Quantum-inspired joint optimization method for fluid antenna ISAC systems

Abstract

The application belongs to the technical field of wireless communication and signal processing, and discloses a quantum heuristic joint optimization method suitable for a flow state antenna ISAC system, which comprises the following steps: 1, a MIMO-FAS model is constructed, and an equivalent downlink channel is given to realize unified expression of a communication link and a sensing link; 2, a weighted sum rate maximization problem P1, a minimum user SINR maximization problem P2 and a sensing signal-to-jamming noise ratio SCNR maximization problem P3 are constructed; 3, the weighted sum rate maximization problem P1 is solved; 4, the minimum user SINR maximization problem P2 is solved; 5, the sensing signal-to-jamming noise ratio SCNR maximization problem P3 is solved; and 6, a quantum heuristic simulation bifurcation method is adopted to perform port selection optimization to obtain a joint optimal port-beam configuration. The application realizes collaborative optimization of port configuration and beam design, effectively improves the communication rate and sensing accuracy of the system while guaranteeing the calculation efficiency and scalability.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication and signal processing technology, specifically relating to a quantum-inspired joint optimization method suitable for fluid antenna ISAC systems. Background Technology

[0002] Integrated Sensing and Communication (ISAC) refers to a technology that shares the same frequency band and hardware for sensing and communication. With the evolution of fifth-generation mobile communication and future sixth-generation networks, communication systems are gradually shifting from a single information transmission mode towards ISAC. By simultaneously achieving data transmission and target detection on the same platform, ISAC can significantly improve spectrum utilization efficiency and overall system performance. However, in complex multi-user environments, channel time-varying characteristics and multi-source interference make coordinating communication and sensing requirements more difficult. Maintaining stable and reliable sensing capabilities while ensuring communication quality has become a crucial problem that urgently needs to be solved.

[0003] In recent years, fluidic antennas have demonstrated significant advantages in enhancing communication reliability and improving anti-interference performance due to their flexible port switching and spatial reconfigurability. By dynamically selecting different radiating ports within a limited size, fluidic antennas can acquire additional diversity gain in multipath environments and provide new spatial degrees of freedom for the synergy of communication and sensing.

[0004] However, in multi-user ISAC systems, port selection and beam control are highly coupled, making the overall optimization problem large in dimension and strongly non-convex. Traditional methods often struggle to balance computational efficiency and global performance in complex scenarios. Summary of the Invention

[0005] To address the shortcomings of existing fluid antenna ISAC systems, such as high coupling between port selection and beamforming, large optimization dimensionality, and difficulty in balancing communication and sensing performance, this application proposes a quantum-inspired joint optimization method suitable for fluid antenna ISAC systems. This method achieves coordinated optimization of port configuration and beam design, effectively improving the system's communication rate and sensing accuracy while ensuring computational efficiency and scalability.

[0006] To achieve the above objectives, this application employs the following technical solution:

[0007] This application presents a quantum-inspired joint optimization method applicable to a fluid antenna ISAC system, which includes a dual-function base station and... One communication user terminal, dual-function base station deployment Root transmitting flow antenna and Each receiving antenna has a transmitting current antenna with The first selectable port, the second Deploy a single user terminal with A selectable-port fluidic antenna, by selecting one port from multiple ports of each fluidic antenna deployed at the user end at a given time to participate in operation, specifically, the quantum heuristic joint optimization method includes the following steps:

[0008] Step 1: Construct a MIMO-FAS model, define the transceiver port structure and port selection vector from the dual-function base station to the user, and provide the equivalent downlink channel to achieve a unified expression of the communication link and the sensing link;

[0009] Step 2: Based on the MIMO-FAS model constructed in Step 1, construct three types of objective optimization models, and give corresponding power constraints, port selection constraints, and communication and sensing performance constraints for each optimization model. Specifically, the three types of objective optimization models are the weighted sum rate maximization problem P1, the minimum user SINR maximization problem P2, and the sensing signal-to-clutter-noise ratio (SCNR) maximization problem P3.

[0010] Step 3: Solve the weighted sum rate maximization problem P1 using the weighted least mean square error algorithm (WMMSE).

[0011] Step 4: Solve the minimum user SINR maximization problem P2 using the uplink / downlink duality principle;

[0012] Step 5: Solve the problem P3, which involves maximizing the sensing signal-to-noise ratio (SCNR), using the semi-definite relaxation (SDR) algorithm.

[0013] Step 6: Use a quantum-inspired simulated bifurcation method to optimize port selection and obtain the joint optimal port-beam configuration.

[0014] A further improvement in this application is that, in step 1, constructing the MIMO-FAS model specifically includes the following steps:

[0015] Step 1.1: Select Port. The base station port selection vector is defined as follows: The user port selection vector is After port selection, a diagonal matrix of dual-function base stations is constructed. The diagonal matrix of the user end is Based on a dual-function base station diagonal matrix and the user-side diagonal matrix Constructing an effective equivalent downlink channel matrix ,in For the full-port channel matrix, the effective equivalent downlink channel matrix is ​​obtained. Perform matrix decomposition to obtain the equivalent downlink channel ;

[0016] Step 1.2, Dual-function base station pair Several users simultaneously send downlink signals, with the transmitted signal vector being... Satisfying the transmit power constraint ,in, For beamforming vectors, At maximum transmission power, For the first The data signals of each user are independent of each other, and beamforming is defined as follows. , No. The received signal for each user is ,in For the first The equivalent channel vector corresponding to each user For additive white Gaussian noise, the noise power is defined as follows: , No. The downlink signal-to-interference-plus-noise ratio (SINR) for each user is:

[0017] ;

[0018] Step 1.3: Dual-function base stations utilize the same transmit signal vector. The opposite direction is Target perception and observation of target reflected signals on the dual-function base station receiver array. for:

[0019] ,

[0020] in, To record the complex reflection coefficient of the target, For clutter, To receive noise, Here is the clutter covariance matrix. The directional response matrix is ​​decomposed into... , For the transmitting array at the angle The guiding vector at that location, For the receiving array at the angle The guidance vector at the location is used to apply the receiving beamforming vector to the target reflected signal. Then form a sensory output Under the minimum variance distortionless response criterion, the optimal receiving beamforming is:

[0021] ,

[0022] in, It is a non-zero scalar. It is the conjugate transpose;

[0023] Perception Output The corresponding sensing clutter-to-noise ratio (SCNR) is:

[0024] ,

[0025] Among them, the perception weight matrix , For channel power gain, It is an identity matrix.

