Radar safety wireless sensing method based on IRS assistance

Abstract

The invention relates to the field of wireless communication, in particular to a radar safety wireless sensing method based on IRS assistance, which is characterized in that an IRS with a sensing function is mounted on a target, and sensing is carried out in two stages, and specifically comprises the following steps: in the first stage, an IRS reflecting element is closed, and signals from a legal radar LRS and a plurality of illegal radar URS are received through an IRS sensing unit; estimating all radar angle information; in the second stage, an IRS reflection coefficient is designed according to an estimation result so as to reduce the illegal radar sensing probability to the maximum extent; according to the scheme provided by the invention, the security of legal radar sensing can be improved.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication, and in particular, to a radar-secure wireless sensing method based on IRS assistance. Background Art

[0002] Due to the broadcast nature of wireless signals, radars face sensing security issues during the process of sensing targets. For example, information such as the angle, position, and speed of a target may be obtained by illegal radars. To make the target invisible to illegal radars, electromagnetic stealth materials have received extensive attention. Currently, electromagnetic stealth mainly focuses on designing advanced materials with ideal absorption performance and multi-layer structures composed of multiple materials. By absorbing the energy of the incident signal, these materials can effectively reduce the reflected signal power in all directions, thereby reducing the signal strength received by the radar receiver. However, due to limitations in coating thickness, incident angle, and the inherent properties of the materials, electromagnetic stealth materials have deficiencies in terms of flexibility and adaptability.

[0003] In recent years, intelligent reflecting surface (IRS) has received extensive attention from the industrial and academic communities as a technology with broad prospects. IRS consists of a large number of low-cost passive reflecting elements. Each element can independently reflect the incident signal by controlling its amplitude and phase and cooperate to reflect the signal in the direction of the target. By dynamically configuring these elements according to system requirements, IRS can reshape the wireless propagation environment, thereby assisting wireless communication systems to improve communication capacity and reliability. In addition, IRS can also assist integrated sensing and communication (ISAC) systems to achieve performance balance between sensing and communication functions, thereby improving the overall efficiency and adaptability of the system, or assist wireless sensing to improve the sensing accuracy of targets. Regarding IRS as an additional anchor node in the wireless network to improve the sensing accuracy of the base station (BS). However, the radar signal needs to be reflected multiple times before being received by the BS receiver / radar, resulting in a certain degree of path loss, which reduces the sensing efficiency and accuracy of the target.

[0004] Since the IRS has the ability to enhance the wireless signals at the desired receiver and suppress the wireless signals at the undesired receiver, the IRS can also be used in a wireless sensing system to enhance radar sensing or weaken radar sensing to make the target stealthy to the radar. The key advantage of the IRS over traditional electromagnetic stealth materials is its real-time reconfigurability, which can flexibly regulate the incident signals within a large frequency and angle range, making up for the deficiencies of electromagnetic stealth materials and providing a new solution for improving sensing security. In the prior art, in a single URS scenario, the IRS is installed on the target and works in cooperation with electromagnetic stealth materials, and the optimization problem of minimizing the signal-to-noise ratio of the illegal radar (URS) is proposed. Extended from a single URS scenario to multiple URS scenarios, the optimization problem of minimizing the sum of the signal powers received by the URSs is proposed. Although the target can be made stealthy to the URS, it also makes the target stealthy to the legitimate radar (LRS). The radar anti-jamming theory and its key technologies are systematically reviewed from the aspects of waveform design, receiver level, and signal and data processing. The prior art also proposes a moving target detection algorithm for airborne radar against dense deceptive jamming based on data fitting. It can improve the sensing ability of the LRS, but these methods can also be applied to the URS. In a single URS and LRS scenario, the IRS is installed on the target to enhance the sensing ability of the LRS while reducing the sensing ability of the URS. The optimization problem of maximizing the signal power received by the legitimate radar is proposed on the premise of ensuring the upper bound constraint of the signal power received by the URS. However, the sensing scenarios considered in the above research work are simple, either only considering the illegal radar, or considering both the legitimate radar and the illegal radar without considering the impact of the number of illegal radars on the security performance of the IRS-assisted wireless sensing system. Summary of the Invention

[0005] The purpose of the present invention is to overcome the deficiencies of the prior art and propose a radar security wireless sensing method based on IRS assistance. By adopting a two-stage sensing scheme, the present invention can improve the security of legitimate radar sensing.

