An energy efficiency optimization method for multi-cell communication systems based on intelligent reflector-assisted rate division multiple access.

Abstract

An energy efficiency optimization method for multi-cell communication systems based on intelligent reflector-assisted rate division multiple access (RSMA) combines RSMA with intelligent reflector RIS. RIS intelligently regulates signal transmission links to enhance signal transmission quality, thereby improving signal quality for users at cell edges. RSMA effectively suppresses co-channel interference between users by flexibly controlling rate division strategies and beamforming. To this end, an optimization problem is constructed with the goal of maximizing system energy efficiency, jointly optimizing the beamforming vector, rate division multiple access matrix, and phase shift matrix on the base station side and the RIS side. To solve this non-convex optimization problem, an iterative solution is proposed for the beamforming, rate division, and phase shift optimization subproblems. Continuous convex optimization is used for the former, and semidefinite programming and penalty function methods are used for the latter. Compared with traditional space division multiple access and non-orthogonal frequency division multiple access, this method can achieve energy efficiency improvements of 28.5% and 10.2% for multi-cell communication systems, respectively.

Description

Technical Field

[0001] This invention belongs to the field of mobile communication technology, specifically relating to an energy efficiency optimization method for multi-cell communication systems based on intelligent reflector-assisted rate division multiple access. Background Technology

[0002] Compared to 4G / 5G communication, 6G will offer greater capacity, lower latency, higher reliability, higher security, and full-space coverage. Furthermore, to meet the demands of emerging applications such as Augmented Reality (AR) and Virtual Reality (VR) for higher speeds and lower latency, the network capacity of 6G wireless communication systems is expected to be 100 times that of 5G wireless communication systems. Simultaneously, exploring and overcoming the constraints of numerous uncontrollable factors in the wireless environment and reconfiguring the wireless transmission environment are new directions for 6G development. Over the past decade, to meet the diverse emerging service needs of users, communication scholars and experts have proposed and thoroughly researched various wireless technologies, most notably Ultra-Dense Networks (UDN), Massive Multiple-Input Multiple-Output (MIMO), cellular networks, and millimeter-wave communication. The rise of these technologies has led to a significant leap in spectrum efficiency, thereby meeting the ultra-high capacity requirements for large-scale communication between numerous wireless devices. However, the accompanying system energy consumption and hardware costs are key issues in practical applications. To realize the green and environmentally friendly concept of sixth-generation wireless communication, the key is to find environmentally friendly and energy-saving technological support.

[0003] Recently, Smart Reflective Surfaces (RIS), a key technology in 6G, has been recognized as a promising green and cost-effective solution. RIS improves the throughput and energy efficiency of wireless networks by reconfiguring the wireless propagation environment. Specifically, RIS consists of a two-dimensional array of numerous passive reflective elements, each of which can independently adjust the phase shift of the incident signal in real time via a RIS controller.

[0004] By segmenting users in the power domain, Non-Orthogonal Frequency Division Multiple Access (NOMA) can simultaneously serve multiple users with the same frequency or time resources. Therefore, NOMA-based access schemes can achieve higher spectral efficiency than traditional Orthogonal Multiple Access (OMA). However, with NOMA, users must decode all interference when receiving messages, which greatly increases the computational complexity required for signal processing. To address this issue, researchers have proposed the concept of Rate Division Multiple Access (RSMA) and applied it to wireless communication. Studies have shown that RSMA can achieve superior system performance. Summary of the Invention

[0005] To address the problems existing in the aforementioned background technology, this invention proposes an energy efficiency optimization method for multi-cell communication systems based on intelligent reflector-assisted rate division multiple access (RSMA), combining RSMA with intelligent reflector RIS. RIS intelligently regulates the signal transmission link to enhance signal transmission quality, thereby improving the signal quality for users at the cell edge. RSMA effectively suppresses co-channel interference between users by flexibly controlling the rate division strategy and beamforming. To this end, an optimization problem is constructed with the goal of maximizing system energy efficiency, and the beamforming vector, rate division multiple access matrix, and phase shift matrix on the base station side are jointly optimized. To solve this non-convex optimization problem, an iterative solution is proposed for the beamforming, rate division, and phase shift optimization subproblems. Continuous convex optimization is used for the former, and semidefinite programming and penalty function methods are used for the latter.

