A method, system, equipment, and media for optimizing security resources in an IRS-NOMA system based on graph neural networks.

Abstract

A method, system, device, and medium for security resource optimization in an IRS-NOMA system based on graph neural networks are disclosed. The method includes: a base station first transmits the signal to be transmitted; the signal is received by legitimate users through direct and indirect links; the transmission rate of the legitimate user signal is calculated; then, the eavesdropping rates of external and internal eavesdroppers in the IRS-assisted NOMA system are calculated; a confidentiality rate formula is obtained by analyzing and deriving the communication process; the optimization objective function of the IRS-assisted NOMA system is then clarified; finally, performance optimization is performed on the objective function using a graph neural network to obtain the optimization result. The system, device, and medium are used to implement the method. This invention has the advantages of low communication cost, low design complexity, comprehensive design, simple and easy optimization implementation, high optimization efficiency, strong adaptability and flexibility, and high confidentiality rate.

Description

Technical Field

[0001] This invention relates to the fields of wireless communication security, intelligent reflectors, and non-orthogonal multiple access technologies, specifically to a method, system, device, and medium for security resource optimization in an IRS-NOMA system based on graph neural networks. Background Technology

[0002] With technological advancements, the demand for high-capacity, low-latency wireless communication has surged, driving continuous technological innovation. Non-Orthogonal Multiple Access (NOMA) technology, with its efficient spectrum utilization, widespread connectivity, and enhanced user fairness, has garnered significant attention in both academia and industry. However, while improving spectrum efficiency and user fairness, NOMA also faces security threats such as eavesdropping, challenging transmission confidentiality and integrity, and posing a severe test to the security of wireless communication systems. Physical layer security (PLS) technologies utilize wireless channel characteristics (such as attenuation, multipath, and interference) to enhance information confidentiality and integrity. Among these, Intelligent Reflecting Surface (IRS) technology is particularly crucial. It can dynamically adjust signal phase, enhancing the received signal for legitimate users through optimized configuration, while simultaneously weakening the interception capabilities of eavesdroppers, opening up new avenues for improving the transmission performance and security of NOMA systems.

[0003] Combining NOMA and IRS technologies can construct efficient and secure wireless communication systems that meet the requirements of high capacity and low latency while effectively resisting eavesdropping and ensuring information confidentiality and integrity. However, current research on resource optimization and performance analysis of IRS-assisted NOMA systems for secure communication is limited and faces several challenges. First, many studies have a one-sided view of security performance, considering only malicious external eavesdroppers while neglecting potentially defective internal eavesdroppers, thus failing to provide a comprehensive security analysis. Second, many studies rely on additional artificial noise to interfere with eavesdroppers and suppress their signal acquisition, increasing system cost and complexity. Furthermore, most studies employ traditional optimization methods to obtain resource allocation and performance results, which are not only inefficient but also lack adaptability and flexibility.

[0004] Chinese patent application CN115002802A discloses a method for maximizing the security rate of an IRS-assisted NOMA drone network. However, because this patent application only considers external eavesdroppers and uses traditional optimization methods, the system is not comprehensive and has low optimization efficiency.

[0005] Wei Wang et al. published "Beamforming and Jamming Optimization for IRS-Aided Secure NOMA Networks" (W.Wang et al., "Beamforming and Jamming Optimization for IRS-Aided Secure NOMA Networks," in IEEE Transactions on Wireless Communications, vol.21, no.3, pp.1557-1569, March 2022, doi:10.1109 / TWC.2021.3104856.). This paper proposes to enhance the secure communication of IRS-aided NOMA systems with artificial noise. The base station needs to combine the NOMA signal with artificial noise and send them together to the user. The user can separate the artificial noise to obtain the NOMA signal, while the eavesdropper cannot obtain the NOMA signal due to the interference of the artificial noise. However, this design requires a lot of communication resources and has the disadvantages of high complexity and high communication cost. Summary of the Invention

[0006] To overcome the shortcomings of the prior art, the present invention aims to provide a method, system, device, and medium for optimizing security resources in an IRS-NOMA system based on graph neural networks. By designing external and internal eavesdropper models and constructing graph neural networks, a method for optimizing resources and analyzing performance of an IRS-assisted NOMA system is realized. This method features low communication cost, low design complexity, comprehensive design, simple and easy optimization, high optimization efficiency, strong adaptability and flexibility, and high security rate.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] A method for optimizing security resources in an IRS-NOMA system based on graph neural networks includes the following steps:

[0009] Step 1: The base station sends the signal to be transmitted. The signal is received by the legitimate user through the direct link and the indirect link. The transmission rate of the legitimate user's signal is calculated. The direct link is when the base station broadcasts directly to the user. The indirect link is when the base station transmits to the IRS. The IRS decodes the signal and then broadcasts it to the user.

[0010] Step 2: Calculate the eavesdropping rate of the signal broadcast in Step 1 by external and internal eavesdroppers in the IRS-assisted NOMA system.

[0011] Step 3: By analyzing and deriving the communication process in Steps 1 and 2, the confidentiality rate formula is obtained, and the optimization objective function of the IRS-assisted NOMA system is then clarified.

[0012] Step 4: Optimize the performance of the objective function analyzed and derived in Step 3 using a graph neural network to obtain the optimization result.

