A method for integrated sensing and privacy protection suitable for dual-selected channels

Abstract

The application discloses a kind of suitable for dual selection channel's communication and sensing integrated sensing privacy protection method, comprising: (1) communication sensing integrated equipment transmitting end adopts orthogonal time-frequency space OTFS modulation multiple frame length is the super-Gaussian constellation symbol vector of It is continuously transmitted multiple frame discrete time domain communication sensing integrated signal to communication user and sensing target;(2) communication sensing integrated equipment processor utilizes the transmission signal known in step (1) to build each frame corresponding overcomplete sensing dictionary;(3) communication sensing integrated equipment receiver continuously receives the observation signal of multiple OTFS frames;(4) receiver judges whether there is distance migration according to system parameter and the speed range of to be sensed target;(5) communication sensing integrated equipment processor utilizes the overcomplete dictionary built in step (2) and adopts corresponding sparse bayesian method to process multiple frame received signal in the foregoing step (3), and obtains delay-doppler diagram, extracts target delay-doppler parameter.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication and radar sensing technology, and particularly relates to a privacy protection method for integrated sensing and communication suitable for dual-selection channels. Background Technology

[0002] Integrated Sensing and Communication (ISAC) significantly improves spectrum efficiency and hardware utilization by simultaneously implementing communication and sensing functions on a unified signal waveform and hardware platform, making it one of the key enabling technologies for sixth-generation mobile communication (6G). In a communication-centric ISAC architecture, the transmitted signal is used by legitimate receivers (such as base stations) for target detection, parameter estimation, and environmental perception while simultaneously transmitting the communication payload. However, due to the broadcast nature of wireless channels, this signal may also be intercepted by non-cooperative third-party receivers and used for passive sensing, thereby leaking sensitive information such as target location and speed, posing a serious threat to sensing privacy and security.

[0003] To address the above problems, existing technologies mainly focus on the following three directions:

[0004] First, artificial noise and beamforming methods. These methods reduce the perceived signal-to-interference-plus-noise ratio (SNR) in the direction of eavesdropping by injecting artificial noise in a specific direction into the transmitted signal or designing beamforming weight vectors. However, these methods all assume that the transmitter knows the eavesdropper's Channel State Information (CSI) in order to precisely align the artificial noise with the null space of the eavesdropping channel or suppress the beam gain in the direction of eavesdropping. However, in passive sensing scenarios, the eavesdropper only receives signals and does not actively transmit, making it impossible to obtain their CSI; therefore, the above solutions are not applicable.

[0005] Second, fuzzy function engineering methods. These methods optimize the frequency domain characteristics of the transmitted signal, causing its fuzzy function to exhibit high-peak sidelobes in a specific time delay region, thereby reducing the eavesdropper's ability to perceive and distinguish it. However, at the legitimate receiver, fuzzy function engineering typically suppresses artificial sidelobes based on reciprocal filtering, i.e., deconvolving the received signal with the known transmitted signal. The essence of reciprocal filtering is to take the linear inverse operation of the channel mapping. In a double-selective channel (DSC), the channel simultaneously exhibits delay spread and Doppler spread; the dimension of its delay-Doppler channel mapping is... ,in The number of signs is [number], while the length of a single observation is only [number]. The corresponding inverse problem is severely ill-conditioned, and at this point, the reciprocity filter does not have a closed-form solution. Therefore, the fuzzy function engineering method fails under dual-selection channels.

[0006] Third, multi-frame coherent accumulation methods. These methods improve the sensing signal-to-noise ratio by extending the observation time and coherently accumulating the received signals from multiple consecutive frames. To reduce the impact of randomness on sensing performance, traditional coherent accumulation requires the target to remain stationary and the echo complex amplitude to remain approximately constant during the accumulation period. However, in high-mobility scenarios such as Vehicle-to-Everything (V2X) communication, high-speed rail, and low-Earth orbit satellites, the target echo complex amplitude typically follows a Swerling type II or IV distribution, which changes between frames, causing traditional coherent accumulation to fail due to incoherent echo amplitudes. Furthermore, when inter-frame range migration occurs—that is, when the radial displacement of the target exceeds one range resolution cell between multiple frames—phase mismatch occurs, further degrading performance.

[0007] In summary, existing technologies struggle to achieve effective perception privacy protection in passive sensing scenarios with dual-selection channels. Therefore, there is an urgent need for a signal design and parameter estimation method that is independent of eavesdropper CSI, fully compatible with communication payloads, and adaptable to high-mobility scenarios. Summary of the Invention

[0008] The purpose of this invention is to address the problems existing in the prior art. This invention proposes a privacy protection method for integrated sensing and communication in dual-selection channels. The method includes two parts: signal design and target delay-Doppler estimation, and can be used in single-station integrated sensing and communication scenarios.