[0026] A further improvement in this application is that, in step 2, constructing the three types of objective optimization models specifically includes the following steps:

[0027] Step 2.1: Based on the diagonal matrix of dual-function base stations User-side diagonal matrix and beamforming vector Under the joint design, with the multi-user weighted sum rate (Sum-rate) as the objective, construct the weighted sum rate maximization problem P1:

[0028]

[0029]

[0030]

[0031]

[0032] in, For the first The weighting coefficient of each user For the first Communication rate per user;

[0033] Step 2.2: To enhance the communication reliability of edge users, a diagonal matrix of dual-function base stations is used. User-side diagonal matrix and beamforming vector Under joint design, maximize the worst-case user weighted SINR.

[0034] The minimum user SINR maximization problem P2 is constructed as follows:

[0035]

[0036]

[0037]

[0038]

[0039] in, For the first Weighting coefficients for each user;

[0040] Step 2.3: In a perception-priority scenario, to improve perception performance in the target direction, the problem P3 is constructed to maximize the perception signal-to-clutter-to-noise ratio (SCNR) in the target direction.

[0041]

[0042]

[0043]

[0044]

[0045]

[0046] in, This represents the minimum SINR threshold.

[0047] A further improvement in this application lies in: for the weighted sum-rate maximization problem P1, under the constraint of transmit power... The weighted least mean square error (WMMSE) algorithm is used to evaluate the beamforming vectors of each user. The iterative update process includes the following steps:

[0048] Step 3.1, Beamforming Initialization: Under transmit power constraints Given the given information, initialize the transmit power to an average value based on the number of users. , Given an equivalent downlink channel matrix Under the premise of obtaining the initial beamforming direction matrix using the RZF method. and the initial beamforming direction matrix Scaling is performed according to the transmit power constraint to obtain the initial beamforming matrix. ;

[0049] Step 3.2, in the... In the next iteration, at the current initial beamforming matrix Next, calculate the corresponding linear receiver filter coefficients for each user. and mean square error weight ,in For the user at the current iteration Beamforming vector;

[0050] Step 3.3, Linear Receiver Filter Coefficients Construct a diagonal matrix and error matrix The closed-form update of the initial beamforming matrix is ​​derived based on the weighted least mean square error (WMMSE) algorithm. ,in, The regularization factor is used to satisfy the transmit power constraint. The normalized and updated beamforming matrix: ;

[0051] Step 3.4: Calculate the mean square error (MSE);

[0052] Step 3.5: After each iteration, construct the convergence error based on the changes in the weights of the mean squared error (MSE) before and after the update. When convergence error When the number of iterations is less than the preset threshold or the maximum number of iterations is reached, the weighted minimum mean square error algorithm (WMMSE) is determined to have converged, and the current beamforming matrix is ​​output as the result of weighted sum rate optimization.

[0053] A further improvement in this application is that, in step 4, the uplink-downlink duality is used to solve the minimum user SINR maximization problem P2, specifically including the following steps:

[0054] Step 4.1: Introduce an uplink power allocation vector for each communication user. ;

[0055] Step 4.2: Construct the corresponding uplink interference plus noise covariance matrix for each communication user. , And based on the minimum mean square error (MMSE) criterion, the user's... Normalized uplink receive beamforming vector And calculate the uplink interference plus noise ratio (SINR):

[0056]

[0057] Step 4.3: Introduce a diagonal matrix and interference matrix Constructing the interference coupling matrix Block matrix corresponding to the generalized eigenvalue problem ,

[0058]

[0059]

[0060] in, , Given a column vector consisting entirely of 1s, solve for the partitioned matrix. The largest real eigenvalue and corresponding feature vectors To obtain the optimal uplink power allocation that satisfies the power constraints. ;

[0061] Step 4.4: Using the largest real eigenvalue The relative change of the error is used as the convergence criterion. If the relative change of the error is less than the threshold, the uplink optimization converges.

[0062] Step 4.5, Uplink / Downlink Dual Mapping and Minimum SINR Calculation: After uplink optimization convergence, based on the uplink / downlink duality principle, the obtained uplink receive beamforming vector is... and optimal uplink power allocation Mapped to downlink transmit beamforming vector and downlink power allocation : Based on this, the downlink interference plus noise ratio (SINR) of each communication user is calculated using the downlink SINR expression given in the communication system, and the optimization objective of problem P2 is taken. The target value is to improve the link quality for the weakest users.

[0063] A further improvement in this application is as follows: In step 5, for the problem of maximizing the perceived signal-to-clutter-noise ratio (SCNR) P3, under the premise of ensuring that the communication SINR of each user at a given port is not lower than a given threshold, a covariance matrix optimization method based on the semi-definite relaxation algorithm SDR is used for beam design to solve the problem of maximizing the perceived signal-to-clutter-noise ratio (SCNR) P3. The specific steps in step 5 include the following steps:

[0064] Step 5.1: The beamforming vector already given in the communication system Based on this, introduce users The corresponding emission covariance matrix , The original SCNR maximization problem, which used the beamforming vector as a variable, was rewritten using the trace of each transmit covariance matrix to characterize the corresponding transmit power. To optimize the form of the variables;

[0065] Step 5.2: Use the uplink-downlink dual method to determine whether the system parameter settings of problem P3, which aims to maximize the perceived signal-to-noise ratio (SCNR), meet the constraint of the minimum SINR for communication.

[0066] Step 5.3: Transform the SCNR optimization problem into a convex semidefinite programming problem, and then recover the beamforming vector from the optimal covariance: combine this with the sensing weight matrix given in the communication system. Construct an objective function to maximize SCNR in the covariance domain. Meanwhile, transmit power constraints Written as The minimum communication SINR constraint for each user is expressed in the covariance domain as a linear matrix inequality. This forms a standard semidefinite programming problem;

[0067] Step 5.4: Use a convex optimization solver to numerically solve the semidefinite programming problem and obtain the optimal set of covariance matrices. When the optimal matrix If the rank is 1, then the beam can be recovered through eigenvalue decomposition.

[0068] A further improvement in this application is that step 6 involves a port selection and beamforming optimization method, specifically including the following steps:

[0069] Step 6.1: In a multi-user fluid antenna ISAC system, for each fluid antenna, its port index is represented by binary encoding, requiring the following number of bits: By concatenating the port indices of all base station transmitters and user antennas, we obtain a dimension of... Global port selection Boolean vector Given a port, select a Boolean vector. The equivalent channel is constructed by using the corresponding port selection matrix, and the performance index of the three types of optimization objective models is calculated.

