A scalable communication method based on ris-assisted rate-splitting multiple access

Abstract

The application discloses a scalable communication method based on RIS-assisted rate splitting multiple access, which comprises the following steps: establishing a multi-device downlink communication system model of RIS-assisted rate splitting multiple access, transmitting data from a base station to a user through rate splitting multiple access, constructing a public data stream and a private data stream, transmitting a signal by the base station using a precoding matrix, receiving the signal by the user after reflection of the RIS, training a neural network RISnet to learn the mapping of available CSI to a reflection phase shift matrix of the RIS, constructing a target function to maximize the weighted sum rate of the model, decoding the public data stream and the private data stream separately at a receiving end, and merging information symbols of the public data stream and the private data stream after decoding to reconstruct an original message. The application expands the transmission range of the signal while ensuring the communication quality, and solves the high-dimensional scalability problem caused by a large number of RIS reflection units.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wireless communication, in particular to a scalable communication method based on RIS-assisted rate division multiple access. BACKGROUND

[0002] Reconfigurable Intelligent Surface (RIS) has attracted much attention in recent years. It is a two-dimensional thin layer of artificial electromagnetic surface structure with programmable electromagnetic properties, composed of many passive and reconfigurable units. By adjusting the electromagnetic parameters (such as phase and amplitude) of the incident electromagnetic wave, it can reflect the incident electromagnetic wave with complex moving phase, enhance or attenuate the reflected signal, optimize the channel characteristics, and improve the communication quality and expand the coverage of the signal. In most typical scenarios, in order to establish sufficient link budget, a large RIS with several hundred to several thousand reflecting units needs to be deployed. The deployment of a large number of RIS reflecting units leads to cost problems, and also makes it difficult to obtain Channel State Information (CSI).

[0003] The traditional method of obtaining CSI is mainly based on the pilot method. Even if it is considered that the channel state is only changing slowly, the user CSI obtained by the transmitter may change dramatically due to the user's high speed or occasional large obstacles. Therefore, obtaining complete channel information requires high pilot overhead and computational complexity, resulting in outdated or incomplete CSI.

[0004] Many researchers use machine learning, deep learning, meta-learning, etc. to train models to predict channels to optimize RIS configuration, but the complexity of the network model will also increase with the number of RIS reflecting units, and scalability remains an open question.

[0005] Existing technologies related to Reconfigurable Intelligent Surface-assisted rate division multiple access communication include:

[0006] For example, the optimization method for a dual-IRS-assisted wireless power-carrying rate division communication network disclosed in patent publication CN115103375A considers maximizing the minimum user communication rate, thereby maximizing the user communication rate under the user energy harvesting requirement and taking into account fairness. In addition, due to the deployment strategy of dual RIS, greater communication rates can be achieved under limited available resources, improving communication resource utilization and communication coverage.

[0007] The method is used for solving the beamforming, rate segmentation and phase shift optimization sub-problems iteratively for the constructed target non-convex optimization problem of maximizing system energy efficiency.

[0008] However, the above-mentioned disclosed schemes have the problem that all the reflecting elements of the RIS in the system model can obtain ideal CSI, or the number of the set RIS reflecting elements is far different from the number of RIS required to achieve the necessary link budget in many scenarios. In practice, the incompleteness of the CSI makes the beamformer designed based on the CSI not accurate enough to match the actual channel, thereby causing the decline of system data transmission performance and safety performance. The existing technology has not simultaneously considered the extension of the RIS reflecting elements and the rate segmentation-based scalable communication technology based on the limited CSI that can be obtained. SUMMARY

[0009] In order to overcome the defects and deficiencies existing in the prior art, the present application provides a scalable communication method based on RIS-assisted rate segmentation multiple access, which is used to solve the high-dimensional scalability problem caused by a large number of RIS reflecting elements, and through the neural network architecture to optimize the configuration of the CSI to significantly expand the coverage of the network under the condition of limited CSI acquisition in the actual system. At the same time, considering the robustness of the rate segmentation multiple access technology to non-ideal CSI, RSMA is introduced into the RIS-assisted multi-device downlink communication system to further reduce the difficulty of channel estimation, thereby effectively improving the reliability of communication and system rate.