[0009] Technical Solution: To achieve the objectives of this invention, this invention proposes a sensing privacy protection method suitable for dual-selection channels, which includes the following steps:

[0010] (1) The transmitting end of the integrated communication and sensing equipment adopts orthogonal time-frequency-space (OTFS) modulation with a multi-frame length of [missing information]. The super-Gaussian constellation symbol vector, and continuously transmit multiple frames of length. The discrete-time domain communication and sensing integrated signal is transmitted to the communication user and the sensing target;

[0011] (2) The processor of the integrated communication and sensing device uses the known transmission signals in step (1) to construct an overcomplete sensing dictionary corresponding to each frame;

[0012] (3) The receiver of the integrated communication and sensing equipment continuously receives observation signals of multiple OTFS frames;

[0013] (4) The receiver of the integrated communication and sensing equipment determines whether there is distance migration based on the system parameters and the speed range of the target to be sensed;

[0014] (5) Based on the aforementioned distance migration judgment, the processor of the communication sensing integrated device uses the overcomplete dictionary constructed in step (2) to process the multi-frame received signal in step (3) using the corresponding sparse Bayesian method, and obtains the time delay-Doppler map and extracts the target time delay-Doppler parameters.

[0015] Furthermore, in step (1), the transmitting end continuously transmits multiple frames with a length of The super-Gaussian OTFS notation is as follows:

[0016] (1.1) Super-Gaussian constellations are generated using probabilistic constellation shaping (PCS). The kurtosis of its probability distribution ,in, This indicates the calculation of mathematical expectation. The modulus of a complex number The super-Gaussian constellation is a symbol in the constellation. It adopts a multi-ring amplitude phase shift keying structure. The constellation points on each ring have a uniform phase distribution. The probability of each ring is distributed according to a preset ratio, so that the probability of the constellation points appearing follows a super-Gaussian distribution.

[0017] One specific implementation method is to adopt... A concentric amplitude phase shift keying (APSK) structure is used, where the constellation points on each ring have a uniform phase distribution. The probability of each ring is allocated according to a preset ratio, so that the probability of constellation point occurrence follows a super-Gaussian distribution. For example, a four-ring 64-APSK structure is used, where each of the four rings contains 16 constellation points, and the ring radii are... , , and The probability ratios are respectively , , , The kurtosis of the obtained super-Gaussian constellation is Alternatively, a double-ring 4-APSK structure can be adopted, with an inner ring radius of... Outer ring radius The inner ring probability is 12 times the outer ring probability, and the kurtosis of the resulting super-Gaussian constellation is... .

[0018] (1.2) Each frame of transmission contains Independent and identically distributed communication symbol vectors ,in , and The grid numbers for the time delay axis and the Doppler axis, respectively, and the sign in the vector. Constellations generated in step (1.1) The symbol vector is transmitted through the OTFS modulation base. Mapped to time-domain transmitted signal ,in, for Point discrete Fourier transform matrix, Represents the Kronecker product. for An identity matrix of order 1. This indicates the transpose operation. This indicates the conjugate transpose operation.

[0019] In each frame, the communication symbol vector Each transmission is performed independently, therefore the transmitted signal in each frame... Also, implement it independently once, and the first The implementation of a frame is denoted as .

[0020] (1.3) In the first Frame time-domain transmitted signal A cyclic prefix (CP) is added before the transmission frame to form a complete transmit frame, which is then up-converted to the carrier frequency by the radio frequency front end. Post-transmission, system parameters include: carrier frequency ,bandwidth Symbol period CP length Frame period .

[0021] Furthermore, in step (2), the processor of the integrated communication and sensing device constructs an overcomplete sensing dictionary corresponding to each frame using the known transmission signals from step (1), as follows: based on the accurate first... Frame transmission signal Building an ultra-complete perceptual dictionary Its column vector is defined as:

[0022] ;

[0023] in, Cyclic delay matrix, Doppler frequency shift matrix, For two-dimensional time-delay-Doppler indexing Mapping to one-dimensional dictionary column indexes.

[0024] Furthermore, in step (3), the integrated communication sensing device receiver continuously receives observation signals from multiple OTFS frames, as follows: the integrated communication sensing device receiver and transmitter are co-located and receive target echo signals reflected through a dual-selection channel, wherein the first... The echo of each target is represented by a cyclic delay operator. and Doppler operator The cascading effect, i.e.

[0025] ;

[0026] in, For the target total number, For the first The complex reflection coefficient of a target, and The first Discrete time delay index and Doppler index of each target, Cyclic delay matrix, Doppler frequency shift matrix, The vector is an additive complex Gaussian white noise. For the first The observed signal of the frame.

[0027] Furthermore, in step (4), the receiver of the integrated communication and sensing device determines whether there is distance migration based on the system parameters and the speed range of the target to be sensed, as follows:

[0028] If the target's maximum radial velocity satisfy

[0029] ;

[0030] in, For wave speed, For distance resolution, If the number of OTFS frames received in step (3) is less than the number of OTFS frames received, then it is determined that there is no distance migration; otherwise, it is determined that there is distance migration.