[0070] Step 6.2, under a given beamforming Subsequently, for the weighted sum rate maximization problem P1, the minimum user SINR maximization problem P2, and the perceived signal-to-clutter-noise ratio (SCNR) maximization problem P3, the negative value of the final objective value after solving the weighted sum rate maximization problem P1, the minimum user SINR maximization problem P2, and the perceived signal-to-clutter-noise ratio (SCNR) maximization problem P3 is represented by a unified high-order port selection objective function. This means that the original weighted sum rate maximization problem P1, the minimum user SINR maximization problem P2, and the perceived signal-to-clutter-noise ratio (SCNR) maximization problem P3 are equivalently transformed into minimization problems. ;

[0071] Step 6.3, at the reference point Nearby objective function for selecting higher-order ports Perform a local second-order Taylor approximation:

[0072]

[0073] in, For reference point The gradient vector at that point, For reference point Hessian estimation at the given location, introducing a Hamming distance regularization term, and constructing a joint objective function. :

[0074]

[0075] in, For regularization parameters, the joint objective function Rearranged into the standard quadratic unconstrained binary optimization (QUBO) form: ,in, This is the standard quadratic unconstrained binary optimization (QUBO) coefficient matrix;

[0076] Step 6.4: Obtain the standard quadratic unconstrained bivariate optimization (QUBO) coefficient matrix. Then, the quantum heuristic SB solver is invoked to solve the problem. Solve the matrix to obtain candidate solutions. ;

[0077] Step 6.5: After obtaining candidate solutions Later, random bit flipping was introduced. A better solution is retrieved from the neighborhood of the bit index set;

[0078] Step 6.6: Use the acceptance criterion based on Armijo rules to determine the candidate solutions for each iteration.

[0079] A further improvement to this application is that step 6.5 specifically includes the following steps:

[0080] Step 6.5.1, let the first... The candidate solution obtained in the second iteration is Randomly select a set of bit indices Construct a set of bit indices neighborhood ,in This indicates a bitwise XOR operation. In the first Standard basis vectors with one component of 1 and the rest of the components of 0;

[0081] Step 6.5.2: For the bit index set The higher-order port selection objective function is calculated for each of the candidate solutions. If there exists a solution that reduces the objective value, then the best one is selected as the updated candidate solution.

[0082] A further improvement in this application is that, in step 6.6, an acceptance criterion based on Armijo rules is used to determine the candidate solutions for each iteration, specifically as follows:

[0083] Step 6.6.1, let the first... The current solution in the next iteration is The candidate solution is Check the Armijo guidelines at each update step: If and only if When accepting candidate solutions, among which For Armijo parameters, The current step size, , direction vector The normalized gradient direction;

[0084] Step 6.6.2, let And increase the regularization parameter , Simultaneously update step size If the conditions are not met, then the candidate solution is rejected and the current solution is maintained. And reduce the regularization parameter , ;

[0085] Step 6.6.3: Through adaptive adjustment of the Armijo criterion and regularization parameters, the port selection vector sequence is optimized. Converging to a certain Boolean optimal solution Boolean optimal solution Together with the solutions to the weighted sum rate maximization problem P1, the minimum user SINR maximization problem P2, and the perceived signal-to-clutter-to-noise ratio (SCNR) maximization problem P3, they constitute a joint optimization solution, achieving joint optimization.

[0086] The beneficial effects of this application are:

[0087] By jointly optimizing the selection of the fluid antenna port and beamforming, this application can simultaneously achieve a multi-user communication rate (Sum-rate), minimum downlink user SINR, and sensing signal-to-clutter-noise ratio (SCNR) within the same framework. This results in a significant improvement in both communication quality and sensing capability, achieving a dual enhancement in communication and sensing performance.

[0088] This application employs the quantum-heuristic SB method to solve the port selection problem. Compared with traditional exhaustive search and heuristic random methods, it has a faster convergence speed and stronger global search capability in high-dimensional combinatorial spaces, and has higher solution efficiency and stability in high-dimensional discrete-continuous coupling problems.

[0089] As the number of antenna ports, users, or sensing constraints increases, the joint optimization framework of this invention can still maintain good convergence performance and overall complexity control, making MIMO-FAS for ISAC more feasible in large-scale deployment and significantly enhancing the scalability of the system in complex scenarios. Attached Figure Description

[0090] Figure 1 This is a schematic diagram of the MIMO-FAS scenario for ISAC in this application.

[0091] Figure 2 This is a flowchart of the WMMSE-based Sum-rate optimization constructed in this application under a given port selection.

[0092] Figure 3 This is a flowchart of SINR optimization based on uplink / downlink duality under a given port selection, as constructed in this application.

[0093] Figure 4 This is a flowchart of the SDR-based SCNR optimization constructed under a given port selection, as described in this application.

[0094] Figure 5 This is a flowchart illustrating the port selection optimization method constructed in this application using a quantum-inspired simulated bifurcation approach.

[0095] Figure 6 This is a schematic diagram of the convergence performance of three problems under different antenna and port combinations in the embodiments of this application, where (a) is the convergence performance of Sum-rate; (b) is the convergence performance of SINR; and (c) is the convergence performance of SCNR.

[0096] Figure 7 This is a schematic diagram illustrating the optimized performance of the Sum-rate problem in the embodiments of this application, where (a) represents the change of Sum-rate with the number of antenna ports; and (b) represents the change of Sum-rate with the number of users.

[0097] Figure 8 This is a schematic diagram illustrating the optimized performance of the SINR problem in the embodiments of this application, where (a) represents the change of minimum downlink SINR with the number of antenna ports; and (b) represents the change of minimum downlink SINR with the number of users.

[0098] Figure 9 This is a schematic diagram illustrating the optimized performance of the present application embodiment on the SCNR problem, where (a) shows the change of SCNR with the minimum communication SINR constraint; and (b) shows the change of SCNR with the number of users. Detailed Implementation

[0099] The embodiments of the present invention will be disclosed below with reference to the drawings. For clarity, many practical details will be described in the following description. However, it should be understood that these practical details are not intended to limit the invention. That is, in some embodiments of the invention, these practical details are not essential.

[0100] This application presents a quantum-inspired joint optimization method for flow-mode antenna ISAC systems, specifically for surface-mount ISAC systems, aiming to achieve a dual improvement in communication and sensing performance. Figure 1 As shown, the ISAC system includes a dual-function base station and Each communication user terminal and the base station simultaneously provide multi-user downlink communication and single-target sensing functions; dual-function base station deployment. Root transmitting flow antenna and Each receiving antenna has a transmitting current antenna with The first selectable port, the second Deploy a single user terminal with A selectable-port fluidic antenna can select one port from multiple ports of each fluidic antenna deployed at the user end to participate in operation at a given time, thereby achieving spatial reconstruction of the equivalent array in a limited space and improving communication and sensing performance.

