Interference exploitation method for reconfigurable intelligent surface assisted cognitive non-orthogonal multiple access network

Abstract

The application discloses a method for interference utilization of a reconfigurable intelligent surface assisted cognitive non-orthogonal multiple access network, and comprises the following steps: constructing a reconfigurable intelligent surface assisted downlink cognitive radio non-orthogonal multiple access system; a superposition coding is adopted by a base station to send signals to all users, and a joint optimization of an active beamforming vector of the base station and a passive reflection phase shift matrix of the reconfigurable intelligent surface is performed to construct a constructive superposition signal for direct detection of a secondary user; the secondary user directly demodulates own data through a single user detector without performing serial interference cancellation (SIC) based on the constructive superposition signal; by utilizing the intelligent reconstruction of a wireless propagation environment by the reconfigurable intelligent surface, the application adopts a constructive interference technology to convert strong interference of a primary user into a constructive signal of a secondary user; the mechanism realizes direct detection through the constructive interference design, greatly simplifies the receiving and processing of the secondary user and reduces the time delay.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and more specifically to a method for interference exploitation in reconfigurable smart surface-assisted cognitive nonorthogonal multiple access networks. Background Technology

[0002] With the evolution of 5G mobile communication and future wireless networks, the contradiction between scarce spectrum resources and massive device access is becoming increasingly prominent. To address this challenge, the industry has proposed several key technologies:

[0003] Non-orthogonal multiple access (NOMA) technology offers significant advantages in improving spectrum efficiency, supporting massive connectivity, and ensuring user fairness by serving multiple users within the same time, frequency, or code resources. Specifically, it achieves this by superimposing multiple user signals at the transmitter and decoding them using continuous interference cancellation (SIC) at the receiver. However, the achievable performance of NOMA systems largely depends on the quality of the wireless link; particularly in non-line-of-sight (NLoS) propagation scenarios characterized by obstacles, the quality of service (QoS) for users is severely degraded.

[0004] Reconfigurable Smart Surfaces (RIS): As a remedy, reconfigurable smart surfaces (RIS) have become an effective example of reconfiguring wireless propagation environments. By dynamically adjusting the phase shift of their reflective elements, RIS can establish virtual line-of-sight (LoS) links to bypass obstacles, thereby significantly improving channel conditions.

[0005] Cognitive Radio: To further alleviate the spectrum shortage problem, cognitive radio (CR) technology has become a key technology for improving spectrum utilization through intelligent sensing and dynamic access. The core principle of CR is to enable users to detect and utilize idle resources within licensed frequency bands, thereby establishing communication without causing harmful interference to the primary user.

[0006] Constructive interference: To overcome the performance limitations of traditional interference suppression techniques, constructive interference (CI) has received considerable attention in recent years. Specifically, CI is based on the principle of symbol-level precoding (SLP) and employs advanced beamforming design to generate a constructive superposition between multi-user interference and the desired signal at the receiver. The received symbols are then pushed deeper into the constructed region, away from the decision boundary of the signal constellation.

[0007] Combining CR, NOMA, and RIS technologies has become a research hotspot in the field of wireless communication. The integration of CR and NOMA generates powerful synergistic advantages. Specifically, CR's dynamic spectrum access capability provides NOMA systems with richer transmission opportunities and a larger resource pool; while NOMA's efficient multiplexing technology optimizes the capacity of identified spectrum opportunities, improving access fairness and service quality for disadvantaged secondary users. The introduction of RIS is expected to further address the performance bottlenecks caused by poor channel conditions in CR-NOMA networks.

[0008] However, the following key drawbacks still exist in RIS-assisted CR-NOMA networks:

[0009] 1. While Reconfigurable Smart Surface (RIS)-assisted NOMA networks have shown great potential for improving spectral and energy efficiency, most existing schemes focus on utilizing the passive beamforming capabilities of RIS to orthogonalize user channels or maximize the power difference between the desired signal and interference. These approaches aim to facilitate continuous interference cancellation (SIC) at the receiver. However, such a design philosophy inevitably hinders further performance improvements. On the one hand, strong signals from the primary user (PU) are consistently considered harmful co-channel interference to the secondary user (SU). Therefore, the base station (BS) and the RIS are forced to consume valuable spatial degrees of freedom (DoF) to suppress this interference, resulting in significant waste of transmit power and DoF.

[0010] 2. The realization of the expected gain in traditional NOMA heavily relies on the assumption of ideal successive interference cancellation (SIC) processing at the receiver. However, in practical deployments, especially in cognitive radio networks where the primary user (PU) signal may be several orders of magnitude stronger, the traditional NOMA paradigm faces severe challenges. SIC processing at the secondary user (SU) must first decode and cancel the strong PU interference, requiring extremely high computational accuracy. Due to hardware limitations, imperfect cancellation is unavoidable. Furthermore, channel estimation errors cause residual interference to propagate in the SIC cascade, severely degrading the decoding performance of the desired SU signal. In addition, accurate execution of SIC is highly dependent on the receiver possessing perfect channel state information, which incurs significant signaling overhead and is susceptible to inaccurate estimation or quantization errors.

