process; the ultimate goal is to maximize the sum rate of the system. Therefore, the sum rate is treated as an additional part of the overall loss function, i.e. A hybrid training approach is adopted, with downlink CSI, power features, and phase shift vector trained in a supervised manner, and rate trained in an unsupervised manner.

[0085] Step 204: Send feedback information to the electronic device according to the transmission power.

[0086] This application also provides a wireless signal transmission device, which includes computer instructions. When the computer instructions are executed on the wireless signal transmission device, the wireless signal transmission device performs the various functions or steps in the above method embodiments. For example, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), magnetic tape, floppy disk, and optical data storage device, etc.

[0087] When using integrated units, Figure 4 A possible structural diagram of the wireless signal transmission device 400 involved in the above embodiments is shown. The wireless signal transmission device 400 may include: a processing module 401, a storage module 402, and a communication module 403. The processing module 401 is used to control and manage the actions of the server. The storage module 402 is used to store the server's program code and data, such as methods for creating chat groups. The communication module 403 is used to support communication between the server and other network entities to achieve data interaction functions, such as supporting communication between the server and electronic devices to achieve data interaction functions.

[0088] The processing module 401 can be a processor or a controller. The communication module 403 can be a transceiver, RF circuit, or communication interface, etc. The storage module 402 can be a memory.

[0089] Other embodiments of this application provide a server that may include a memory and one or more processors. The touchscreen, memory, and processors are coupled. The memory stores computer program code, including computer instructions. When the processor executes the computer instructions, the server can perform various functions or steps performed by the server in the above method embodiments.

[0090] This application also provides a computer-readable storage medium including computer instructions that, when executed on the electronic device, cause the electronic device to perform various functions or steps performed by the mobile phone in the above method embodiments.

[0091] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0092] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0093] The units described as separate components may or may not be physically separate. A component shown as a unit can be one or more physical units; that is, it can be located in one place or distributed in multiple different locations. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0094] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0095] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, essentially or in other words, the parts that contribute to the prior art, or all or part of the technical solutions, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0096] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A signal transmission method, characterized by, The application is applied to a wireless communication system, the wireless communication system comprises a neural network, the neural network comprises a preset beamforming structure obtained by deep learning, and the neural network is used for determining a beamforming matrix according to received uplink channel information; the neural network comprises data driving and model driving; the data driving is used for receiving request information of the uplink channel transmission, and determining downlink channel state information, RIS reflection phase shift and transmission power according to channel state information of the uplink channel; the model driving comprises the preset beamforming structure; the method comprises the following steps: The method comprises the following steps: receiving request information from an electronic device, wherein the request information is transmitted through an uplink channel; determining downlink channel state information, RIS reflection phase shift and transmission power based on the uplink channel; wherein the downlink channel state information comprises direct channel state information and cascaded channel state information; generating feedback information based on the preset beamforming structure, the downlink channel state information and the RIS reflection phase shift in response to the request information; sending the feedback information to the electronic device according to the transmission power; wherein the preset beamforming structure is used for recovering a beamforming matrix based on equivalent downlink channel information and power characteristics, the equivalent downlink channel information is obtained according to downlink channel state information and RIS reflection phase shift, and the recovery of the beamforming matrix adopts the following formula: The neural network comprises a first channel state network, a second channel state network, a phase network and a power network; wherein is a beamforming vector for user k, is an identity matrix, K is the total number of users, is the noise power, is the equivalent downlink channel information; is the optimal downlink power vector; is an auxiliary vector can be used to determine the beamforming direction.

2. The method of claim 1, wherein, The first channel state network outputs predicted downlink direct channel state information according to the uplink channel; The second channel state network outputs predicted downlink cascaded channel state information according to the uplink channel; The phase network outputs a reconfigurable intelligent surface (RIS) phase shift according to the uplink channel; The power network outputs power characteristics according to the uplink channel, and the power characteristics are used for recovering a beamforming matrix. The wireless communication system comprises a base station, a reconfigurable intelligent surface (RIS) and the electronic device; the method further comprises the following steps:

3. The method according to claim 1 or 2, characterized in that, The base station receives the request information from the electronic device, wherein the request information is transmitted by the uplink channel; The base station determines downlink channel state information, RIS reflection phase shift and transmission power according to the uplink channel; The base station generates response information based on the request information, the downlink channel state information and the RIS reflection phase shift; The base station sends reflection control information and the feedback information to the RIS according to the transmission power; The RIS adjusts a reflection element in response to the reflection control information, and sends the feedback information to the electronic device. ​ 4. A signal transmission device, characterized by comprising: The application is applied to a wireless communication system, the wireless communication system comprises a neural network, the neural network comprises a preset beamforming structure obtained by deep learning, and the neural network is used for determining a beamforming matrix according to received uplink channel information; the neural network comprises data driving and model driving; the data driving is used for receiving request information of the uplink channel transmission, and determining downlink channel state information, RIS reflection phase shift and transmission power according to channel state information of the uplink channel; the model driving comprises the preset beamforming structure; the device comprises a receiving module, a determining module, a generating module and a sending module; The receiving module is used for receiving request information from an electronic device, wherein the request information is transmitted through an uplink channel; The determining module is used for determining downlink channel state information, RIS reflection phase shift and transmission power based on the uplink channel; The generating module is used for generating feedback information based on the preset beamforming structure, the downlink channel state information and the RIS reflection phase shift in response to the request information; The sending module is used for sending the feedback information to the electronic device according to the transmission power; The preset beamforming structure is used for recovering a beamforming matrix based on equivalent downlink channel information and power characteristics, the equivalent downlink channel information is obtained according to downlink channel state information and RIS reflection phase shift, and the recovery of the beamforming matrix adopts the following formula: wherein is a beamforming vector for user k, is a transmit power, is an identity matrix, K is the total number of users, is a power characteristic parameter, is a noise power, is equivalent downlink channel information.

5. The apparatus of claim 4, wherein, The device further comprises a beamforming recovery module; The beamforming recovery module is used for determining equivalent downlink channel information based on the downlink channel state information and the RIS reflection phase shift, and recovering a beamforming matrix according to the equivalent downlink channel information and transmission power.

6. A signal transmission device, characterized by comprising: Comprise: A processor and a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the method of any one of claims 1-3.

7. A server, characterized by Comprise: A processor; A memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the method of any one of claims 1-3.

8. A computer readable storage medium having stored thereon computer instructions, wherein, The computer instructions realize the method of any one of claims 1-3 when running on an electronic device.

Images