[0101] Specifically, the heuristic joint optimization method of this application includes the following steps:

[0102] Step 1: Construct the MIMO-FAS model, define the transceiver port structure and port selection vectors (including both base station and user selection vectors) from the dual-function base station to the user, and provide the equivalent downlink channel under its influence, thus achieving a unified expression of the communication link and the sensing link. The specific steps for constructing the MIMO-FAS model are as follows:

[0103] Step 1.1: Select Port. The base station port selection vector is defined as follows: The user port selection vector is After port selection, a diagonal matrix of dual-function base stations is constructed. The diagonal matrix of the user end is Based on a dual-function base station diagonal matrix and the user-side diagonal matrix Constructing an effective equivalent downlink channel matrix ,in For the full-port channel matrix, the effective equivalent downlink channel matrix is ​​obtained. Perform matrix decomposition to obtain the equivalent downlink channel ;

[0104] Step 1.2, Dual-function base station pair Several users simultaneously send downlink signals, with the transmitted signal vector being... Satisfying the transmit power constraint ,in, For beamforming vectors, At maximum transmission power, For the first The data signals of each user are independent of each other, and beamforming is defined as follows. , No. The received signal for each user is ,in For the first The equivalent channel vector corresponding to each user For additive white Gaussian noise, the noise power is defined as follows: , No. The downlink signal-to-interference-plus-noise ratio (SINR) for each user is:

[0105] ;

[0106] Step 1.3: In the perception model, dual-function base stations utilize the same transmitted signal vector. The opposite direction is Target perception and observation of target reflected signals on the dual-function base station receiver array. for:

[0107] ,

[0108] in, To record the complex reflection coefficient of the target, For clutter, To receive noise, Here is the clutter covariance matrix. The directional response matrix is ​​decomposed into... , For the transmitting array at the angle The guiding vector at that location, For the receiving array at the angle The steering vector at the location is determined by the array geometry, and the receiving beamforming vector is used to target reflected signals. Then form a sensory output Under the minimum variance distortionless response criterion, the optimal receiving beamforming is:

[0109] ,

[0110] in, It is a non-zero scalar. For conjugate transpose, using ,

[0111] Perception Output The corresponding sensing clutter-to-noise ratio (SCNR) is:

[0112] ,

[0113] Among them, the perception weight matrix , For channel power gain, It is an identity matrix.

[0114] Step 2: Based on the MIMO-FAS model constructed in Step 1, three types of objective optimization models are constructed, and corresponding power constraints, port selection constraints, and communication and sensing performance constraints are given for each optimization model. Specifically, the three types of objective optimization models are the weighted sum rate maximization problem P1, the minimum user SINR maximization problem P2, and the sensing signal-to-clutter-noise ratio (SCNR) maximization problem P3. The construction of these three types of objective optimization models in this step includes the following steps:

[0115] Step 2.1: To improve the overall communication throughput of the system, while ensuring total power and port constraints, a diagonal matrix of dual-function base stations is used. User-side diagonal matrix and beamforming vector Under the joint design, with the multi-user weighted sum rate (Sum-rate) as the objective, construct the weighted sum rate maximization problem P1:

[0116]

[0117]

[0118]

[0119]

[0120]

[0121] in, For the first The weighting coefficient of each user For the first The communication rate of each user; the constraints are total power constraint, single-port activation of the base station transmitter and user terminal streaming antenna, and binary port selection constraint that the port selection needs to satisfy after being encoded.

[0122] The weighted sum-rate maximization problem P1 is oriented towards a communication scenario. Constraint 1 is to ensure that the base station's transmit power does not exceed the set maximum value. Constraints 2 and 3 are to ensure that the base station and the user only use one port respectively. Constraint 4 is to ensure that the port selection of the base station and the user satisfies the binary constraint, that is, to encode the port number as a binary 0 and 1 vector.

[0123] Step 2.2: To improve system fairness and enhance communication reliability for edge users, a diagonal matrix of dual-function base stations is used. User-side diagonal matrix and beamforming vector Under joint design, maximize the worst-case user weighted SINR.

[0124] The minimum user SINR maximization problem P2 is constructed as follows:

[0125]

[0126]

[0127]

[0128]

[0129]

[0130] in, For the first The constraints are: weight coefficients for each user; total transmit power limit, single-port activation of transmit and user-end flow antennas, and binary port selection constraint; minimum user SINR maximization problem P2 is oriented towards communication scenarios. The minimum user SINR maximization problem P2 is to maximize the minimum user SINR, which is to maximize the worst communication conditions of the system. The constraints are the same as those in the above problem.

[0131] Step 2.3: In the perception-first scenario, to further improve the perception performance in the target direction while ensuring the quality of communication services, the problem P3 is constructed to maximize the perception signal-to-clutter-to-noise ratio (SCNR) in the target direction, with the goal of maximizing the SCNR in the target direction.

[0132]

[0133]

[0134]

[0135]

[0136]

[0137]

[0138] in, The minimum SINR threshold is represented by the following constraints: communication SINR constraint, total transmit power constraint, single-port activation constraint of transmit and user-end flow antennas, and binary port selection constraint. The problem of maximizing the perceived signal-to-clutter-noise ratio (SCNR) is P3-oriented, which maximizes the perceived SCNR. The difference is that constraint 2 ensures that the worst communication quality of the system is not less than the threshold, that is, to maximize the sensing capability while ensuring the basic communication quality of the system.

[0139] Step 3: Given the port configuration, solve the weighted sum-rate maximization problem P1 using the Weighted Least Mean Square Error (WMMSE) algorithm. Given the port selection matrix, for the weighted sum-rate maximization problem P1, under the transmit power constraint... The weighted least mean square error (WMMSE) algorithm is used to evaluate the beamforming vectors of each user. The process of iterative updates is as follows: Figure 2 As shown, the specific steps include the following:

[0140] Step 3.1, Beamforming Initialization: Under transmit power constraints Given the given information, initialize the transmit power to an average value based on the number of users. , Given an equivalent downlink channel matrix Under the premise of obtaining the initial beamforming direction matrix using the RZF method. and the initial beamforming direction matrix Scaling is performed according to the transmit power constraint to obtain the initial beamforming matrix. ;

[0141] Step 3.2, in the... In the next iteration, at the current initial beamforming matrix Next, calculate the corresponding linear receiver filter coefficients for each user. and mean square error weight ,in For the user at the current iteration Beamforming vector;

[0142] Step 3.3, Linear Receiver Filter Coefficients Construct a diagonal matrix and error matrix The closed-form update of the initial beamforming matrix is ​​derived based on the weighted least mean square error (WMMSE) algorithm. ,in, The regularization factor is related to the current MSE weights and the receiver filter, and can be set to a value that satisfies the transmit power constraint. The normalized and updated beamforming matrix: ;

[0143] Step 3.4: Calculate the mean square error (MSE);

[0144] Step 3.5: After each iteration, construct the convergence error based on the changes in the weights of the mean squared error (MSE) before and after the update. When convergence error When the number of iterations is less than the preset threshold or the maximum number of iterations is reached, the weighted minimum mean square error algorithm (WMMSE) is determined to have converged. The current beamforming matrix is ​​output as the result of weighted sum rate optimization, thereby improving the spectral efficiency under given port selection conditions.