[0006] The purpose of the present invention is achieved by the following measures: A radar security wireless sensing method based on IRS assistance, in which an IRS with sensing function is installed on the target and a two-stage sensing is adopted, specifically including:

[0007] The first stage: Turn off the IRS reflection elements, and receive the signals from the legitimate radar LRS and multiple illegal radars URS through the IRS sensing unit to estimate all radar angle information;

[0008] The second stage: Design the IRS reflection coefficient according to the estimation result to minimize the illegal radar sensing probability to the greatest extent.

[0009] Preferably, in the first stage, a system model is constructed;

[0010] Assume that the channel matrix between the illegal radar and the IRS at time t is The channel matrix between the legitimate radar and the IRS at time t is The distance between the radar and the target is relatively far. Assume that the channels between the LRS / URS and the IRS are LoS channels.

[0011] Define the one-dimensional steering vector of the uniform linear array ULA as:

[0012]

[0013] where φ is the signal phase shift difference between adjacent two antennas / elements, is the number of antennas / elements in the ULA;

[0014] Express the AoA / AoD pair of the IRS as Express the AoA / AoD pairs of the LRS and URS as and Let denote the array response vector of the IRS. Let and denote the array response vectors of the URS and LRS respectively;

[0015] The array response vectors of the IRS, URS, and LRS are respectively expressed as:

[0016]

[0017] where λ is the signal wavelength, Δe is the element spacing of the IRS, and Δa is the antenna spacing at each radar;

[0018] Assume that the target moves at a constant speed v. The channel link is affected by the Doppler frequency. Therefore, the far-field LoS channel between the radar and the IRS is modeled as the outer product of the array response vectors on both sides:

[0019]

[0020] where is the path gain corresponding to time t, β represents the reference path gain at a distance of 1 meter, represents the distance between the two nodes at time t, is the Doppler frequency, T c is the channel coherence interval. Assume that the channel between the radar and the IRS remains approximately constant within the coherence time, then

[0021] The reflection coefficient vector of the IRS at time t is and Denote the reflection amplitude and phase shift of the nth IRS element respectively. The signal reflected from the IRS received by radar k can be expressed as:

[0022]

[0023] where Θ = diag(θ [t] ) is the phase shift matrix of the IRS at time t. The transmitted radar waveform at time t is satisfying and P > M (L is the number of subcarriers), is an additive white Gaussian noise matrix with zero mean and variance σ 2 . The radar sensing process within each channel coherence time can be formulated as a binary hypothesis testing problem as follows:

[0024]

[0025] where H0 and H1 denote the absence and presence of the target respectively. According to the NP criterion, the detection probability is

[0026]

[0027] Q(·) is the Marcum-Q function. Considering the monotonic relationship between the signal-to-noise ratio and the detection probability in Equation (8), the signal-to-noise ratio is used as an index to evaluate the system performance;

[0028] Assume that there is cooperation among URSs and non - cooperation between USR and LRS. Then the signal-to-noise ratios received by URS and LRS can be expressed as:

[0029]

[0030] Preferably, in the first stage, the IRS reflection elements are turned off, and an L-shaped array consisting of L = Lx + Ly - 1 sensing devices is embedded on the IRS to estimate the K angle information of all radars; denotes the array response vector of the sensing elements, and the vectors are respectively:

[0031]

[0032] The channel between the L-shaped sensing array and radar k can be expressed as which is modeled as the outer product of the array response vectors on its two sides. During the channel coherence time, omitting the time index [t], the signal received by the L-shaped sensing array is:

[0033]

[0034] where is an additive white Gaussian noise with zero mean and variance σ 2The zero-mean additive white Gaussian noise matrix, is the array response matrix of the L-shaped sensing array, F = [F1,..., F K T represents the transmitted beamforming signal matrix of K radars,

[0035] Preferably, deleting the time index [t], the corresponding optimization problem is expressed as:

[0036]

[0037] where ε is the minimum signal-to-noise ratio required for the LRS to achieve the target sensing performance. Using the channel structure between the radar and the target, the signal-to-noise ratio expressions in (9) and (10) can be simplified. Substituting (5) into (9) and (10), the signal-to-noise ratios of the URS and LRS can be rewritten as:

[0038]

[0039] where

[0040] and are all constants, and equation (14) can be simplified to

[0041]

[0042] Equation (17) is a typical non-convex optimization problem.