[0013] Step 1 specifically includes:

[0014] All channel coefficients are assumed to exhibit small-scale fading with path loss, following the Rayleigh fading model. The coefficients of the channel from the base station to the IRS are... The coefficients of the channel from the base station to user k are: The coefficient of the channel from the base station to the external eavesdropper is The coefficient of the channel from the base station to the internal eavesdropper is The coefficients of the channel from IRS to user k are: The coefficient of the channel from the IRS to the external eavesdropper is The coefficient of the channel from the IRS to the insider is set up If the signal to be transmitted is given, then the signal that the k-th user among all legal users can receive is:

[0015]

[0016] Where K is the number of users, k is the number of users, T is the matrix transpose operation, and w k For the beamforming matrix, a k x is the NOMA power allocation factor. k For the signal to be transmitted, P t This refers to the base station's transmission power. Let θ be the phase shift matrix of the IRS, where θ n It is the nth reflection coefficient of the phase shift matrix, n k It is Gaussian white noise;

[0017] In an IRS-assisted NOMA system, the user employs Successive Interference Cancellation (SIC) technology for signal decoding, and the user's demodulation sequence is ||h1|| 2 ≥||h2|| 2 ≥…≥||h K || 2The k-th ranked user corresponds to the k-th strongest channel. The k-th strongest user decodes the signal by treating the signals of other weaker users as interference. Therefore, the signal-to-interference-plus-noise ratio (SINR) of the k-th user when decoding the signal is expressed as:

[0018]

[0019] Among them, a k For NOMA power allocation factor, w k For the beamforming matrix, σ k The noise at user k;

[0020] The transmission rate of the kth legitimate user is expressed as:

[0021] R k =log2(1+γ) k (1-3).

[0022] Step 2 specifically includes:

[0023] The signal received by the external eavesdropper can be deduced as follows:

[0024]

[0025] Where e represents the external eavesdropper, K represents the number of users, k represents the number of users, T represents the matrix transpose operation, and h represents the number of users. b,e h is the coefficient of the channel from the base station to the external eavesdropper. r,e Let G be the coefficient of the channel from the IRS to the external eavesdropper, and G be the coefficient of the channel from the base station to the IRS. Let θ be the phase shift matrix of the IRS, where θ n It is the nth reflection coefficient of the phase shift matrix, w k For the beamforming matrix, P t For the base station's transmission power, a k x is the NOMA power allocation factor. k For the signal to be transmitted, n e Gaussian white noise at the eavesdropper's location;

[0026] The SINR of an external eavesdropper intercepting the k-th user's signal is expressed as:

[0027]

[0028] Among them, a k For NOMA power allocation factor, w k For the beamforming matrix, σ k The noise at user k;

[0029] The eavesdropping rate of an external eavesdropper is expressed as:

[0030] R e→k =log2(1+γ) e→k (2-3)

[0031] The signal received by the inside eavesdropper is represented as:

[0032]

[0033] Where f is the insider eavesdropper, and n f For the eavesdropper, the Gaussian white noise is h. b,f h is the coefficient of the channel from the base station to the inside eavesdropper. r,k The coefficients of the channel from the IRS to user k;

[0034] The SINR of an insider eavesdropping on the k-th user is represented as:

[0035]

[0036] Among them, w f Beamforming at point f for the internal eavesdropper, σ f Noise at the location of the eavesdropper;

[0037] The eavesdropping rate of an insider can then be expressed as:

[0038] R f→k =log2(1+γ) f→k (2-6).

[0039] Step 3 specifically includes:

[0040] The formula for calculating the secrecy rate in an IRS-assisted NOMA system is expressed as follows:

[0041]

[0042] Among them, R k Let k be the transmission rate of the k-th user. Let e ​​be the eavesdropping rate at which the eavesdropper eavesdrops on user k, e be the external eavesdropper, and f be the internal eavesdropper.

[0043] The input signal Y is mapped to the optimized beamforming, power allocation factor, and phase shift matrix using the mapping function g. The objective function for this optimization is written as:

[0044]

[0045] Among them, a and a k For NOMA power allocation factors, w and w kLet Φ be the beamforming matrix, Φ be the IRS phase shift matrix, K be the number of users, k be the number of users, and h be the number of users. bk h is the coefficient of the channel from the base station to user k. rk P represents the coefficients of the channel from the IRS to user k. t This refers to the base station's transmission power. Let θ be an element in the IRS phase shift matrix. n It is the nth reflection coefficient of the phase shift matrix.

[0046] Step 4 specifically includes:

[0047] Step 4.1: Set up the network;

[0048] The Cooperative Graph Neural Network (CO-GNN) aims to learn the received signal Y through a multi-layer architecture containing an input layer, two message passing layers, and an output layer. The received signal as input contains information needed for joint optimization as initial input values. These inputs are propagated through two message passing layers containing aggregation and combination functions. Finally, at the output layer, the representation is directly converted into optimized beamforming w, power allocation factor a, and phase shift matrix Φ through a fully connected layer.