[0006] An energy efficiency optimization method for a multi-cell communication system based on intelligent reflector-assisted rate division multiple access includes the following steps:

[0007] Step S1: Establish a multi-cell system model, determine the base station, intelligent reflective surface (RIS), channel model between users, and RIS deployment scheme;

[0008] Step S2: Based on the actual service needs of users, construct user groups; determine the public data stream and private data stream in the rate segmentation multiple access (RSMA) and encode them on the base station side;

[0009] Step S3: Design a RIS-RSMA mechanism to improve system performance, construct an objective function to maximize system energy efficiency; determine the optimization variables as the beamforming vector and rate segmentation matrix on the base station side and the phase shift matrix on the RIS side; formulate constraints, including common signal decoding constraints, minimum user rate constraints, RIS phase shift unity mode constraints, maximum base station transmit power constraints, and user rate requirement constraints.

[0010] Step S4: The objective function constructed above is converted into a convex optimization problem, which is decomposed into beamforming and rate splitting matrix optimization subproblems and RIS phase shift optimization subproblems. The former is solved by continuous convex optimization SCA, and the latter is solved by semidefinite programming and penalty function method.

[0011] Step S5: Find the optimal solutions for the beamforming vector, phase shift matrix, and rate splitting matrix through the above steps, substitute them into the objective function to obtain the system energy efficiency, provide the algorithm flow and analyze the algorithm complexity, and finally compare the system energy efficiency of traditional SDMA, NOMA and the proposed scheme through simulation to verify the feasibility of the model and algorithm.

[0012] The beneficial effects achieved by this invention are as follows:

[0013] (1) In this method, in the multi-cell edge user scenario, the gain of system energy efficiency brought about by the combination of intelligent reflective surface (RIS) and rate division multiple access (RSMA) technology was investigated. The final analysis results show that compared with traditional space division multiple access (SDMA) and non-orthogonal frequency division multiple access (NOMA), RSMA can achieve energy efficiency improvement of 28.5% and 10.2% respectively in multi-cell communication system (how these values ​​are obtained and whether they can be supplemented in the specific implementation in the following text to reflect them).

[0014] (2) Compared with NOMA-based access schemes, it controls and reduces the computational complexity required for signal processing, thereby improving system efficiency.

[0015] (3) By utilizing the RIS reflection link, a high-quality wireless transmission channel is created collaboratively, thereby improving the user's signal power, meeting the user's higher rate requirements, and reducing the base station's transmission power, thus improving the system's energy efficiency. Attached Figure Description

[0016] Figure 1 This is an overall structural diagram of the scene in an embodiment of the present invention.

[0017] Figure 2 This is a schematic diagram illustrating the structural principle of RSMA operation in an embodiment of the present invention.

[0018] Figure 3 This is a system simulation diagram in an embodiment of the present invention.

[0019] Figure 4 These are system performance simulation diagrams under different mechanisms in the embodiments of the present invention. Detailed Implementation

[0020] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings.

[0021] This invention is based on Figure 1 The study scenario consists of three parts: base station, multiple RIS, and multiple users. The energy efficiency of the system is studied, with a focus on analyzing and optimizing the overall energy efficiency of the system brought about by RSMA and RIS.

[0022] Specifically, an energy efficiency optimization method for a multi-cell communication system based on intelligent reflectors and rate segmentation multiple access includes the following steps:

[0023] Step S1: Establish a multi-cell system model, determine the channel model between base stations, RIS, and users, and the deployment scheme of RIS;

[0024] Step S2: Based on the actual business needs of users, construct user groups; determine the public data stream and private data stream in RSMA, and encode them on the base station side;

[0025] Step S3: Design a RIS-RSMA mechanism to improve system performance, construct an objective function to maximize system energy efficiency; determine the optimization variables as the beamforming vector and rate segmentation matrix on the base station side and the phase shift matrix on the RIS side; formulate constraints, mainly including base station transmit power constraints, RIS phase shift unity mode constraints, and user rate requirement constraints.

