Downlink network physical layer security optimization method based on ISAC system

Abstract

The application proposes a downlink network physical layer security optimization method based on an ISAC system. First, the application proposes a physical layer security modeling method combining ISAC sensing capability under the condition of non-perfect eavesdropper CSI, which combines the norm boundary of ISAC sensing capability with the non-perfect CSI of the eavesdropper. Secondly, the application uses an alternating optimization algorithm to decouple the optimization variables and simplify them into a simpler form. Finally, the application uses a continuous convex approximation algorithm to solve the non-convex characteristics in the optimization problem of the base station beamforming auxiliary variable and the signal-to-noise ratio auxiliary variable parameter optimization of the eavesdropper, and realizes the physical layer security optimization in the downlink network. The application can enhance the security of the downlink network based on the ISAC system, and has excellent performance under different sensing performance requirements and total transmit power conditions.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of wireless network security, and particularly relates to a physical layer security optimization method in a communication-sensing integrated system, which is an optimization method for realizing physical layer security of a communication-sensing integrated system under the condition of non-perfect eavesdropper channel state information (CSI) by means of wireless communication technology, optimization and computer science and technology. BACKGROUND

[0002] With the application of next-generation wireless communication technology in various emerging industries, the integrated sensing and communication (ISAC) technology plays an increasingly important role in the Internet of Things, the Internet of Vehicles and secure communication. At the same time, sensitive data will face more threats in the transmission of future wireless communication networks, such as zero-day attacks, quantum attacks, and physical layer attacks in visible light or terahertz communication. In order to ensure data security, the physical layer security (PLS) technology is considered as a lightweight and adaptive solution. The PLS technology utilizes the scattering and superposition of electromagnetic wave channels to counter potential eavesdroppers. However, the acquisition method of non-cooperative or passive eavesdropper channel state information (CSI) has not been properly solved. The sensing capability of the ISAC system provides a new paradigm for realizing robust secure communication under the condition of non-perfect eavesdropper CSI.

[0003] The existing security communication research based on ISAC mainly focuses on two aspects. The first is the physical layer security enhancement based on signal to noise ratio (SNR). For example, in a multiple-input multiple-output radar, the covariance matrix is optimized to dynamically adjust the SNR of echo signals and communication signals. Some scholars also use artificial noise to reduce the quality of the eavesdropper's received signal, or improve the beamforming gain of the communication system through intelligent reflecting surfaces. This kind of processing method will bring additional software and hardware overhead to the system. The second is the physical layer security enhancement based on beam pattern, such as realizing beamforming design by minimizing the root mean square error between the actual beam pattern and the theoretical beam pattern. This method can also achieve wide-area coverage and enhance communication security by designing a reasonable beam width. However, neither the physical layer security enhancement method based on SNR nor the physical layer security enhancement method based on beam pattern can theoretically depict the Pareto boundary of the secure communication system. Moreover, the above optimization models do not integrate the actual sensing capability into the eavesdropper channel uncertainty. In summary, it is still an open problem to enhance the physical layer security of the downlink communication network based on the ISAC system under the condition of non-perfect eavesdropper CSI. SUMMARY

[0004] The application aims to enhance the security of the downlink network under the condition of non-perfect CSI of the eavesdropper, and proposes a downlink network physical layer security optimization method based on the ISAC system.

[0005] The application is realized by the following technical scheme, and the application proposes a downlink network physical layer security optimization method based on the ISAC system, and the specific process of the method is as follows:

[0006] Step one: initialize parameters, including the number of transmitting antennas N t , the number of receiving antennas N r , the number of users N B , the channel of the base station-Bob , the channel estimation value of the base station-eavesdropper , the communication noise variance , the radar noise variance , the eavesdropper reflection factor α, the iteration number N BCD of the alternating optimization algorithm , the eavesdropper azimuth angle estimation value , the base station beamforming auxiliary variable , the eavesdropper signal-to-interference-noise ratio auxiliary variable , the sensing Cramer-Rao bound auxiliary variable t 0 , the S-procedure auxiliary variable , the maximum iteration number N SCA of the successive convex approximation algorithm

