Optimization method for safe multi-target sensing and communication in IRS-assisted ISAC system

Abstract

The invention relates to a physical layer security assurance method and a physical layer security assurance device for an integrated sensing and communication (ISAC) system based on the assistance of an intelligent reflecting surface (IRS), in particular to a physical layer security assurance method and a physical layer security assurance device for an integrated sensing and communication (ISAC) system based on the assistance of the IRS. The system can realize multi-target detection and multi-user safety communication in a complex electromagnetic propagation environment. In order to improve communication secrecy performance and sensing precision at the same time, a joint optimization framework is provided and is used for collaborative design of active beam forming of a transmitting end and a phase control strategy of an IRS. In a multi-target identification process, when reflection echoes of a plurality of targets are incident through an IRS at a similar angle, the targets are difficult to distinguish. In order to solve the problem, the invention introduces a signature sequence (SS) modulation scheme, and high-precision target identification is realized by encoding a reflected signal. Under the constraint conditions of transmitting power, interference suppression, IRS configuration and the like, an optimization model taking secrecy rate maximization as a target is established. In order to efficiently solve the optimization problem, an algorithm integrating multiple optimization technologies such as alternating optimization, a penalty function method, successive convex approximation and positive semidefinite relaxation is provided. Simulation results show that the method provided by the invention is superior to an existing comparison scheme in the aspects of sensing precision and communication security, and better communication and sensing integration performance can be realized in a complex channel environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of wireless communication and target sensing fusion, and particularly relates to an IRS-assisted ISAC system physical layer security enhancement method and device, which can be applied to multi-target detection, secure multi-user transmission and signal optimization design in complex electromagnetic propagation environments. BACKGROUND

[0002] With the rapid evolution of 6G wireless communication technology, integrated sensing and communication (ISAC) has become one of the core directions to achieve efficient spectrum utilization and intelligent environment sensing. This technology simultaneously realizes wireless communication and target sensing functions under the same hardware and spectrum resources, which can significantly reduce system energy consumption, improve time-frequency resource utilization, and realize deep integration of communication links and sensing information. However, in actual complex propagation environments, traditional ISAC systems still face many challenges. On the one hand, due to the randomness and multipath effect of wireless channels, the signal resolution and sensing accuracy of the receiving end are limited, especially under non-line-of-sight (NLOS) propagation conditions, the target detection and positioning accuracy significantly decreases. On the other hand, to achieve multi-user parallel communication, the system usually needs complex beamforming and power allocation strategies, which are difficult to simultaneously consider sensing performance and communication security under limited hardware conditions.

[0003] Intelligent reflecting surface (IRS) as a new generation of reconfigurable electromagnetic environment technology, is widely used to enhance wireless link performance due to its low power consumption, high controllability and easy deployment. IRS can realize programmable control of the phase, amplitude and propagation direction of incident electromagnetic waves through a reflecting array composed of a large number of adjustable phase units, thereby realizing directional enhancement and spatial reconstruction of signal energy under passive conditions. Existing research shows that introducing IRS into the ISAC system can effectively improve signal coverage and energy distribution, making it possible to optimize communication links and sensing links collaboratively.

[0004] However, the existing IRS-assisted ISAC technology still has the following outstanding problems:

[0005] Insufficient multi-target discrimination capability: when the echo signals of multiple targets are incident at similar angles via IRS, their reflection paths are highly correlated, and traditional angle-discrimination-based target detection methods are difficult to accurately distinguish, resulting in decreased identification accuracy.

[0006] Limited communication security: Due to the passive nature of IRS and the predictability of reflected signals, the system is vulnerable to signal capture by potential eavesdroppers, and traditional encryption or random scrambling methods are difficult to balance real-time and complexity requirements in this scenario.

[0007] High complexity of joint optimization: The joint design of the transmit beamforming and the IRS phase matrix is essentially a non-convex optimization problem, and existing algorithms often fall into local optima, making it difficult to achieve efficient solutions while ensuring security performance.