[0043] Preferably, equation (17) is further simplified to equation (18):

[0044]

[0045] where is the cascaded array response, represents the selection matrix with the nth diagonal element being 1 and the remaining elements being 0,

[0046] Introducing a slack variable z, the problem is reformulated as the following form:

[0047]

[0048] Due to the fractional constraint C1 and the non-convex constraint C2, the fractional constraint C1 can be converted into a polynomial expression. Introducing an auxiliary variable q k , q k the optimal value of the optimization problem can be further expressed as:

[0049] ​

[0050] Constraints C1 and C2 are non-convex with respect to the variable θ. By definition

[0051] X = θθ H (21)

[0052] The quadratic term θθ H is transformed into the main variable X. The rank-1 Hermitian positive semi-definite matrix X satisfies

[0053] X = X H , X ≥ 0 (22)

[0054] Rank(X) = 1 (23)

[0055] Define the set of all N×N dimensional Hermitian positive semi-definite matrices as Transform equation (20) into:

[0056]

[0057] Using the SDR algorithm, the original problem can be simplified to

[0058]

[0059] Preferably, an iterative optimization algorithm combining the Dinkelbach algorithm and the semi-definite relaxation (SDR) algorithm:

[0060] 1) Initialization

[0061] 2)

[0062] 3) Let Solve the optimization problem 25 to obtain X (l+1) ;

[0063] 4) Use EVD decomposition to obtain θ (l+1) ;

[0064] 5)

[0065] 6) end while;

[0066] 7) Output θ * = θ (l) .

[0067] Advantages of the present invention: The present invention deploys an IRS with sensing capabilities on the target and designs a two-step sensing protocol. In the first step, the IRS reflection elements are turned off, and the IRS sensing elements estimate the angle information of all radars. In the second step, based on the estimated information, the IRS reflection coefficient is designed to enhance the received SNR of the LRS and suppress the received SNR of the URS, thereby achieving secure sensing of the target. Due to the monotonic relationship between the radar sensing probability and the radar received SNR, a problem of minimizing the maximum URS SNR is constructed under the premise of ensuring the LRS SNR constraint and the IRS reflection phase shift modulus constraint. For the above non-convex optimization problem, an iterative optimization algorithm based on Dinkelbach and semidefinite relaxation (SDR) techniques is used. Due to the complexity of the objective function, slack variables are introduced to linearize the fractional objective function, and then successive convex approximation and SDR are used to transform the original non-convex problem into a convex optimization problem. Finally, the solution of the original problem is obtained through iterative optimization. The solution of the present invention can greatly improve the security of legal radar sensing. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 is the model of the radar security sensing system based on IRS;

[0069] Figure 2 is the two-stage sensing based on IRS sensing and reflection;

[0070] Figure 3 is the relationship between the LRS SNR and the number N of reflection elements;

[0071] Figure 4 is the relationship between the LRS / URS SNR and the shortest distance from the URSs to the target;

[0072] Figure 5 is the influence of the angle estimation error of the IRS sensing element on the LRS / URS SNR. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0073] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0074] Embodiment 1: A method for radar security wireless sensing assisted by IRS, which installs an IRS with sensing capabilities on the target and adopts two-stage sensing, specifically including:

[0075] The first stage: Turn off the IRS reflection element, receive signals from the legitimate radar LRS and multiple illegal radars URS through the IRS sensing unit, and estimate all radar angle information;

[0076] The second stage: Design the IRS reflection coefficient according to the estimation result to minimize the illegal radar sensing probability.

[0077] As Figure 1 shown, the radar sensing system model consists of an LRS, K - 1 URSs, and a target. An IRS with sensing function is installed on the target surface to reflect the signals sent from the radar to the target. By designing the IRS reflection coefficient, its reflection effects on the LRS and URS can be optimized respectively. In addition, the rest of the target surface is covered with electromagnetic stealth materials, so that the LRS and URS can only receive the signals reflected by the IRS, and thus the IRS can completely control the radar signals received by the target.

