Joint optimization method for RIS-aided isac system against eavesdropping

Abstract

The RIS assisted ISAC system joint optimization method facing eavesdropping confrontation belongs to the field of communication, comprising: taking potential eavesdroppers as sensing targets and modeling the system; taking the RIS reflection phase matrix, the secret signal beam vector and the artificial noise beam vector as joint configuration variables, taking the communication secrecy rate and the target direction sensing accuracy as joint optimization targets, and constructing a multi-objective optimization problem; adopting a random grouping and density sensing non-dominated sorting genetic algorithm, combining a non-dominated sorting method, a random grouping mutation mechanism and a density sensing end front selection strategy to solve the multi-objective optimization problem; and optimizing the joint configuration variables. The method breaks through the limitation of the prior art that only serves legal users, realizes active confrontation to eavesdroppers, avoids the artificial fragmentation caused by variable splitting, can directly find the optimal solution in a multi-dimensional continuous space, and improves the solving efficiency and solution set distribution uniformity under high-dimensional strong coupling variables.

Description

Technical Field

[0001] This invention belongs to the field of communication technology, specifically relating to a joint optimization method for RIS-assisted ISAC systems for eavesdropping countermeasures. Background Technology

[0002] As mobile communications evolve towards B5G / 6G, the system places extreme demands on transmission rates, coverage depth, and inherent security, while simultaneously endowing base stations with radar-like environmental awareness capabilities. Integrated Communication and Sensing (ISAC) technology achieves dual gains in communication and sensing through shared spectrum, hardware, and waveform design. However, while ISAC signals detect targets, they are highly susceptible to interception by eavesdroppers nearby, or the detected target itself may be an eavesdropping node. This potential risk of leakage through sensing makes the physical layer security of ISAC systems a key constraint on their commercialization.

[0003] Reconfigurable smart surfaces (RIS), as artificial electromagnetic metamaterials capable of reconfiguring the wireless propagation environment, offer a new dimension for solving the aforementioned challenges. Through digitally controlled reflective elements, RIS can perform near-arbitrary phase modulation of spatial beams, thereby constructing programmable wireless channels at the physical layer. Introducing RIS into ISAC systems not only compensates for the perception blind spots caused by line-of-sight links, but more importantly, it provides additional spatial degrees of freedom. By precisely designing the RIS's reflection matrix, signals can be superimposed in phase at legitimate users to enhance strength, while being canceled out of phase at eavesdroppers to suppress interception, thus establishing a natural barrier against eavesdropping at the physical layer.

[0004] Existing RIS-assisted wireless communication technologies have been widely applied in scenarios such as channel enhancement, beamforming, and wireless coverage optimization. However, in RIS-assisted ISAC systems aimed at countering eavesdropping, related research typically attempts to improve the system's security and detection capabilities by jointly designing parameters such as the RIS reflection coefficient, transmit beam, and artificial noise. However, these methods still have significant limitations in RIS-assisted ISAC systems aimed at countering eavesdropping:

[0005] (1) Limited functional scope and lack of countermeasures: The current RIS-assisted ISAC system can only focus on secure communication and location services for legitimate users, and cannot counter eavesdroppers. Its optimization logic mainly revolves around enhancing the receiving gain of legitimate targets or reducing perception errors, but fails to fully consider how to actively suppress the interception capabilities of eavesdroppers in the joint resource configuration, resulting in the system lacking necessary defensive countermeasures when facing actual eavesdropping threats.

[0006] (2) Local Optimality Problem of Alternating Optimization (AO) Algorithm: To handle complex non-convex constraints, existing technologies often employ alternating optimization methods, splitting and iterating the RIS reflection phase, the secure signal beam, and the artificial noise (AN) beam. However, in the ISAC scenario for eavesdropping countermeasures, the duality of the ISAC signal (it must be used for both information transmission and target detection) leads to a strong nonlinear coupling relationship between the secure signal beam, the artificial noise beam, and the RIS reflection phase. Furthermore, the system must simultaneously consider the two highly competitive objectives of secrecy and perception. Such methods are highly dependent on the selection of initial values, easily fall into local optima, and cannot provide a Pareto optimal solution set that reflects the trade-off between countermeasure performance and target direction perception accuracy.