[0011] 3. Traditional constructive interference (CI) precoding mainly relies on active beamforming at the base station (BS) to manipulate the signal phase. Under adverse channel conditions, phase adjustment at the BS alone is often insufficient to meet the strict geometric constraints of constructive interference (CI), resulting in a limited feasible solution space or excessive energy consumption.

[0012] In summary, existing technical solutions in RIS-assisted CR-NOMA networks suffer from problems such as low resource utilization, excessive receiver complexity, inadequate interference cancellation, and limited performance under adverse channel conditions. A novel transmission framework is urgently needed to overcome these bottlenecks. Summary of the Invention

[0013] To address the problems in the prior art, this invention provides an interference exploitation method for reconfigurable smart surfaces-assisted cognitive nonorthogonal multiple access networks. The aim is to utilize constructive interference (CI) and reconfigurable smart surfaces (RIS) to achieve the coexistence of primary users (PU) and secondary users (SU) in cognitive networks, and to improve the performance of secondary users (SU) without compromising the needs of primary users (PU).

[0014] A method for exploiting interference in reconfigurable smart surface-assisted cognitive nonorthogonal multiple access networks includes the following steps:

[0015] Construct a downlink cognitive radio nonorthogonal multiple access system assisted by a reconfigurable smart surface, the system comprising a base station, a reconfigurable smart surface, a primary user, and at least one secondary user;

[0016] The base station uses superposition coding to send signals to all users, and constructs a constructive superposition signal for direct detection by secondary users by jointly optimizing the active beamforming vector of the base station and the passive reflection phase shift matrix of the reconfigurable smart surface.

[0017] The secondary user, based on the constructive superimposed signal, can directly demodulate its own data through a single-user detector without performing serial interference cancellation (SIC).

[0018] Furthermore, the constructive superimposed signal is formed by the in-phase superposition of the primary user signal component and the secondary user's own signal component at the secondary user's receiving end, so that the primary user signal manifests as constructive interference to the secondary user.

[0019] Furthermore, the joint optimization includes: constructing an optimization problem with the objective of minimizing the total transmit power of the base station, and introducing geometric constraints to ensure that the constructive superimposed signal falls within the constructive region of the constellation diagram.

[0020] Furthermore, the optimization problem is solved using the alternating optimization (AO) algorithm, including:

[0021] Step A: Fix the passive reflection phase shift matrix of the reconfigurable smart surface and optimize the active beamforming vector;

[0022] Step B: Fix the optimized active beamforming vector and optimize the passive reflection phase shift matrix;

[0023] Iteratively execute steps A and B until the preset convergence condition is met.

[0024] Further, step A includes: using the continuous convex approximation technique, the non-convex signal-to-interference-plus-noise ratio constraint, power allocation constraint, and power ranking constraint are transformed into a linear approximation form through a first-order Taylor expansion, thereby transforming the original problem into a convex optimization problem to be solved.

[0025] Further, step B includes: introducing a reflection coefficient vector. And construct a positive semidefinite matrix. This transforms the original optimization problem into a semidefinite programming problem involving a positive semidefinite matrix. Represents the reflection coefficient matrix of RIS; Indicates taking The diagonal elements form a vector, and then the conjugate transpose is taken; For vectors The conjugate transpose of;

[0026] Using semidefinite relaxation techniques to relax non-convex rank-one constraints And solve it using convex optimization methods.

[0027] Furthermore, the rank-one constraint is relaxed using the difference between the trace norm and the spectral norm, expressed as: And a regularization term is introduced to ensure numerical stability and convergence.

[0028] Furthermore, the objective function of the optimization problem is the sum of the Euclidean norms of the active beamforming vectors of the base station, and the constraints include at least: primary user signal-to-interference-plus-noise ratio (SINNR) constraints, secondary user SINNR constraints, constructive interference geometry constraints, unit mode constraints of the reconfigurable smart surface reflector, power allocation constraints between primary and secondary users, and power ranking constraints among secondary users.

[0029] Furthermore, the power ranking constraint is linearized by introducing non-negative relaxation variables and employing a first-order Taylor expansion, specifically expressed as follows:

[0030]

[0031] in, Let the current iteration point be the... User precoding For the k-th sub-user precoding vector to be optimized, It is a non-negative slack variable. This indicates the operation of taking the real part.

[0032] Furthermore, the computational complexity of the secondary user directly detecting and demodulating its own data is linearly related to the modulation order of the secondary user, and there is no need to perform serial interference cancellation operations.

[0033] The beneficial effects of this invention are:

[0034] 1. By utilizing Reconfigurable Smart Surfaces (RIS) to intelligently reconstruct the wireless propagation environment, we employ Constructive Interference (CI) technology to transform strong interference from the primary user (PU) into constructive signals from the secondary user (SU). This mechanism achieves direct detection through Constructive Interference (CI) design, greatly simplifying the receiving processing of the secondary user (SU) and reducing latency.

[0035] 2. A joint optimization framework was established to jointly design the active beamforming vector of the base station (BS) and the passive reflection phase shift matrix of the reconfigurable smart surface (RIS). The goal is to minimize the total system transmit power while meeting the communication needs of all primary users (PU) and secondary users (SU).