[0026] Step S4: Propose a low-complexity algorithm to transform the non-convex optimization problem constructed above into a convex optimization problem. It is mainly decomposed into beamforming and rate splitting matrix optimization subproblems and RIS phase shift optimization subproblems. The former is solved by continuous convex optimization (SCA), and the latter is solved by semidefinite programming and penalty function method.

[0027] Step S5: Find the optimal solutions for the beamforming vector, phase shift matrix, and rate splitting matrix through the above steps, substitute them into the objective function to obtain the system energy efficiency, give the algorithm flow and analyze the algorithm complexity, and finally compare the system energy efficiency of traditional SDMA, NOMA and the proposed scheme through simulation to verify the feasibility of the model and algorithm.

[0028] First, in step S1, a multi-cell system model is established, such as... Figure 1 As shown, a RIS-assisted multi-cell communication system model is constructed. A system equipped with N... t A 5G base station with K antennas serves K cell edge users with single antennas. These users will be subject to co-channel interference from neighboring cells, where multiple intelligent reflectors (RIS) equipped with N reflector elements act as cooperative nodes to reflect signals transmitted by the base station. For ease of description, ... Represents a collection of neighborhoods. This represents the set of all users at the edge of the community. Represents a set of RIS. For example... Figure 2 The incident signal is emitted by multiple RIS (Reflection Base Stations), where each RIS contains N reflection units. The reflection coefficient matrix of the r-th RIS is represented as:

[0029]

[0030] For simplicity, assume that the coefficient β of all reflecting units is... n =1, meaning the reflecting unit only reflects the signal and does not bring any gain. For user k, assume its group number is The required signal for this group is represented as follows: The signal received by user k can be represented as:

[0031]

[0032] in This represents the direct complex channel coefficient matrix between the l-th cell base station and user k, where the superscript H represents H l,k The conjugate transpose of x l This represents the signal transmitted by the l-th cell base station. This represents the complex channel coefficient matrix between the r-th RIS and user k. Let n represent the complex channel coefficient matrix between the l-th cell base station and the r-th RIS. k This indicates that additive white Gaussian noise satisfies in This represents the noise variance. Direct and indirect channels can be combined to form a cascaded channel. but Let represent the downlink channel link of the system. Note that the channel coefficient matrix changes in real time. For ease of study, it is assumed that all channel coefficients remain constant within a time block, but change independently in different time blocks.

[0033] Next, in step S2, as Figure 2 Users at the cell edge are divided into B multicast groups based on the content of their requests. For ease of representation, let's use... Indicates a multicast group set. Let W represent the set of all users in group b. Users in the same group request the same information from the base station. Assume that the information requested by all users in group b is W. b At the base station transmitter, based on the RSMA design principle, Information W required in the group b Divided into common parts W c,b and private part W p,b , where all common parts {W c,1 ,…,W c,B The code is jointly encoded using a codebook shared by all users into a common data stream s0 that all users expect, while the private part {W}... p,1 ,…,W p,B} are encoded into their respective user-expected data streams {s1,…,s B All data streams are normalized to power, i.e. The superscript H represents the conjugate transpose of the signal. At the base station, these data streams are linearly superimposed at different power levels to generate the transmitted signal vector x. Assuming the l-th cell is the serving cell, the signal transmitted by its base station is represented as:

[0034]

[0035] in N represents the private data stream beamforming vector of group b. t This refers to the base station's transmitting antenna. The beamforming vector represents the common data stream, s0 represents the common data stream, s b This represents the private data stream of group b. The maximum transmit power of the base station in the l-th cell is denoted as P. l,max Then we have:

[0036]

[0037] Furthermore, in step S3, a RIS-RSMA mechanism is designed to improve system performance. To achieve this goal, an optimization problem aimed at maximizing system energy efficiency is modeled. For RSMA technology, the common data stream s0 is first broadcast to all users and decoded by them using serial interference cancellation (SIC). Therefore, the achievable common rate of the entire system is limited by the minimum common rate among all users, i.e.:

[0038]

[0039] Among them, R c c represents the achievable rate of decoding the common data stream s0. k For users, the common rate is allocated. Note that the core of RSMA technology is how to allocate the common rate. In the group... In the middle, the common rate is only It is a group The part that the user needs, including |W p,b | indicates the length of information W. For ease of representation, ... Define as a group The rate at which public messages are decoded. Therefore, it must meet the following requirements.