[0007] (3) Search failure and solution set degradation in high-dimensional decision space: As the size of RIS units increases, the variable dimension of the joint configuration problem of the system explodes. Existing multi-objective optimization methods show a significant decrease in search efficiency in high-dimensional decision space when dealing with such strongly coupled and high-dimensional ISAC security configuration problems, often resulting in slow convergence. At the same time, when generating the non-dominated solution set, due to the lack of an effective distribution control mechanism, it is easy to have uneven solution set distribution, poor diversity, getting trapped in local optima, or failing to cover the complete Pareto front. This makes it difficult for the system to flexibly select the optimal configuration scheme according to the real-time adversarial requirements, and it is difficult to support dynamic adversarial scenarios with high real-time requirements.

[0008] The evolution of B5G / 6G wireless communication systems, and the combination of Integrated Communication-Sensing (ISAC) technology and Reconfigurable Smart Surfaces (RIS), has become a key path to improve spectrum efficiency and environmental awareness. However, in complex electromagnetic environments, the openness of wireless channels exposes systems to severe eavesdropping threats. Existing security solutions for RIS-assisted ISAC systems mostly focus on ensuring the secure communication quality and positioning service accuracy of legitimate users, essentially remaining passive defenses. These solutions often neglect to actively suppress or counter potential eavesdroppers using the detection characteristics of ISAC signals, failing to achieve precise attacks or deep countermeasures against eavesdropping through fine-grained control of spatial resources while ensuring sensing performance.

[0009] Therefore, researching a joint optimization scheme that can balance high-performance countermeasures, high-precision sensing, and efficient convergence is of great theoretical and engineering significance for building the next generation of secure and intelligent ISAC wireless communication networks. Summary of the Invention

[0010] To address the limitations of existing technologies, such as limited functional scope, lack of adversarial capabilities, local optima issues in Alternating Optimization (AO) algorithms, and search failures and solution set degradation in high-dimensional decision spaces, this invention provides a joint optimization method for RIS-assisted ISAC systems aimed at countering eavesdropping.

[0011] The technical solution adopted by this invention to solve the technical problem is as follows:

[0012] This invention provides a joint optimization method for RIS-assisted ISAC systems aimed at countering eavesdropping, comprising the following steps:

[0013] In the RIS-assisted secure communication and sensing system, potential eavesdroppers are considered as sensing targets and the system is modeled accordingly. A multi-objective optimization problem is constructed using the RIS reflection phase matrix, the secure signal beam vector, and the artificial noise beam vector as joint configuration variables, and communication security rate and target direction sensing accuracy as joint optimization objectives. A random grouping and density-sensing non-dominated sorting genetic algorithm is employed, combined with non-dominated sorting methods, random grouping mutation mechanisms, and density-sensing terminal front selection strategies, to solve the multi-objective optimization problem. The joint configuration variables are then optimized to obtain system parameter configuration results that meet the requirements of communication security rate and target direction sensing accuracy.

[0014] Furthermore, the RIS-assisted secure communication and sensing system includes a base station, a communication user, a potential eavesdropper, and the RIS; both the communication user and the potential eavesdropper are single-antenna devices; mathematical expressions are constructed for the RIS reflection phase matrix, the base station transmitted signal, the communication model, the communication user received signal, the communication user received signal-to-interference-plus-noise ratio, the communication rate of the communication user, the eavesdropping rate of the potential eavesdropper, the communication confidentiality rate, the sensing model, and the target direction sensing accuracy.