[0008] To solve the above problems, the present application proposes a physical layer security enhancement method for intelligent reflecting surface assisted integrated sensing and communication system based on signature sequence modulation. This method introduces a distinguishable signature sequence into the sensing signal, which effectively identifies the angle-related target. At the same time, a joint optimization strategy is used to design the transmit beam and the IRS phase control, to improve the security rate and the sensing accuracy of the system. Compared with existing solutions, the present application can significantly improve the anti-interference performance, target discrimination ability and security communication reliability of the IRS assisted ISAC system. SUMMARY

[0009] The present application aims to overcome the problems of insufficient multi-target discrimination ability, limited communication security and high complexity of joint optimization in existing IRS assisted ISAC systems, and proposes a physical layer security enhancement method for intelligent reflecting surface assisted integrated sensing and communication system based on signature sequence modulation. This method constructs a multi-dimensional joint optimization framework to realize the coordinated control of communication beam and IRS reflection phase, and simultaneously improves the sensing accuracy and security performance of the system in complex propagation environment, thereby providing a low complexity and high reliability technical solution for multi-user secure communication and high-precision multi-target sensing.

[0010] To achieve the above object, the application adopts the following technical solutions: a sensing and communication integrated system structure containing a multi-antenna transmitting end and an intelligent reflecting surface is constructed. The transmitting end simultaneously performs communication and sensing tasks using the same signal carrier, and the IRS controls the reflected waves through a programmable phase array to achieve spatial redistribution of signal energy, thereby enhancing the overall performance of the sensing link and the communication link. To solve the problem of identification ambiguity caused by the similar angles of multiple target echoes, the application introduces a signature sequence (SS) modulation mechanism in the sensing signal. This mechanism assigns a unique sequence code to different targets, allowing the reflected signal to carry target identity information. The base station can separate and identify targets based on sequence characteristics, thereby effectively distinguishing echoes with high correlation angles. While considering the communication needs of multiple users, the application constructs an optimization model based on maximum secrecy rate, taking the transmit beamforming matrix and the IRS phase matrix as joint optimization variables, and ensuring the physical feasibility of the transmit power, interference suppression, and IRS reflection amplitude under the constraint conditions. By introducing a hybrid joint optimization algorithm proposed by the application, the system secrecy rate is maximized under the premise of meeting the communication reliability, realizing physical layer protection against potential eavesdropping users. Specifically, the application solves the problems in the prior art and realizes more efficient sensing and communication security through the following steps.

[0011] The system architecture includes a full-duplex base station (FDBS) and an IRS, the IRS has a MxM uniform planar array (UPA), the FDBS is equipped with a dual-function uniform linear array (ULA) with N elements, the azimuth angles (θj ) and elevation angles (φj ) of J sensing targets, the spatial coordinates of I single-antenna users, the FDBS transmit power, the target reflection coefficient , the time delay , the noise power spectral density , the total number of system time slots L, and the mathematical models of system transmit signals, sensing echoes, and communication reception are constructed; the mathematical model of the system average secrecy rate (ASR) is established according to Shannon's theorem and the SINR expression; the optimization problem model is constructed according to the maximization of ASR; the constraint conditions of the optimization problem model are established.

[0012] Step 2: The maximization of the optimization problem model under the constraint conditions is taken as the optimization objective, and an iterative algorithm combining the alternating optimization (AO) framework is used for optimization solution, obtaining the communication beamforming vector under the maximum ASR of the optimized system. , radar beamforming vectors and IRS phase shift matrix .

[0013] Step 3: After the channel state or user / target position changes, the optimized and are obtained through step 1 and step 2, and the IRS reflection link is reconfigured to make the system maintain the sensing accuracy and communication security in the dynamic environment.

[0014] Initialize the number of AO iterations T, the upper bound of the power budget of the optimization variable and the lower bound of the minimum communication rate , define the dimensions (including communication dimension, radar dimension and dimension of phase shift matrix); initialize the penalty parameter , convex approximation point (such as initial random beam and , where u is the IRS phase shift vector); the initial system parameters are defined as:

[0015] where, is the communication beam of the i-th user, is the corresponding communication signal, is the sensing beam of the j-th target, is the sensing signal allocated to the j-th target. The statistics of communication and sensing signals are not related.

[0016] Let t represent the current iteration number, represent the i-th user communication beamforming vector at the t-th iteration, represent the j-th target radar beamforming vector at the t-th iteration, represent the SS sequence assigned to the j-th target (satisfying , ), .

[0017] Start the first sub-iteration (fix , optimize the beam), update the semi-definite matrix , . Define the DC decomposition: , where , similarly define , introduce the first-order Taylor expansion to approximate the concave term and :

[0018] where, denotes the last feasible point, is the current sub-iteration number.