[0010] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows:

[0011] The present application provides a scalable communication method based on RIS-assisted rate segmentation multiple access, comprising the following steps:

[0012] Establishing a RIS-assisted multi-device downlink communication system model based on rate segmentation multiple access;

[0013] The base station sends a message to the user, splits the message into a private part and a public part, encodes the public part into a public data stream, and encodes the private part into a private data stream;

[0014] The base station transmits a signal using a precoding matrix, and the user receives the signal after reflection by the RIS;

[0015] Construct a neural network, RISnet, and generate training and test sets with different random user locations based on ray tracing channel data. Train the neural network, RISnet, to learn the mapping from available CSI to the reflection phase shift matrix of the RIS. Construct an objective function to maximize the model's weighted sum rate.

[0016] Calculate the public data stream precoding matrix and private data stream precoding matrix under the current RIS configuration, obtain the reflection phase shift matrix of each data sample, and calculate the mixing channel and objective function of each data sample. After iterative training, output the optimized public data stream precoding matrix, private data stream precoding matrix and reflection phase shift matrix;

[0017] At the receiving end, the public data stream and the private data stream are decoded separately, and then the information symbols of the public data stream and the private data stream are combined to reconstruct the original message.

[0018] As a preferred technical solution, the multi-device downlink communication system model includes a base station equipped with M antennas, K single-antenna users, and a reconfigurable intelligent reflective surface (RIS) with N reflective units;

[0019] The base station transmits data to users through Rate Splitting Multiple Access (RSMA).

[0020] As a preferred technical solution, the base station uses a precoding matrix to transmit signals, and the user receives the signals after reflection through the RIS. Specifically, it includes:

[0021] The base station uses the precoding matrix V = [V c ,V1,V2,..,V K ]Transmission signal s=[s c ,s1,s2,..,s K ] T , where V c represents the common data stream precoding matrix, V k represents the private data stream precoding matrix, s c Indicates public data flow, s k represents the private data stream of the kth user, k∈K;

[0022] The signal received by the user after reflection from the RIS is:

[0023] y=(GΦL+D)Vs+n=HVs+n

[0024] Where G represents the communication channel from RIS to user, L represents the communication channel from base station to RIS, Φ represents the reflection phase shift matrix of RIS, D represents the communication channel from base station to user, n represents the noise interference matrix, and H = GΦL + D is the mixed channel.

[0025] As a preferred technical solution, the common data stream precoding matrix is specifically represented as:

[0026]

[0027] The private data stream precoding matrix is specifically represented as:

[0028]

[0029] wherein K represents the number of users, P c = (1-t)P t , respectively, t∈(0,1] is a power allocation factor, P t represents the maximum transmit power, I N represents an identity matrix of size N×N, h k represents the mixing channel of the kth user, h i represents the mixing channel of the ith user.

[0030] As a preferred technical solution, the common data stream and the private data stream are both standardized power, and E represents expectation.

[0031] As a preferred technical solution, the neural network RISnet learns the mapping of available CSI to the reflection phase shift matrix of the RIS, specifically comprising:

[0032] Taking the channel feature Γ as the equivalent CSI representation, the neural network RISnet maps the equivalent CSI to the reflection phase shift matrix of the RIS;

[0033] The feature between the user u and the RIS reflection unit n is set as γ un The complete channel feature is a three-dimensional tensor, and the element corresponding to the second dimension index is the user u, and the third dimension index is the reflection unit n. un

[0034] As a preferred technical solution, the neural network RISnet includes multiple layers of ordinary layers and extended layers;

[0035] The output feature of the user u and the RIS reflection unit n after being processed by the ith ordinary layer is represented as:

[0036]

[0037] wherein, is the trainable bias of the ith layer of the cc type, is the trainable weight of the ith layer of the cc type,​ is the trainable bias of the i-th layer of the ca class, is the trainable weight of the i-th layer of the ca class, is the trainable bias of the i-th layer of the oc class, is the trainable weight of the i-th layer of the oc class, is the trainable bias of the i-th layer of the oa class, is the trainable weight of the i-th layer of the oa class, N represents the number of layers, U represents the total number of users, the cc class is the current user and the current RIS reflection element, the ca class is the current user and all available RIS reflection elements, the oc class is the other user and the current RIS reflection element, and the oa class is the other user and all available RIS reflection elements;