[0015] Furthermore, the mathematical expression for the RIS reflection phase matrix is: ; , Let j be the phase offset of the l-th RIS reflector unit, where j is the imaginary unit, satisfying... .

[0016] Furthermore, the mathematical expression for the communication security rate is: , For communication user communication rate, The rate at which the eavesdropper listens.

[0017] Furthermore, the target direction perception accuracy is measured using the Cramer-Rao bound, and its mathematical expression is as follows:

[0018] ;

[0019] in, , For the signal wavelength, To sense channel noise power, For the target angle of arrival, For echo signal power, To superimpose the number of transmitted signals onto the snapshot, The number of receiving antennas configured for the base station.

[0020] Furthermore, the multi-objective optimization problem is expressed as follows:

[0021]

[0022] in, For secure communication speed, To improve the accuracy of target direction perception. For the secure signal beam vector, For artificial noise beam vectors, This represents the phase shift of the l-th RIS reflector unit. This indicates the calculation of the L2 norm. This is the maximum allowed CRB threshold.

[0023] Furthermore, the specific implementation process of the random grouping and density-aware non-dominated sorting genetic algorithm is as follows:

[0024] (1) Randomly generate a set of candidate solutions, each of which contains the RIS reflection phase matrix, the secret signal beam vector and the artificial noise beam vector, and construct an initial population; evaluate the system performance of each individual in the initial population to obtain the objective function vector corresponding to each individual;

[0025] (2) Sort all candidate solutions by non-dominated sorting method, determine the Pareto level of each individual in the population, and obtain the non-dominated solution set in the current population;

[0026] (3) Population update based on random grouping and mutation mechanism: In each generation of evolution, the current population is updated and offspring individuals are generated through crossover, mutation and random grouping operations;

[0027] (4) Implement the density-sensing front selection strategy to select uniformly distributed solution sets as the next generation population;

[0028] (5) Through multiple generations of iterative search, the Pareto optimal solution set of the multi-objective optimization problem is gradually approximated;

[0029] (6) Select the RIS reflection phase configuration and transmit beamforming parameters that meet the system requirements from the obtained Pareto optimal solution set, and apply them to the RIS-assisted security communication and sensing system.

[0030] Furthermore, in step (3), at the beginning of each generation of evolution, the index set consisting of all decision variables is represented as follows: D represents the total number of decision variables; the index set is divided into G subgroups using a random grouping mutation mechanism. The average size of each subgroup is When generating offspring individuals, for a pair of parent solutions... and Choose only one group of variables Perform crossover and mutation operations while keeping the other variables unchanged.

[0031] Furthermore, in step (4), during the environmental selection phase, the parent population and the offspring population are merged, and the non-dominated sorting method is executed to obtain the non-dominated solution set; all complete Pareto fronts are directly entered into the next generation population, while for the last Pareto front that cannot be completely retained, the reference point mechanism is used for screening.

[0032] Furthermore, a set of uniformly distributed reference points are generated in the target space. The objective function vector is translated and normalized according to the ideal points, and the vertical distance between the solution and the reference points is calculated. A density-aware mechanism is introduced in the selection process of the last leading edge. The nearest neighbor distance of any solution in the last leading edge is defined. When the same reference point corresponds to multiple candidate solutions, the solution with the largest nearest neighbor distance is selected first. If multiple candidate solutions have the same nearest neighbor distance, the solution with the smaller vertical distance to the reference point is selected.

[0033] The beneficial effects of this invention are:

[0034] (1) This invention breaks through the limitation of existing technologies that only serve legitimate users and achieves proactive countermeasures against eavesdroppers;

[0035] This invention innovatively treats potential eavesdroppers directly as the system's sensing target, actively acquiring spatial information about the eavesdropping end by utilizing the detection characteristics of the ISAC signal. By jointly controlling the RIS reflection phase, the secure signal beam, and the artificial noise (AN) beam, it can actively suppress the eavesdropper's interception capability while ensuring target direction perception accuracy (CRB), filling the gap in existing technologies' eavesdropping countermeasure capabilities and truly achieving effective countermeasures against eavesdroppers at the physical layer.