[0049] (1) Input layer: The input layer receives the feature vector Y of the signal from each user. k Where k = 1, ..., K, for a user node, the feature representation through the input layer is as follows:

[0050]

[0051] in, It is a multilayer perceptron (MLP) containing two fully connected hidden layers, where T is the matrix transpose operation;

[0052] IRS is represented by the features of the input layer as follows:

[0053]

[0054] Where K is the number of users, and k is the number of users;

[0055] (2) Message Passing Layer: The message passing layer extracts information from nodes using neighbor aggregation and combination strategies to generate a representation vector containing beamforming, power allocation factors, and phase shift matrices, thereby achieving joint optimization. The aggregation function of the user node is expressed as:

[0056]

[0057] Where i represents the i-th aggregation layer, N(k) represents the set of neighboring nodes near node k, and Θ is the mean pooling function. It is a fully connected hidden layer;

[0058] The combination function of user nodes is expressed as:

[0059]

[0060] The representation vector of an IRS node is:

[0061]

[0062] Combining aggregation and composition, the message passing layer can be represented as:

[0063]

[0064] Where △ is the max pooling function;

[0065] (3) Output layer: After completing I information transmission layers, the output layer generates a jointly optimized phase shift matrix. Beamforming and power allocation factor a∈R K ;

[0066] The phase shift matrix is ​​derived from the IRS representation vector:

[0067]

[0068] Beamforming and power allocation factors are derived from the representation vectors of user nodes:

[0069]

[0070] Step 4.2: Train and optimize the network built in Step 4.1;

[0071] Step 4.2.1: Set the loss function for network training optimization as the optimization objective function. The negative value;

[0072] Step 4.2.2: Randomly generate the coefficients for the channel from the base station to the IRS. The coefficients of the channel from the base station to user k are: The coefficient of the channel from the base station to the external eavesdropper is The coefficient of the channel from the base station to the internal eavesdropper is The coefficients of the channel from IRS to user k are: The coefficient of the channel from the IRS to the external eavesdropper is The coefficient of the channel from the IRS to the insider is And the network input, i.e., the received signal Y;

[0073] Step 4.2.3: Set the number of training optimization rounds for the network to 100 rounds, input the channel coefficients and the received signal Y into the network, and start training;

[0074] Step 4.2.4: Based on the network input and training optimization parameters set in Steps 4.2.1 to 4.2.3, perform 100 rounds of training optimization on the input layer, message passing layer and output layer of the network built in Step 4.1 to obtain the optimization results of beamforming, power allocation factor and phase shift matrix;

[0075] Step 4.2.5: Calculate the optimized objective function based on the beamforming, power allocation factor, and phase shift matrix obtained in Step 4.2.4. result.

[0076] A resource optimization system for an IRS-assisted NOMA system with a presence of eavesdroppers, based on graph neural networks, includes:

[0077] Security Model Module: The base station sends the signal to be transmitted. The signal is received by the legitimate user through direct and indirect links. The module calculates the transmission rate of the legitimate user's signal and the eavesdropping rate of the signal received by external or internal eavesdroppers in the IRS-assisted NOMA system. The direct link is the base station broadcasting directly to the user, and the indirect link is the base station transmitting to the IRS, which decodes the signal and then broadcasts it to the user.

[0078] Optimization Target Confirmation Module: By analyzing and deriving the confidentiality rate formula through the communication process, the optimization objective function of the IRS-assisted NOMA system is then clarified;

[0079] Network training module: Optimizes the performance of the objective function using a neural network to obtain the optimization result.

[0080] A security resource optimization device for an IRS-NOMA system based on graph neural networks, comprising:

[0081] Memory: Used to store the computer program that implements the aforementioned method for optimizing security resources in an IRS-NOMA system based on a graph neural network;

[0082] Processor: Used to implement the aforementioned IRS-NOMA system security resource optimization method based on graph neural networks when executing the computer program.

[0083] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned IRS-NOMA system security resource optimization method based on a graph neural network.

[0084] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0085] 1. This invention, by analyzing the signal reception behavior of legitimate users, external eavesdroppers, and internal eavesdroppers, realizes the design of an IRS-assisted NOMA system with eavesdroppers present. It not only provides a comprehensive and systematic perspective for analysis and design, but also has low communication costs and low design complexity.

[0086] 2. This invention achieves various resource optimizations for IRS-assisted NOMA systems by constructing a collaborative graph neural network (CO-GNN), which has the advantages of being simple to implement, having high optimization efficiency, and being highly adaptable and flexible.

[0087] 3. This invention maximizes the total security rate by establishing an optimization objective for an IRS-assisted NOMA system and using CO-GNN for network optimization training, and can provide a resource optimization performance analysis scheme with high security rate.

[0088] In summary, this invention has the advantages of low communication cost, low design complexity, comprehensive design, simple and easy optimization, high optimization efficiency, strong adaptability and flexibility of optimization, and high security rate. Attached Figure Description

[0089] Figure 1 This is a diagram of the IRS-assisted NOMA system security model based on graph neural networks.

[0090] Figure 2 This is a simulation result diagram of the relationship between security rate and transmission power in the external eavesdropping scenario of this embodiment.

[0091] Figure 3 This is a simulation result diagram of the relationship between security rate and transmission power in the internal eavesdropping scenario of this embodiment.

[0092] Figure 4 This is a simulation result diagram showing the relationship between the security rate and the number N of reflective elements equipped in the IRS under the external eavesdropping scenario in this embodiment.

[0093] Figure 5 This is a simulation result diagram showing the relationship between the security rate and the number N of reflective elements equipped in the IRS under the internal eavesdropping scenario in this embodiment.

[0094] Figure 6 This is a simulation result diagram showing the relationship between the security rate and the number of antennas M equipped in the base station under the external eavesdropping scenario in this embodiment.

[0095] Figure 7 This is a simulation result diagram showing the relationship between the security rate and the number of antennas M equipped in the base station under the internal eavesdropping scenario in this embodiment. Detailed Implementation

[0096] The present invention will now be described in detail with reference to the accompanying drawings.