[0038] For the cc class in the i-th layer, after the user u and the RIS element n are locally feature-extracted by the convolutional layer, the corresponding global output feature is calculated by the fully connected layer, and the weighted average value of the result is taken when the other classes process information;

[0039] In the expansion layer, multiple information processing units are used for feature processing for each class, the RIS reflection element with CSI is defined as an anchor element, the same information processing unit is applied to the adjacent RIS reflection element with the same relative position as the anchor element, and the output of the RIS reflection element n using the information processing unit j is calculated as follows:

[0040]

[0041] wherein v(n, j) refers to the index of the other RIS reflection element corresponding to the processing when the input RIS reflection element is n, one information processing unit is used to process information in the last layer, and the features of different users are attributed to the phase shift.

[0042] As a preferred technical solution, the target function is represented as:

[0043]

[0044] s.t.tr(VV H )≤P t ,

[0045] |φ nn |=1,

[0046] |φ nn' | n≠n' =0.

[0047] wherein a k represents the weight of the user k, R ck represents the instantaneous achievable rate of the common data stream, and R kdenotes the instantaneous achievable rate of the private data stream, K denotes the number of users, V c denotes the precoding matrix of the common data stream, V k denotes the precoding matrix of the private data stream, denotes the reflection phase shift matrix of the RIS, V denotes the precoding matrix, P t denotes the maximum transmit power, tr denotes the trace, | denotes the diagonal element of the nth row and the nth column of nn | denotes the diagonal element of the nth row and the nth column of nn′ | n≠n′ denotes the off-diagonal element.

[0048] As a preferred technical solution, the instantaneous achievable rate of the common data stream and the instantaneous achievable rate of the private data stream are calculated according to the Shannon formula, and are specifically represented as:

[0049] R ck = log2(1 + SINR c,k )

[0050] R k = log2(1 + SINR p,k )

[0051]

[0052] SINR c,k denotes the signal-to-interference-and-noise ratio of the common data stream, SINR p,k denotes the signal-to-interference-and-noise ratio of the private data stream, P c = (1-t)P t , are the power allocated to the common message and the power of the private message respectively, t e (0, 1] is a power allocation factor, P t denotes the maximum transmit power, denotes the noise interference received by the kth user, K denotes the number of users, h k denotes the mixing channel of the kth user, H = GFL + D is the mixing channel, G denotes the communication channel from the RIS to the user, L denotes the communication channel from the base station to the RIS, denotes the reflection phase shift matrix of the RIS, and D denotes the communication channel from the base station to the user.

[0053] As a preferred technical solution, the common data stream and the private data stream are decoded separately at the receiving end, and specifically include:

[0054] First, the common data stream is decoded, all private data streams are regarded as interference or noise, the common data stream is removed through the serial interference cancellation technology, and then the private data stream sent to each user's own private data stream is decoded, and the private data streams of other users are regarded as interference or noise.

[0055] Compared with the prior art, the present application has the following advantages and beneficial effects:

[0056] (1) The scalable communication method based on RIS-assisted rate-splitting multiple access can combine the robust superiority of RSMA to partial CSI and RISnet for optimizing RIS configuration, optimize the phase shift of all RIS reflection units and the precoding matrix on the base station side under the condition of limited CSI acquisition, expand the transmission range of the signal while ensuring the communication quality, and achieve a good compromise between hardware complexity and performance.

[0057] (2) Compared with the traditional SDMA, under different channel conditions knowing only partial CSI, the present application introduces RSMA into the RIS-assisted multi-device downlink communication system to further reduce the difficulty of channel estimation, and RSMA can effectively improve the weighting and rate of the system.

[0058] (3) The present application is applicable to the acquisition requirements of any number of RIS reflection units and any number of users, has practical applicability and flexibility, and is suitable for various communication application scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0059] Figure 1 It is an implementation architecture schematic diagram of the scalable communication method based on RIS-assisted rate-splitting multiple access of the present application.

[0060] Figure 2 It is a neural network RISnet architecture schematic diagram of the present application.

[0061] Figure 3 It is a flowchart of RISnet training of the present application.