[0036] (2) This invention overcomes the defect that the traditional Alternating Optimization (AO) algorithm is prone to getting trapped in local optima;

[0037] This invention constructs a multi-objective optimization problem from a global perspective, avoiding the artificial fragmentation caused by variable splitting. It can directly find the optimal solution in a multi-dimensional continuous space and provide a Pareto optimal solution set that accurately reflects the trade-off between adversarial performance and target direction perception accuracy.

[0038] (3) This invention significantly improves the solution efficiency and solution set distribution uniformity under high-dimensional strongly coupled variables;

[0039] This invention proposes a random grouping and density-aware non-dominated sorting genetic algorithm (RGDNSGA-III), which has the following advantages:

[0040] 2) In terms of search efficiency: The RGDNSGA-III algorithm introduces a random grouping mutation mechanism, which divides high-dimensional variables into multiple low-dimensional subgroups and performs local perturbation during the evolution process, which greatly reduces the scale of variables in a single evolution, effectively alleviates the complexity of high-dimensional optimization, and improves the convergence speed and algorithm stability.

[0041] 3) Regarding the quality of the solution set: The RGDNSGA-III algorithm introduces a density-aware front selection strategy, which prioritizes retaining sparser candidate solutions in the target space during environment selection. This strategy effectively addresses the pain points of uneven solution set distribution and poor diversity in density-aware front selection strategies, ensuring that the final generated Pareto front is widely and uniformly distributed. This provides high-quality decision support for the system to flexibly select the optimal parameter configuration scheme according to real-time adversarial requirements. Attached Figure Description

[0042] Figure 1 This is a structural diagram of a RIS-assisted secure communication and sensing system.

[0043] Figure 2 The flowchart of a joint optimization method for a RIS-assisted ISAC system for eavesdropping countermeasures provided by the present invention is shown.

[0044] Figure 3 The joint optimization method of RIS-assisted ISAC system for eavesdropping countermeasures provided by this invention shows the performance of a RIS-assisted secure communication and sensing system with different numbers of RIS reflection units. The vertical axis is the Cramer-Rao boundary, and the horizontal axis is the system communication security rate. Detailed Implementation

[0045] The present invention will be further described in detail below with reference to the accompanying drawings.

[0046] This invention provides a joint optimization method for a RIS-assisted secure communication and sensing system for eavesdropping countermeasures. Based on a RIS-assisted secure communication and sensing system, and targeting potential eavesdroppers, it achieves coordinated secure communication and target direction sensing through RIS reflection modulation, secure signal transmission, and artificial noise injection. A multi-objective optimization problem is constructed to address the joint configuration of the RIS reflection phase, secure signal beam, and artificial noise beam in the system. A randomized grouping and density-sensing non-dominated sorting genetic algorithm (RGDNSGA-III) is used to solve this problem, combining non-dominated sorting methods, randomized grouping mutation mechanisms, and density-sensing terminal front selection strategies. Simultaneously, a training and solution foundation is established by combining communication and sensing models. By optimizing the joint configuration variables, system parameter configuration results that meet the target direction sensing accuracy requirements and achieve a high communication security rate are obtained.

[0047] This invention provides a joint optimization method for RIS-assisted ISAC systems aimed at countering eavesdropping, mainly comprising three parts: system modeling, joint optimization problem construction, and parameter optimization design. For example... Figure 2 As shown, the specific implementation process is as follows:

[0048] Step 1: System Modeling;

[0049] Consider a RIS-assisted secure communication and sensing system. The system consists of a base station (BS), a communication user (CU), a potential eavesdropper (Eve), and a RIS. The base station is configured with... One transmitting antenna and There are one receiving antenna; both the communication user and the potential eavesdropper are single-antenna devices. The RIS contains L reflector elements to assist signal propagation. The potential eavesdropper is modeled as a sensing target, such as... Figure 1 As shown.