[0031] Furthermore, in step (5), based on the aforementioned distance migration judgment, the processor of the integrated communication and sensing device uses the corresponding sparse Bayesian method to process the multi-frame received signals in step (3). and the overcomplete dictionary constructed in step (2) The time delay-Doppler image is obtained, and the target time delay-Doppler parameters are extracted from it, as follows:

[0032] (5.1) Set the target parameter covariance vector for each frame The relationship between them When there is no distance migration, the relationship is:

[0033] ;

[0034] in, Let be the target parameter covariance vector of the first frame; when distance migration exists, let be the discrete delay index in the first frame. And Doppler Index The corresponding distance-velocity pair is Then in the first In the frame, due to the target's motion, the distance-velocity pair evolves into ,in, The frame period is used to obtain the target parameter support covariance vector. The relationship between them is

[0035] ;

[0036] in, Let covariance vector of target parameters in frame 1 be the permutation matrix. (where the permutation matrix corresponding to the first frame is) An identity matrix of order 1, denoted as )satisfy

[0037] ;

[0038] in, For the first Standard basis vectors For the first A standard basis vector, mapping Delay-Doppler indexing of the first frame Mapped to the Frame index:

[0039] ;

[0040] in, No. The index of the target parameter in the frame corresponds to the range-velocity pair. Specifically, for the single-station sensing scenario considered in this invention, the first... Discrete Delay Index in Frame And Doppler Index The corresponding distance-velocity pair is : Then in the first frame( In the above, due to the target's movement, the distance becomes... ,get ;

[0041] (5.2) Calculate the received signal for each frame. covariance matrix , ;

[0042]

[0043] in, This represents a diagonal matrix with the input vector as its diagonal elements. For the first Frame noise variance;

[0044] (5.3) Solve the similarity maximization problem to obtain the result in step (5.1). Estimate:

[0045] ;

[0046] in, Represents the trace of a matrix. Represents the determinant of a matrix. Represents the natural logarithm. This represents the Moore-Penrose pseudoinverse of the matrix. The above problem can be solved using Mackay iteration. For the case where there is no distance migration, the following iterative solution is used:

[0047] ;

[0048] in, For the first The first frame dictionary List, To control the exponent of convergence speed, express The result of the last iteration, express The result of iterative updates. The part of the above method corresponding to distance-free migration is collectively referred to as the MD-SBL method;

[0049] For cases involving distance migration, the following iterative solution is used:

[0050]

[0051] in, For the first The first frame dictionary List, To control the exponent of convergence speed, express The result of the last iteration, express The result of iterative updates. The parts of the above methods that involve distance migration are collectively referred to as the RMC-MD-SBL method;

[0052] (5.4) After iterative convergence, the covariance estimate of the target parameters in the first frame is obtained. Constructing a time-delay-Doppler plot:

[0053]

[0054] Extracting the peak position from the time-delay-Doppler map yields the target's time-delay index estimate. And Doppler index estimation .

[0055] Beneficial effects: Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0056] (1) The integrated sensing signal design scheme based on super-Gaussian constellation and OTFS modulation base proposed in this invention significantly broadens the difference in sensing performance between legitimate sensing end and illegal eavesdropping end, thereby providing protection for sensing privacy;

[0057] (2) While suppressing the perception performance of illegal eavesdropping terminals, reliable perception of legitimate terminals under dual-selection channels is achieved: In traditional methods, reliable perception requires a long coherence time and stable target intensity fluctuation characteristics to ensure sufficient coherence accumulation times. Therefore, it cannot be applied to scenarios with drastic target intensity fluctuations and short coherence times in dual-selection channels. This invention achieves coherence accumulation within short coherence intervals through a multi-dictionary sparse Bayesian method, and proposes corresponding compensation methods for scenarios with distance migration, thus achieving coherence accumulation in such high-dynamic scenarios and ensuring reliable perception performance;

[0058] (3) The entire solution of the present invention does not rely on the eavesdropper's CSI, geographical location or any prior information. It only utilizes the statistical characteristics of communication signals and the precise known prior information of the transmission signals by the integrated sensor device processor to achieve perceptual privacy protection at the physical layer, and has inherent universality against passive eavesdropping attacks. Attached Figure Description

[0059] Figure 1 This is a schematic diagram of the integrated sensing scenario for privacy protection under dual-channel sensing as described in this invention;

[0060] Figure 2 This is the overall signal processing flowchart of the present invention;

[0061] Figures 3-4 This is a comparison curve of the successful recovery probability of the legitimate receiver and the eavesdropper as described in this invention, as a function of the reference channel SINR.

[0062] Figures 5-6 This is a comparison curve of the successful recovery probability of the legitimate receiver and the eavesdropper as described in this invention, as a function of channel SNR.

[0063] Figures 7-8 This is a comparison curve of the speed and distance estimation RMSE of the RMC-MD-SBL described in this invention under a super-Gaussian constellation as a function of channel SNR in a high-dynamic scenario. Detailed Implementation

[0064] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0065] Reference Figure 1 This invention considers a typical single-site integrated sensing scenario, comprising an integrated sensing device transmitter, an integrated sensing device receiver (Alice), and an eavesdropping node (Eve). The transmitter and Alice are co-located at the base station, while Eve is a passive eavesdropping node equipped with an independent receiving channel. The transmitted signal is reflected by a dual-selection channel and simultaneously received by Alice and Eve. Alice has precise knowledge of the transmitted signal, while Eve can only obtain a noisy copy of the reference signal through a reference channel.

[0066] like Figure 2 The diagram shows the overall signal processing flowchart of the technology in this patent application. This invention proposes a sensing privacy protection method suitable for dual-selection channels. First, the transmitter generates a super-Gaussian constellation symbol through probabilistic constellation shaping and transmits a sensing integrated signal via OTFS modulation. Then, the processor constructs an overcomplete sensing dictionary using the known transmitted signal. The receiver receives echo signals from multiple OTFS frames. Afterward, the processor determines whether distance migration exists based on system parameters and the target velocity range, and selects the corresponding sparse Bayesian processing method for parameter estimation to obtain the time delay-Doppler map, extracting the target time delay and Doppler parameters. Two complete embodiments are given below, one for scenarios with and without distance migration, each containing specific implementation steps and corresponding Monte Carlo simulation verification.