[0007] , the iteration number index n=0 of the alternating optimization algorithm and the iteration number index m=0 of the successive convex approximation algorithm;

[0008] The CSI is channel state information;

[0009] The ISAC is a communication and sensing integrated system;

[0010] Step three: solve the eavesdropper signal-to-interference-noise ratio auxiliary variable γ k and the S-procedure auxiliary variable p k under the condition of fixed t n , by using the successive convex approximation algorithm, let m=m+1 and , and execute step four;

[0011] , wherein, is the base station beamforming auxiliary variable in the n-th iteration of the alternating optimization algorithm, and t nthe eavesdropper-to-interference-plus-noise ratio auxiliary variable in the m-th successive convex approximation algorithm iteration, the eavesdropper-to-interference-plus-noise ratio auxiliary variable in the m-th successive convex approximation algorithm iteration;

[0012] Step four: if the successive convex approximation algorithm iteration index satisfies m≥N SCA , let m = 0, and proceed to Step five, if m < N SCA , proceed to Step three;

[0013] wherein, the eavesdropper-to-interference-plus-noise ratio auxiliary variable in the n-th alternating optimization algorithm iteration, the S-procedure auxiliary variable in the n-th alternating optimization algorithm iteration;

[0014] Step five: solve the base station beamforming auxiliary variable W and the perception Cramér-Rao bound auxiliary variable t using the successive convex approximation algorithm with W k and t fixed, let m = m + 1 and t m = t; proceed to Step six;

[0015] wherein, the eavesdropper-to-interference-plus-noise ratio auxiliary variable in the n-th alternating optimization algorithm iteration, the S-procedure auxiliary variable in the n-th alternating optimization algorithm iteration, the base station beamforming auxiliary variable in the m-th successive convex approximation algorithm iteration;

[0016] Step six: if the successive convex approximation algorithm iteration index satisfies m≥N SCA , let t n+1 = t m , m = 0, n = n + 1 and proceed to Step seven, if m < N SCA , proceed to Step five;

[0017] wherein, the base station beamforming auxiliary variable in the n+1-th alternating optimization algorithm iteration, t n+1 the perception Cramér-Rao bound auxiliary variable in the n+1-th alternating optimization algorithm iteration;

[0018] Step seven: if the alternating optimization algorithm iteration index satisfies n≥N BCD , the algorithm ends and outputs otherwise proceed to Step three.

[0019] Further, the step two is a mathematical modeling method for the downlink physical layer security problem of the ISAC system under the condition of non-perfect eavesdropper CSI, and the specific process is as follows:

[0020] Assume that the base station sends a signal Where ω k is the beamforming vector, s k is the transmitted symbol; a total of L slots can be used for azimuth estimation of the eavesdropper; a total of K users, one eavesdropper; the channel between the base station and Bob is denoted as The channel between the base station and the eavesdropper is denoted as And assume that is a line-of-sight propagation channel, then

[0021]

[0022] Where β E is the fading at the reference distance, d AE is the distance from the base station to the eavesdropper, α E is the path loss exponent, θ is the azimuth angle, is the direction vector of the transmitting antenna; since there is an estimation error in the azimuth angle, the azimuth angle is re-expressed as Where is the estimated value of the azimuth angle, Δθ is the estimation error of the azimuth angle, which is related to the sensing ability of the ISAC system; therefore, the achievable rate of the legitimate user and the eavesdropper can be expressed as:

[0023]

[0024]

[0025] Then the secrecy rate can be expressed as From the perspective of sensing, the Cramer-Rao bound of the azimuth angle can be expressed as:

[0026]

[0027] Where, is the covariance matrix of beamforming, A(θ)=b(θ)a H (θ), is the gradient of A(θ), b(θ) is the direction vector of the receiving antenna, is the radar receiving noise;

[0028] In summary, the uncertainty of the eavesdropper CSI is caused by the estimation error of the azimuth angle, and the estimation error of the azimuth angle is related to its Cramer-Rao bound. If it is assumed that Δθ obeys a Gaussian distribution, then the probability that θ falls in the interval is 99.73%, and a(θ) can be re-expressed as:

[0029]