[0050] (1) The RIS reflection phase matrix is ​​represented as: ;in, , denoted as the phase offset of the l-th RIS reflector unit, where j is the imaginary unit.

[0051] (2) The base station transmits the signal as follows: ;in, For communication signals, It is an artificial noise signal. For the secure signal beam vector, For artificial noise (AN) beam vector; To maintain the confidentiality of the signal beam transmission power, Let be the artificial noise (AN) beam transmit power, and satisfy: , This is the maximum transmission power.

[0052] (3) Constructing a communication model: Downlink communication is completed with the assistance of RIS.

[0053] Let the channel from the base station to the RIS be... The channel from RIS to CU is The corresponding cascaded channel is represented as: The superscript H indicates the conjugate transpose of the matrix.

[0054] (4) The CU received signal is represented as: The signal-to-interference-plus-noise ratio (SIR) received by the CU is expressed as: ;CU communication rate is expressed as: Similarly, Eve's eavesdropping rate can be expressed as: ,in, , For users to receive channel noise, For users to receive channel noise power, This allows eavesdroppers to listen to the channel noise power.

[0055] (5) The system communication security rate is defined as: .in, The eavesdropper's eavesdropping rate is represented by the superscript "+", which indicates that the non-negative part is taken.

[0056] (6) Constructing a perception model: The system perceives the direction of potential eavesdroppers while communicating.

[0057] Let the channel from the base station to the RIS be... The channel from RIS to Eve is The corresponding cascaded channel is represented as: .

[0058] Assuming the base station receiver is a uniform linear array, the echo channel is represented as follows: ;in, The target reflectance coefficient, The array direction vector, This is the equivalent concatenated channel from the base station through the RIS to the eavesdropper. The angle of incidence is relative to the base station receiving array.

[0059] If in If the transmitted signal is superimposed on each snapshot, the received echo matrix is ​​represented as: ;in, For the emission matrix, This is a Gaussian noise matrix.

[0060] (7) The accuracy of target direction perception is measured using the Cramer-Rao bound (CRB), and its mathematical expression is as follows:

[0061] ;

[0062] in, , For the signal wavelength, To sense channel noise power, The power of the echo signal is expressed mathematically as follows:

[0063] .

[0064] This invention not only constructs a system model covering base stations (BS), communication users (CU), potential eavesdroppers (Eve), and RIS, but more importantly, it changes the previous passive logic that only focused on the service of legitimate communication users, and achieves proactive countermeasures and control against potential eavesdroppers through integrated perception feedback.

[0065] Step 2: Constructing the joint optimization problem;

[0066] Confidential signal beam transmission power Artificial noise (AN) beam transmit power With the RIS reflection phase matrix fixed, the present invention will... and secure signal beam vector Artificial noise (AN) beam vector As joint configuration variables, communication security rate and target direction perception accuracy are used as joint optimization objectives to achieve coordinated configuration of system parameters.

[0067] Given the approximate direction of the eavesdropper, to simultaneously improve communication security performance and target direction perception accuracy, the following multi-objective optimization problem is established:

[0068]

[0069] in, This indicates the calculation of the L2 norm. The maximum allowable CRB threshold is used to guarantee basic sensing performance. Since the RIS reflection phase, the secure signal beam vector, and the artificial noise (AN) beam vector are coupled in the rate and CRB expression, this multi-objective optimization problem is a non-convex optimization problem.

[0070] Step 3: System parameter optimization design;

[0071] To solve the constructed multi-objective optimization problem, this invention proposes a random grouping and density-aware non-dominated sorting genetic algorithm, named RGDNSGA-III. The RGDNSGA-III algorithm improves upon the classic NSGA-III multi-objective optimization framework by using a random grouping mutation mechanism and a density-aware terminal front selection strategy to enhance the efficiency of joint parameter optimization under high-dimensional decision variable conditions and the uniformity of the Pareto solution set distribution.