[0019] In the first sub-iteration, the algorithm updates iteratively, and according to the SDR relaxation rank-one constraint, when and , the algorithm recovers by Cholesky decomposition. The specific problem can be expressed as:

[0020] where is the minimum sensing beam gain of the target, is the interference constraint threshold, which is about of the maximum beam pattern gain. .

[0021] After all the beams complete the update in the first sub-iteration, the optimal is output, and its ASR value is calculated and input as the initial value of the second sub-iteration, i.e., .

[0022] Initialize the parameters in the second sub-iteration, keep the upper bound , lower bound , and dimension of the variable, and set the maximum sub-iteration number to ; initialize the penalty function in the second sub-iteration algorithm .

[0023] In the second sub-iteration (fixed beam, optimize ), the augmented matrix is updated, and the diagonal unit length constraint is applied, where . The rank-one constraint is equivalent to the difference between the kernel norm and the spectral norm and is punished:

[0024] Integrate it into the objective function, and iteratively update the algorithm, and the specific expression is as follows:

[0025] where , , , , .

[0026] In the process of updating the second sub-iteration, the algorithm updates the current optimal , calculate its ASR value, and compare it with the ASR value of the optimal beam obtained in the first sub-iteration. After completing the second sub-iteration After the second update, output the optimal , that is, the optimal solution, represents the IRS phase shift And output the ASR value under the optimal solution, that is, the maximum secrecy rate of the system.

[0027] Compared with the prior art, the present application has the following technical advantages:

[0028] The present application improves the traditional ISAC optimization framework. In view of the problem that the traditional method ignores the multi-target angle overlap and safety constraints, which easily leads to perception failure and low secrecy rate, the SS modulation and AO-SCA-SDR hybrid mechanism are introduced to improve the global optimization ability. The optimal solution obtained after the beam sub-iteration is used as the initial solution of the phase shift sub-iteration, the rank-one constraint is processed in combination with the penalty function, the calculation complexity is reduced, the sub-iteration optimization is solved according to the given convex proxy formula, the local optimum in the beam-phase shift coupling can be corrected, and the solution can be verified as a global optimal solution as a verification mechanism, which can effectively improve the convergence speed (usually less than 10 iterations) and optimization accuracy (ASR is improved by 15%-25%).

[0029] In the case of changes in the channel state in the system propagation environment, the improved AO iteration algorithm is used to optimize and solve the maximum ASR that can be reached by the current ISAC system, and then the IRS phase shift and beam configuration are adjusted, so that the stable perception-communication link state is maintained after the channel changes, and the robustness and usability of the system are improved. BRIEF DESCRIPTION OF DRAWINGS

[0030] Figure 1 is the system model diagram described in the present application.

[0031] Figure 2 is the optimization algorithm flowchart for solving the system ASR described in the present application.

[0032] Figure 3 is the simulation diagram for explaining the system convergence described in the present application.

[0033] Figure 4 is the simulation diagram for explaining the relationship between ASR and maximum transmit power described in the present application.

[0034] Figure 5 is the simulation diagram for explaining the relationship between ASR and the number of IRS elements described in the present application.

[0035] Figure 6 is the simulation diagram for explaining the relationship between ASR and the number of transmit antennas described in the present application. DETAILED DESCRIPTION

[0036] The application will be described in further detail below in conjunction with the accompanying drawings and specific embodiments.

[0037] As shown in the figure, the system model proposed by the application includes the following contents: Figure 1

[0038] An FDBS and an IRS, a plurality of single-antenna ground users, and a plurality of to-be-sensed targets. The FDBS is configured with a dual-function uniform linear array (ULA); the IRS adopts a uniform planar array (UPA) with a total of reflector units. The users and the targets are distributed in the coverage area of the IRS; there are potential eavesdroppers in the scene with unknown channel characteristics. The cascaded channel between the FDBS and the IRS is denoted as , the channel between the IRS and the i-th user is denoted as , the direct channel between the FDBS and the i-th user is denoted as , and the phase shift matrix of the IRS is denoted as . The system operates according to a frame-slot structure. Each frame contains slots, and the FDBS transmits communication signals and sensing beams simultaneously in the same time slot. The transmitted signal is denoted as:

[0039]

[0040] wherein is the communication beam of the i-th user, is the corresponding communication signal, is the sensing beam of the j-th target, is the sensing signal allocated to the j-th target. The communication and sensing signals are statistically independent.