[0030] and

[0031]

[0032] where m(n) = (N t -1) / 2-n and The set expression of the non-perfect CSI two-norm boundary can be obtained by Taylor formula, that is:

[0033]

[0034]

[0035] where,

[0036] The Pareto boundary of the downlink physical layer security optimization model based on ISAC system is defined:

[0037]

[0038] Then according to the above Pareto boundary, the following optimization problem can be formed:

[0039]

[0040] where Γ CRB is the maximum Cramer Rao bound, P t is the total power of base station transmission; let H AE = h AE (h AE ) H , Then the problem can be simplified as:

[0041]

[0042] where

[0043]

[0044] The rank 1 constraint is relaxed, and a relaxation variable γ k is introduced. By using S-procedure theorem and combining with Schur complement theorem, the following optimization problem form can be obtained:

[0045]

[0046] s.t.t≥t0

[0047]

[0048]

[0049]

[0050]

[0051]

[0052]

[0053] Where p k For S-procedure parameters,

[0054] Furthermore, in step three, a continuous convex approximation algorithm is used to fix... t n , Solve for the auxiliary variable γ of the signal-to-interference-plus-noise ratio of the eavesdropper. k and S-procedure auxiliary variable p k The specific process is as follows:

[0055] Considering only the auxiliary variables of the eavesdropper's information drying ratio and the S-procedure, we can obtain the following from the first-order Taylor formula:

[0056]

[0057] The problem can then be simplified to:

[0058]

[0059] The problem is a convex problem, which can be solved using the convex optimization toolbox.

[0060] Furthermore, in step five, a continuous convex approximation algorithm is used to fix... Solving for the base station beamforming auxiliary variable W under the following conditions k And the auxiliary variable t for perceiving the Cramerau bound; the specific process is as follows:

[0061] Considering only the auxiliary variables of base station beamforming and the auxiliary variables of the sensing Cramerau boundary, we can obtain the following from the first-order Taylor formula:

[0062]

[0063] Regarding W k The rank-1 constraint is equivalent to ||W k || * -||W k ||2≤0, where||·|| *||·|| 2 is the nuclear norm and spectral norm; in order to obtain a higher quality rank 1 solution, the equivalent condition of the rank 1 constraint can be utilized by a first-order Taylor expansion and punished into the objective function, and then the problem can be simplified as:

[0064]

[0065] Where, ρ>0 is a penalty factor, μ max (·) is the eigenvector corresponding to the maximum eigenvalue; since the problem is a convex problem, a convex optimization toolbox is used for efficient solution.

[0066] The beneficial effects of the present application are:

[0067] The present application proposes a downlink network physical layer security optimization method based on the ISAC system under the condition of non-perfect eavesdropper CSI. First, the present application proposes a physical layer security modeling method combining the ISAC sensing ability under the condition of non-perfect eavesdropper CSI, which combines the ISAC sensing ability with the norm boundary of non-perfect CSI of the eavesdropper. Secondly, the present application decouples the optimization variables by using an alternating optimization algorithm, and simplifies it into a simpler form. Finally, the present application uses a continuous convex approximation algorithm to solve the non-convex characteristics of the base station beamforming auxiliary variable optimization problem and the eavesdropper signal-to-noise ratio auxiliary variable parameter optimization, and realizes the physical layer security optimization in the downlink network. The present application can enhance the security of the downlink network based on the ISAC system, and has excellent performance under different sensing performance requirements and total transmit power conditions. The significance of the present application is to enhance the security of the downlink network based on the ISAC system, and to reduce the hardware overhead under the condition of the same security index. BRIEF DESCRIPTION OF DRAWINGS

[0068] Figure 1 is a system model schematic diagram;

[0069] Figure 2 is an algorithm convergence schematic diagram;

[0070] Figure 3 is a minimum secure communication rate and Crámer-Rao upper bound relationship schematic diagram, Γ CRB is the maximum Crámer-Rao bound;

[0071] Figure 4 is a minimum secure communication rate and transmit power relationship schematic diagram. DETAILED DESCRIPTION