[0036] 3. We apply the Continuous Convex Approximation (SCA) technique to transform the non-convex active beamforming problem into a convex problem. For the passive beamforming problem involving unit mode constraints, we adopt a penalty-based iterative optimization method. Finally, we develop an Alternating Optimization (AO) algorithm and use the CVX tool to iteratively solve these two convex subproblems. This solves the problem that the constructed problem is mathematically difficult to solve directly due to the high coupling of variables and the non-convex constraints. Attached Figure Description

[0037] Figure 1 This is a flowchart of the present invention;

[0038] Figure 2 A system model for RIS-assisted CI-NOMA networks;

[0039] Figure 3 A geometrical diagram illustrating the principle of constructive interference under quadrature phase shift keying (QPSK) modulation;

[0040] Figure 4 The graph shows the convergence behavior of the total transmit power as a function of the number of iterations.

[0041] Figure 5 This is a graph showing the relationship between total transmit power and the number of base station antennas.

[0042] Figure 6 The graph shows the relationship between total transmit power and the number of RIS reflector units;

[0043] Figure 7 A graph showing the relationship between total transmit power and primary user target SINR requirements;

[0044] Figure 8 A graph showing the relationship between total transmit power and secondary user target SINR requirements;

[0045] Figure 9 This is a graph showing the relationship between total transmit power and the number of secondary users. Detailed Implementation

[0046] The present invention will now be described in detail with reference to the accompanying drawings. Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. The directional terms such as left, center, right, top, and bottom in the embodiments of the present invention are only relative concepts or referenced to the normal use state of the product, and should not be considered restrictive.

[0047] Interference exploitation methods for reconfigurable smart surface-assisted cognitive nonorthogonal multiple access networks, such as Figure 1 As shown, it includes the following steps:

[0048] Step 1: Construct a reconfigurable smart surface (RIS)-assisted downlink cognitive radio nonorthogonal multiple access (CR-NOMA) system. Consider a reconfigurable smart surface (RIS)-assisted downlink multiple-input single-output cognitive radio network, such as... Figure 2 As shown; the system consists of a base station equipped with M antennas, a reconfigurable smart surface (RIS) with N reflector elements, a single-antenna primary user (PU), and K single-antenna secondary users (SUs); the set of secondary users is denoted as K; in the system under consideration, we assume that the direct line-of-sight link between the base station and the users is blocked by obstacles, so communication relies entirely on the reflective link provided by the reconfigurable smart surface (RIS); for ease of representation, the baseband channel matrix from the base station to the reconfigurable smart surface (RIS), the channel vector from the reconfigurable smart surface (RIS) to the primary user (PU), and the channel vector from the RIS to the k-th secondary user are respectively expressed as: ;

[0049] Step 2: The base station uses superposition coding to send signals to all users and jointly optimizes the active beamforming vector of the base station and the passive reflection phase shift matrix of the reconfigurable smart surface to construct a constructive superposition signal for direct detection by secondary users.

[0050] In this invention, the base station employs superposition coding technology to transmit signals to all users. Based on the principle of constructive interference (CI), the active beamforming vector of the base station and the passive reflection phase shift matrix of the RIS are jointly designed. The purpose is to strategically align the signal interference of the primary user (PU) with the desired signal of the secondary user (SU) through reflection by the RIS, so that the interference is transformed into a constructive component that enhances the signal power at the secondary user's receiver. The core idea of ​​this invention is to utilize known interference information in the signal space and, through precise beamforming design, phase align the interference with the desired signal of the secondary user. Therefore, the primary user's signal is no longer considered harmful interference; instead, it is transformed into useful signal energy, thereby improving the received signal-to-interference-plus-noise ratio of the secondary user. This step constructs a constructive superposition signal for direct detection by the secondary user through the following sub-steps.

[0051] Step 2.1: Calculate the primary user (PU) and the second user under conventional NOMA precoding and constructive interference NOMA (CI-NOMA) precoding respectively. The signal-to-interference-plus-noise ratio (SINR) of each user is given, and the mathematical definition of constructive superimposed signals is provided.

[0052] According to the NOMA protocol, the base station uses an overlay coding scheme to simultaneously transmit signals to both the primary user (PU) and the secondary user (SU). This signal can be represented as: ,in, , These represent the primary user and the secondary user, respectively. and These respectively represent being sent to the primary user and the second user. The data symbols of each user satisfy the normalized power constraint: and ; and These are respectively with the main user and the first The precoding vectors corresponding to each user; the total transmit power of the base station is limited by: ,in, This indicates the power occupied by the primary user signal. This represents the sum of the power occupied by all secondary user signals. Indicates the maximum available power budget;

[0053] Therefore, the received signal at the primary user can be written as: ;in, Representing vectors (Conjugate transpose), the effective concatenated channel vector from the base station through the reconfigurable smart surface to the primary user can be represented as: ,in, The reflection coefficient matrix of RIS is a diagonal matrix defined as follows: ; Represents the imaginary unit; here, and Representing the first The reflection amplitude coefficient and phase shift of each RIS unit; The additive white Gaussian noise at the primary user location has a variance of . Similarly, the first The received signal at each user can be expressed as: Similarly, Representing vectors (Conjugate transpose), from base station to the... The cascaded channel vector for each user is given by the following formula: ,and The additive white Gaussian noise at the secondary user location has a variance of ;

[0054] Conventional NOMA precoding:

[0055] In the traditional NOMA framework, the signal used by the secondary user (SU) is considered interference to the primary user (PU); therefore, the signal-to-interference-plus-noise ratio (SINR) received at the primary user (PU) is:

[0056]

[0057] in, This indicates the power of the signal that the primary user receives, which is the power of the signal it expects. It is the sum of the interference power caused by all secondary user signals to the primary user. It is the additive white Gaussian noise power at the main user receiver.