[0097] Example

[0098] See Figure 1 This paper presents a security resource optimization method for an IRS-NOMA system based on graph neural networks, applied to communication in a wireless network scenario. The communication utilizes an IRS-assisted NOMA system, which includes a base station, an IRS, K users, and two types of eavesdroppers: a malicious external eavesdropper or a "rebellious" internal eavesdropper. The base station is equipped with M antennas, the IRS with N reflective elements, and users and eavesdroppers with a single antenna. The method includes the following steps:

[0099] Step 1: The base station sends the signal to be transmitted. The signal is received by the legitimate user through the direct link and the indirect link. The transmission rate of the legitimate user's signal is calculated. The direct link is when the base station broadcasts directly to the user. The indirect link is when the base station transmits to the IRS. The IRS decodes the signal and then broadcasts it to the user.

[0100] Step 2: Calculate the eavesdropping rate of the signal received by the external and internal eavesdroppers in the IRS-assisted NOMA system;

[0101] Step 3: By analyzing and deriving the communication process in Steps 1 and 2, the confidentiality rate formula is obtained, and the optimization objective function of the IRS-assisted NOMA system is then clarified.

[0102] Step 4: Optimize the performance of the objective function analyzed and derived in Step 3 using a graph neural network to obtain the optimization result.

[0103] Step 1 specifically includes:

[0104] All channel coefficients are assumed to exhibit small-scale fading with path loss, following the Rayleigh fading model. The coefficients of the channel from the base station to the IRS are... The coefficients of the channel from the base station to user k are: The coefficient of the channel from the base station to the external eavesdropper is The coefficient of the channel from the base station to the internal eavesdropper is The coefficients of the channel from IRS to user k are: The coefficient of the IRS to the external eavesdropper is The coefficient of the channel from the IRS to the insider is set up If the signal to be transmitted is given, then the signal that the k-th user can receive among all legal users is:

[0105]

[0106] Where K is the number of users, k is the number of users, T is the matrix transpose operation, and w k For the beamforming matrix, a k x is the NOMA power allocation factor. k For the signal to be transmitted, P t This refers to the base station's transmission power. Let θ be the phase shift matrix of the IRS, where θ n It is the nth reflection coefficient of the phase shift matrix, n k It is Gaussian white noise;

[0107] In an IRS-assisted NOMA system, users employ Successive Interference Cancellation (SIC) for signal decoding. The demodulation order in SIC is determined based on the equivalent channel gain of the combined channels. Users with the strongest channel gain are demodulated first, while users with weaker channel gains receive information directly. The user demodulation order is ||h1|| 2 ≥||h2|| 2 ≥…≥||h K || 2 The k-th ranked user corresponds to the k-th strongest channel. The k-th strongest user decodes the signal by treating the signals of other weaker users as interference. Therefore, the signal-to-interference-plus-noise ratio (SINR) of the k-th user when decoding the signal is expressed as:

[0108]

[0109] Among them, a k For NOMA power allocation factor, w k For the beamforming matrix, σ k The noise at user k;

[0110] The transmission rate of the kth legitimate user is expressed as:

[0111] R k =log2(1+γ) k (1-3).

[0112] Step 2 specifically includes:

[0113] Considering the adverse effects of complex electromagnetic environments on secure communication via exposed radio signals, external and internal eavesdropping scenarios were set up to analyze the security and confidentiality performance of an IRS-assisted NOMA system.

[0114] The external eavesdropper is unauthorized and not part of the NOMA pairing attacker. The broadcast nature of the wireless network allows the external eavesdropper to eavesdrop on the signals broadcast by the IRS base station. Therefore, the signal received by the external eavesdropper can be deduced as follows:

[0115]

[0116] Where e represents the external eavesdropper, K represents the number of users, k represents the number of users, T represents the matrix transpose operation, and h represents the number of users. b,e h is the coefficient of the channel from the base station to the external eavesdropper. r,e Let G be the coefficient of the channel from the IRS to the external eavesdropper, and G be the coefficient of the channel from the base station to the IRS. Let θ be the phase shift matrix of the IRS, where θ n It is the nth reflection coefficient of the phase shift matrix, w k For the beamforming matrix, P t For the base station's transmission power, a k x is the NOMA power allocation factor. k For the signal to be transmitted, n e Gaussian white noise at the eavesdropper's location;

[0117] An external eavesdropper decodes the information of the k-th user by applying serial interference cancellation. Therefore, the SINR of the external eavesdropper eavesdropping on the k-th user's signal is expressed as:

[0118]

[0119] Among them, a k For NOMA power allocation factor, w k For the beamforming matrix, σ k The noise at user k;

[0120] The eavesdropping rate of an external eavesdropper is expressed as:

[0121] R e→k =log2(1+γ) e→k (2-3)

[0122] An insider eavesdropper is a legitimate user capable of NOMA pairing, but due to relatively weaker channel conditions compared to other users, it can act as an insider eavesdropper to listen to other legitimate users. The signal received by the insider eavesdropper is represented as follows:

[0123]

[0124] Where f is the insider eavesdropper, and n f For the eavesdropper, the Gaussian white noise is h. b,f h is the coefficient of the channel from the base station to the inside eavesdropper. r,kThe coefficients of the channel from the IRS to user k;

[0125] The SINR of an insider eavesdropping on the k-th user is represented as:

[0126]

[0127] Among them, w f Beamforming at point f for the internal eavesdropper, σ f Noise at the location of the eavesdropper;

[0128] The eavesdropping rate of an insider can then be expressed as:

[0129] R f→k =log2(1+γ) f→k (2-6).