[0078] Assume that both the LRS and URS are monostatic radars (i.e., their transmitters and receivers are located at the same position), and both are equipped with a uniform planar array (UPA) composed of M = M x ×M y antenna elements, the IRS is equipped with N = N x ×N y reflection elements. In addition, in order to estimate the angle of arrival (AoA) of the radar signal, L = L x +L y -1 sensing devices are integrated into the RIS, where L x and L y represent the number of sensing devices along the x-axis and y-axis respectively.

[0079] Assume that the channel matrix between the illegal radar and the IRS at time t is The channel matrix between the legitimate radar and the IRS at time t is The distance between the radar and the target is relatively far. Assume that the channel between the LRS / URS and the IRS is a line-of-sight (LoS) channel.

[0080] Define the one-dimensional steering vector of the uniform linear array (ULA) as:

[0081]

[0082] where φ is the signal phase shift difference between adjacent two antennas / elements, is the number of antennas / elements in the ULA;

[0083] Represent the AoA / AoD pair of the IRS as Represent the AoA / AoD pairs of the LRS and URS as and Let Denote the array response vector of the IRS as \(\mathbf{a}_{IRS}\), and let and represent the array response vectors of the URS and LRS respectively;

[0084] The array response vectors of the IRS, URS, and LRS are respectively expressed as:

[0085]

[0086] where \(\lambda\) is the signal wavelength, \(\Delta e\) is the element spacing of the IRS, and \(\Delta a\) is the antenna spacing at each radar;

[0087] Assume that the target moves at a constant speed \(v\), and the channel link is affected by the Doppler frequency. Therefore, the far-field LoS channel between the radar and the IRS is modeled as the outer product of the array response vectors on both sides, i.e.,

[0088]

[0089] where is the path gain corresponding to time \(t\), \(\beta\) represents the reference path gain at a distance of 1 meter, represents the distance between the two nodes at time \(t\), is the Doppler frequency, \(T\) c is the channel coherence interval. Assume that the channel between the radar and the IRS remains approximately constant within the coherence time, then

[0090] The reflection coefficient vector of the IRS at time \(t\) is and represent the reflection amplitude and phase shift of the \(n\)-th IRS element respectively. Therefore, the signal received by radar \(k\) reflected from the IRS can be expressed as:

[0091]

[0092] where \(\Theta=\text{diag}(\theta [t] )\) is the phase shift matrix of the IRS at time \(t\), and the transmitted radar waveform at time \(t\) is satisfies and \(P > M\) (\(L\) is the number of subcarriers), is a zero-mean additive Gaussian white noise matrix with variance \(\sigma 2 . The radar sensing process within each channel coherence time can be represented as a binary hypothesis testing problem, as follows:

[0093]

[0094] where \(H_0\) and \(H_1\) represent the absence and presence of the target respectively. Transforming the problem of sensing the presence of the target into a binary hypothesis problem, using the NP criterion (the signal-to-noise ratio SNR at a given \(t\)[t] and the false alarm probability The required detection probability (maximum) to make an optimal decision for this binary hypothesis problem. According to the NP criterion, the detection probability is

[0095]

[0096] Q(·) is the Marcum-Q function. For a given A higher signal-to-noise ratio will result in a higher target detection probability. Considering the monotonic relationship between the signal-to-noise ratio and the detection probability in Equation (8), the signal-to-noise ratio is used as an index to evaluate the system performance.

[0097] Assume that there is cooperation among URSs and non-cooperation between USR and LRS. Then the signal-to-noise ratios received by URS and LRS can be expressed as follows:

[0098]

[0099] As Figure 2 shown, this protocol defines a coherent time-aware block, in which the positions and directions of the target and the radar remain constant within the channel coherent time-aware block, while they may change in different blocks. Each sensing block is divided into two steps. The first step: The IRS reflection elements are turned off, and the angle information of all radars is estimated based on the signals received by the IRS sensing elements. The second step: According to the estimated information, the reflection coefficients of the IRS are designed.