[0072] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0073] Specific Implementation Method 1: The specific process of the downlink network physical layer security optimization method based on the ISAC system is as follows:

[0074] Step 1: Initialize parameters, including the number of transmit antennas N. t Number of receiving antennas N r Number of users N B Base station-Bob channel Base station-eavesdropper channel estimation Communication noise variance Radar noise variance The eavesdropper's reflection factor α, and the number of iterations N in the alternating optimization algorithm. BCD Estimated azimuth of the eavesdropper Base station beamforming auxiliary variables Eavesdropper signal-to-interference-plus-noise ratio (SINR) auxiliary variable Perception of the Cramerau boundary auxiliary variable t 0 S-procedure auxiliary variables Maximum number of iterations N for continuous convex approximation algorithm SCA The iteration index of the alternating optimization algorithm is n=0, and the iteration index of the continuous convex approximation algorithm is m=0.

[0075] Step 2: Mathematically model the downlink network physical layer security problem of the ISAC system under the condition of imperfect eavesdropper CSI. This model integrates the ISAC sensing capability into the norm boundary of imperfect CSI.

[0076] The CSI stands for Channel State Information;

[0077] ISAC is an integrated communication and sensing system.

[0078] Step 3: Use the continuous convex approximation algorithm on a fixed... t n , Solve for the auxiliary variable γ of the signal-to-interference-plus-noise ratio of the eavesdropper. k and S-procedure auxiliary variable p k Let m = m + 1 and And proceed to step four;

[0079] in, Wn n is the eavesdropper-to-signal-plus-noise ratio auxiliary variable in the mth successive convex approximation algorithm iteration, is the eavesdropper-to-signal-plus-noise ratio auxiliary variable in the mth successive convex approximation algorithm iteration,

[0080] Step four: if the successive convex approximation algorithm iteration index satisfies m≥N SCA , let m = 0, and go to Step five if m < N SCA , go to Step three;

[0081] where is the eavesdropper-to-signal-plus-noise ratio auxiliary variable in the n th alternating optimization algorithm iteration, is the S-procedure auxiliary variable in the n th alternating optimization algorithm iteration;

[0082] Step five: solve the base station beamforming auxiliary variable Wn and the perceived Cramér-Rao bound auxiliary variable t using the successive convex approximation algorithm with Wn k and t fixed, let m = m + 1 and t m = t; go to Step six;

[0083] where is the eavesdropper-to-signal-plus-noise ratio auxiliary variable in the n th alternating optimization algorithm iteration, is the S-procedure auxiliary variable in the n th alternating optimization algorithm iteration, is the base station beamforming auxiliary variable in the mth successive convex approximation algorithm iteration;

[0084] Step six: if the successive convex approximation algorithm iteration index satisfies m≥N SCA , let t n+1 = t m , m = 0, n = n + 1 and go to Step seven if m < N SCA , go to Step five;

[0085] where is the base station beamforming auxiliary variable in the n + 1th alternating optimization algorithm iteration, t n+1 is the perceived Cramér-Rao bound auxiliary variable in the n + 1th alternating optimization algorithm iteration;

[0086] Step seven: if the alternating optimization algorithm iteration index satisfies n≥N BCD , the algorithm ends and outputs otherwise go to Step three.

[0087] Specific implementation two: the difference between this implementation and specific implementation one is that the step two is as shown in the formula (1) based on the mathematical modeling method of the downlink network physical layer security problem of the ISAC system under the non-perfect eavesdropper CSI condition; The specific process is: Figure 1 The specific process is:

[0088] Assuming that the base station sends a signal Where ω k is the beamforming vector, s k [l] is the transmitted symbol. A total of L time slots can be used for the azimuth estimation of the eavesdropper. There are a total of K users and one eavesdropper. The channel between the base station and Bob is denoted as The channel between the base station and the eavesdropper is denoted as And assuming is the line-of-sight propagation channel, then

[0089]

[0090] Where β E is the fading at the reference distance, d AE is the distance from the base station to the eavesdropper, α E is the path loss exponent, θ is the azimuth angle, is the direction vector of the transmitting antenna. Since there is an estimation error in the azimuth angle, the azimuth angle is re-expressed as Where is the estimated value of the azimuth angle, and Δθ is the azimuth angle estimation error, which is related to the sensing ability of the ISAC system. Therefore, the achievable rate of the legitimate user and the eavesdropper can be expressed as:

[0091]

[0092]

[0093] Then the secrecy rate can be expressed as From the perspective of sensing, the Cramer-Rao bound of the azimuth angle can be expressed as:

[0094]

[0095] Where, is the covariance matrix of beamforming, A(θ)=b(θ)a H (θ), is the gradient of A(θ), b(θ) is the direction vector of the receiving antenna, is the radar receiving noise.