[0040] In private data streams b Send to the group via multicast All users and groups Decoding private data streams b The achievable rate is limited by the minimum private decoding rate of this group of users, i.e. Where R p,b Indicates user The rate at which private data is decoded. Therefore, based on the above analysis, the achievable data rate is the same for all users in the same group, and can be defined as the group rate. For a group... In this context, the group rate can be defined as the common rate. and private rate (corresponding group) ) and:

[0041]

[0042] To represent interference between cells, Substitution Zhongde:

[0043]

[0044] in, This indicates interference with data streams from other groups within the serving cell. Represents the l-th base station for group The corresponding beamforming vector, s l,b n represents the signal corresponding to group b for the l-th base station. k This indicates additive noise interference. This indicates interference from other cell base stations. s represents the channel from the nth base station to the user k. n,b Let represent the signal from the nth base station corresponding to group b. To achieve quantization, assume unit bandwidth. According to Shannon's formula, the common data rate received by user k is expressed as:

[0045]

[0046] in, The signal-to-noise ratio (SNR) for receiving public data is represented by the denominator, where the denominator represents interference from other data streams within the cell and co-channel interference from other cells. After removing public signal interference using Sequential Interference Cancellation (SIC) technology, the private data rate received by user k is expressed as:

[0047]

[0048] in The signal-to-noise ratio of the received private signal is represented by the denominator, which represents the interference from other private data streams after removing interference from the public data stream, as well as co-channel interference and noise interference from other cells.

[0049] The total power consumption of the RIS-assisted RSMA multi-cell system under consideration includes the base station transmit power, the circuit power consumption of the base station and all users, and the power consumption of all RIS. Therefore, the total power consumption of the system is expressed as:

[0050]

[0051] in v l =μ l -1 , where μ l P represents the amplifier efficiency of the l-th cell base station. lP represents the power consumption of the base station's circuitry. k P represents the power consumed by the user circuit. R This represents the power consumed by each reflector unit of the RIS.

[0052] Given the system model under consideration, the objective of this study is to jointly optimize the beamforming vector of the base station. (i.e., the beamforming vector on the base station side), the phase shift matrix of RIS Φ = diag(Φ1,…,Φ R ), rate segmentation matrix To maximize the system's energy efficiency, the optimization problem modeled is:

[0053]

[0054]

[0055]

[0056]

[0057]

[0058]

[0059] Constraint 1 ensures that all users can decode the common signal, and the minimum rate limit for all users is specified in Constraint 2 (which defines the user rate R). k It needs to be greater than or equal to its minimum set value to avoid negative rates. Constraint 3 indicates the RIS phase shift limit. Let represent the nth RIS unit of the r-th RIS. The fourth constraint represents the maximum transmit power constraint of the base station, and the fifth constraint represents the non-negative rate allocated to each user. Since the optimization objective is a fractional programming problem and the optimization variables are coupled with each other, and since the first, second, and third constraints are all non-convex, it is a non-convex optimization problem. Next, we propose an alternating optimization algorithm, a continuous convex optimization algorithm, an SDR algorithm, and a penalty function method to transform the non-convex problem into a convex problem.