[0100] In the first stage (i.e., the first step), the IRS reflection elements are turned off, and an L-shaped array consisting of L = Lx + Ly - 1 sensing devices is embedded on the IRS to estimate the K angle information of all radars; denotes the array response vector of the sensing elements, and the vectors are respectively:

[0101]

[0102] The channel between the L-shaped sensing array and radar k can be expressed as is modeled as the outer product of the array response vectors on both sides of it; within the channel coherent time, omitting the time index [t], the signal received by the L-shaped sensing array is:

[0103]

[0104] where is a zero-mean additive Gaussian white noise matrix with variance σ 2 and is the array response matrix of the L-shaped sensing array, F = [F1,..., F K T ​Denote the transmit beamforming signal matrix of K radars. The transmit radar waveform at this moment is S = [s1,..., s P which satisfies E[SS H = I M and P > M (P is the number of subcarriers).

[0105] Based on (13), existing AoA estimation algorithms such as the Multiple Signal Classification (MUSIC) algorithm

[28] can be used for the estimation of AoAs pair.

[0106] In the second step, according to the estimated angle information above, design the reflection coefficients of the IRS, so as to enhance the received signal-to-noise ratio (SNR) of the LRS while reducing the SNR of the URS received.

[0107] To improve the security of the IRS-aided wireless sensing system, under the constraints of the LRS SNR constraint and the IRS reflection phase shift modulus constraint, this paper minimizes the SNR of the maximum URS by optimizing the IRS reflection coefficients to reduce the information eavesdropped by the URS as much as possible. In addition, this paper processes in units of sensing blocks, so the time index [t] is removed, and the corresponding optimization problem is expressed as:

[0108]

[0109] where ε is the minimum SNR required for the LRS to achieve the target sensing performance. Using the channel structure between the radar and the target, the SNR expressions in (9) and (10) can be simplified. Substituting (5) into (9) and (10), the SNRs of the URS and LRS can be rewritten as:

[0110]

[0111] where

[0112] and are all constants, and equation (14) can be simplified to

[0113]

[0114] Equation (17) is a typical non-convex optimization problem. To solve this problem, an iterative optimization algorithm based on the Dinkelbach and SDR techniques is proposed for solution.

[0115] Equation (17) is further simplified to equation (18):

[0116]

[0117] where is the cascaded array response. Denote the selection matrix where the n-th diagonal element is 1 and the rest of the elements are 0

[0118] Introduce a slack variable z and reformulate the problem as follows:

[0119]

[0120] Due to the fractional constraint C1 and the non-convex constraint C2, the above problem is still difficult to solve. Notice that equation (19) has a similar form to the max-min ratio fractional programming problem, and the Dinkelbach transformation can be applied to transform it into a more tractable form. Specifically, the fractional constraint C1 can be converted into a polynomial expression. By introducing an auxiliary variable q k , which essentially represents the signal-to-noise ratio SNR of the k-th URS k , and it can be updated with the reflection coefficient vector θ. The optimal value of q k The optimization problem can be further expressed as:

[0121]

[0122] The constraints C1 and C2 are non-convex with respect to the variable θ. By defining

[0123] X = θθ H (21)

[0124] Transform the quadratic term θθ H into the main variable X. At the same time, the rank-1 Hermitian positive semi-definite matrix X satisfies

[0125] X = X H , X ≥ 0 (22)

[0126] Rank(X) = 1 (23)

[0127] Define the set of all N×N dimensional Hermitian positive semi-definite matrices as Transform equation (20) into:

[0128]

[0129] Using the SDR algorithm, the original problem can be simplified to

[0130]

[0131] The above is a semi-definite programming problem, which can be directly solved using the CVX toolbox to obtain the optimal solution X * . If the rank of the obtained X * is not 1, the EVD decomposition or Gaussian randomization is usually used to obtain the optimal solution θ * .​

[0132] The specific algorithm is summarized as follows, where the superscript (·) represents a certain iteration of the algorithm. Set the iteration accuracy parameter to Γ and the maximum number of iterations to T max 。

[0133] An iterative optimization algorithm combining the Dinkelbach algorithm and the semidefinite relaxation (SDR) algorithm:

[0134] 1) Initialization

[0135] 2)

[0136] 3) Another Solve the optimization problem 25 to obtain X (l+1) ;

[0137] 4) Use EVD decomposition to obtain θ (l+1) ;

[0138] 5)