[0062] Figure 4 It is a schematic diagram of the weighting and rate in the scheme proposed by the present application under different channel conditions.

[0063] Figure 5 It is a schematic diagram of the weighting and rate comparison between the scheme based on RSMA and the scheme based on SDMA proposed by the present application under different channel conditions. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0065] This embodiment provides a scalable communication method based on RIS-assisted rate division multiple access. With the assistance of RIS, some users that do not have a direct communication link with a base station or whose direct communication link is unreliable can receive signals or obtain enhanced signals, thereby expanding the coverage and reliability of communication and enabling large-scale RIS deployment with low complexity. The method specifically includes the following steps:

[0066] S1: Establish a multi-device downlink communication system model assisted by Rate Splitting Multiple Access (RIS) based on rate splitting multiple access. This system includes a base station, RIS, and several wireless users. There is no direct line-of-sight link between the base station and the users. RIS is used to help improve signal transmission between the base station and the users.

[0067] like Figure 1 As shown in the figure, the multi-device downlink communication system includes a base station equipped with M antennas, K single-antenna users, and a reconfigurable smart reflective surface (RIS) with N reflective units. The base station transmits data to the users through rate splitting multiple access (RSMA), but there is no direct communication link between the base station and the users. Each RIS reflective unit can receive the signal from the base station and reflect the signal to the user end through phase adjustment without changing the signal amplitude to assist communication, thereby expanding the signal's reach. are the communication channels from base station to user, base station to RIS and RIS to user respectively. It is the phase shift matrix of RIS, and only a few RIS reflection units are equipped with the RF chain for channel estimation.

[0068] S2: Based on the downlink user's received signal, combined with the statistical characteristics of the channel, the RIS phase shift matrix, and the base station-side precoding matrix, the data transmission rate of the public data stream and the data transmission rate of all user private data streams are derived. The objective function is constructed to maximize the weighted sum rate of the system model.

[0069] Determine the optimization variables as the phase shift matrix of the RIS and the precoding matrix on the base station side;

[0070] Forming constraints: including RIS phase shift unit mode constraints, base station maximum transmit power constraints, and public signal decoding constraints;

[0071] S3: Generate training and test sets with different random positions and build the neural network RISnet architecture;

[0072] In this embodiment, the input of RISnet in step S3 is channel characteristics and partial CSI information;

[0073] Since the original partial CSI cannot be directly taken as the input of RISnet, a channel feature Γ is defined as the equivalent CSI representation, specifically, a feature γ is defined for user u and RIS element n un There are three channel matrices in the RIS-assisted communication system considered in this embodiment: Where L can be assumed as a constant since the positions of the BS and RIS are fixed, while D and G are N θ The input of RISnet and will change with the position of the user. Since D cannot be directly mapped to the elements of RIS, J = DL + Therefore, the feature between user u and RIS reflecting element n can be defined as:

[0074]

[0075] Where |a| and arg(a) are the amplitude and phase of complex number a, and the complete channel feature is a three-dimensional tensor, and the element corresponding to the second index u and the third index n is γ un ;

[0076] S4: Train RISnet to learn the mapping relationship between available CSI, optimal precoding and phase adjustment, alternately optimize the precoding matrix and the phase shift matrix, and test the performance based on the test set after each training iteration;

[0077] In the offline training phase, a neural network N θ (Actually, it is RISnet) is first defined with θ as the parameter and is mapped from available CSI to the reflecting phase shift matrix Φ of RIS, i.e. Φ = N θ (Γ), where the channel feature Γ is the equivalent CSI representation, and the objective function is determined by the channel feature Γ and the reflecting phase shift matrix Φ, which can be written as f(Γ, Φ) = f(Γ, N θ (Γ); θ).

[0078] Preferably, parallel computing is used in step S4 to speed up training and testing.