[0058] Assuming the units are ordered according to channel gain, the decoding order is from user 1 to... User arrangement, i.e., user 1 has the weakest channel conditions, while user... The channel conditions are strongest; for (No. (For each user), a stronger signal from the PU must first be decoded; therefore, it can represent The signal-to-interference-plus-noise ratio (SINR) of the decoded PU signal is:

[0059]

[0060] in, express The power of the received PU signal, It is all secondary user signals (including) The sum of the interference power caused by its own signal to the current decoding process. yes Additive white Gaussian noise power at the receiver;

[0061] After successful removal via SIC Continue decoding your own signal; specifically, when decoding When a signal is received, signals from users at later positions in the decoding sequence still exist as interference; correspondingly, the signal from the first user in the decoding sequence... Each user ( The signal-to-interference-plus-noise ratio (SINR) of the device is:

[0062]

[0063] in, express Received Signal power, It is the sum of the interference power from all secondary user signals with an index greater than i;

[0064] Constructive interference NOMA (CI-NOMA) precoding:

[0065] Without loss of generality, such as Figure 3 As shown, taking Quadrature Phase Shift Keying (QPSK) as an example, let O represent the origin of the complex plane. Point A represents the boundary point when the noise-free received signal meets the SINR requirement; point D is the orthogonal projection point from point B to line OA; furthermore, let point C represent the boundary of the CI region, and point B represent the actual received symbol point after interference superposition; we denote... Let the desired noise-free received signal vector have an amplitude of ( :No. The target signal-to-interference-plus-noise ratio (SINR) of each secondary user corresponds to the quadrature phase shift keying (QPSK) modulation under consideration; the ideal constructive region is defined as a fan-shaped region centered on the target constellation point and far from the decision boundary;

[0066] Based on geometric relationships, the received signal must satisfy specific phase and amplitude constraints to fall within the constructive region; we decompose it by rotating the received signal along the direction of the target symbol; specifically, the rotated real projection is defined as the effective signal component, while the imaginary projection represents the residual interference component, where... and Let represent the real and imaginary parts of the received signal after rotation, respectively; to ensure that the interference is constructive, i.e., the received signal is located within the correct decision region, the condition is... This should be satisfied; based on this derivation, the constructive interference (CI) constraint can be expressed as:

[0067]

[0068] in, This represents the half-angle width of the decision region related to the modulation scheme. For QPSK, the decision boundary is perpendicular to the coordinate axes. ,therefore ;

[0069] For the sake of simplicity, let and These represent primary users (PU) and secondary users (PU), respectively. The data symbols, where, and These represent the corresponding modulation phases; therefore, the equivalent received signal This can be deduced as:

[0070]

[0071] in, It's noise after rotation;

[0072] The equivalent received signal This refers to the mathematical definition of a constructive superimposed signal formed by the superposition of the primary user signal and its own signal at the secondary user receiver. Figure 3 The red arrows indicate that in the proposed CI-NOMA scheme, the primary user signal is strategically aligned to achieve constructive superposition with the secondary user's desired signal. At the secondary user end, the signal from the primary user is no longer considered harmful interference or noise; instead, it contributes to the useful received signal power. Therefore, the primary user (PU) and the secondary user (N) signal are mutually exclusive. Sub-users ( The signal-to-interference-plus-noise ratio (SIR) of ) is given by the following formulas:

[0073]

[0074]

[0075] Among them, in the molecule This indicates that the superposition of the primary user signal and the secondary user signal in the precoding domain is transformed into a constructive superposition signal at the receiving end after passing through the concatenated channel. After phase-aligned precoding, the PU and SU signals are constructively superimposed at the receiver, so the PU signal power can be considered as the useful signal power and included in the numerator.

[0076] Step 2.2: Construct an optimization problem (Optimization Problem 1) with the objective of minimizing the total transmit power of the base station, and introduce geometric constraints to ensure that the superimposed signals are constructive. By introducing the basic geometric constraints of the constructive interference (CI) region, the transmit power minimization problem can be formulated as follows, where the geometric constraints directly guarantee the equivalent received signal. (i.e., the constructive superimposed signal) falls within the constructive area:

[0077]

[0078] Constraints (st):

[0079]

[0080]

[0081]

[0082]

[0083]

[0084]

[0085]

[0086] Where, expression It is the sum of the Euclidean norms (i.e., moduli) of each precoding vector; although the transmit power is usually expressed as the square of the norm, the first power is used here to reflect the minimization of the total transmit amplitude and to maintain convexity, which is convenient for optimization. and They are the main user and the first The target signal-to-interference-plus-noise ratio for each secondary user; This is a constructive interference geometric constraint, which forces the equivalent received signal. It must fall within a constructive region on the constellation chart to ensure that the superimposed signal is constructive; The power allocation constraint between primary and secondary users reflects that the signal power of the primary user should not be less than the sum of the signal power of all secondary users;

[0087] Step 2.3: The Alternating Optimization (AO) algorithm is used to solve the optimization problem, and the generation of the constructive superposition signal is physically realized;

[0088] Because the active and passive beamforming variables are tightly coupled, the resulting optimization problem is non-convex and computationally difficult to solve directly. To address this challenge, we decompose the original problem into two subproblems and then use the Alternating Optimization (AO) algorithm for iterative solution. Specifically, we first fix the passive phase shift matrix and use the Continuous Convex Approximation (SCA) technique to optimize the active beamforming vector. Then, using the optimized active beamforming vector, we further optimize the passive phase shift using the Semidefinite Relaxation (SDR) method. Through iterative solution, we finally obtain the beamforming vector and RIS phase shift matrix that satisfy the constructive constraints, thereby realizing the generation of constructive superimposed signals at the physical layer.