[0067] Example 1: Verification and Implementation of Perception Performance in a Distance-Free Migration Scenarios

[0068] This embodiment describes the specific implementation process of the method of the present invention in a distance-free migration scenario, and is used to illustrate how the method of the present invention can maintain the performance of legitimate perception and suppress eavesdropping perception under the conditions of low target radial velocity and consistent sparse support in each frame, through super constellation signal design and MD-SBL multi-frame coherent accumulation.

[0069] In accordance with step (1) of the corresponding technical solution, in this embodiment, the transmitting end constructs a multi-frame OTFS integrated communication and sensing signal based on the super-Gaussian constellation, and specifically performs the following steps:

[0070] Step (1.1) Generate the super-Gaussian constellation.

[0071] Super-Gaussian constellations are generated using Probabilistic Constellation Shaping (PCS). The kurtosis of its probability distribution The specific implementation uses a four-ring 64-APSK structure: each of the four rings contains 16 constellation points, and the constellation points on each ring have a uniform phase distribution. The ring radii are set as follows: , , and The probability ratio of each ring is approximately The resulting super-Gaussian constellation kurtosis It is greater than the kurtosis value of the Gaussian distribution. It belongs to a typical super-Gaussian distribution.

[0072] The mathematical description of constellation point generation is: in each frame, the first... Communication symbols Among them, amplitude The discrete probability distribution mentioned above is taken from... phase exist It is evenly distributed on the surface.

[0073] As an alternative, a double-ring 4-APSK structure can also be used: inner ring radius Outer ring radius The inner ring probability is equal to the outer ring probability. Times, the resulting constellation kurtosis This embodiment uses the four-ring 64-APSK as the main solution.

[0074] Step (1.2): Modulate using the OTFS modulation base to generate a time-domain transmit signal.

[0075] Each frame of transmission contains an independent and identically distributed vector of communication symbols. ,in and These are the grid numbers for the time delay axis and the Doppler axis, respectively. Total number of symbols Each symbol in the vector comes from the super-Gaussian constellation generated in step (1.1). symbol vector Through OTFS modulation base Mapped to time-domain transmitted signals:

[0076] ;

[0077] in, for Point discrete Fourier transform matrix, for An identity matrix of order 1. Represents the Kronecker product. This represents the conjugate transpose operation. In each frame, the communication symbol vector is generated independently and randomly once; therefore, the transmitted signal in each frame... Independent of each other .

[0078] Step (1.3): Append the cyclic prefix and fire.

[0079] In the Frame time-domain transmitted signal The preceding additional length is The cyclic prefix (CP) is used to form a complete transmit frame. CP length The number of sampling points should be greater than the number corresponding to the maximum target delay to ensure that there is no inter-frame interference after the receiver removes the CP, and that each frame is up-converted to the carrier frequency by the RF front-end. Launched later.

[0080] The system parameter settings for this embodiment are as follows: carrier frequency GHz, bandwidth MHz, sampling interval s, CP length Frame period s, number of transmission frames .

[0081] Step (2), construct the perception dictionary.

[0082] Alice, the legitimate receiver, utilizes the known first... Frame transmission signal , build Dimensional overcomplete perception dictionary Each column of the dictionary corresponds to a grid point in the time-delay-Doppler domain, defined as follows:

[0083] ;

[0084] Where k is the discrete time delay index and q is the discrete Doppler index. Cyclic delay matrix, Doppler frequency shift matrix, This is the Doppler frequency shift matrix. This represents the mapping from a two-dimensional time-delay-Doppler index to a one-dimensional dictionary column index.

[0085] The number of columns in the dictionary is Much larger than the number of rows This constitutes a typical underdetermined sparse recovery problem. Because The communication symbols for each frame are generated independently, and the dictionary for each frame... They are also different from each other. Node Eve obtains a noisy copy of the reference signal through the reference channel. ,in, Eve's dictionary is constructed using the reference channel noise as a reference, but due to the noise in the reference signal, Eve's dictionary is biased.

[0086] Step (3): Receive multiple frames of observation signals:

[0087] Alice continuously receives The observed signals of OTFS frames. There are a total of [number] frames in the scene. One target to be perceived (of which) Strong goals and (The complex reflection coefficient amplitude of a weak target is half that of a strong target). Delay index of each target And Doppler Index Randomly generated within the valid range, the first... The received signal after CP removal is:

[0088] ;

[0089] in, For the target total number, For the first The complex reflection coefficient of a target, and The first The discrete time delay index and Doppler index of each target are used as the cyclic time delay and Doppler frequency shift, respectively. Cyclic delay matrix, Doppler frequency shift matrix, It is an additive complex Gaussian white noise vector. For the first The observed signal of the frame.

[0090] In this embodiment, the target is in Within the short coherence interval of a frame, its delay index And Doppler Index The complex reflection coefficients remain unchanged (i.e., sparse support is consistent), but the complex reflection coefficients of each frame are... It allows for frame-dependent variation and conforms to Swerling Type II and IV target fluctuation models.