[0072] like Figure 2 As shown, the specific implementation process of the RGDNSGA-III algorithm is as follows:

[0073] (1) Initialize and optimize the population;

[0074] A set of candidate solutions is randomly generated based on the system model. Each individual represents a set of candidate solutions for system parameters. Each candidate solution contains parameters such as the RIS reflection phase matrix, the secure signal beam vector, and the artificial noise (AN) beam vector, and an initial population is constructed.

[0075] The system performance is evaluated for each individual in the initial population. Based on the RIS reflection phase matrix, the secure signal beam vector, and the artificial noise (AN) beam vector corresponding to the current individual, the system's communication security rate and target direction perception accuracy are calculated, and the objective function vector for each individual is obtained.

[0076] (2) Perform a non-dominated sorting method;

[0077] For the current population, based on the system's communication security rate and target direction perception accuracy, all candidate solutions are sorted using a non-dominated sorting method to determine the Pareto level of each individual in the population and obtain the non-dominated solution set in the current population.

[0078] (3) Population renewal based on random grouping mutation mechanism;

[0079] In each generation of evolution, the current population is updated, and offspring individuals are generated through operations such as crossover, mutation, and random grouping, thereby achieving joint updates of the RIS reflection phase matrix, the secure signal beam vector, and the artificial noise (AN) beam vector.

[0080] Specifically, to reduce the complexity of optimizing high-dimensional decision variables, at the beginning of each generation of evolution, the index set consisting of all decision variables is represented as: Where D is the total number of decision variables. The index set is then divided into G subgroups using a random grouping mutation mechanism: ;satisfy At the same time, maintain , For the i-th subgroup, Let j be the j-th subgroup. The above random grouping operation is achieved by randomly shuffling the decision variable indices and then dividing them sequentially; therefore, the average size of each subgroup is approximately [missing information]. When generating offspring individuals, for a pair of parent solutions... and Choose only one group of variables Crossover and mutation operations are performed, while the remaining variables remain unchanged. The mathematical expression for how the offspring variables are generated is: At the same time, maintain , , Let be the sub-vectors obtained by mutating the two offspring on the g-th variable group. , Let be the child vectors of the two parents on the g-th variable group; where, , The set of indices of the remaining variables that were not selected. This represents the mutation operator.

[0081] Through this random grouping mutation mechanism, the number of variables that actually change in each evolutionary process is reduced from the original number. The number decreased to approximately This method reduces the search difficulty of high-dimensional decision variable optimization problems by eliminating the need for large-scale adjustments to a large number of RIS reflection phase and beam parameters simultaneously in the same generation of evolution, thereby improving the stability of the search process.

[0082] (4) Implement density-sensing frontier selection strategy;

[0083] The density-aware terminal front selection strategy can be used to select uniformly distributed solution sets as the next generation population. Then, through multiple generations of iterative search, the Pareto optimal solution set of the multi-objective optimization problem can be gradually approximated.

[0084] During the environmental selection phase, the parent and offspring populations are first merged, and a non-dominated sorting method is performed to obtain the non-dominated solution set. All complete Pareto fronts are directly incorporated into the next generation population, while the last Pareto front that cannot be completely preserved is excluded. Then, a reference point mechanism is used for selection. Specifically, firstly, a set of uniformly distributed reference points is generated in the objective space to maintain the uniform distribution of the Pareto solution set. Then, the multi-objective optimization problem is translated and normalized according to the ideal points, and the normalized objective function vector is denoted as... Then, the perpendicular distance between the solution and the reference point is calculated:

[0085] ;

[0086] in, z is a reference point in the target space. For the normalized target vector The angle between the reference point vector z.