[0089] Step 2.3.1: Active beamforming optimization;

[0090] For a fixed phase shift matrix The original optimization problem can be simplified to only concerning the preencoder. and The optimization problem is to transform each non-convex constraint into a convex constraint one by one.

[0091] It is about A convex function; obtained by passing through a given point Applying a first-order Taylor expansion to the master user precode at the current iteration point, we can obtain the following linear approximation:

[0092]

[0093] Similarly, The linear approximation is:

[0094]

[0095] in, This represents the k-th sub-user precode at the current iteration point. Indicates taking the real part;

[0096] For secondary users (SUs), the signal and interference terms involved in the effective cascaded channel should be processed first; by applying a first-order Taylor expansion to the constraints, we can further obtain a series of linear approximations:

[0097]

[0098]

[0099]

[0100] In addition, non-convex constraints The absolute value inequality in the equation essentially represents the intersection of two linear inequalities; therefore, it can be directly decomposed into the following two linear constraints:

[0101]

[0102]

[0103] To address the non-convex power constraints, we linearize the inequalities by performing a first-order Taylor expansion on the power terms. By performing Taylor expansions at local points, we can derive the power constraints between primary and secondary users, as well as the total power constraint. Regarding the power ranking constraint, we introduce a non-negative relaxation variable. The constraints are then reconstructed. Based on the above constraint processing, the transmit power minimization problem can be reformulated as a convex optimization problem, which can be solved using the Convex Optimization Toolbox (CVX). Simultaneously, the active precoding vector is also obtained. and These precoding vectors determine the amplitude and phase of the base station's transmitted signal and are the active component in the formation of the constructive superposition signal; specifically:

[0104] To solve the following three nonconvex power constraints , and The inequality is linearized by performing a first-order Taylor expansion on the power term; the linear approximation of the power term is:

[0105]

[0106]

[0107] Therefore, by local points and By performing a Taylor expansion at the point, the power constraints between the non-convex primary user and secondary user, as well as the total power constraint, can be derived:

[0108]

[0109]

[0110] Regarding the power ranking constraint, we introduce a nonnegative relaxation variable. The constraints are refactored as follows:

[0111]

[0112] Based on the above constraint processing, the transmit power minimization problem (optimization problem 2) can be restated as follows:

[0113]

[0114] Constraints (st):

[0115]

[0116]

[0117]

[0118]

[0119]

[0120]

[0121]

[0122] Clearly, the above is a convex optimization problem, which can be solved using the Convex Optimization Toolbox (CVX). Simultaneously, the active precoding vector can be obtained by solving for this problem. and ;

[0123] Step 2.3.2: Passive beamforming optimization;

[0124] In active precoding vectors and Once these variables are determined, we fix them and focus on optimizing the phase shift matrix of RIS. ;because The diagonal elements are subject to non-convex unit modulus constraints. We employ a positive semidefinite relaxation technique to obtain an efficient solution. To simplify the mathematical expression, we introduce the following intermediate auxiliary vector: , , , Define the RIS reflection coefficient vector as: ;in, Therefore The elements are a diagonal matrix with diagonal elements; Represents the reflection coefficient matrix (diagonal matrix) of RIS; Indicates taking The diagonal elements form a vector, and then the conjugate transpose is taken;

[0125] Therefore, in order to find a feasible solution The original optimization problem (Optimization Problem 3) can be restated as:

[0126] about The problem is a semidefinite programming problem; the signal-to-interference-plus-noise ratio (SIR) constraints for primary and secondary users can be reformulated as linear inequalities. Geometric constraints on the CI region are equivalent to linear constraints; furthermore, the semidefinite matrix is ​​applied... Non-convex rank-1 constraint on This can be equivalently transformed into the difference between the trace norm and the spectral norm. To ensure the numerical stability and convergence of the SCA algorithm, we introduce a quadratic regularization term. The reconstructed problem constitutes a standard convex semidefinite programming (SDP) problem, which can be efficiently solved using CVX. Given an initial point, we can iteratively solve the problem to obtain the required rank-1 solution until the decrease in the objective function is below a preset threshold. The obtained RIS phase shift matrix By adjusting the phase of the reflected signal, the primary user signal and the secondary user signal are ensured to be superimposed in phase at the receiving end, thereby completing the construction of a constructive superimposed signal.