[0091] Step (4): Distance migration determination:

[0092] The integrated communication and sensing device determines whether distance migration exists based on system parameters and the velocity range of the target to be sensed. The judgment criterion is: if the target's maximum radial velocity... satisfy:

[0093]

[0094] Then it is determined that there is no distance migration, where, m / s is the speed of electromagnetic wave propagation. For bandwidth, For frame number, The frame period. Substituting the parameters into this embodiment: MHz, , s, the critical velocity can be obtained as m / s. In most high-mobility ground scenarios (including high-speed rail with a maximum speed of approximately...), m / s, V2X communication scenario approximately In m / s), this condition can be met. When the condition is met, the sparse support of each frame is consistent, and the MD-SBL processing flow of step (5) is entered. If the condition is not met (i.e. there is distance migration), the process is transferred to the RMC-MD-SBL flow of step (5).

[0095] Step (5): Multi-frame coherent accumulation and delay-Doppler parameter extraction based on MD-SBL.

[0096] Step (5.1) establish the common covariance prior of the multi-dictionary SBL.

[0097] In the scenario of no-distance migration, the sparse support of the target is consistent across frames, therefore a common covariance prior is established: let the covariance matrix of the sparse coefficient vectors of each frame be... satisfy

[0098]

[0099] in, for Frame sharing A common covariance vector. The physical meaning of this prior is: although the complex reflection coefficients of each frame are different due to target fluctuations (manifested as...) The non-zero value can vary), but the target's position in the time-delay-Doppler domain (i.e., the ...) The position of non-zero elements remains consistent across frames.

[0100] Step (5.2) Calculate the covariance matrix of the received signal for each frame.

[0101] No. Frame received signal In a given The covariance matrix is ​​as follows:

[0102] ;

[0103] in, For the first frame noise variance for An identity matrix of order 1.

[0104] Step (5.3) solves for the common covariance vector through the Mackay iteration of MD-SBL.

[0105] Common covariance vector The estimate is obtained by maximizing the joint evidence (Type-II likelihood), which is equivalent to minimizing the following objective function:

[0106] ;

[0107] in, Represents the trace of a matrix. Represents the determinant of a matrix. Represents the natural logarithm. This represents the Moore-Penrose pseudoinverse of the matrix. The above problem can be solved using Mackay iteration; the specific steps are as follows:

[0108] Calculate the received signal for each frame covariance matrix , ;

[0109]

[0110] in, This represents a diagonal matrix with the input vector as its diagonal elements. For the first The noise variance of the frame; for the case where there is no distance migration, the following method is used for iterative solution:

[0111] ;

[0112] in, For the first The first frame dictionary List, To control the exponent of convergence speed, express The result of the last iteration, express The results of iterative updates, the part of the above methods corresponding to distance-free migration, are collectively referred to as the MD-SBL method;

[0113] Step (5.4): Construct the time delay-Doppler plot and extract the parameters.

[0114] After iterative convergence, the common covariance estimate is obtained. Reorganize it into Delay-Doppler Map (DDMap):

[0115]

[0116] Extract from DDMap By identifying the peak positions, the target's time delay index can be estimated. And Doppler index estimation Convert discrete indices to physical parameters:

[0117]

[0118] in, For the first The estimated distance to each target. To estimate the radial velocity.

[0119] The performance of the method described in this embodiment is verified by Monte Carlo simulation. The simulation parameters are set as follows: OTFS symbol count. ( ), CP length Carrier frequency GHz, bandwidth MHz. The scene contains 2 strong targets and 3 weak targets, with the complex reflection coefficient amplitude of the weak targets being half that of the strong targets. Target delays and Doppler indices are randomly generated within the effective range for each frame, and the complex reflection coefficients vary independently with each frame, following a Swerling II / IV fluctuation model. Monte Carlo simulations are performed at each point. Each independent trial is used, with the probability of successful recovery as the performance evaluation metric. A trial is considered successful if and only if the probability of successful recovery is achieved for all trials. Each goal has and .

[0120] Figures 3-4 The fixed channel SNR is shown to be At dB, the probability of successful recovery for Alice and Eve varies with the reference channel SINR (from dB to The curve shows the variation of dB. This figure verifies the perceived privacy protection effect of the super-Gaussian constellation from the perspective of reference channel SINR—even though Eve uses the same SINR as Alice. The frame MD-SBL algorithm still has a much lower success rate in recovery than Alice under the super-Gaussian constellation, and the required reference channel SNR increases further when the number of available frames decreases. In contrast, the performance gap between the two is significantly narrowed under the sub-Gaussian constellation (16-QAM).

[0121] Figures 5-6 The fixed reference channel SNR is shown to be At dB, the probability of successful recovery for Alice and Eve varies with channel SNR (from dB to The curve shows the variation of dB. This figure verifies the effect of MD-SBL multi-frame coherent accumulation from the perspective of channel SNR. Under the super-Gaussian constellation, using Alice's channel SNR is [value] during frame MD-SBL accumulation. dB can achieve this Successful recovery; in a single-frame scenario, neither side can perfectly recover. (In the same...) Under frame accumulation conditions, Eve's successful recovery probability is still lower than .