[0060] Secondly, in step S4, to solve the energy efficiency optimization problem in (P1), after decoupling the variables, it is transformed into two sub-problems: alternating optimization of the phase shift vector and beamforming vector. A low-complexity iterative algorithm is proposed. First, given the RIS phase shift matrix, the SCA method is used to optimize the beamforming vector and rate segmentation matrix. Then, based on the obtained variable values, the SDR and penalty function methods are used to optimize the RIS phase shift matrix. Finally, the alternating optimization algorithm is used until the objective function converges. First, the beamforming vector and rate segmentation matrix of the serving cell base station are studied, and then the beamforming vector and rate segmentation matrix of the interfering cell are designed. For convenience, let... An alternating optimization algorithm is used to optimize the objective function. The phase shift of RIS and w on other cells are fixed. m,m≠l The optimization problem described above can be reformulated as:

[0061]

[0062]

[0063]

[0064]

[0065]

[0066] The objective function of the optimization problem (P2) is a fractional programming problem, which is easily proven to be non-convex; the first and second constraints are also non-convex. Next, the SCA method is used to transform the non-convex constraints into convex ones, as follows:

[0067] First, we introduce slack variables. The first constraint can be equivalent to the following two constraints, where β k The slack variable representing the k-th user introduced is:

[0068]

[0069]

[0070] Similarly, the second constraint can be equivalent to the following two constraints, δ k Slack variables representing k users:

[0071]

[0072]

[0073] Since the above constraints are fractional constraints, they are still non-convex constraints. Therefore, slack variables are further introduced. Therefore, it can be equivalent to the following two constraints, γ k It is a slack variable, which can be regarded as a variable, η k Also a variable:

[0074]

[0075]

[0076] Similarly, the first two constraints in (P2) are equivalent to the following two constraints.

[0077]

[0078]

[0079] The above constraint is still a non-convex problem, which can be approximated using a first-order Taylor expansion:

[0080]

[0081]

[0082] Among them, such as (x) (n) The form represents the value of x in the nth iteration. From the two constraints above, it can be seen that this constraint is about w. l,0 ,w l,u(k) ,γ k ,η k The expression is a first-order linear polynomial, therefore the constraint is a convex constraint. Finally, the optimization problem is equivalent to...

[0083]

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090] It can be seen that the optimization objective is a concave-convex fractional programming problem with respect to the objective function, which can be solved using the classic Dinkelbach method. To better illustrate the algorithm flow above, Algorithm 1 summarizes the detailed process of the SCA algorithm.

[0091]

[0092] For the design of RIS reflection phase shift, given the base station-side precoding matrix... and rate segmentation matrix After removing the constant, the original optimization problem can be equivalent to a maximum sum rate problem, that is:

[0093]

[0094]

[0095]

[0096]

[0097] Considering that the first and second constraints implicitly contain the optimization variable Φ, and the third constraint contains the continuous phase shift constraint, this optimization problem is non-convex and cannot be solved using convex programming methods. Since the optimization involves only one variable, the problem can be simplified to a feasibility test problem of finding the phase shift Φ. The optimal value is obtained by searching for solutions within the feasible region. The specific process is as follows: Let... Will use Substituting, we get:

[0098]

[0099] make The above formula can be expressed as:

[0100]

[0101] Similarly, we can obtain

[0102]

[0103] make It can be obtained

[0104]

[0105] Therefore, the optimization problem can be equivalent to...

[0106]

[0107]

[0108] This optimization problem cannot be directly transformed into an SOCP problem, but it can be approximated using the SDR technique:

[0109]

[0110] Represent the above equation in matrix form.

[0111]

[0112] make Where t is an auxiliary complex variable satisfying |t|=1, then we can obtain:

[0113]

[0114] Similarly, we can obtain

[0115]

[0116] make It can be obtained

[0117]

[0118] make It can be obtained

[0119]

[0120] Then, through transformation, we can obtain:

[0121]

[0122] because make Where V > 0 and rank(V) = 1, the original optimization problem is equivalent to:

[0123] (P6)Find V

[0124]

[0125] V n,n =1,n=1,…,RN

[0126] V≥0

[0127] rank(V) = 1

[0128] To obtain a better convergent solution, the problem (P6) is further transformed into an optimization problem with a clear objective, in order to obtain a phase-shift solution that is usually more effective.