[0145] Step 4: Given the port configuration, solve the minimum user SINR maximization problem P2 using the uplink / downlink duality principle;

[0146] In step 4, given the port selection matrix, the minimum user SINR maximization problem P2 is solved using uplink-downlink duality. The optimal or suboptimal downlink beamforming vector is obtained by constructing an equivalent uplink power allocation problem, such as... Figure 3 As shown, the specific steps include the following:

[0147] Step 4.1, under a given equivalent downlink channel Based on this, an uplink power allocation vector is introduced for each communication user. ;

[0148] Step 4.2: Construct the corresponding uplink interference plus noise covariance matrix for each communication user. , And based on the minimum mean square error (MMSE) criterion, the user's... Normalized uplink receive beamforming vector And calculate the uplink interference plus noise ratio (SINR):

[0149]

[0150] Step 4.3: To characterize the interference relationships among multiple users, a diagonal matrix is ​​introduced. and interference matrix Constructing the interference coupling matrix Block matrix corresponding to the generalized eigenvalue problem ,

[0151]

[0152]

[0153] in, , Given a column vector consisting entirely of 1s, solve for the partitioned matrix. The largest real eigenvalue and corresponding feature vectors To obtain the optimal uplink power allocation that satisfies the power constraints. ;

[0154] Step 4.4: Using the largest real eigenvalue The relative change of the error is used as the convergence criterion. If the relative change of the error is less than the threshold, the uplink optimization converges.

[0155] Step 4.5, Uplink / Downlink Dual Mapping and Minimum SINR Calculation: After uplink optimization convergence, based on the uplink / downlink duality principle, the obtained uplink receive beamforming vector is... and optimal uplink power allocation Mapped to downlink transmit beamforming vector and downlink power allocation : Based on this, the downlink interference plus noise ratio (SINR) of each communication user is calculated using the downlink SINR expression given in the communication system, and the optimization objective of problem P2 is taken. The target value in this embodiment is to improve the link quality for the weakest user.

[0156] Step 5: Given the port configuration, solve the perceived signal-to-clutter-to-noise ratio (SCNR) maximization problem P3 using the semi-definite relaxation (SDR) algorithm. In Step 5, for the perceived SCNR maximization problem P3, under the premise of ensuring that the communication SINR of each user at a given port is not lower than a given threshold, use the covariance matrix optimization method based on the semi-definite relaxation algorithm (SDR) for beam design to solve the perceived SCNR maximization problem P3. The process is as follows: Figure 4 As shown, specific step 5 includes the following steps:

[0157] Step 5.1: The beamforming vector already given in the communication system Based on this, introduce users The corresponding emission covariance matrix , Furthermore, the trace of each transmit covariance matrix is ​​used to characterize the corresponding transmit power. Thus, the original SCNR maximization problem with the beamforming vector as the variable is rewritten as a problem with the set of covariance matrices. To optimize the form of the variables;

[0158] Step 5.2: Use the uplink-downlink dual method to determine whether the system parameter settings of problem P3, which aims to maximize the perceived signal-to-noise ratio (SCNR), meet the constraint of the minimum SINR for communication.

[0159] Step 5.3: Transform the SCNR optimization problem into a convex semidefinite programming problem, and then recover the beamforming vector from the optimal covariance: combine this with the sensing weight matrix given in the communication system. Construct an objective function to maximize SCNR in the covariance domain. Meanwhile, transmit power constraints Written as The minimum communication SINR constraint for each user is expressed in the covariance domain as a linear matrix inequality. The above constraints no longer explicitly include rank 1 constraints, thus forming a standard semidefinite programming problem, achieving the convexification of the original non-convex beamforming problem.

[0160] Step 5.4: Use a convex optimization solver to numerically solve the semidefinite programming problem and obtain the optimal set of covariance matrices. When the optimal matrix If the rank is 1, the beam can be recovered through eigenvalue decomposition. This method is suitable for scenarios involving coordinated optimization of communication and sensing.

[0161] In typical scenarios, the optimal solution satisfies At this time, it can be done by... Perform singular value decomposition or eigenvalue decomposition to extract the unique non-zero eigenvalues ​​and corresponding eigenvectors. Scale the eigenvectors according to power to recover the corresponding optimal or suboptimal beamforming vectors. This allows for the completion of SCNR priority beamforming design under a given port configuration, achieving a balanced optimization of communication and sensing performance.

[0162] Step 6: Use the quantum heuristic simulated bifurcation (SB) method to optimize port selection and obtain the joint optimal port-beam configuration.

[0163] like Figure 5 As shown, the specific steps for port selection and beamforming optimization include the following:

[0164] Step 6.1: In a multi-user fluid antenna ISAC system, for each fluid antenna, its port index is represented by binary encoding, requiring the following number of bits: By concatenating the port indices of all base station transmitters and user antennas, we obtain a dimension of... Global port selection Boolean vector Given a port, select a Boolean vector. The equivalent channel is constructed by using the corresponding port selection matrix, and the performance index of the three types of optimization objective models is calculated.