[0122] comprehensive Figures 3-4 and Figures 5-6 In the context of no-distance migration, the super-Gaussian constellation significantly widens the perception performance gap between legitimate receivers and eavesdroppers through deceptive grating lobes in DP-AF, while MD-SBL achieves reliable coherent accumulation within short coherent intervals through multi-frame joint sparse support priors.

[0123] Example 2: Performance Verification and Implementation of Distance Migration Compensation in Distance Migration Scenarios

[0124] This embodiment describes the specific implementation process of the method of the present invention in a high dynamic scene with inter-frame distance migration. It is used to illustrate how the method of the present invention compensates for inter-frame distance migration through the RMC-MD-SBL algorithm in a high dynamic scene with high target radial velocity and significant inter-frame distance migration between frames, thereby extending multi-frame joint sparse recovery to the high dynamic scene while maintaining the perceived privacy protection effect brought by the super-Gaussian constellation.

[0125] This embodiment considers an underwater acoustic communication scenario, where the speed of sound wave propagation is relatively low ( m / s), even if the target radial velocity is low (e.g. The distance migration (m / s) can also generate significant distance migration during multi-frame accumulation, making it a typical application scenario for verifying distance migration compensation methods.

[0126] Solution Step (1): OTFS modulation and super-Gaussian constellation signal design, same as in Example 1.

[0127] Step (1) and its sub-steps (1.1) to (1.3) are consistent with those in Example 1. The Alice transmitter is constructed based on a super-Gaussian constellation. Frame OTFS communication sensing integrated signal. Constellation generation adopts the four-ring 64-APSK scheme (kurtosis) described in Example 1. ) or dual-ring 4-APSK scheme (kurtosis) The OTFS modulation parameters are the same as in Example 1.

[0128] The system parameters in this embodiment are adjusted as follows based on the underwater acoustic scenario: carrier frequency kHz, bandwidth kHz, sampling interval ms, CP length OTFS symbol count ( ), distance resolution m, frame period ms.

[0129] Solution Step (2): Perceptual dictionary construction, same as in Implementation Example 1

[0130] Step (2) is consistent with Example 1, Alice uses a precisely known transmission signal Construct the corresponding overcomplete perception dictionary ( The dictionary construction method is exactly the same as in Example 1.

[0131] Corresponding technical solution step (3): Multi-frame observation signal reception

[0132] Alice continuously receives Observation signals of one OTFS frame The received signal model is the same as step (3) in Example 1.

[0133] This embodiment includes the scenario. The targets have radial velocities of... , , , and m / s, distances are respectively , , , and m. The transmitted signal for each frame is generated independently, and the target complex reflection coefficient is allowed to vary with each frame.

[0134] Due to the extremely low propagation speed of underwater sound ( (m / s), far lower than the speed of electromagnetic waves, and even if the target's radial velocity is only... m / s, at During frame accumulation (total duration approximately) (ms) can also cause distance movement to reach the meter level (e.g., target speed is). At m / s Distance changes approximately during frame m, corresponding to approximately (each distance resolution unit) produces significant inter-frame distance migration.

[0135] Step (4): Distance migration determination:

[0136] ;

[0137] Substitute the parameters into this embodiment: m / s, kHz, , ms, the calculated critical velocity is approximately m / s. In this embodiment, the maximum radial velocity of the target is The velocity is m / s, which is much greater than the critical velocity, therefore it is determined that distance migration exists.

[0138] Due to distance migration, the sparse support of each frame is no longer consistent. The MD-SBL method based on common covariance prior in Implementation 1 cannot be directly applied and needs to be transferred to the RMC-MD-SBL process described below.

[0139] Step (5): Distance migration compensation and time delay-Doppler parameter extraction based on RMC-MD-SBL

[0140] Step (5.1), assuming the target is in Maintain a constant radial velocity during frame accumulation. (For a duration of approximately The assumption of a ms underwater acoustic communication frame sequence is reasonable. Discrete Delay Index in Frame And Doppler Index The corresponding distance-velocity pair is : Then in the first frame( In the above, due to the target's movement, the distance becomes... .

[0141] According to this definition, from the first Frame delay - Doppler grid points To the Frame mapping :

[0142] ;

[0143] Based on this mapping, construct the permutation matrix. ( ,in ):

[0144] ;

[0145] in For the first A standard basis vector, defined . for A dimensional matrix, its function is to group the 1st dimension matrix. Frame delay - prior covariance at Doppler grid points transferred to the first The actual location of the target in the frame is determined to compensate for the inter-frame support offset caused by the target's movement.

[0146] Unlike Example 1, where all frames share a common covariance, this example employs deterministic evolutionary support priors: the covariance vector of each frame. With the Frame covariance vector The relationship is:

[0147] ;

[0148] in , ( The covariance vector of each frame is determined by the mapping relationship described above. The above equation shows that the covariance vector of each frame is determined by the first... The frame covariance vector is obtained through deterministic permutation.

[0149] Step (5.2) calculates the covariance matrix of the received signal in each frame. The covariance matrix of the received frame signal is:

[0150] ;

[0151] For step (5.3), solve the similarity maximization problem to obtain the result in step (5.1). Estimate:

[0152] ;

[0153] in, Represents the trace of a matrix. Represents the determinant of a matrix. Represents the natural logarithm. The Moore-Penrose pseudoinverse of the matrix is ​​represented; the above problem can be solved using Mackay iteration.