[0041] In view of the problem that the echo of multiple targets reflected from the IRS cannot be separated due to similar target angles, the application modulates the sensing beam by SS to make the reflection of each target carry a unique coded fingerprint, and completes identification and detection by matching filtering and threshold decision. The specific implementation method of SS modulation and sensing processing is as follows:

[0042] First, the signature sequence is designed and allocated, and a complex sequence with a length of L is allocated to each target j . The sequence is one-to-one corresponding to the target direction, and satisfies the energy constraint and low cross-correlation characteristics.

[0043] Let ​​​, which represents the equivalent main lobe gain of the jth target direction, is set by setting the minimum sensing beam gain of the target , and the sidelobes are controlled to suppress the cross-target leakage and interference to the communication link, the sensing beam directionality and energy control.

[0044] FDBS constructs a matched filter for each target j , and performs filtering and demodulation on the received vector to obtain the statistics:

[0045] wherein , represents all interference and clutter.

[0046] The present application adopts binary hypothesis testing , wherein is a threshold value that meets the expected false alarm probability, so that the target can still be reliably distinguished and detected in the multi-target angle overlapping scene.

[0047] To adapt the convex optimizer, the present application introduces a semi-definite replacement in sub-problem one , temporarily relaxes the rank-one constraint by using SDR, and rewrites the user / eavesdropper rate term into a differential convex form of the difference of logarithmic functions, obtains the convex proxy target and the linearized rate lower bound, interference and sensing gain constraint by performing SCA at the current feasible point, forms a semi-definite programming that can be directly solved by the interior point method, and obtains which can be directly recovered by eigenvalue decomposition.

[0048] In sub-problem two, the present application augments the IRS phase vector to , and constructs (satisfying ). The difference between the nuclear norm and the spectral norm ( ) is used to equivalently characterize the rank-one property and is incorporated into the objective function as a penalty term, so as to control the penalty strength. Then, the SCA is performed on the concave part at the current solution, so as to obtain a convex sub-problem containing the sensing gain, interference threshold, linearization of the user minimum rate, unit modulus and semi-definite constraint, and the principal eigenvector of is taken after solving to recover and obtain .

[0049] After the above update, the system average secrecy rate corresponding to the current outer iteration is calculated , and then enters the next iteration. The termination of the outer loop follows the following criterion: when the relative change of the target value of the adjacent two outer iterations satisfies When the number of outer iterations exceeds a preset limit, the algorithm is considered to have converged; otherwise, the algorithm is considered to have converged. If the iteration stops, the iteration also stops. Upon termination, the optimal parameters for each time slot obtained in the last outer iteration are output. IRS phase shift matrix and the corresponding average confidentiality rate .

[0050] The above process achieves the joint design of active and passive arrays by first optimizing the beam under a given phase, then optimizing the phase under a given beam, and then unifying the solutions of each time slot through a global update. This ensures continuous improvement and stable convergence of security performance while meeting power and constraint conditions.

[0051] like Figure 3 As shown, this invention is compared with random beamforming and random IRS phase in terms of convergence and ASR. The ASR of all methods increases with the number of iterations and tends to stabilize. This invention maintains a significantly higher ASR and a faster convergence speed, effectively improving the convergence speed of the algorithm (usually less than 10 iterations) and optimization accuracy (ASR improvement of 15%-25%).

[0052] like Figure 4 , Figure 5 and Figure 6 As shown, this invention outperforms two other benchmark schemes under different maximum transmit power, IRS component numbers, and antenna numbers conditions in terms of ASR performance. The algorithm proposed in this invention has better performance advantages in solving this optimization problem.