[0127] The proposed Alternating Optimization (AO) algorithm is specifically an alternating optimization algorithm for RIS-assisted CI-NOMA systems. Its process is as follows: First, the active beamforming vector is randomly initialized. , RIS phase shift matrix sum matrix and set the outer iteration index. With convergence tolerance Then it enters the outer loop, with the RIS phase shift matrix fixed. Solve optimization problem 1 under the given conditions and update the active beamforming vector. , and slack variables Next, proceed to the inner iteration and set the inner index. And order In the inner loop, when the condition is met... or When calculating the sub-gradient And solve the optimization problem get ,renew The inner loop ends when the rank condition is met or the maximum number of inner iterations is reached, after which let And update Then update the outer iteration index. The entire process continues until the decrease in the objective function value of optimization problem 1 is less than... or the number of outer iterations The process terminates at a certain time; the entire process gradually reduces the total transmit power of the system by alternately optimizing active and passive beamforming.

[0128] For the Alternating Optimization (AO) algorithm, given the passive phase shift matrix... In the active beamforming problem, the target value is obtained through a first-order Taylor expansion in SCA. It is non-incremental during the iteration process; subsequently, at a fixed point... In the passive beamforming problem, we employ a penalty-based iterative method to relax the SDR matrix. We impose a rank-one constraint; specifically, we minimize the penalized, enhanced objective function and update... This iterative process continues until convergence. This process ensures that the objective function value is non-increasing, thus yielding... Therefore, the total transmit power decreases monotonically with the number of iterations. Since the transmit power is physically bounded to zero and is constrained by the minimum SINR requirement, the algorithm guarantees that iterative convergence to a stable point.

[0129] Constraints (st):

[0130]

[0131]

[0132]

[0133]

[0134]

[0135]

[0136] in, express It is a positive semi-definite matrix. This means that the reflection coefficient of each RIS unit must satisfy the unity modulus constraint; , , , , ;in, , , , , Both are positive semi-definite matrices with rank 1; therefore, we can further obtain:

[0137]

[0138]

[0139]

[0140]

[0141] in, , , , All are represented as the trace of a matrix;

[0142] Therefore, the signal-to-interference-plus-noise ratio (SIR) constraint for the primary user can be reformulated as the following linear inequality:

[0143]

[0144] Similarly, the CI-SINR constraint for secondary users is transformed into:

[0145]

[0146] Geometric constraints on the CI (constructive interference) region are equivalent to linear constraints; furthermore, the constraints applied to the positive semi-definite matrix... Non-convex rank-1 constraint on This can be equivalently transformed into the form of the difference between the trace norm and the spectral norm:

[0147]

[0148] For a positive semi-definite matrix, its rank is one if and only if its trace is equal to its largest eigenvalue; we define a penalty function. To ensure the numerical stability and convergence of the SCA algorithm, we introduce a quadratic regularization term. ,Will Decompose into the difference of two strongly convex functions:

[0149] in, Represented as F-norm, coefficient It is a regularization parameter;

[0150] because The term introduces nonconvexity, so we further linearize it; by applying a first-order Taylor expansion, the convex function... Feasible points obtained in the previous iteration The approximation is:

[0151]

[0152] in, express exist The subgradient at point is expressed as:

[0153]

[0154] Here, Representation matrix The normalized eigenvector corresponding to the largest eigenvalue; based on the above transformation, regarding The optimization problem (optimization problem) This can be restated as:

[0155]

[0156] Constraints (st):

[0157]

[0158]

[0159]

[0160]

[0161]

[0162] Clearly, the reconstructed problem constitutes a standard convex semidefinite programming (SDP) problem, which can be solved efficiently using CVX; given an initial point... We can obtain the desired rank-1 solution by iteratively solving the problem until the decrease in the objective function falls below a preset threshold. ;

[0163] Step 3: The secondary user (SU) demodulates its own data directly through the single-user detector based on the constructive superimposed signal constructed in Step 2, without performing serial interference cancellation (SIC);

[0164] By comparing the receiver architectures of traditional NOMA and the proposed CI-NOMA scheme, we found that: Constructive interference (CI)-based precoding, by achieving direct detection, greatly simplifies the secondary user (SU) processing flow and shifts most of the processing burden to the base station (BS) side. This is achieved through joint optimization of beamforming and reconfigurable smart surface (RIS) phase. Specifically, in step 2, the constructive superposition signal (i.e., the equivalent received signal that satisfies the constructive interference condition) constructed through mathematical definition, geometric constraints, and alternating optimization... When the signal reaches the secondary user receiver, the primary user signal component and the secondary user's own signal component are already in phase superimposed. Therefore, the secondary user does not need to perform a complex serial interference cancellation (SIC) process. It can directly demodulate its own data from this constructive superimposed signal using a simple single-user detector.

[0165] Specifically, for the first For each secondary user, in traditional NOMA, the primary user signal needs to be decoded first, the component subtracted from the received signal, and then the user's own information is detected. Sub-users ( The computational complexity of ) can be expressed as:

[0166]

[0167] in, and They represent and the kth The modulation order, This indicates that the complexity of the subtraction operation is constant and independent of the modulation order.