[0129]

[0130]

[0131] V n,n =1,n=1,…,RN

[0132] V≥0

[0133] The rank-one constraint in the rank(V) = 1 optimization problem (P7) can be equivalent to the following anti-convex constraint:

[0134] χ(V)-tr(V)=0

[0135] Where χ(V) represents the largest eigenvalue of V, therefore we can obtain Where v max Let χ(V) represent the eigenvector corresponding to the unit modulus of V. According to the properties of matrices, χ(V) ≤ tr(V) always holds true. To maximize χ(V) - tr(V), a penalty function method is introduced, and the optimization objective is transformed into... Where τ≥0 represents the penalty factor. Finally, the rank semidefinite programming optimization problem (P7) is further transformed into the following optimization problem, where τ represents the penalty factor:

[0136]

[0137] V n,n =1,n=1,…,RN

[0138] V≥0

[0139] Because the trace function is convex, it can be seen that all the above constraints are convex constraints, and the objective function is in the form of a convex difference. A first-order approximation method is used to solve this problem. In the j-th iteration, the point {V} can be used. {j} The objective function is approximated by a first-order low-order approximation at}, i.e., the objective function is equivalent to... Based on the above analysis, in the j-th iteration, the original optimization problem can be approximated as the following convex optimization problem:

[0140]

[0141] V n,n =1,n=1,…,RN

[0142] V≥0

[0143] It can be seen that the original problem is reduced to a standard convex optimization problem, which can be solved using the CXV tool. Iterative calculations can yield a local optimum with rank-one constraints. For better description, Algorithm 2 outlines the algorithm flow of the penalty function method:

[0144]

[0145]

[0146] Finally, to better describe the process of the alternating optimization algorithm used to solve the optimization problem (P1), Algorithm 3 presents the overall flowchart of the algorithm:

[0147]

[0148] Finally, in step S5, Algorithm 3 provides an iterative algorithm for solving the energy efficiency maximization problem in (P1). From Algorithm 3, it can be seen that the complexity of solving problem (P1) is mainly determined by the complexities of (P3) and (P9). Specifically, the complexity of the Continuous Convex Optimization (SCA) algorithm used for the beamforming vector and rate segmentation matrix is:

[0149] C1=O(I1(N t +BN t +2B+3K) 3.5 log(1 / ∈1)

[0150] Where O is the space complexity label, and N t Let N represent the number of transmit antennas, B represent the number of user groups, K represent the number of users, and I1 represent the number of iterations required for the algorithm to converge. t +BN t +2B+3K) represents the total number of variables

[14] , ∈1 represents the accuracy of the SCA method used to solve problem (P3); the complexity of the semidefinite relaxation algorithm and penalty function method used to solve the RIS phase-shift semidefinite programming problem is:

[0151] C2=O(I2(R(N+1)) 3.5 log(1 / ∈2)

[0152] Where (N+1) represents the dimension of the semidefinite programming matrix, R represents the number of RIS, and I² represents the number of iterations required for Algorithm 2 to converge. In summary, the total complexity of the proposed optimization algorithm is:

[0153] C = O(I) tot (C1+C2))

[0154] Where I tot Represents the total number of iterations in the algorithm.

[0155] To illustrate the effectiveness of the method proposed in this invention, an example is given below. The performance of the proposed algorithm is evaluated through simulation. For example... Figure 3 As shown, the system model under study is mapped to a real-world scenario with 3 base stations, 6 users randomly located at the cell edge, and 3 RIS (Reflection Nodes) acting as cooperative nodes to reflect signals emitted by the base stations. The base stations are located at (100, 400), (400, 100), and (400, 700). The 6 users are randomly located within a circle with a radius of 10m centered at (250, 400), and are divided into 3 groups of two users each. The RIS are positioned between the base stations and the users to provide high-quality signal transmission. Table 1 summarizes the system parameters.

[0156] Table 1 Simulation Parameters

[0157]

[0158]

[0159] Assume a direct link channel h d,k The RIS-assisted channel follows Rayleigh fading, and assumes that the antenna elements at the base station and RIS are half-wavelength uniform linear arrays. Therefore, channels G and H... r,k It can be modeled as:

[0160]

[0161]

[0162] Among them, L1, L 2,k This represents the corresponding path loss, σ represents the Ricaian factor (set to 10), and a represents the steering vector (a is used in the formula). N and a M θ, ψ, and φ represent steering vectors (the subscript is used to distinguish between different steering vectors). k These represent the azimuth angles of the base station transmitting antenna, the RIS reflector unit, and the user receiving signal, respectively. and This represents the non-line-of-sight component that follows CN(0,1).