[0087] To further improve the uniformity of the solution set distribution, this invention introduces a density-sensing mechanism during the selection of the final leading edge. For any solution in the last leading edge... The nearest neighbor distance is defined as:

[0088] ;

[0089] in, Let q be the normalized target vector of the candidate solution q.

[0090] When multiple candidate solutions correspond to the same reference point, the solution with the largest nearest neighbor distance is selected first, i.e.:

[0091] ;

[0092] in, For the last layer of non-dominated frontier In the middle, with reference point The set of candidate solutions associated with the problem.

[0093] If multiple candidate solutions have the same nearest neighbor distance, the solution with the smaller vertical distance to the reference point is selected. This density-aware front selection strategy can preferentially retain solutions that are sparsely distributed in the target space, thereby improving the uniformity and diversity of the final Pareto solution set.

[0094] This invention effectively reduces the scale of variable perturbation in high-dimensional decision space and improves search efficiency by introducing a random grouping mutation mechanism; and ensures the uniformity of Pareto solution distribution by combining a density-aware front selection strategy.

[0095] This invention utilizes the RGDNSGA-III algorithm to proactively suppress the reception quality of potential eavesdroppers, improve communication security rate, and significantly enhance convergence and solution set diversity in high-dimensional multi-objective optimization processes, while meeting stringent constraints on target direction perception accuracy.

[0096] (5) Determine whether the number of iterations has reached the preset upper limit. If not, return to step (2) to continue population update (merging the parent population and the offspring population as the new search basis) and performance evaluation; if it has reached the upper limit, output the final Pareto optimal solution set.

[0097] (6) Select the RIS reflection phase configuration and transmit beamforming parameters that meet the system requirements from the obtained Pareto optimal solution set, and apply them to the RIS-assisted security communication and sensing system to achieve synergistic optimization of system communication security performance and environmental sensing performance.

[0098] like Figure 3 As shown, the joint optimization method for RIS-assisted ISAC system for eavesdropping countermeasures provided by this invention exhibits good system parameter estimation performance under different numbers of RIS reflection units in RIS-assisted secure communication and sensing systems.

[0099] This invention uses the RGDNSGA-III algorithm to iteratively search for system parameters, which can obtain the optimal RIS reflection phase configuration and transmit beam parameters that meet the requirements of communication security and target direction perception accuracy, thereby achieving joint optimization of the performance of RIS-assisted secure communication and perception systems.

[0100] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A joint optimization method for RIS-assisted ISAC systems for eavesdropping countermeasures, characterized in that, Includes the following steps: In the RIS-assisted secure communication and sensing system, potential eavesdroppers are considered as sensing targets and the system is modeled accordingly. A multi-objective optimization problem is constructed using the RIS reflection phase matrix, the secure signal beam vector, and the artificial noise beam vector as joint configuration variables, and communication security rate and target direction sensing accuracy as joint optimization objectives. A random grouping and density-sensing non-dominated sorting genetic algorithm is employed, combined with non-dominated sorting methods, random grouping mutation mechanisms, and density-sensing terminal front selection strategies, to solve the multi-objective optimization problem. The joint configuration variables are then optimized to obtain system parameter configuration results that meet the requirements of communication security rate and target direction sensing accuracy.

2. The joint optimization method for RIS-assisted ISAC systems for eavesdropping countermeasures as described in claim 1, characterized in that, The RIS-assisted secure communication and sensing system includes a base station, a communication user, a potential eavesdropper, and the RIS. Both the communication user and the potential eavesdropper are single-antenna devices; mathematical expressions are constructed for the RIS reflection phase matrix, base station transmitted signal, communication model, communication user received signal, communication user received signal-to-interference-plus-noise ratio, communication user communication rate, potential eavesdropper eavesdropping rate, communication security rate, perception model, and target direction perception accuracy.

3. The joint optimization method for RIS-assisted ISAC systems for eavesdropping countermeasures as described in claim 2, characterized in that, The mathematical expression for the RIS reflection phase matrix is: ; , Let j be the phase offset of the l-th RIS reflector unit, where j is the imaginary unit, satisfying... .