[0165] Step 6.2, under a given beamforming Subsequently, for the weighted sum rate maximization problem P1, the minimum user SINR maximization problem P2, and the perceived signal-to-clutter-noise ratio (SCNR) maximization problem P3, the negative value of the final objective value after solving the weighted sum rate maximization problem P1, the minimum user SINR maximization problem P2, and the perceived signal-to-clutter-noise ratio (SCNR) maximization problem P3 is represented by a unified high-order port selection objective function. This means that the original weighted sum rate maximization problem P1, the minimum user SINR maximization problem P2, and the perceived signal-to-clutter-to-noise ratio (SCNR) maximization problem P3 are equivalently transformed into minimization problems. In the joint optimization framework, the outer port selection problem to be solved in this step is uniformly represented as... ;

[0166] Step 6.3, at the reference point Nearby objective function for selecting higher-order ports Perform a local second-order Taylor approximation:

[0167]

[0168] in, For reference point The gradient vector at that point, For reference point Hessian estimation at the given location, introducing a Hamming distance regularization term, and constructing a joint objective function. :

[0169]

[0170] in, To the current iteration reference point The relevant regularization parameters will be used in conjunction with the objective function. Rearranged into the standard quadratic unconstrained binary optimization (QUBO) form: ,in, This is the standard quadratic unconstrained binary optimization (QUBO) coefficient matrix;

[0171] because For high-dimensional, non-convex higher-order functions defined on Boolean space, direct solutions are difficult to obtain. Therefore, this embodiment employs a local second-order approximation method based on Taylor expansion, revolving around a reference point in each iteration. Approximate the objective function. Let the reference point be... The gradient vector and Hessian matrix at point are respectively , ,but exist The nearby second-order Taylor approximation can be expressed as: To constrain the search range and enhance iterative stability, this embodiment superimposes a Hamming distance-based regularization term onto the local approximation function to construct a joint objective function. ,in, The regularization parameter is related to the current iteration reference point and is used to penalize and Solutions with excessively large differences are used to keep the search process within a controlled neighborhood in each iteration, achieving a gradual approximation from coarse to fine.

[0172] By organizing the above joint objective function and collecting the quadratic and linear terms, we can... Written in standard quadratic form: ,in It is a symmetric coefficient matrix. Ignoring... After irrelevant constant terms, it can be Explicit construction is ,in This represents an operator that generates a diagonal matrix from a vector. It is an all-one vector. The local subproblem can be represented by the standard QUBO model: .

[0173] To characterize the search scale, this embodiment defines the first... The step size parameter for the next iteration is And through adaptive updates To indirectly regulate Size: when When the step size is large, Smaller, the search focuses on local fine-tuning; when When the step size is small, increasing the step size helps to escape shallow local minima and enhances the global search capability.

[0174] Step 6.4: Obtain the standard quadratic unconstrained bivariate optimization (QUBO) coefficient matrix. Then, the quantum heuristic SB solver is invoked to solve the problem. Solve the matrix to obtain candidate solutions. After obtaining the coefficient matrix Subsequently, this embodiment employs the quantum-inspired SB algorithm as the core solution engine. The simulated bifurcation algorithm maps the discrete QUBO problem to the dynamic evolution of a set of coupled nonlinear oscillator systems by constructing an equivalent classical Hamiltonian. Let the continuous state variables be... , Let the position and momentum of each virtual oscillator be represented respectively. Then the corresponding standard SB Hamiltonian can be expressed as:

[0175] ,

[0176] in For detuning parameters, These are nonlinear coefficients. Pump intensity as a function of time, The coefficient is used to control the degree of influence of the objective function on the system dynamics.

[0177] The evolution equation for each oscillator can be obtained from the Hamiltonian:

[0178]

[0179] The above ordinary differential equations are numerically solved using the fourth-order Runge-Kutta explicit method, starting from near-zero initial conditions and gradually increasing the pump parameters. This causes the oscillator system to gradually bifurcate from the symmetrical phase, eventually resulting in each phase... Converging to a positive or negative stable value. The continuous solution is mapped back to the Boolean port selection solution using the following symbolic function: The candidate port selection vector can then be obtained. .

[0180] Step 6.5: After obtaining candidate solutions Later, random bit flipping was introduced. The process involves retrieving a better solution from the neighborhood of the bit index set; specifically, it includes the following steps:

[0181] Step 6.5.1, let the first... The candidate solution obtained in the second iteration is Randomly select a set of bit indices Construct a set of bit indices neighborhood ,in This indicates a bitwise XOR operation. In the first Standard basis vectors with one component of 1 and the rest of the components of 0;

[0182] Step 6.5.2: For the bit index set The higher-order port selection objective function is calculated for each of the candidate solutions. If a solution exists that reduces the objective value, the best one is selected as the updated candidate solution, thereby improving the convergence speed and solution diversity while keeping the computational complexity under control.

[0183] Step 6.6: To ensure the monotonic convergence of the outer iteration, this embodiment uses an acceptance criterion based on Armijo's rule to determine the candidate solutions for each iteration, specifically as follows:

[0184] Step 6.6.1, let the first... The current solution in the next iteration is The candidate solution is Check the Armijo guidelines at each update step: If and only if When accepting candidate solutions, among which For Armijo parameters, The current step size, , direction vector The normalized gradient direction;

[0185] Step 6.6.2, let And increase the regularization parameter , Simultaneously update step size If the conditions are not met, then the candidate solution is rejected and the current solution is maintained. And reduce the regularization parameter , ;

[0186] Step 6.6.3: Select the objective function at the higher-order port through adaptive adjustment of the Armijo criterion and regularization parameters. Under the premise of monotonically non-increasing, a larger global exploration step size is achieved in the early stage of iteration, and a finer local convergence is achieved in the later stage of iteration, thereby enabling the port selection vector sequence. Converging to a certain Boolean optimal solution Boolean optimal solution Together with the solutions to the weighted sum rate maximization problem P1, the minimum user SINR maximization problem P2, and the perceived signal-to-clutter-to-noise ratio (SCNR) maximization problem P3, they constitute a joint optimization solution, achieving joint optimization.

[0187] The quantum-inspired joint optimization method presented in this application exhibits good convergence and performance under multiple simulation conditions. For example... Figure 6 As shown, under different antenna and port combinations, all three types of optimization tasks achieve stable convergence within a relatively small number of iterations. Figure 6 (a) is the convergence curve of the Sum-rate. Figure 6 (b) is the convergence curve of the minimum user SINR. Figure 6 (c) is the convergence curve of SCNR.

[0188] like Figure 7 As shown, this invention achieves a higher overall system rate compared to the baseline method under different port and user count conditions in the Sum-rate maximization task. Figure 7 (a) demonstrates the performance improvement with changes in the number of ports. Figure 7 (b) shows the performance under changes in the number of users.

[0189] like Figure 8 As shown, in the task of maximizing the minimum user SINR, this invention can significantly improve the downlink SINR of the worst user. Figure 8 (a) shows the growth trend of minimum SINR as the number of ports changes. Figure 8 (b) shows how the minimum SINR varies with the number of users.