[0168] Therefore, the total complexity of the traditional NOMA scheme is:

[0169]

[0170] In the proposed CI-NOMA scheme, the primary user's signal components are designed to be constructive for secondary users. Therefore, each secondary user can recover data by directly detecting its own signal, requiring minimal additional processing. In the CI-NOMA scheme, the primary user's signal components are designed to be constructive for secondary users. The complexity of an individual user can be expressed as:

[0171]

[0172] in The computational cost of the CI decision operation is represented; the total complexity of the CI-NOMA scheme can be written as:

[0173]

[0174] This approach embodies a shift in computational complexity, where the base station undertakes the computationally intensive tasks of jointly optimizing beamforming and RIS phase, while constructing a constructive superimposed signal at the transmitter using interference from the primary user (PU), thereby significantly reducing the complexity at the secondary user end.

[0175] Example

[0176] The technical effects of the present invention will be verified through specific simulation experiments below;

[0177] Simulation conditions: Number of base station antennas RIS unit number Secondary user count noise power Main user target signal-to-interference-plus-noise ratio Secondary user target signal-to-interference-plus-noise ratio Path loss (base station - RIS) = 0.6, Path loss (RIS - primary user) = 0.6, Path loss (RIS - secondary user 1) = 0.5, Path loss (RIS - secondary user 2) = 0.3, Maximum transmit power Convergence tolerance .

[0178] Figure 4The convergence behavior of the traditional NOMA scheme and the proposed CI-NOMA scheme is demonstrated. It can be clearly observed that the proposed alternating optimization (AO) algorithm for CI-NOMA exhibits good convergence characteristics. Specifically, the total transmit power of the base station shows a monotonically decreasing trend with the number of iterations and quickly reaches a stable state within about 8 iterations. By comparing the two curves, it can be clearly seen that under the same system configuration and convergence criteria, the transmit power required by the proposed CI-NOMA scheme is significantly lower than that of the traditional NOMA baseline scheme. In addition, the traditional NOMA scheme converges faster because the CI constraint in the considered CI-NOMA scheme introduces an additional optimization burden.

[0179] Figure 5 The curves showing the total transmit power of the base station as a function of the number of antennas M were plotted. The results demonstrate that the proposed CI-NOMA scheme consistently exhibits superior energy efficiency, significantly outperforming other schemes. CI assistance reduces power consumption by 25%, while phase shift optimization contributes an additional 35% energy gain compared to random phase shift configuration. Specifically, compared to the traditional NOMA-RIS scheme using optimized phase shift and the CI-NOMA-RIS scheme using random phase shift, the proposed CI-assisted scheme achieves transmit power savings of approximately 3.5 dBm and 4.5 dBm, respectively. This is attributed to the fact that the traditional NOMA scheme requires additional power to suppress interference from the PU to the SUs, while the CI-NOMA scheme eliminates this need. This type of interference is transformed into constructive and useful signals, thereby significantly alleviating the power budget pressure on base stations. It can be seen that the jointly optimized CI-NOMA brings a huge performance gain compared to the random phase shift scheme. In addition, although OMA can effectively avoid the interference inherent in traditional NOMA, its lower spectral efficiency requires the use of the highest transmit power within a limited time-frequency resource block to meet the same SINR requirement, which further confirms the superiority of CI-NOMA. Finally, it can be concluded that the gain provided by CI to the NOMA-RIS scheme is insufficient to offset the power cost of using random phase shift compared to optimized phase shift.

[0180] Figure 6The study depicts the variation of the base station's average transmit power with the number of RIS reflector units N. The transmit power of all schemes shows a significant decreasing trend with increasing N. The power savings of the CI-NOMA scheme are particularly significant when N increases from 9 to 49. CI-assisted technology achieves a 75% power reduction, while phase-shift optimization provides an additional 95% energy gain compared to random phase alignment. In contrast, the traditional OMA scheme requires twice the transmit power to achieve comparable performance. Furthermore, with an increase in RIS units, interference signals from the PU can be projected onto the constructive area of ​​the SU with greater flexibility and accuracy, further widening the performance gap between the proposed scheme and traditional NOMA and OMA schemes.

[0181] Figure 7 The study demonstrates how the total transmit power of the base station varies with the primary user's target SINR threshold. As primary user communication requirements increase, all schemes inevitably need to increase transmit power to maintain link quality. The proposed CI-NOMA scheme significantly outperforms random phase-shift configuration and traditional NOMA benchmark schemes. Furthermore, as the primary user's target SINR increases, the power consumption gap between the proposed CI-NOMA scheme and traditional NOMA schemes gradually widens. Moreover, as primary user SINR requirements become more stringent, the power consumption of the proposed CI-NOMA scheme exhibits a stabilizing effect. The underlying mechanism is that the interference generated by the extra power allocated to the primary user is received constructively by the secondary user. This effectively converts a portion of the power used for the primary link into signal energy useful for the secondary link, forming a power redistribution that alleviates the overall pressure on the base station's power budget.

[0182] Figure 8 The study demonstrates how the total transmit power of the base station varies with the target SINR threshold of the secondary user. The transmit power of all transmission schemes increases monotonically with the target SINR of the SU. The power consumption of the CI-NOMA scheme is lower than that of the traditional NOMA scheme. This is attributed to the fact that the CI-NOMA scheme converts the interference signal from the PU into a constructive component in the numerator of the SINR expression.