[0139] 6) end while;

[0140] 7) Output θ * = θ (l) 。

[0141] The performance of the proposed scheme is evaluated through simulation analysis. Assume that the number of URSs is 3. Assume that the LRS, URSs, and IRS are all uniform linear arrays (ULA). Unless otherwise specified, the number of antennas of the LRS and URSs is M = 64, and the number of reflecting elements and sensing elements of the IRS is N = 49 and L = 13 respectively. The distance between the LRS and the target is set to d_LT = 20m, and the shortest distance between the URS and the target is d_UT = 30m. The wavelength λ = 0.05m, and the antenna spacing of each radar and the IRS unit spacing are set to Δa = λ / 2 = 0.025m and Δe = λ / 4 = 0.0125m respectively. In all simulations, let the signal-to-noise ratio of the URS represent the sum of the signal-to-noise ratios of 3 URSs. First, assume that the IRS sensing device perfectly estimates the AOAs of the LRS and URSs to the target / IRS, and then evaluate the impact of its imperfect estimation on radar performance.

[0142] The scheme of the present invention is compared and analyzed with the following two benchmark schemes:

[0143] 1) Random phase shift: Assume that each IRS reflecting element is independently and uniformly distributed in (0; 2π).

[0144] 2) The target is not equipped with an IRS: The target is not equipped with any reflecting elements.

[0145] Figure 3 It depicts the relationship between the LRS signal-to-noise ratio and the number N of reflecting elements. It can be found that as the number of IRS elements increases, the received signal-to-noise ratio of LRS also increases. This is because installing more reflecting elements in the IRS can enhance the passive beamforming gain. Secondly, this scheme is superior to the schemes without IRS and with random phase shifts in terms of increasing the received signal-to-noise ratio. Finally, the signal-to-noise ratio under the random phase shift design decreases with the increase in the number of IRS elements. This indicates that if the signals reflected by the IRS are not properly designed, the performance obtained by using the IRS is even worse than that without using the IRS.

[0146] In Figure 4 , the AOAs of the fixed LRS and URSs to the target / IRS and the distance between the LRS and the target are kept unchanged, and only the distance from the URSs to the target is changed. The relationship between the LRS / URS signal-to-noise ratio and the shortest distance from the URSs to the target is plotted. The results show that within a wide range of distances, compared with the case where the target is not equipped with IRS, equipping the target with IRS can significantly reduce the received signal-to-noise ratio of the URS while increasing the received signal-to-noise ratio of the LRS.

[0147] The present invention evaluates the influence of the angle estimation error of the IRS sensing elements on the LRS / URS signal-to-noise ratio, as Figure 5 shown. As the AoA estimation error increases, the LRS signal-to-noise ratio decreases and the URS signal-to-noise ratio increases. This indicates that the AoA estimation error will reduce the signal superposition / cancellation ability of the IRS. However, if the angle estimation error is less than 1° (corresponding to an error of less than 1.7 m for a radar-target distance of 100 m), the performance loss is small. The simulation results show that the scheme proposed in the present invention can greatly improve the security of legitimate radar sensing.

[0148] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A radar safety wireless sensing method based on IRS assistance, characterized in that: An IRS with perception function is installed on the target and a two-stage perception is adopted, including: Phase 1: Turn off the IRS reflective element, receive signals from the legal radar LRS and multiple illegal radar URSs through the IRS perception unit, and estimate all radar angle information; Phase 2: Design the IRS reflection coefficient based on the estimation results to minimize the probability of illegal radar perception.