[0079] As shown in Figure 2 , considering the trade-off between performance and complexity, the neural network RISnet architecture has a total of 8 layers, including normal layers and extended layers, of course, the number of network layers and the number of extended layers can also be adjusted according to specific application requirements;

[0080] Because the phase shift of each RIS reflecting element will depend on all users and all RIS reflecting elements, all users and available RIS reflecting elements are divided into the following four categories:

[0081] (1) cc class: current user and current RIS reflecting elements;

[0082] (2) ca class: current user and all available RIS reflecting elements;

[0083] (3) oc class: other users and current RIS reflecting elements;

[0084] (4) oa class: other users and all available RIS reflecting elements;

[0085] Multiple information processing units are used to process local and global information, and the outputs are stacked as the inputs of the next layer, thus creating proper information flow for complex decisions on Φ;

[0086] In the normal layer, one information processing unit is adopted for each class, then the output features of user u and RIS reflecting element n after the i-th layer normal layer processing can be calculated by the following formula:

[0087]

[0088] where, is the trainable bias of the i-th layer for the cc class, is the trainable weight of the i-th layer for the cc class, P i represents the input feature dimension of the cc class, Q i represents the output feature dimension, similar definitions and the same dimensions are applicable to the ca class, oc class and oa class. For the cc class in the i-th layer, after the local feature extraction of user u and RIS element n by the convolution layer, a traditional fully connected layer is applied to calculate the corresponding global output feature, while the other classes take the weighted average of the results to enhance the nonlinear expression ability of the network.

[0089] Unlike the normal layer, in the expansion layer, 9 information processing units are used for each class to process features. Specifically, the RIS reflecting element with CSI (that is, available) is defined as an anchor element, and by applying the same information processing unit to the adjacent RIS reflecting element with the same relative position as the anchor element, each anchor element in a layer is expanded to 9 anchor elements, as shown in Figure 2 , that is, the features of the same RIS reflecting element and the adjacent 8 reflecting elements are output. By defining two such expansion layers, the number of anchor elements in the row and column is increased to 9 times the original. If there are only 16 RIS reflecting elements with CSI initially, after such two expansion layers, the phase shift of 1296 RIS reflecting elements can be obtained.

[0090] The output of RIS reflecting element n using information processing unit j is calculated as follows:

[0091]

[0092] Here, v(n,j) refers to the index of the other RIS reflection units processed when the input RIS reflection unit is n. In the last layer, only one information processing unit is used to process the information. Because all users share the same phase shift, the characteristics of different users are attributed to the phase shift.

[0093] S5: Save model parameters, optimal precoding, and some information about training and test sets at regular intervals.

[0094] like Figure 3 As shown in Figure 2, in the data preparation phase, the open source DeepMIMO ray tracing dataset is used to obtain real ray tracing channel data, and then the training set and test set with different random user positions are generated. In the training phase, RISnet is first randomly initialized, a batch of data samples are randomly selected, and then the precoding matrix V of the batch of samples under the current RIS configuration is calculated according to the precoding calculation formula. c 、V k , get the reflection phase shift matrix Φ of each data sample, calculate the mixing channel H of each data sample, and calculate the objective function of each data sample By calculating the objective function Regarding the gradient of the neural network, a stochastic gradient ascent step is performed to update the relevant parameters of the network. After each training round, the performance of the model is evaluated using the test set to adjust the learning rate. Finally, the saved model parameters can be used to obtain the objective function. Maximize the optimized phase shift matrix Φ and precoding matrix V c 、V k .

[0095] As shown in Table 1 below, the relevant system and model parameters are summarized. ADAM (Adaptive Moment Estimation) is an optimization algorithm commonly used in deep learning. It can adjust the learning rate of each parameter by calculating the first-order moment estimate and second-order moment estimate of the gradient.

[0096] Table 1 Related system and model parameters

[0097]

[0098] S6: Based on the optimal model parameters and precoding method obtained in the above steps, the system weighted sum rate is maximized while ensuring the expansion of the communication range.

[0099] The application combines the robustness of RSMA for partial CSI and the management ability of interference and RISnet for optimizing RIS configuration to configure the phase shift matrix of RIS and the precoding matrix of the base station side under the condition of limited CSI acquisition, so that the performance of the system is stronger.