[0183] Figure 9The study demonstrates the trend of total base station transmit power as the number of sub-users K increases. The proposed CI-NOMA-RIS scheme achieves approximately 8.0 dB of transmit power reduction compared to the OMA-RIS baseline scheme, corresponding to a significant energy saving of up to 84% in the linear domain. Furthermore, compared to the traditional NOMA-RIS scheme, the proposed method brings a 3.4 dB gain, equivalent to a 54% reduction in required transmit power. This result stems from the orthogonal resource allocation method of the OMA scheme, which imposes strict spectral efficiency constraints, thus requiring the base station to provide much higher transmit power to serve all users. Moreover, as K increases, the performance gap between the CI-NOMA scheme and the traditional NOMA scheme widens further. This is because, in the traditional NOMA architecture, the SIC decoding chain at the receiver becomes increasingly complex with the increase in the number of users. To ensure successful SIC execution, the base station must meet strict power domain sequencing and interference cancellation constraints, which severely compresses the optimization space for beamforming. In contrast, the CI-NOMA scheme provides greater design flexibility to promote constructive superposition of signals.

[0184] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for interference utilization in reconfigurable intelligent surface-assisted cognitive non-orthogonal multiple access networks, characterized in that, Includes the following steps: Construct a downlink cognitive radio nonorthogonal multiple access system assisted by a reconfigurable smart surface, the system comprising a base station, a reconfigurable smart surface, a primary user, and at least one secondary user; The base station uses superposition coding to send signals to all users, and constructs a constructive superposition signal for direct detection by secondary users by jointly optimizing the active beamforming vector of the base station and the passive reflection phase shift matrix of the reconfigurable smart surface. The secondary user, based on the constructive superimposed signal, can directly demodulate its own data through a single-user detector without performing serial interference cancellation (SIC).

2. The interference utilization method for reconfigurable intelligent surface-assisted cognitive non-orthogonal multiple access networks according to claim 1, characterized in that: The constructive superposition signal is formed by the in-phase superposition of the primary user signal component and the secondary user's own signal component at the secondary user's receiver, so that the primary user signal behaves as constructive interference to the secondary user.

3. The interference utilization method for reconfigurable intelligent surface-assisted cognitive non-orthogonal multiple access networks according to claim 1, characterized in that: The joint optimization includes: constructing an optimization problem with the goal of minimizing the total transmit power of the base station, and introducing geometric constraints to ensure that the constructive superimposed signal falls within the constructive region of the constellation diagram.

4. The interference utilization method for reconfigurable intelligent surface-assisted cognitive non-orthogonal multiple access networks according to claim 3, characterized in that: The optimization problem is solved using the alternating optimization (AO) algorithm, including: Step A: Fix the passive reflection phase shift matrix of the reconfigurable smart surface and optimize the active beamforming vector; Step B: Fix the optimized active beamforming vector and optimize the passive reflection phase shift matrix; Iteratively execute steps A and B until the preset convergence condition is met.

5. The interference utilization method for reconfigurable intelligent surface-assisted cognitive non-orthogonal multiple access networks according to claim 4, characterized in that: Step A includes: using the continuous convex approximation technique, the non-convex signal-to-interference-plus-noise ratio constraint, power allocation constraint, and power ranking constraint are transformed into a linear approximation form through a first-order Taylor expansion, thereby transforming the original problem into a convex optimization problem to be solved.

6. The interference utilization method for reconfigurable intelligent surface-assisted cognitive non-orthogonal multiple access networks according to claim 4, characterized in that: Step B includes: introducing a reflection coefficient vector. And construct a positive semidefinite matrix. This transforms the original optimization problem into a semidefinite programming problem involving a positive semidefinite matrix. Represents the reflection coefficient matrix of RIS; Indicates taking The diagonal elements form a vector, and then the conjugate transpose is taken; For vectors The conjugate transpose of; Using semidefinite relaxation techniques to relax non-convex rank-one constraints And solve it using convex optimization methods.

7. The interference utilization method for reconfigurable intelligent surface-assisted cognitive non-orthogonal multiple access networks according to claim 6, characterized in that: The rank-one constraint is relaxed by the difference between the trace norm and the spectral norm, expressed as: And a regularization term is introduced to ensure numerical stability and convergence.

8. The interference utilization method for reconfigurable intelligent surface-assisted cognitive non-orthogonal multiple access networks according to claim 3, characterized in that: The objective function of the optimization problem is the sum of the Euclidean norms of the active beamforming vectors of the base station. The constraints include at least: primary user signal-to-interference-plus-noise ratio (SINR) constraints, secondary user SINR constraints, constructive interference geometry constraints, unit mode constraints of the reconfigurable smart surface reflector, power allocation constraints between primary and secondary users, and power ranking constraints among secondary users.

9. The interference utilization method for reconfigurable intelligent surface-assisted cognitive non-orthogonal multiple access networks according to claim 8, characterized in that: The power ranking constraint is linearized by introducing non-negative relaxation variables and using a first-order Taylor expansion, specifically expressed as follows: in, Let the current iteration point be the... User precoding For the k-th sub-user precoding vector to be optimized, It is a non-negative slack variable. This indicates the operation of taking the real part.

10. The interference utilization method of the reconfigurable intelligent surface-assisted cognitive non-orthogonal multiple access network according to claim 1, characterized in that: The computational complexity of the secondary user directly detecting and demodulating its own data is linearly related to the modulation order of the secondary user, and there is no need to perform serial interference cancellation operations.

Images