[0130] Step 3 specifically includes:

[0131] The objective of this invention is to maximize the overall security rate of an IRS-assisted NOMA system within an IRS-NOMA network security model by jointly optimizing the base station beamforming w, power allocation factor a, and IRS phase shift matrix Φ, and to obtain corresponding performance analysis. By directly mapping the received signal Y to the optimized beamforming, power allocation factor, and phase shift matrix to maximize utility, channel estimation can be bypassed, and optimized transmission strategies can be obtained more efficiently. The formula for calculating the security rate of an IRS-assisted NOMA system is expressed as:

[0132]

[0133] Among them, R k Let k be the transmission rate of the k-th user. Let e ​​be the eavesdropping rate at which the eavesdropper eavesdrops on user k, e be the external eavesdropper, and f be the internal eavesdropper.

[0134] The input signal Y is mapped to the optimized beamforming, power allocation factor, and phase shift matrix using the mapping function g. The objective function for this optimization is written as:

[0135]

[0136] Among them, a and a k For NOMA power allocation factors, w and w k Let Φ be the beamforming matrix, Φ be the IRS phase shift matrix, K be the number of users, k be the number of users, and h be the number of users. bk h is the coefficient of the channel from the base station to user k. rk P represents the coefficients of the channel from the IRS to user k. t This refers to the base station's transmission power. Let θ be an element in the IRS phase shift matrix.n It is the nth reflection coefficient of the phase shift matrix.

[0137] Because the objective function is non-convex, solving it computationally presents a significant challenge. To effectively address this issue and circumvent the need for channel estimation, the mapping function g(·) is visualized using a neural network, and the network parameters are trained using the received signal Y to achieve the optimal overall security rate.

[0138] Step 4 specifically includes:

[0139] Due to the non-convexity of the objective function, traditional optimization methods often present challenges. Furthermore, applying traditional optimization techniques to dynamically changing environments to formulate effective security joint optimization strategies is usually impractical. Therefore, this invention introduces a neural network—Cooperative Graph Neural Network (CO-GNN)—to simulate the mapping function g(·). Using the CO-GNN, the received signal Y can be directly mapped to the base station beamforming w, power allocation factor a, and intelligent reflector phase shift matrix Φ. Through continuous training and optimization, the overall system security rate can be maximized. By using the inverse of the objective function as the network's loss function, joint optimization can be achieved by suppressing malicious eavesdroppers, thereby ensuring that the signal is transmitted in a direction favorable to legitimate users, based on the indication of the objective function. Specifically, this step mainly includes network construction and network training processes:

[0140] Step 4.1: Set up the network;

[0141] The Cooperative Graph Neural Network (CO-GNN) aims to learn the received signal Y through a multi-layer architecture containing an input layer, two message passing layers, and an output layer. The received signal as input contains information needed for joint optimization as initial input values. These inputs are propagated through two message passing layers containing aggregation and combination functions. Finally, at the output layer, the representation is directly converted into optimized beamforming w, power allocation factor a, and phase shift matrix Φ through a fully connected layer.

[0142] (1) Input layer: The input layer receives the feature vector Y of the signal from each user. k Where k = 1, ..., K, the signal Y contains rich information related to beamforming, power allocation, and phase shift matrix, which is beneficial for feature learning and joint optimization of the CO-GNN network. Since the signal contains real and imaginary parts, for user nodes, the feature representation through the input layer is:

[0143]

[0144] in, It is a multilayer perceptron (MLP) containing two fully connected hidden layers, where T is the matrix transpose operation;

[0145] For the IRS, since all signals are reflected through the IRS, it is necessary to ensure that the received signals contain an equivalent amount of information related to the IRS. Therefore, the IRS is represented by the features of the input layer as follows:

[0146]

[0147] Where K is the number of users, and k is the number of users;

[0148] (2) Message Passing Layer: The message passing layer extracts information from nodes using neighbor aggregation and combination strategies to generate a representation vector containing beamforming, power allocation factors, and phase shift matrices for joint optimization. The aggregation function compiles the features of neighboring nodes into a message vector and passes it to the central node. The combination function updates the node's representation at the current time step, merging the node's current representation with the messages collected from the aggregation function; this is essentially an information passing process. The aggregation function for a user node is represented as follows:

[0149]

[0150] Where i represents the i-th aggregation layer, N(k) represents the set of neighboring nodes near node k, and Θ is the mean pooling function. It is a fully connected hidden layer;

[0151] The combination function of user nodes is expressed as:

[0152]

[0153] The representation vector of an IRS node is:

[0154]

[0155] Combining aggregation and composition, the message passing layer can be represented as:

[0156]

[0157] Where △ is the max pooling function;

[0158] (3) Output layer: After completing I information transmission layers, the output layer generates a jointly optimized phase shift matrix. Beamforming and power allocation factor a∈R K ;

[0159] The phase shift matrix is ​​derived from the IRS representation vector:

[0160]

[0161] Beamforming and power allocation factors are derived from the representation vectors of user nodes:

[0162]

[0163] Step 4.2: Train and optimize the network built in Step 4.1;

[0164] Step 4.2.1: Set the loss function for network training optimization as the optimization objective function. The negative value;