[0100] In the embodiment, the specific working process of the rate splitting multiple access RSMA adopted is as follows:

[0101] (1) The base station sends K independent messages to K users, and the message sent to user k is represented by W k , According to the design principle of rate splitting, each user message W k is split into a private part W pk and a common part W ck , the common part {W c1 ,W c2 ,...,W cK} is jointly encoded into a common data stream s c using a codebook shared by all users, and the private part of user k is independently encoded into a private data stream s k , all data streams are standardized power, and E represents expectation;

[0102] (2) Under the constraint of maximum transmission power P t , the base station transmits a signal s = [s c ,s1,s2,..,s K ] T using a precoding matrix V = [V c ,V1,V2,..,V K ] c , wherein V k represents a common data stream precoding matrix, and V 2 represents a private data stream precoding matrix, K represents the total number of users;

[0103] (3) After reflection by the RIS, the signal received by the user end is y = (GΦL + D)Vs + n = HVs + n, wherein, is a noise interference matrix, n follows a distribution with a variance σ 2 and a mean of 0, that is, n ~ CN(0, σ 2 ), and H = GΦL + D is a mixed channel;

[0104] In the embodiment, the common stream adopts maximum ratio transmission (MRT) precoding, and the private stream adopts minimum mean square error (MMSE) precoding, that is, where and P c = (1-t)P t , are the power allocated to the common message and the power allocated to the private message, respectively, t ∈ (0, 1] is the power allocation factor, I N denotes an N x N identity matrix, whose diagonal elements are 1 and the rest of the elements are 0, N is the number of reflecting elements, h k denotes the mixed channel of the k-th user;

[0105] (4) At the receiver, each user first decodes the common data stream s c and regards all the private data streams as interference or noise, then cancels the common data stream through the serial interference cancellation (SIC) technique, and decodes the private data stream s k which is sent to the k-th user while regarding the private data streams of other users as interference or noise, so as to achieve the purpose of rate splitting by partially canceling the interference. The signal to interference plus noise ratio (SINR) of the common data stream at the k-th user is: and the SINR of the private data stream is According to the Shannon formula, the corresponding instantaneous achievable rates of the common data stream and the private data stream are R ck = log2(1 + SINR c,k ) and R k = log2(1 + SINR p,k ), respectively. denotes the noise interference received by the k-th user, i.e., the k-th part of σ 2 in the foregoing.

[0106] (5) After successfully decoding the common data stream and the private data stream, the k-th user combines the private part and the common part information symbols to reconstruct the original message.

[0107] Given the above-mentioned system model, the main goal of the embodiment scheme is to jointly optimize the phase shift matrix Φ of the RIS, the precoding matrices V c and V k at the base station side, and the common rate set C of all users to maximize the weighted sum rate (WSR) of all users, and the constructed optimization problem is as shown in

[0108] ​

[0109] tr(VV H )≤P t ,

[0110] |φ nn |=1,

[0111] |φ nn' | n≠n' =0.

[0112] where a k denotes the weight of user k, C k is the portion of the common rate allocated to user k, and R c is shared by users. Constraint 1 is the common rate constraint, which ensures that the common message can be decoded successfully by each user; constraint 2 indicates that the common rate allocated to user k should be non-negative; constraint 3 specifies that the total transmit power cannot exceed the maximum transmit power P t ; and constraints 4 and 5 ensure that the RIS does not amplify the received signal, |Φ nn | denotes the diagonal element of the nth row and the nth column of Φ, |Φ nn′ | n≠n′ denotes the off-diagonal element because the column is not equal to the row, and tr denotes the trace;

[0113] The optimization problem can be further simplified to by the following inference where the optimization variable C is no longer considered.

[0114]

[0115] s.t.tr(VV H )≤P t ,

[0116] |φ nn |=1,

[0117] |φ nn' | n≠n' =0.

[0118] The purpose of solving the original equivalent optimization problem is to jointly optimize the precoding matrices V c , V k and the phase shift matrix Φ of the RIS to maximize the WSR, but the high-dimensional complexity of this problem makes it difficult to find the optimal solution, so the present embodiment learns the mapping between the available CSI, the optimal precoding and phase shift matrix by training a neural network RISnet,

[0119] To verify the effectiveness of the scheme, three different channels are considered in the simulation process, a deterministic ray tracing channel, an independent and identically distributed complex Gaussian gain channel, and a deterministic ray tracing channel plus a complex Gaussian gain, denoted as channel A, channel B and channel C respectively. Under the simulation conditions of channel A, since all RIS reflection units share the same propagation path, the CSI of all reflection units can be inferred by a small number of RIS reflection units; under the simulation conditions of channel B, since the channel gain is an independent and identically distributed complex Gaussian random variable in the RIS reflection unit, part of the CSI cannot provide sufficient information about the complete channel, so it is difficult to infer the complete CSI; channel C is between the two, determined by the strong ray tracing channel and the complex Gaussian gain caused by weak scattering. It is hoped that the introduction of RSMA will have the advantage of "robust performance in the case of partial CSI" compared with SDMA, and through simulation, it is found that under the conditions of channel B and C, the system based on the RSMA transmission scheme can also obtain good performance.