4. The joint optimization method for RIS-assisted ISAC systems for eavesdropping countermeasures as described in claim 1, characterized in that, The mathematical expression for the communication security rate is: , For communication user communication rate, The rate at which the eavesdropper listens.

5. The joint optimization method for RIS-assisted ISAC systems for eavesdropping countermeasures according to claim 1, characterized in that, The accuracy of target direction perception is measured using the Cramer-Rao bound, and its mathematical expression is as follows: ； in, , For wavelength, To sense channel noise power, The angle of arrival in the direction of the target. For echo signal power, To superimpose the number of transmitted signals onto the snapshot, The number of receiving antennas configured for the base station.

6. The joint optimization method for RIS-assisted ISAC systems for eavesdropping countermeasures according to claim 1, characterized in that, The multi-objective optimization problem is expressed as follows: ； in, For secure communication speed, To improve the accuracy of target direction perception. For the secure signal beam vector, For artificial noise beam vectors, This represents the phase shift of the l-th RIS reflector unit. This indicates the calculation of the L2 norm. This is the maximum allowed CRB threshold.

7. The joint optimization method for RIS-assisted ISAC systems for eavesdropping countermeasures according to claim 1, characterized in that, The specific implementation process of the random grouping and density-aware non-dominated sorting genetic algorithm is as follows: (1) Randomly generate a set of candidate solutions, each of which contains the RIS reflection phase matrix, the secret signal beam vector and the artificial noise beam vector, and construct an initial population; evaluate the system performance of each individual in the initial population to obtain the objective function vector corresponding to each individual; (2) Sort all candidate solutions by non-dominated sorting method, determine the Pareto level of each individual in the population, and obtain the non-dominated solution set in the current population; (3) Population update based on random grouping and mutation mechanism: In each generation of evolution, the current population is updated and offspring individuals are generated through crossover, mutation and random grouping operations; (4) Implement the density-sensing front selection strategy to select uniformly distributed solution sets as the next generation population; (5) Through multiple generations of iterative search, the Pareto optimal solution set of the multi-objective optimization problem is gradually approximated; (6) Select the RIS reflection phase configuration and transmit beamforming parameters that meet the system requirements from the obtained Pareto optimal solution set, and apply them to the RIS-assisted security communication and sensing system.

8. The joint optimization method for RIS-assisted ISAC systems for eavesdropping countermeasures according to claim 7, characterized in that, In step (3), at the beginning of each generation of evolution, the index set consisting of all decision variables is represented as follows: D represents the total number of decision variables; the index set is divided into G subgroups using a random grouping mutation mechanism. The average size of each subgroup is When generating offspring individuals, for a pair of parent solutions... and Choose only one group of variables Perform crossover and mutation operations while keeping the other variables unchanged.

9. The joint optimization method for RIS-assisted ISAC systems for eavesdropping countermeasures according to claim 7, characterized in that, In step (4), during the environmental selection phase, the parent population and the offspring population are merged, and the non-dominated sorting method is executed to obtain the non-dominated solution set; all complete Pareto fronts are directly entered into the next generation population, while the last Pareto front that cannot be completely retained is screened using the reference point mechanism.

10. The joint optimization method for RIS-assisted ISAC systems for eavesdropping countermeasures according to claim 9, characterized in that, A set of uniformly distributed reference points is generated in the target space. The objective function vector is translated and normalized according to the ideal points, and the vertical distance between the solution and the reference points is calculated. A density-aware mechanism is introduced in the selection process of the last leading edge. The nearest neighbor distance of any solution in the last leading edge is defined. When the same reference point corresponds to multiple candidate solutions, the solution with the largest nearest neighbor distance is selected first. If multiple candidate solutions have the same nearest neighbor distance, the solution with the smaller vertical distance to the reference point is further selected.

Images