[0165] Step 4.2.2: Randomly generate the coefficients for the channel from the base station to the IRS. The coefficients of the channel from the base station to user k are: The coefficient of the channel from the base station to the external eavesdropper is The coefficient of the channel from the base station to the internal eavesdropper is The coefficients of the channel from IRS to user k are: The coefficient of the channel from the IRS to the external eavesdropper is The coefficient of the channel from the IRS to the insider is And the network input, i.e., the received signal Y;

[0166] Step 4.2.3: Set the number of training optimization rounds for the network to 100 rounds, input the channel coefficients and the received signal Y into the network, and start training;

[0167] Step 4.2.4: Based on the network input and training optimization parameters set in Steps 4.2.1 to 4.2.3, perform 100 rounds of training optimization on the input layer, message passing layer and output layer of the network built in Step 4.1 to obtain the optimization results of beamforming, power allocation factor and phase shift matrix;

[0168] Step 4.2.5: Calculate the optimized objective function based on the beamforming, power allocation factor, and phase shift matrix obtained in Step 4.2.4. result.

[0169] In network training, the loss function is defined as the negative of the total security rate. In each training iteration, the loss is iteratively fed back into the network to guide the optimization of beamforming, power allocation factors, and phase shift matrices, thereby maximizing the transmission rate for legitimate users while minimizing the eavesdropping rate. As the loss function value decreases, the system's total security rate performance also improves, ultimately achieving joint optimization of the parameters to obtain the optimal total security rate. Network training is conducted separately for external and internal eavesdropping scenarios.

[0170] A system for optimizing security resources in an IRS-NOMA system based on graph neural networks, comprising:

[0171] Security Model Module: The base station sends the signal to be transmitted. The signal is received by legitimate users through direct and indirect links. The module calculates the transmission rate of the legitimate user signal and the eavesdropping rate of external or internal eavesdroppers in the IRS-assisted NOMA system. The direct link is the base station broadcasting directly to the user, and the indirect link is the base station transmitting to the IRS, which decodes and then broadcasts to the user. This module is used to implement steps 1 to 2 of a security resource optimization method for an IRS-NOMA system based on graph neural networks.

[0172] Optimization Target Confirmation Module: By analyzing and deriving the confidentiality rate formula through the communication process, the optimization objective function of the IRS-assisted NOMA system is clarified, which is used to implement step 3 of an IRS-NOMA system security resource optimization method based on graph neural network;

[0173] Network training module: The network trainer performs performance optimization on the objective function using a neural network to obtain the optimization results, which are used to implement step 4 of an IRS-NOMA system security resource optimization method based on graph neural networks.

[0174] A security resource optimization device for an IRS-NOMA system based on graph neural networks, comprising:

[0175] Memory: Used to store the computer program that implements the aforementioned method for optimizing security resources in an IRS-NOMA system based on a graph neural network;

[0176] Processor: Used to implement the aforementioned IRS-NOMA system security resource optimization method based on graph neural networks when executing the computer program.

[0177] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the aforementioned method for optimizing security resources in an IRS-NOMA system based on a graph neural network.

[0178] Simulation Analysis

[0179] Figure 2This paper describes the relationship between the security rate and transmit power using the proposed method and other optimized scenarios in the presence of external eavesdroppers. Specific simulation conditions are as follows: the base station is equipped with M=5 antennas, the IRS has N=100 reflectors, and the number of legitimate users is K=2. As shown in the figure, the security rate increases continuously with increasing transmit power. This is because, during the joint optimization of beamforming, power allocation factor, and phase shift matrix, the increased power biases the transmission channel towards legitimate users, minimizing eavesdropping by unauthorized external eavesdroppers. Furthermore, the proposed method achieves the highest security rate among all schemes, demonstrating the rationality and effectiveness of this optimization scheme.

[0180] Figure 3 This paper describes the relationship between the security rate and transmit power using the method proposed in this invention and other optimized scenarios in the presence of an internal eavesdropper. Specific simulation conditions are as follows: the base station is equipped with M=5 antennas, the IRS has N=100 reflectors, and the number of legitimate users is K=2. Figure 2 In comparison, the security rate under internal eavesdropping is higher than that under external eavesdropping. This is due to the inherent characteristics of the internal eavesdropping node—it was once a legitimate node, and the neural network can more effectively improve its transmission performance. Therefore, when it exists as an eavesdropper, the neural network can also better suppress its transmission, thereby achieving higher system security performance.

[0181] Figure 4 This paper describes the relationship between the security rate and the number N of reflective elements equipped in the IRS under different optimization conditions using the method proposed in this invention, in the presence of external eavesdroppers. Specific simulation conditions are as follows: the number of antennas equipped in the base station is M=5, the number of legitimate users is K=2, and the transmit power P... t =30. Except for the "random IRS phase shift matrix" scheme, which maintains a relatively stable rate and is insensitive to changes in the number of IRS elements, all other schemes with IRS assistance significantly improve security performance with increasing N. By deploying IRSs with more reflection elements, user links can be effectively enhanced and external eavesdropping suppressed, thereby ensuring the security of legitimate users.

[0182] Figure 5 This paper describes the relationship between the security rate and the number of reflective elements N in the IRS under different optimization conditions, using the method proposed in this invention, in the presence of an internal eavesdropper. Specific simulation conditions are as follows: number of transmission antennas M = 5, number of legitimate users K = 2, and transmit power P... t =30. The proposed solution demonstrates the highest security rate. Furthermore, the trend observed in this figure is consistent with... Figure 3The curves depicted are very similar, indicating that in an internal eavesdropping scenario, the transmit power P t The number of reflective elements equipped in the IRS has a similar impact on the security rate.