[0154] For cases involving distance migration, the following iterative solution is used:

[0155] ;

[0156] in, For the first The first frame dictionary List, To control the exponent of convergence speed, express The result of the last iteration, express The results of iterative updates, the parts of the above methods that correspond to distance migration, are collectively referred to as the RMC-MD-SBL method.

[0157] in, The core difference between this formula and the MD-SBL update formula in Example 1, which is a convergence control exponent, lies in the introduction of... When a time-delay Doppler grid point moves out of the observation window in a frame due to distance migration, that frame does not contribute information to that grid point; however, when the position of that grid point differs across frames due to distance migration, It is accurately associated with the corresponding position in each frame.

[0158] For step (5.4), construct the delay-Doppler map for each frame and extract the parameters.

[0159] After iterative convergence, the th The covariance of the frame is estimated as follows , No. The covariance of the frame is estimated by Obtained. Construct delay-Doppler plots for each frame:

[0160] ;

[0161] By extracting the peak locations from the DDMap of each frame, the target delay index estimate for each frame can be obtained. And Doppler index estimation Due to the introduction of distance migration compensation, the first The frame estimation results can be used as precise parameter estimates of the target and can be converted into physical distance and velocity.

[0162] The performance of the method described in this embodiment is verified by Monte Carlo simulation. The simulation parameters are set as follows: underwater acoustic communication scenario, propagation speed... m / s, carrier frequency kHz, bandwidth kHz, CP length Distance resolution m. The scene includes The targets have radial velocities of [number] and [number] respectively. , , , and m / s, distances are respectively , , , and m. Adopted Frame RMC-MD-SBL accumulation, Monte Carlo simulation at each point This is an independent trial.

[0163] Figures 7-8 The graphs show the changes in distance estimation RMSE and velocity estimation RMSE as a function of channel SNR for Alice and Eve under super-Gaussian and sub-Gaussian constellations. Under the super-Gaussian constellation, the channel SNR is... At dB, Alice's distance estimate RMSE is less than m, speed estimate RMSE is less than m / s. Eve, however, suffers from both reference signal degradation and grating lobe spurious peak interference, failing to achieve ideal range and velocity estimation accuracy even at higher signal-to-noise ratios. This result validates the effectiveness of RMC-MD-SBL in compensating for inter-frame range migration through a deterministic evolutionary model, and the continuous awareness and privacy protection effect of the super-Gaussian constellation in high-dynamic scenes.

[0164] In summary, this invention combines a super-Gaussian constellation generated by probabilistic constellation shaping with OTFS modulation to produce controllable deceptive grating lobes in the time-delay-Doppler domain, causing eavesdroppers to generate numerous spurious peaks when the reference signal is impaired. Multi-frame coherent accumulation is achieved through MD-SBL, enabling reliable joint time-delay-Doppler estimation within short coherence intervals. Inter-frame distance migration is compensated through RMC-MD-SBL, extending multi-frame joint sparse recovery to high-dynamic scenarios. This invention requires no additional system overhead, is fully compatible with communication-centric ISACs, possesses inherent robustness against passive eavesdropping, and is suitable for various high-mobility scenarios such as V2X, high-speed rail, low-Earth orbit satellites, and underwater acoustic communications.

Claims

1. A method for privacy protection through integrated sensing induction suitable for dual-selection channels, characterized in that, The method includes the following steps: (1) The transmitting end of the integrated communication and sensing equipment adopts orthogonal time-frequency-space OTFS modulation with a multi-frame length of [missing information]. The super-Gaussian constellation symbol vector, and continuously transmit multiple frames of length. The discrete-time domain communication and sensing integrated signal is transmitted to the communication user and the sensing target; (2) The processor of the integrated communication and sensing device uses the known transmission signals in step (1) to construct an overcomplete sensing dictionary corresponding to each frame; (3) The receiver of the integrated communication and sensing equipment continuously receives observation signals of multiple OTFS frames; (4) The receiver of the integrated communication and sensing equipment determines whether there is distance migration based on the system parameters and the speed range of the target to be sensed; (5) Based on the aforementioned distance migration judgment, the processor of the communication sensing integrated device uses the overcomplete dictionary constructed in step (2) to process the multi-frame received signal in step (3) using the corresponding sparse Bayesian method, and obtains the time delay-Doppler map and extracts the target time delay-Doppler parameters.

2. The method for privacy protection of integrated sensing in a dual-selection channel according to claim 1, characterized in that, In step (1), the method by which the transmitting end modulates multi-frame super-Gaussian constellation symbol vectors using OTFS and transmits discrete time-domain signals is as follows: (1.1) Super-Gaussian constellations are generated using probabilistic constellation shaping (PCS). The kurtosis of its probability distribution ,in, This indicates the calculation of mathematical expectation. The modulus of a complex number The super-Gaussian constellation is a symbol in the constellation. It adopts a multi-ring amplitude phase shift keying structure. The constellation points on each ring have a uniform phase distribution. The probability of each ring is distributed according to a preset ratio, so that the probability of the constellation points appearing follows a super-Gaussian distribution. (1.2) Each frame of transmission contains Independent and identically distributed communication symbol vectors ,in , and The grid numbers for the time delay axis and the Doppler axis, respectively, and the sign in the vector. Constellations generated in step (1.1) The symbol vector Through OTFS modulation base Mapped to time-domain transmitted signal ,in, for Point discrete Fourier transform matrix, Represents the Kronecker product. for An identity matrix of order 1. This indicates the transpose operation. This represents the conjugate transpose operation; in each frame, the communication symbol vector Implement the transmitted signal independently once, in each frame Also, implement it independently once, and the first The implementation of a frame is denoted as ; (1.3) In the first Frame time-domain transmitted signal A cyclic prefix (CP) is added before the transmission frame to form a complete transmit frame, which is then up-converted to the carrier frequency by the radio frequency front end. Post-transmission, system parameters include: carrier frequency ,bandwidth Symbol period CP length Frame period .