[0120] As shown in Figure 4 , under the above simulation conditions, the weighted sum rate of the scheme proposed in the application in channel A is obtained, and it can be seen that the scalable communication scheme based on rate splitting multiple access of the application achieves good system weighted sum rate results, and limited CSI obtains similar performance to complete CSI.

[0121] As shown in Figure 5 , the comparison results of the weighted sum rate of the scheme proposed in the application and SDMA under three channel simulation conditions are obtained, and the RISnet model used is Partial CSI, and it can be seen that: under the condition of known Partial CSI, RSMA is significantly better than SDMA, indicating that compared with the traditional SDMA, under different channel conditions where only part of the CSI is known, RSMA can effectively improve the weighted sum rate of the system.

[0122] The above embodiments are the preferred embodiments of the application, but the embodiments of the application are not limited by the above embodiments, and any changes, modifications, substitutions, combinations and simplifications made without departing from the spirit and principles of the application shall be equivalent replacement methods, and all shall be included in the protection scope of the application.

Claims

1. A scalable communication method based on RIS-aided rate-splitting multiple access, characterized in that, The method comprises the following steps: A multi-device downlink communication system model based on RIS-assisted RSMA is established; The base station transmits a message to the user, splits the message into a private part and a public part, encodes the public part into a public data stream, and encodes the private part into a private data stream; The base station transmits a signal using a precoding matrix, and the user receives the signal after reflection by the RIS; A neural network RISnet is constructed, training sets and test sets with different random user positions are generated based on ray tracing channel data, the neural network RISnet is trained to learn the mapping of available CSI to the reflection phase shift matrix of the RIS, and a target function is constructed to maximize the weighted sum rate of the model; The neural network RISnet is trained to learn the mapping of available CSI to the reflection phase shift matrix of the RIS, specifically including: The channel feature Γ is taken as an equivalent CSI representation, and the neural network RISnet maps the equivalent CSI to the reflection phase shift matrix of the RIS; Let the feature between user u and RIS reflecting element n be denoted as γ un , the complete channel feature is a three-dimensional tensor, the element corresponding to the second dimension index user u and the third dimension index reflecting element n is γ un ; The neural network RISnet includes multiple layers of ordinary layers and extended layers; The output feature of the user u and the RIS reflection element n after processing by the i-th ordinary layer is represented as: wherein, is a trainable bias of the i-th layer for the cc-class, is a trainable weight of the i-th layer for the cc-class, is a trainable bias of the i-th layer for the ca-class, is a trainable weight of the i-th layer for the ca-class, is a trainable bias of the i-th layer for the oc-class, is a trainable weight of the i-th layer for the oc-class, is a trainable bias of the i-th layer for the oa-class, is a trainable weight of the i-th layer for the oa-class, N denotes the number of layers, U denotes the total number of users, the cc-class is the current user and the current RIS reflecting element, the ca-class is the current user and all available RIS reflecting elements, the oc-class is the other users and the current RIS reflecting element, and the oa-class is the other users and all available RIS reflecting elements. For the cc class in the i-th layer, after local feature extraction of the user u and the RIS element n by the convolution layer, the corresponding global output feature is calculated by the fully connected layer, and the weighted average value of the result is taken when other classes process information; In the extended layer, multiple information processing units are used for feature processing for each class, the RIS reflection element with CSI is defined as an anchor element, the same information processing unit is applied to adjacent RIS reflection elements with the same relative position as the anchor element, and the output of the RIS reflection element n using the information processing unit j is calculated as follows: Where v(n,j) refers to the index of the other RIS reflection element corresponding to the processing when the input RIS reflection element is n, and in the last layer, one information processing unit is used to process information, and the features of different users are summarized as phase shifts; The target function is represented as: s.t.tr(VV H )≤P t , | φ nn | = 1, | φ nn' | n≠n' = 0. where a k denotes the weight of user k, R ck denotes the instantaneous achievable rate of common data stream, R k denotes the instantaneous achievable rate of private data stream, K denotes the number of users, V c denotes the precoding matrix of common data stream, V k denotes the precoding matrix of private data stream, Φ denotes the reflection phase shift matrix of RIS, V denotes the precoding matrix, P t denotes the maximum transmit power, tr denotes the trace, |Φ nn | denotes the diagonal element of the nth row and the nth column of Φ, |Φ nn′ | n≠n′ denotes the off-diagonal element; The instantaneous achievable rate of the public data stream and the instantaneous achievable rate of the private data stream are calculated according to the Shannon formula, and are specifically represented as: R ck = log2(1 + SINR c,k ) R k = log2(1 + SINR p,k ) where SINR c,k denotes the signal-to-interference-and-noise ratio of the common data stream, SINR p,k denotes the signal-to-interference-and-noise ratio of the private data stream, P c = (1 - t)P t , are the power allocated to the common message and the power allocated to the private message, respectively, t ∈ (0, 1] is the power allocation factor, denotes the noise interference suffered by the kth user, h k denotes the hybrid channel of the kth user, H = GΦL + D is the hybrid channel, G denotes the communication channel from the RIS to the user, L denotes the communication channel from the base station to the RIS, and D denotes the communication channel from the base station to the user; The public data stream precoding matrix and the private data stream precoding matrix under the current RIS configuration are calculated, the reflection phase shift matrix of each data sample is obtained, the hybrid channel of each data sample is calculated, and the target function is calculated, and after iterative training, the optimized public data stream precoding matrix, private data stream precoding matrix and reflection phase shift matrix are output. The public data stream and the private data stream are decoded separately at the receiving end, and after decoding, the information symbols of the public data stream and the private data stream are merged to reconstruct the original message.