[0190] like Figure 9As shown, under the task of maximizing SCNR, this invention achieves higher target direction perception performance while satisfying communication SINR constraints. Figure 9 (a) shows the trend of SCNR as a function of communication SINR constraints. Figure 9 (b) shows the performance of SCNR as the number of users changes.

[0191] The above description is merely an embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A quantum-inspired joint optimization method suitable for a flow state antenna ISAC system, characterized in that: The ISAC system comprises a dual-function base station end and a communication user end, the dual-function base station end is provided with a root transmitting stream antenna and a root receiving antenna, each transmitting stream antenna has a selectable port, the first user end is provided with a stream antenna having a selectable port, by selecting one port from the multiple ports of each stream antenna deployed by the user end to participate in work at a given moment, specifically, the quantum heuristic joint optimization method comprises the following steps: Step 1, constructing a MIMO-FAS model, defining the structure of the transceiving ports of the dual-function base station to the user and the port selection vector, and giving an equivalent downlink channel to realize the unified expression of the communication link and the sensing link; Step 2, on the basis of the MIMO-FAS model constructed in step 1, three types of target optimization models are constructed, and the corresponding power constraints, port selection constraints and communication and sensing performance constraints are given for each optimization model, wherein the three types of target optimization models are specifically a weighted sum rate maximization problem P1, a minimum user SINR maximization problem P2 and a sensing signal-to-jamming noise ratio SCNR maximization problem P3; Step 3, a weighted minimum mean square error algorithm WMMSE is used to solve the weighted sum rate maximization problem P1; Step 4, an uplink-downlink duality principle is used to solve the minimum user SINR maximization problem P2; Step 5, a semidefinite relaxation algorithm is used to solve the sensing signal-to-jamming noise ratio SCNR maximization problem P3; Step 6, a quantum heuristic simulation bifurcation method is used for port selection optimization to obtain a joint optimal port-beam configuration, wherein the port selection and beamforming optimization method specifically includes the following steps: Step 6.1, in the multi-user streaming antenna ISAC system, for each streaming antenna, its port index is represented by binary coding, and the number of required bits is , the port index of all base station transmitting ends and user end antennas are concatenated to obtain a global port selection Boolean vector with dimension , given a certain port selection Boolean vector , the equivalent channel is constructed through the corresponding port selection matrix, and the performance indicators of the three types of optimization target models are calculated; Step 6.2, in the given beamforming The final target value of the weighted sum rate maximization problem P1, the minimum user SINR maximization problem P2 and the perceived signal-to-co- noise ratio SCNR maximization problem P3 after the solution of the weighted sum rate maximization problem P1, the solution of the minimum user SINR maximization problem P2 and the solution of the perceived signal-to-co- noise ratio SCNR maximization problem P3 is denoted as -P, and the negative value of the final target value is used as a unified high-order port selection objective function The original weighted sum rate maximization problem P1, the minimum user SINR maximization problem P2 and the perceived signal-to-co- noise ratio SCNR maximization problem P3 are equivalent transformed into a minimization problem ; Step 6.3, At the reference point Nearby to high order port selection objective function Performing a local second order Taylor approximation: where, is the gradient vector at the reference point , is the Hessian estimate at the reference point , a joint objective function is constructed by introducing a Hamming distance regularization term wherein, is a regularization parameter, the joint objective function is arranged into a standard quadratic unconstrained binary optimization form: wherein, is a standard quadratic unconstrained binary optimization coefficient matrix; Step 6.4, obtain the standard quadratic unconstrained binary optimization coefficient matrix After that, call the quantum heuristic SB solver to solve the matrix and obtain the candidate solution After that, call the quantum heuristic SB solver to solve the matrix and obtain the candidate solution ; Step 6.5, obtaining candidate solutions Post-incorporation random bit flips Retrieving better solutions in the neighborhood of the bit index set Step 6.6, an acceptance criterion based on the Armijo rule is used to judge the candidate solution of each iteration.

2. The quantum-inspired joint optimization method suitable for an ISAC system of a streaming antenna according to claim 1, characterized in that: In step 1, the MIMO-FAS model is constructed, which specifically includes the following steps: Step 1.1: Select Port. The base station port selection vector is defined as follows: The user port selection vector is After port selection, a diagonal matrix of dual-function base stations is constructed. The diagonal matrix of the user end is Based on a dual-function base station diagonal matrix and the user-side diagonal matrix Constructing an effective equivalent downlink channel matrix ,in For the full-port channel matrix, the effective equivalent downlink channel matrix is ​​obtained. Perform matrix decomposition to obtain the equivalent downlink channel ; Step 1.2, Dual-function base station pair Several users simultaneously send downlink signals, with the transmitted signal vector being... Satisfying the transmit power constraint ,in, For beamforming vectors, At maximum transmission power, For the first The data signals of each user are independent of each other, and beamforming is defined as follows. , No. The received signal for each user is ,in For the first The equivalent channel vector corresponding to each user , For additive white Gaussian noise, the noise power is defined as follows: , No. The downlink signal-to-interference-plus-noise ratio (SINR) for each user is: ； Step 1.

3. The dual-function base station utilizes the same transmit signal vector to sense targets in the direction of the target reflection signal observation on the dual-function base station receive array is: ， wherein, is the target complex reflection coefficient, is the clutter, is the receive noise, is the clutter covariance matrix, is the directional response matrix, the directional response matrix decomposed as , is the steering vector of the transmit array at angle , is the steering vector of the receive array at angle , the receive beamforming vector is applied to the target reflection signal to form the perception output under the minimum variance distortionless response criterion, the optimal receive beamforming is: ， wherein is a non-zero scalar, is the conjugate transpose; Sensing output The corresponding sensing clutter to noise ratio (SCNR) is: ， where the perception weight matrix , is the channel power gain, is the identity matrix.

3. The quantum-inspired joint optimization method suitable for an ISAC system of a streaming antenna according to claim 1, characterized in that: In step 2, the three types of target optimization models are constructed, which specifically include the following steps: Step 2.1, dual-function base station diagonal matrix , user diagonal matrix and beamforming vector Under joint design, taking the multi-user weighted sum rate as the target, the weighted sum rate maximization problem P1 is constructed: wherein, is the weight coefficient for the th user, is the communication rate for the th user; Step 2.2, To enhance the communication reliability of edge users, the dual-function base station pair diagonal matrix , User pair diagonal matrix and beamforming vector Joint design, maximize the worst user weighted SINR, The minimum user SINR maximization problem P2 is constructed as follows: wherein, is the weight coefficient constraint for the th user; Step 2.3, in the sensing priority scenario, in order to improve the sensing performance of the target direction, the SCNR maximization problem P3 of the target direction is constructed to maximize the SCNR of the target direction: wherein, denotes a minimum SINR threshold, B is beamforming, is a sensing weight matrix, is a downlink signal-to-interference-plus-noise ratio, SINR, of the th user.