[0183] Figure 6 and Figure 7 This paper describes the relationship between the security rate and the number of transmission antennas M under scenarios with external and internal eavesdroppers, using the method proposed in this invention and other optimized conditions. Specific simulation conditions are as follows: the number of legitimate users is K=2, the number of reflectors equipped in the IRS is N=100, and the transmit power P... t =30. As shown in the figure, the security rate monotonically increases with the number of transmit antennas. This is because a larger number of antennas strengthens the main link, providing a more focused approach to confuse potential eavesdroppers by optimizing beamforming, power allocation, and phase shift matrices. Even with a limited number of antennas M, the security rate remains high in all experimental scenarios. This robustness is attributed to the combined effect of the specified power level and a sufficient number of reflective elements N=100 in the IRS, which together ensure strong system security.

[0184] Obviously, those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device, or fabricating them separately as individual integrated circuit modules, or fabricating multiple modules or steps as a single integrated circuit module. Thus, the present invention is not limited to any particular hardware and software combination.

[0185] The embodiments described above are merely illustrative. Units described as separate components may or may not be physically separate; components shown as units may or may not be physical units, meaning they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any inventive effort.

[0186] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. These modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

[0187] This invention is not limited to the optional embodiments described above, and anyone can derive other various forms of products based on the teachings of this invention. The specific embodiments described above should not be construed as limiting the scope of protection of this invention; the scope of protection of this invention should be determined by the claims, and the specification can be used to interpret the claims.

Claims

1. A method for optimizing security resources in an IRS-NOMA system based on graph neural networks, characterized in that, Includes the following steps: Step 1: The base station sends the signal to be transmitted. The signal is received by the legitimate user through the direct link and the indirect link. The transmission rate of the legitimate user's signal is calculated. The direct link is when the base station broadcasts directly to the user. The indirect link is when the base station transmits to the IRS. The IRS decodes the signal and then broadcasts it to the user. Step 2: Calculate the eavesdropping rate of the signal broadcast in Step 1 by external and internal eavesdroppers in the IRS-assisted NOMA system. Step 3: By analyzing and deriving the communication process in Steps 1 and 2, the confidentiality rate formula is obtained, and the optimization objective function of the IRS-assisted NOMA system is then clarified. Step 4: Optimize the performance of the objective function analyzed and derived in Step 3 using a graph neural network to obtain the optimization result. The specific implementation steps are as follows: Step 4.1: Set up the network; The Cooperative Graph Neural Network (CO-GNN) aims to learn the received signal Y through a multi-layer architecture containing an input layer, two message passing layers, and an output layer. The received signal as input contains information needed for joint optimization as initial input values. These inputs are propagated through two message passing layers containing aggregation and combination functions. Finally, at the output layer, the representation is directly converted into optimized beamforming w, power allocation factor a, and phase shift matrix Φ through a fully connected layer. (1) Input layer: The input layer receives the feature vector Y of the signal from each user. k Where k = 1, ..., K, for a user node, the feature representation through the input layer is as follows: in, It is a multilayer perceptron (MLP) containing two fully connected hidden layers, where T is the matrix transpose operation; IRS is represented by the features of the input layer as follows: Where K is the number of users, and k is the number of users; (2) Message Passing Layer: The message passing layer extracts information from nodes using neighbor aggregation and combination strategies to generate a representation vector containing beamforming, power allocation factors, and phase shift matrices, thereby achieving joint optimization. The aggregation function of the user node is expressed as: Where i represents the i-th aggregation layer, N(k) represents the set of neighboring nodes near node k, and Θ is the mean pooling function. It is a fully connected hidden layer; The combination function of user nodes is expressed as: The representation vector of an IRS node is: Combining aggregation and composition, the message passing layer can be represented as: Where △ is the max pooling function; (3) Output layer: After completing I information transmission layers, the output layer generates a jointly optimized phase shift matrix. Beamforming and power allocation factor a∈R K ; The phase shift matrix is ​​derived from the IRS representation vector: Beamforming and power allocation factors are derived from the representation vectors of user nodes: Step 4.2: Train and optimize the network built in Step 4.1; Step 4.2.1: Set the loss function for network training optimization as the optimization objective function. The negative value; Step 4.2.2: Randomly generate the coefficients for the channel from the base station to the IRS. The coefficients of the channel from the base station to user k are: The coefficient of the channel from the base station to the external eavesdropper is The coefficient of the channel from the base station to the internal eavesdropper is The coefficients of the channel from IRS to user k are: The coefficient of the IRS to the external eavesdropper is The coefficient of the channel from the IRS to the insider is And the network input, i.e., the received signal Y; Step 4.2.3: Set the number of training optimization rounds for the network to 100 rounds, input the channel coefficients and the received signal Y into the network, and start training; Step 4.2.4: Based on the network input and training optimization parameters set in Steps 4.2.1 to 4.2.3, perform 100 rounds of training optimization on the input layer, message passing layer and output layer of the network built in Step 4.1 to obtain the optimization results of beamforming, power allocation factor and phase shift matrix; Step 4.2.5: Calculate the optimized objective function based on the beamforming, power allocation factor, and phase shift matrix obtained in Step 4.2.

4. result.