2. The IRS-assisted radar safety wireless sensing method according to claim 1 is characterized in that: In the first stage, a system model is constructed; Assume that the channel matrix between the illegal radar and the IRS at time t is The channel matrix between the legal radar and the IRS at time t is: The distance between the radar and the target is far. Assuming that the channel between LRS / URS and IRS is LoS, The one-dimensional steering vector of the uniform linear array ULA is defined as: Where φ is the signal phase shift difference between two adjacent antennas / units, is the number of antennas / units in the ULA; The AoA / AoD pair of IRS is expressed as The AoA / AoD pairs of LRS and URS are expressed as and make represents the array response vector of IRS, let and represent the array response vectors of URS and LRS respectively; The array response vectors of IRS, URS and LRS are expressed as: Where λ is the signal wavelength, Δe is the element spacing of the IRS, and Δa is the antenna spacing at each radar; Assuming the target moves at a constant speed v, the channel link is affected by the Doppler frequency. Therefore, the far-field LoS channel between the radar and the IRS is modeled as the outer product of the array response vectors on both sides: In the formula is the path gain corresponding to time t, β represents the reference path gain when the distance is 1 meter, represents the distance between two nodes at time t, is the Doppler frequency, T c is the channel coherence interval. Assuming that the channel between the radar and the IRS remains approximately constant during the coherence time, The IRS reflection coefficient vector at time t is and denote the reflection amplitude and phase shift of the nth IRS element respectively. The signal reflected from the IRS received by radar k can be expressed as: Where, θ = diag(θ [t] ) is the phase shift matrix of IRS at time t, and the radar waveform transmitted at time t is satisfy And P>M, L is the number of subcarriers, The variance is σ 2 The zero-mean additive white Gaussian noise matrix, the radar perception process within the coherence time of each channel can be expressed as a binary hypothesis testing problem, as follows: Among them, H0 and H1 represent the target does not exist and the target exists respectively. According to the NP criterion, the detection probability is Q(·) is the Marcum-Q function. Considering the monotonic relationship between the signal-to-noise ratio and the detection probability in equation (8), the signal-to-noise ratio is used as an indicator to evaluate the system performance; Assuming that there is cooperation between URSs and non-cooperation between USRs and LRSs, the signal-to-noise ratios received by URSs and LRSs can be expressed as:

3. The IRS-assisted radar safety wireless sensing method according to claim 1 or 2, characterized in that: In the first stage, the IRS reflective element is turned off, and an L-shaped array consisting of L = Lx + Ly-1 sensing devices is embedded on the IRS to estimate K angle information of all radars; represents the array response vector of the sensing element, the vector They are: The channel between the L-shaped sensing array and the radar k can be expressed as is modeled as the outer product of the array response vectors on both sides; within the channel coherence time, omitting the time index [t], the signal received by the L-shaped sensing array is: in The variance is σ 2 The zero-mean additive white Gaussian noise matrix is is the array response matrix of the L-shaped sensing array, F = [F1, K, F K ] T represents the transmit beamforming signal matrix of K radars, 4. The IRS-assisted radar safety wireless sensing method according to claim 3 is characterized in that: Deleting the time index [t], the corresponding optimization problem is expressed as: Where e is the minimum signal-to-noise ratio required by LRS to achieve target perception performance. The signal-to-noise ratio expressions in (9) and (10) can be simplified by using the channel structure between the radar and the target. Substituting (5) into (9) and (10), the signal-to-noise ratio of URS and LRS can be rewritten as: In the formula and are all constants, and formula (14) can be simplified to Formula (17) is a typical non-convex optimization problem.

5. The IRS-assisted radar safety wireless sensing method according to claim 4 is characterized in that: Formula (17) is further simplified to formula (18): in is the cascade array response, represents a selection matrix where the nth diagonal element is 1 and the rest are 0. Introducing a slack variable z, the problem can be reformulated as follows: Due to the fractional constraint C1 and the non-convex constraint C2, the fractional constraint C1 can be converted into a polynomial expression by introducing an auxiliary variable q k ,q k The optimal value of The optimization problem can be further expressed as: Constraints C1 and C2 are non-convex with respect to the variable θ, by defining X=θθ H (21) The quadratic term θθ H Transformed into the main variable X, the rank 1 Hermitian positive semidefinite matrix X satisfies X=X H ,X≥0 (22) Rank(X)=1 (23) Define the set of all N′N Hermitian positive semidefinite matrices as Transform formula (20) into: Using the SDR algorithm, the original problem can be simplified to 6. The IRS-assisted radar safety wireless sensing method according to claim 5 is characterized in that: Iterative optimization algorithm combining Dinkelbach algorithm and semidefinite relaxation SDR algorithm: 1) Initialization 2)while 3) Other Solve the optimization problem 25 to get X (l+1) ; 4) Use EVD decomposition to get θ (l+1) ; 5) 6)endwhile; 7) output * =θ (l) 。

Images