4. The quantum-inspired joint optimization method suitable for an ISAC system of a streaming antenna according to claim 1, characterized in that: For the weighted sum-rate maximization problem P1, under the transmit power constraint The weighted minimum mean square error algorithm WMMSE is used to update the beamforming vectors of the users in each iteration, specifically including the following steps: Step 3.1, Beamforming initialization: under transmit power constraint The transmit power is initialized as , Under the premise of given equivalent downlink channel matrix , the initial beamforming direction matrix is obtained by using the RZF method, and the initial beamforming direction matrix is scaled under the transmit power constraint to obtain the initial beamforming matrix ; Step 3.2, in the... In the next iteration, at the current initial beamforming matrix Next, calculate the corresponding linear receiver filter coefficients for each user. Mean square error weights ,in For the user at the current iteration Beamforming vector; Step 3.3, linear receive filter coefficients Constructing a diagonal matrix and an error matrix and derive the closed-form update of the initial beamforming matrix according to the Weighted Minimum Mean Square Error algorithm (WMMSE) wherein, is a regularization factor, and is to satisfy the transmit power constraint , the normalized updated beamforming matrix is: ; Step 3.4, calculate the mean square error MSE; Step 3.5: After each iteration, construct the convergence error based on the changes in the weights of the mean squared error (MSE) before and after the update. When convergence error When the number of iterations is less than the preset threshold or the maximum number of iterations is reached, the weighted minimum mean square error algorithm (WMMSE) is determined to have converged, and the current beamforming matrix is ​​output as the result of weighted sum rate optimization.

5. The quantum-inspired joint optimization method suitable for an ISAC system of a streaming antenna according to claim 1, characterized in that: In step 4, for the minimum user SINR maximization problem P2, the uplink-downlink duality is used for solving, which specifically includes the following steps: Step 4.1, Introduce uplink power allocation vector for each communicating user , is the transpose; Step 4.2: Construct the corresponding uplink interference plus noise covariance matrix for each communication user. , And based on the minimum mean square error (MMSE) criterion, the user's... Normalized uplink receive beamforming vector And calculate the uplink interference plus noise ratio (SINR): Step 4.3, Introducing a diagonal matrix and the interference matrix , constructing the interference coupling matrix Block matrix corresponding to the generalized eigenvalue problem , wherein, is a full 1 column vector, the largest real eigenvalue of the block matrix is solved and the corresponding eigenvector , the optimal uplink power allocation satisfying the power constraint is obtained ; Step 4.4: Using the largest real eigenvalue The relative change of the error is used as the convergence criterion. If the relative change of the error is less than the threshold, the uplink optimization converges. Step 4.5, uplink-downlink dual mapping and minimum SINR calculation: after the convergence of the uplink optimization, according to the uplink-downlink dual principle, the obtained uplink receive beamforming vector and the uplink optimal power allocation are mapped to the downlink transmit beamforming vector and the downlink power allocation : , , On this basis, the downlink interference plus noise ratio SINR of each communication user is calculated by using the downlink SINR expression given in the communication system, and the optimization objective of problem P2 is taken as the target value, so as to improve the link quality of the weakest user.

6. The quantum-inspired joint optimization method suitable for an ISAC system of a streaming antenna according to claim 1, characterized in that: In step 5, for the sensing signal-to-jamming noise ratio SCNR maximization problem P3, on the premise that the communication SINR of each user at the given port is not lower than a given threshold, a covariance matrix optimization method based on the semidefinite relaxation algorithm SDR is used for beam design to solve the sensing signal-to-jamming noise ratio SCNR maximization problem P3, and the specific step 5 includes the following steps: Step 5.

1. Beamforming vectors given in a communication system On the basis of the user Corresponding transmit covariance matrices , and the trace of each transmit covariance matrix is used to represent the corresponding transmit power, the original SCNR maximization problem with beamforming vectors as variables is rewritten as a form with covariance matrix set as optimization variables; Step 5.2, the uplink-downlink duality method is used to judge whether the system parameter setting of the sensing signal-to-jamming noise ratio SCNR maximization problem P3 meets the communication minimum SINR constraint; Step 5.3, Convert the SCNR optimization problem into a convex semidefinite program, and recover the beamforming vectors from the optimal covariance: Combine the given sensing weight matrix in the communication system Construct the SCNR maximization objective in the covariance domain At the same time, the transmit power constraint is written as The minimum communication SINR constraints for each user are expressed in the covariance domain as linear matrix inequalities Form a standard semidefinite programming problem; Step 5.

4. Numerically solve the semi-definite programming problem using a convex optimization solver to obtain the optimal covariance matrix set When the rank of the optimal matrix is 1, the beam is recovered by eigen-decomposition.

7. The quantum-inspired joint optimization method suitable for an ISAC system of a streaming antenna according to claim 1, characterized in that: Step 6.5 specifically includes the following steps: Step 6.5.1, Set the candidate solution obtained by the second iteration is , randomly select a set of bit index sets , construct the neighborhood of the bit index set , where represents the bitwise XOR operation, is the standard basis vector with the first component being 1 and the remaining components being 0;​ Step 6.5.2, calculating the high-order port selection objective function for each candidate solution in the set of bit index sets If there are solutions that make the objective value decrease, the optimal one is selected as the updated candidate solution.​ 8. The quantum-inspired joint optimization method suitable for an ISAC system of a streaming antenna according to claim 1, characterized in that: In step 6.6, an acceptance criterion based on the Armijo rule is used to judge the candidate solution of each iteration, which is specifically: Step 6.6.1, Set the current solution to and the candidate solution to At each update, check the Armijo rule: Accept the candidate solution if and only if where is the Armijo parameter, is the current step size, and the direction vector is the normalized gradient direction. Step 6.6.2, let , and increase the regularization parameter , , while updating the step size , if the condition is not met, reject the candidate solution, keep , and decrease the regularization parameter , ; Step 6.6.3, the port selection vector sequence is converged to a certain Boolean optimal solution by the Armijo criterion and adaptive adjustment of the regularization parameter converges to a certain Boolean optimal solution , the Boolean optimal solution The joint optimization solution is composed of the solution of the weighted sum rate maximization problem P1, the solution of the minimum user SINR maximization problem P2 and the solution of the perceived signal-to-jamming-and-noise ratio SCNR maximization problem P3, and the joint optimization is realized.

Images