[0163] Figure 4 The comparison diagrams of the various mechanism systems under the above simulation conditions show that the RIS-RSMA method described in this paper achieves the best performance efficiency results.

[0164] The above description is only a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. Any equivalent modifications or changes made by those skilled in the art based on the content disclosed in the present invention should be included within the scope of protection set forth in the claims.

Claims

1. An energy efficiency optimization method for a multi-cell communication system based on intelligent reflector-assisted rate division multiple access, characterized in that: Includes the following steps: Step S1: Establish a multi-cell system model, determine the base station, intelligent reflective surface (RIS), channel model between users, and RIS deployment scheme; in Step S1, one of the multi-cell system models is equipped with N t A 5G base station with one antenna serves cell edge users with K single antennas. These users will be subject to co-channel interference from neighboring cells, where multiple RIS with N reflector units act as cooperative nodes to reflect signals transmitted by the base station. Step S2: Based on the actual service needs of users, construct user groups; determine the public data stream and private data stream in the rate segmentation multiple access (RSMA) and encode them on the base station side; In step S2, the cell edge users are divided into B multicast groups according to the different content of their requests. The information required in each group is divided into a public part and a private part. All the public parts are jointly encoded into a public data stream expected by all users using a codebook shared by all users, while the private parts are encoded into the data stream expected by each user. Step S3: Design a RIS-RSMA mechanism to improve system performance and construct an objective function to maximize system energy efficiency. The optimization variables are determined as the beamforming vector and rate segmentation matrix on the base station side and the phase shift matrix on the RIS side; constraints are formed, including common signal decoding constraints, minimum user rate constraints, RIS phase shift unity mode constraints, maximum base station transmit power constraints, and user rate requirement constraints. Step S4: The objective function constructed above is converted into a convex optimization problem, which is decomposed into beamforming and rate splitting matrix optimization subproblems and RIS phase shift optimization subproblems. The former is solved by continuous convex optimization SCA, and the latter is solved by semidefinite programming and penalty function method. Step S5: Find the optimal solutions for the beamforming vector, phase shift matrix, and rate splitting matrix through the above steps, substitute them into the objective function to obtain the system energy efficiency, give the algorithm flow and analyze the algorithm complexity, and finally verify the feasibility of the model and algorithm through simulation comparison.

2. The energy efficiency optimization method for a multi-cell communication system based on intelligent reflector-assisted rate division multiple access according to claim 1, characterized in that: In step S2, the base station linearly superimposes the data streams at different power levels to generate a transmitted signal.

3. The energy efficiency optimization method for a multi-cell communication system based on intelligent reflector-assisted rate division multiple access according to claim 1, characterized in that: In step S3, the common rate is limited by the minimum common rate among all users, and the achievable rate of decoding the private data stream is limited by the minimum private decoding rate of the group of users.

4. The energy efficiency optimization method for a multi-cell communication system based on intelligent reflector-assisted rate division multiple access according to claim 1, characterized in that: In step S3, for each group, its group rate can be defined as the sum of the public rate and the private rate.

5. The energy efficiency optimization method for a multi-cell communication system based on intelligent reflector-assisted rate division multiple access according to claim 1, characterized in that: In step S3, interference between cells is represented by interference from data streams from other groups within the cell, interference from base stations in other cells, and additive noise interference.

6. The energy efficiency optimization method for a multi-cell communication system based on intelligent reflector-assisted rate division multiple access according to claim 1, characterized in that: In step S4, firstly, given the RIS phase shift matrix, the beamforming vector and rate splitting matrix are optimized using the SCA method; then, based on the obtained variable values, the RIS phase shift matrix is ​​optimized using the SDR and penalty function methods; finally, the objective function is converged through an alternating optimization algorithm.

Images