2. The RIS-aided rate division multiple access based scalable communication method of claim 1, wherein, The multi-device downlink communication system model includes a base station equipped with M antennas, K single-antenna users, and a reconfigurable intelligent reflecting surface RIS with N reflection units; The base station transmits data to the user through RSMA. 3.The RIS-aided rate division multiple access based scalable communication method of claim 1, wherein, The base station transmits a signal using a precoding matrix, and the user receives the signal after reflection by the RIS, specifically including: The base station uses the precoding matrix V = [V c ,V1,V2,..,V K ]Transmission signal s=[s c ,s1,s2,..,s K ] T , where V c represents the common data stream precoding matrix, V k represents the private data stream precoding matrix, s c Indicates public data flow, s k represents the private data stream of the kth user, k∈K; The signal received by the user after reflection by the RIS is: y = (GΦL + D)Vs + n = HVs + n Wherein, G represents a communication channel of the RIS to the user, L represents a communication channel of the base station to the RIS, Phi represents a reflection phase shift matrix of the RIS, D represents a communication channel of the base station to the user, n represents a noise interference matrix, and H=GPhiL+D is a mixed channel.

4. The RIS-aided rate division multiple access based scalable communication method of claim 3, wherein, The public data stream precoding matrix is specifically represented as: The private data stream precoding matrix is specifically represented as: where K represents the number of users, P c = (1 - t)P t , are the power allocated to the common message and the power allocated to the private message, respectively, t ∈ (0, 1] is a power allocation factor, P t represents the maximum transmit power, I N represents an N x N identity matrix, h k represents the mixing channel of the kth user, h i represents the mixing channel of the ith user.

5. The RIS-assisted rate-splitting multiple access based scalable communication method of claim 3, wherein, Both public and private data streams are standardized power, and E denotes expectation.

6. The RIS-assisted rate-splitting multiple access based scalable communication method of claim 1, wherein, The decoding of the public data stream and the private data stream at the receiving end specifically includes: The public data stream is decoded first, all the private data streams are regarded as interference or noise, the public data stream is eliminated through a serial interference cancellation technology, and then the private data stream of each user is decoded and sent, and the private data streams of other users are regarded as interference or noise.

Images