[0096] In summary, the uncertainty of the eavesdropper CSI is caused by the estimation bias of the azimuth angle, and the azimuth angle estimation bias is related to its Cramer-Rao bound. If it is assumed that Δθ follows a Gaussian distribution, then θ falls in the interval with probability 99.73%, and a(0) can be re-expressed as:

[0097]

[0098] and has:

[0099]

[0100] where m(n) = (N t -1) / 2-n and The set expression of the non-perfect CSI two-norm boundary can be obtained from the Taylor formula, i.e.:

[0101]

[0102]

[0103] where,

[0104] The Pareto boundary of the downlink physical layer security optimization model based on the ISAC system is defined as follows:

[0105]

[0106] Then, according to the above Pareto boundary, the following optimization problem can be formed:

[0107]

[0108] where Γ CRB is the maximum Kullback-Leibler bound, P t is the total power of the base station transmission. Let H AE = h AE (h AE ) H , Then the problem can be simplified as:

[0109]

[0110] where

[0111]

[0112] The rank 1 constraint is relaxed, and a relaxation variable γ k is introduced. By using the S-procedure theorem and combining with the Schur complement theorem, the following optimization problem form can be obtained:

[0113]

[0114] s.t.t≥t0

[0115]

[0116]

[0117]

[0118]

[0119]

[0120]

[0121] Where p k For S-procedure parameters,

[0122] The other steps and parameters are the same as in Specific Implementation Method 1.

[0123] Specific Implementation Method Three: This implementation method differs from Specific Implementation Method One or Two in that step three utilizes a continuous convex approximation algorithm on a fixed... t n , Solve for the auxiliary variable γ of the signal-to-interference-plus-noise ratio of the eavesdropper. k and S-procedure auxiliary variable p k The specific process is as follows:

[0124] Considering only the auxiliary variables of the eavesdropper's signal-to-dryness ratio and the S-procedure, we can obtain the following from the first-order Taylor formula:

[0125]

[0126] The problem can then be simplified to:

[0127]

[0128] The problem is a convex problem, which can be solved using the convex optimization toolbox.

[0129] Other steps and parameters are the same as in specific implementation method one or two.

[0130] Specific Implementation Method Four: This implementation method differs from Specific Implementation Methods One to Three in that step five utilizes a continuous convex approximation algorithm on a fixed... Solving for the base station beamforming auxiliary variable W under the following conditions k And the auxiliary variable t for perceiving the Cramerau bound; the specific process is as follows:

[0131] Considering only the auxiliary variables of base station beamforming and the auxiliary variables of the sensing Cramerau boundary, we can obtain the following from the first-order Taylor formula:

[0132]

[0133] Regarding W k The rank-1 constraint is equivalent to ||W k || * -||W k ||2≤0, where||·|| * Let ||·||² be the nuclear norm and the spectral norm, respectively. To obtain a higher quality rank-1 solution, the equivalent conditions of the rank-1 constraint can be expanded using a first-order Taylor series and penalized into the objective function. The problem can then be simplified to:

[0134]

[0135] Where ρ > 0 is the penalty factor, μ max (·) represents the eigenvector corresponding to the largest eigenvalue. Since the problem is convex, it can be solved efficiently using the convex optimization toolbox.

[0136] The other steps and parameters are the same as those in one of the specific implementation methods one to three.