Claims

1. An optimization method for secure multi-target sensing and communication in an IRS-assisted ISAC system, characterized in that, Includes the following steps: Step 1: The system architecture consists of a full-duplex base station (FDBS) and an IRS. The IRS has an M×M uniform planar array (UPA), while the FDBS is equipped with a dual-function uniform linear array (ULA) with N elements, and J azimuth angles (Θ) of the sensed targets. j ) and elevation angle (φ) j Spatial coordinates of I single-antenna users, FDBS transmit power, target reflection coefficient β j Timely delay j Noise power spectral density δ 2 The total number of system time slots L is used to construct a mathematical model for system signal transmission, sensing echo, and communication reception. A mathematical model for the system's average secrecy rate (ASR) is established based on Shannon's theorem and the SINR expression. An optimization problem model is constructed based on maximizing ASR. Constraints on the optimization problem model are established. Step 2: Taking the maximization of the optimization problem model under constraints as the optimization objective, the optimization is performed using an iterative algorithm based on the Alternating Optimization (AO) framework to obtain the communication beamforming vector {w} under the maximum ASR of the optimized system. c,i }、Radar beamforming vector {w r,j } and the IRS phase shift matrix Φ; Step 3: After a change in channel state or user / target location, obtain the optimized {w} through steps 1 and 2. c,i， w r,j } and Φ, and reconfigure the IRS reflection link to ensure that the system maintains perception accuracy and communication confidentiality in dynamic environments.

2. The optimization method for secure multi-target sensing and communication in an IRS-assisted ISAC system according to claim 1, characterized in that, The specific process of the iterative algorithm optimization solution based on the AO framework proposed in step 2 is as follows: Step 1: Initialize the AO iteration count T, and optimize the upper bound power budget P of the variable. max and the lower bound minimum communication rate R min Define dimensions (including N×I communication dimension, N×J radar dimension, and M-dimensional phase shift matrix); initialize the penalty parameter ϱ>0 in the algorithm, and the convex approximation point (such as the initial random beam W). c,i (0) W r,j (0) and U (0) =u (0) u (0)H (where u is the IRS phase shift vector). The initial system parameters are defined as follows: ； Among them, w c,i For the communication beam of the i-th user, a c,i [l] represents the corresponding communication signal, w r,j For the sensing beam of the j-th target, a r,j [l] represents the sensing signal assigned to the j-th target. Communication and sensing signal statistics are uncorrelated. Let t represent the current iteration number. This represents the communication beamforming vector of the i-th user at the t-th iteration. This represents the radar beamforming vector of the j-th target at the t-th iteration. Represents the SS sequence assigned to the j-th target (satisfying...) ); ; Step 2: Start the first sub-iteration (fixed Φ, optimized beam), update the semi-definite matrix. Define DC decomposition: Similar to the definition of L e F i F e Introducing a first-order Taylor expansion to approximate the concave term F i and L e : ； ； in, This represents the previous feasible point, and t is the current sub-iteration number; Step 3: In the first sub-iteration, for ({W c,i W r,j The algorithm iteratively updates the rank constraint based on SDR, and when (Rank(W)... c,i )≤1)and(Rank(W) r,j When )≤1), recover by Cholesky decomposition: The specific problem can be described as follows: ； s.t. ; ； ； ； in It is the minimum sensing beam gain of the target. This is the interference constraint threshold, whose value is approximately 10 relative to the maximum beam pattern gain. −2 Up to 10 −1 ; Step 4: After all beams have completed their updates in the first sub-iteration, output the optimal {W}. c,i *, W r,j *}, calculate its ASR value, and use it as the initial value for the second sub-iteration, based on the result of the first sub-iteration. Let ; Step 5: Initialize the parameters in the second sub-iteration, maintaining the upper bound P of the variables. max Lower bound R min Given the dimension N×(I+J)+M, let the maximum number of sub-iterations be T2; initialize the penalty function ϱ in the second sub-iteration algorithm; Step 6: In the second sub-iteration (fixed beam, optimized Φ), adjust the augmented matrix U. [t] = ũ [t] ũ [t]H And apply a diagonal unit modulus constraint Diag(U)=1, where ũ H = (1, u H The rank-one constraint is equivalently characterized and penalized by the difference between the nuclear norm and the spectral norm: ； Step 7: Incorporate it into the objective function and perform iterative updates. The specific algorithm expression is as follows: ； s.t. ; ； ； in, , , , , ; Step 8: During the second sub-iteration update, the algorithm updates the current optimal U. (t+1) The ASR value is calculated and compared with the ASR value of the optimal beam obtained in the first sub-iteration. After all Us complete the T2 updates in the second sub-iteration, the optimal U* is output, which is the optimal solution, representing the IRS phase shift Φ* = diag(u*) under the maximum ASR. The ASR value under the optimal solution is also output, which is the maximum security rate of the system.

Images