2. The method for optimizing security resources in an IRS-NOMA system based on graph neural networks according to claim 1, characterized in that, Step 1 specifically includes: All channel coefficients are assumed to exhibit small-scale fading with path loss, following the Rayleigh fading model. The coefficients of the channel from the base station to the IRS are... The coefficients of the channel from the base station to user k are: The coefficient of the channel from the base station to the external eavesdropper is The coefficient of the channel from the base station to the internal eavesdropper is The coefficients of the channel from IRS to user k are: The coefficient of the IRS to the external eavesdropper is The coefficient of the channel from the IRS to the insider is set up If the signal to be transmitted is given, then the signal that the k-th user can receive among all legal users is: Where K is the number of users, k is the number of users, T is the matrix transpose operation, and w k For the beamforming matrix, a k x is the NOMA power allocation factor. k For the signal to be transmitted, P t This refers to the base station's transmission power. Let θ be the phase shift matrix of the IRS, where θ n It is the nth reflection coefficient of the phase shift matrix, n k It is Gaussian white noise; In an IRS-assisted NOMA system, the user employs Successive Interference Cancellation (SIC) technology for signal decoding, and the user's demodulation sequence is ||h1|| 2 ≥||h2|| 2 ≥…≥||h K || 2 The k-th ranked user corresponds to the k-th strongest channel. The k-th strongest user decodes the signal by treating the signals of other weaker users as interference. Therefore, the signal-to-interference-plus-noise ratio (SINR) of the k-th user when decoding the signal is expressed as: Among them, a k For NOMA power allocation factor, w k For the beamforming matrix, σ k The noise at user k; The transmission rate of the kth legitimate user is expressed as: R k =log2(1+γ k (1-3).

3. The method for optimizing security resources in an IRS-NOMA system based on graph neural networks according to claim 1, characterized in that, Step 2 specifically includes: The signal received by the external eavesdropper can be deduced as follows: Where e represents the external eavesdropper, K represents the number of users, k represents the number of users, T represents the matrix transpose operation, and h represents the number of users. b,e h is the coefficient of the channel from the base station to the external eavesdropper. r,e Let G be the coefficient of the channel from the IRS to the external eavesdropper, and G be the coefficient of the channel from the base station to the IRS. Let θ be the phase shift matrix of the IRS, where θ n It is the nth reflection coefficient of the phase shift matrix, w k For the beamforming matrix, P t For the base station's transmission power, a k x is the NOMA power allocation factor. k For the signal to be transmitted, n e Gaussian white noise at the eavesdropper's location; The SINR of an external eavesdropper intercepting the k-th user's signal is expressed as: Among them, a k For NOMA power allocation factor, w k For the beamforming matrix, σ k The noise at user k; The eavesdropping rate of an external eavesdropper is expressed as: R e→k =log2(1+γ e→k ) (2-3) The signal received by the inside eavesdropper is represented as: Where f is the insider eavesdropper, and n f For the eavesdropper, the Gaussian white noise is h. b,f h is the coefficient of the channel from the base station to the inside eavesdropper. r,k The coefficients of the channel from the IRS to user k; The SINR of an insider eavesdropping on the k-th user is represented as: Among them, w f Beamforming at point f for the internal eavesdropper, σ f Noise at the location of the eavesdropper; The eavesdropping rate of an insider can then be expressed as: R f→k =log2(1+γ f→k ) (2-6).

4. The method for optimizing security resources in an IRS-NOMA system based on graph neural networks according to claim 1, characterized in that, Step 3 specifically includes: The formula for calculating the secrecy rate in an IRS-assisted NOMA system is expressed as follows: Among them, R k Let k be the transmission rate of the k-th user. Let e ​​be the eavesdropping rate at which the eavesdropper eavesdrops on user k, e be the external eavesdropper, and f be the internal eavesdropper. The input signal Y is mapped to the optimized beamforming, power allocation factor, and phase shift matrix using the mapping function g. The objective function for this optimization is written as: Among them, a and a k For NOMA power allocation factors, w and w k Let Φ be the beamforming matrix, Φ be the IRS phase shift matrix, K be the number of users, k be the number of users, and h be the number of users. bk h is the coefficient of the channel from the base station to user k. rk P represents the coefficients of the channel from the IRS to user k. t This refers to the base station's transmission power. Let θ be an element in the IRS phase shift matrix. n It is the nth reflection coefficient of the phase shift matrix.

5. A system based on the IRS-NOMA system security resource optimization method according to any one of claims 1 to 4, characterized in that, include: Security Model Module: The base station sends the signal to be transmitted. The signal is received by the legitimate user through direct and indirect links. The module calculates the transmission rate of the legitimate user's signal and the eavesdropping rate of the signal received by external or internal eavesdroppers in the IRS-assisted NOMA system. The direct link is the base station broadcasting directly to the user, and the indirect link is the base station transmitting to the IRS, which decodes the signal and then broadcasts it to the user. Optimization Target Confirmation Module: By analyzing and deriving the confidentiality rate formula through the communication process, the optimization objective function of the IRS-assisted NOMA system is then clarified; Network training module: Optimizes the performance of the objective function using a neural network to obtain the optimization result.

6. A security resource optimization device for an IRS-NOMA system based on graph neural networks, characterized in that, include: Memory: for storing a computer program that implements the IRS-NOMA system security resource optimization method based on graph neural networks as described in any one of claims 1 to 4; Processor: Used to implement the IRS-NOMA system security resource optimization method based on graph neural networks as described in any one of claims 1 to 4 when executing the computer program.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the IRS-NOMA system security resource optimization method based on a graph neural network as described in any one of claims 1 to 4.

Images