[0137] Table of Symbols Explanation

[0138]

[0139]

[0140]

[0141]

[0142] The beneficial effects of the present invention are verified using the following embodiments:

[0143] Example 1:

[0144] This example considers three legitimate users and a single eavesdropper. The base station is equipped with 12 transmit antennas and 14 receive antennas. The reference distance fading is set to 46 dB, the Rice factor to 5, the path loss exponent parameter to 2, the noise power to -70 dBm, the maximum transmit power to 30 dBm, and the number of signal frames used to estimate the eavesdropper's azimuth angle is 30. The signal-to-noise ratio (SNR) of the echo signal is assumed to be SNR. R =|α| 2 LP t / σ R This example tests SNR. R = -11dB and SNR R Algorithm performance at two signal-to-noise ratios of -10dB.

[0145] likeFigure 2 As shown in FIG. 6, the algorithm framework proposed in the present application realizes convergence through 10 iterations under various sensing requirements. Figure 3 As shown in FIG. 7, the algorithm proposed in the present application successfully calculates the Pareto boundary of the downlink secure communication system based on ISAC, and the secure communication performance is improved as the sensing requirement decreases. Figure 4 As shown in FIG. 8, the algorithm proposed in the present application exhibits higher secure communication rate performance under higher transmission power conditions.

[0146] The present application can also have other various embodiments, and those skilled in the art can make various corresponding changes and modifications according to the present application without departing from the spirit and essence of the present application, and these corresponding changes and modifications shall all belong to the protection scope of the claims attached to the present application.

Claims

1. A downlink network physical layer security optimization method based on an ISAC system, characterized in that: The method specifically comprises the following steps: Step 1: Initialize parameters, including number of transmit antennas , number of receive antennas , number of users , Bob-to-base station channel , base station-to-eavesdropper channel estimates , communication noise variance , radar noise variance , eavesdropper reflection factor , number of iterations of alternating optimization algorithm , eavesdropper azimuth angle estimate , base station beamforming auxiliary variable , eavesdropper signal-to-jamming-and-noise ratio auxiliary variable , sensing Cramér-Rao bound auxiliary variable , S-procedure auxiliary variable , maximum number of iterations of successive convex approximation algorithm , iteration index of alternating optimization algorithm , and iteration index of successive convex approximation algorithm ; Step two: a mathematical modeling of the downlink network physical layer security problem based on the ISAC system under the non-perfect eavesdropper CSI condition is performed, and the Cramer-Rao bound of the eavesdropper azimuth angle estimation value is integrated into the norm boundary of the non-perfect CSI; The CSI is channel state information; The ISAC is an integrated communication and sensing system; Step three: solve the eavesdropper SINR auxiliary variable , , and the S-procedure auxiliary variable using the successive convex approximation algorithm with the , let and , , and perform step four; where is the th successive convex approximation algorithm iteration within the S-procedure. in, For the first n Base station beamforming auxiliary variables in the iteration of the alternating optimization algorithm For the first n Perceived Cramerau bound auxiliary variables in the iteration of the alternating optimization algorithm For the first m The signal-to-interference-plus-noise ratio (SINR) auxiliary variable for eavesdroppers in the iteration of the subcontinuous convex approximation algorithm; Step four: if the index of the iteration number of the successive convex approximation algorithm satisfies , let , , , and go to step five, if , go to step three; wherein, is the eavesdropper-to-signal-plus-noise ratio auxiliary variable in the kth iteration of the alternating optimization algorithm, n is the eavesdropper-to-signal-plus-noise ratio auxiliary variable in the kth iteration of the alternating optimization algorithm, is the S-procedure auxiliary variable in the kth iteration of the alternating optimization algorithm, n is the S-procedure auxiliary variable in the kth iteration of the alternating optimization algorithm; Step five: solve the base station beamforming auxiliary variables , , and the sensing CRLB auxiliary variables under the condition that , and , ; wherein, is the sensing CRLB auxiliary variable in the th successive convex approximation algorithm iteration; perform step six; in, For the first n The eavesdropper signal-to-interference-plus-noise ratio (SIN / N) auxiliary variable in the alternating optimization algorithm iteration For the first n S-procedure auxiliary variables in the alternating optimization algorithm iteration For the first m Auxiliary variables for base station beamforming in the iteration of the subcontinuous convex approximation algorithm; Step six: if the index of the iteration number of the successive convex approximation algorithm satisfies , let , , , and go to step seven, if , go to step five; wherein, is the base station beamforming auxiliary variable in the kth iteration of the alternating optimization algorithm, n+ 1is the perception Cramér-Rao bound auxiliary variable in the kth iteration of the alternating optimization algorithm, is the base station beamforming auxiliary variable in the kth iteration of the alternating optimization algorithm, n+ 1is the perception Cramér-Rao bound auxiliary variable in the kth iteration of the alternating optimization algorithm, Step seven: if the index of the iteration number of the alternating optimization algorithm satisfies , then the algorithm ends and outputs , otherwise, go to step three.