3. The method for privacy protection of integrated sensing in a dual-selection channel according to claim 2, characterized in that, In step (2), the processor of the integrated communication and sensing device constructs an overcomplete sensing dictionary for each frame using the transmitted signals known in step (1), as follows: Based on accurate Frame transmission signal Building an ultra-complete perceptual dictionary Its column vector is defined as: ； in, Cyclic delay matrix, Doppler frequency shift matrix, For two-dimensional time-delay-Doppler indexing Mapping to one-dimensional dictionary column indexes.

4. The method for privacy protection of integrated sensing in a dual-selection channel according to claim 3, characterized in that, In step (3), the receiver of the integrated communication and sensing device continuously receives observation signals from multiple OTFS frames, as follows: The integrated communication and sensing equipment deploys its receiver and transmitter at the same location to receive target echo signals reflected through a dual-select channel, wherein the first... The echo of each target is represented by a cyclic delay operator. and Doppler operator The cascading effect, i.e. ； in, For the target total number, For the first The complex reflection coefficient of a target, and The first The discrete time delay index and Doppler index of each target are used as the cyclic time delay and Doppler frequency shift, respectively. Cyclic delay matrix, Doppler frequency shift matrix, It is an additive complex Gaussian white noise vector. For the first The observed signal of the frame.

5. A method for privacy protection of integrated sensing in a dual-selection channel as described in claim 4, characterized in that, In step (4), the receiver of the integrated communication and sensing device determines whether there is distance migration based on the system parameters and the speed range of the target to be sensed, as follows: If the target's maximum radial velocity If the following conditions are met, it is determined that there is no distance migration; otherwise, it is determined that there is distance migration: ； in, For wave speed, For distance resolution, The number of OTFS frames received in step (3).

6. The method for privacy protection of integrated sensing in a dual-selection channel according to claim 5, characterized in that, In step (5), based on the aforementioned distance migration judgment, the processor of the integrated communication and sensing device uses the overcomplete dictionary constructed in step (2) to process the multi-frame received signal in step (3) using the corresponding sparse Bayesian method, and obtains the time delay-Doppler map, extracting the target time delay-Doppler parameters as follows: (5.1) Set the target parameter covariance vector for each frame The relationship between them When there is no distance migration, the relationship is as follows: ； in, Let be the target parameter covariance vector of the first frame, where the th is the . element This represents the perceptual dictionary corresponding to the first frame of transmitted signal. No. The covariance of the coefficients corresponding to the column; when distance migration exists, let the discrete delay index in the first frame be... And Doppler Index The corresponding distance-velocity pair is Then in the first In the frame, due to the target's motion, the distance-velocity pair evolves into ,in, The frame period is used to obtain the target parameter support covariance vector. The relationship between them is: ； in, Let covariance vector of target parameters in frame 1 be the permutation matrix. The permutation matrix corresponding to the first frame is: An identity matrix of order 1, denoted as ,satisfy: ； in, For the first Standard basis vectors For the first A standard basis vector, mapping Delay-Doppler indexing of the first frame Mapped to the Frame index: ； in, No. The index of the target parameter in the frame corresponds to the range-velocity pair. For the single-station sensing scenario under consideration, the first... Discrete Delay Index in Frame And Doppler Index The corresponding distance-velocity pair is : ； Then in the In the frame, due to the target's movement, the distance becomes... ,get ; (5.2) Calculate the received signal for each frame covariance matrix , ; ； in, This represents a diagonal matrix with the input vector as its diagonal elements. For the first Frame noise variance; (5.3) Solve the two similarity maximization problem to obtain the result in step (5.1). Estimate: ； in, Represents the trace of a matrix. Represents the determinant of a matrix. Represents the natural logarithm. Represents the Moore-Penrose pseudoinverse of a matrix; For cases where there is no distance migration, the following iterative solution is used: ； in, For the first The first frame dictionary List, To control the exponent of convergence speed, Indicate The result of the last iteration, express The results of iterative updates, the part of the above methods corresponding to distance-free migration, are collectively referred to as the MD-SBL method; For cases involving distance migration, the following iterative solution is used: ； in, For the first The first frame dictionary List, To control the exponent of convergence speed, express The result of the last iteration, express The results of iterative updates, the parts of the above methods that correspond to distance migration, are collectively referred to as the RMC-MD-SBL method; (5.4) After iterative convergence, the covariance estimate of the target parameters in the first frame is obtained. Constructing a time-delay-Doppler plot: ； Extracting the peak position from the time-delay-Doppler map yields the target's time-delay index estimate. And Doppler index estimation .

Images