2. The method of ISAC system based downlink network physical layer security optimization according to claim 1, characterized in that: The mathematical modeling method of the downlink network physical layer security problem based on the ISAC system under the non-perfect eavesdropper CSI condition in step two specifically comprises the following steps: The base station transmits a signal wherein is a beamforming vector, is a transmitted symbol; a total of slots are used for position estimation of the eavesdropper; a total of users, one eavesdropper; Let the channel between the base station - Bob be denoted by the channel between the base station - the eavesdropper be denoted by and set as a line-of-sight propagation channel, （1） where is the fading at the reference distance, is the distance from the base station to the eavesdropper, is the path loss exponent, is the azimuth angle, is the direction vector of the transmit antenna; due to the estimation error in the azimuth angle, the azimuth angle is re-expressed as where is the estimated value of the azimuth angle, is the estimation error of the azimuth angle, which is related to the sensing capability of the ISAC system; thus the achievable rates of the legitimate user and the eavesdropper are expressed as: （2） （3） The secrecy rate is expressed as From the perspective of perception, the Cramér-Rao bound of azimuth is expressed as: （4） wherein a covariance matrix for beamforming, , is a gradient of a direction vector of a receive antenna, a radar noise variance; The uncertainty of the eavesdropper's CSI is caused by the estimation bias of the azimuth angle, which is related to its Cramer-Rao bound, denoted as subject to a Gaussian distribution, The probability of falling into the interval is 99.73%, and is re-written as: （5） And （6） wherein and The set expression of the non-perfect CSI two-norm boundary is obtained by Taylor formula, that is: （7） （8） wherein, ; defining a pareto boundary of downlink physical layer security optimization model based on ISAC system: （9） According to the above Pareto boundary, the following optimization problem is formed: （10） wherein is the maximum Cramer-Rao bound, is the total power transmitted by the base station; let , , , （11） Wherein （12） Rank 1 constraint relaxation, introducing relaxation variables Using S-procedure theorem and combining with Schur complement theorem, the following optimization problem is obtained: （13） wherein is an S-procedure auxiliary variable, .

3. The method of ISAC system based downlink network physical layer security optimization according to claim 2, characterized in that: In step three, the continuous convex approximation algorithm is used to fix... , , Solving for the signal-to-interference-plus-noise ratio auxiliary variable of the eavesdropper. and S-procedure auxiliary variables The specific process is as follows: In the case of only considering the auxiliary variable of the eavesdropper signal-to-noise ratio and the S-procedure auxiliary variable, the first-order Taylor formula can be obtained as follows: （14） （15） Solve by using a convex optimization toolbox.

4. The method of ISAC system based downlink network physical layer security optimization according to claim 3, characterized in that: The step five solves the base station beamforming auxiliary variable , , and the sensing Cramer-Rao bound auxiliary variable under the condition of fixed ; the specific process is as follows: In the case of only considering the auxiliary variable of the base station beamforming and the auxiliary variable of the sensing Cramer-Rao bound, the first-order Taylor formula can be obtained as follows: （16） And the rank 1 constraint on is equivalent to where is the nuclear norm and the spectral norm; to obtain a higher quality rank 1 solution, the equivalent condition of the rank 1 constraint is utilized by a first order Taylor expansion and penalized into the objective function, （17） wherein is a penalty factor, is the eigenvector corresponding to the largest eigenvalue; thus efficiently solved by using the convex optimization toolbox.

Images