RIS-assisted ISAC waveform design based on parallel product complex circle manifold
Through the parallel product complex manifold framework and parallel conjugate gradient method, the waveform and RIS reflection coefficient are optimized, which solves the problem of degradation of multi-objective detection and communication performance in radar communication systems, and achieves low-cost multi-objective detection and communication capabilities.
Patent Information
- Application Number
- CN202411177163.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-26
- Publication Date
- 2024-12-03
AI Technical Summary
The existing integrated radar communication system has the problem of performance degradation in multi-object detection and communication, especially in the joint minimization of beam pattern matching error and multi-user interference, and the optimization problem with constant mode constraints is high calculation cost. , it is difficult to achieve comprehensive performance while multi-object detection and communication.
Using the parallel product complex manifold (P2C2M) framework, the reflection coefficients of waveforms and RIS are optimized through the parallel conjugate gradient method, and converted into unconstrained coupling four-time problem (UCQP) to achieve beam pattern matching error and multi-user interference. Joint minimization to improve multi-objective detection performance.
The radar beam pattern matching error with low computing cost and RIS-assisted multi-user interference is realized, which improves the radar's multi-target detection performance and communication capabilities, and reduces computing costs.
Smart Images

Figure CN119070868A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of radar communication integration, and in particular to a method for integrating radar communication with a parallel product complex circle manifold. 2 C 2 M)-based reconfigurable smart surface (RIS)-assisted integrated sensing and communication (ISAC) waveform design method. Background Art
[0002] ISAC plays a vital role by enabling resource sharing between radar sensing and wireless communications, including spectrum, hardware, and signal processing. In addition, the waveform design of ISAC can not only optimize resource usage in congested environments, but also promote collaborative data sharing and enhance communication and sensing capabilities. In practical applications, in order to improve the utilization efficiency of transmitters, constant modulus constraints (CMC) are usually enforced on the waveform. Therefore, the waveform design of ISAC system with CMC has received extensive attention.
[0003] Existing approaches can be mainly divided into two categories: (1) trade-off waveform design without reconfigurable smart surface (RIS) and (2) RIS-assisted waveform design with optimal performance in sensing or communication.
[0004] In the first category of methods, a trade-off performance is usually obtained. Typically, the literature "Dokhanchi SH, Shankar MR B, Alaee-Kerahroodi M, et al. Adaptive Waveform Design for Automotive Joint Radar-Communication Systems [J]. IEEE Transactions on Vehicular Technology, 2021 (70-5).", "Tsinos CG, Arora A, Chatzinotas S, et al. Joint Transmit Waveform and Receive Filter Design for Dual-Function Radar-Communication Systems [J]. IEEE Journal of Selected Topics in Signal Processing, 2021, 15." studies waveform design that does not involve RIS, where the joint optimization of beamforming for a single target and multi-user interference (MUI) in communication is considered simultaneously. However, multi-target detection cannot be achieved. To solve this problem, the paper "F. Liu, C. Masouros, T. Ratnarajah and A. Petropulu," On Range Sidelobe Reduction for Dual-Functional Radar-Communication Waveforms," in IEEE Wireless Communications Letters, vol. 9, no. 9, pp. 1572-1576, Sept. 2020, doi: 10.1109 / LWC.2020.2997959." studied the proposed gradient descent algorithm by updating the conjugate gradient to minimize the weighted beam pattern matching error and multi-user interference, and consider the power budget of each antenna. However, the comprehensive performance of the above method has declined.
[0005] For the second type of methods, the paper "Wang X, Fei Z, Zheng Z, et al. Joint Waveform Design and Passive Beamforming for RIS-Assisted Dual-Functional Radar-Communication System [J]. IEEE Transactions on Vehicular Technology, 2021, 70 (5): 5131-5136." proposed an alternating optimization (AO) framework based on the conjugate gradient method to simultaneously minimize the weighted sum of RIS-assisted multi-user interference and waveform similarity square errors. In order to improve DOF, the paper "An D, Hu J, Huang C. Joint design of transmit waveform and passive beamforming for RIS-assisted ISAC system [J]. Signal Processing: The Official Publication of the European Association for Signal Processing (EURASIP), 2023." proposed an alternating gradient descent (AGRM) method based on Riemann manifold by jointly optimizing the phase shift of the waveform and RIS. However, this method cannot achieve multi-target detection. Summary of the invention
[0006] The present invention proposes a RIS-assisted ISAC waveform design based on parallel product complex circular manifolds, which realizes the joint minimization of radar beam pattern matching error and RIS-assisted communication multi-user interference with low computational cost, and realizes simultaneous multi-target detection and communication with comprehensive performance.
[0007] The technical solution adopted by the present invention is:
[0008] The RIS-assisted ISAC waveform design based on parallel product complex circular manifolds performs the following steps in a RIS-assisted ISAC system including a base station with M antennas, K single-antenna users and L RIS elements:
[0009] Step 1: Construct the optimization problem of RIS-assisted ISAC waveform design:
[0010]
[0011] in, is the set of feasible reflection coefficients on the waveform s=vec(S), the scaling parameter η and RIS The objective function is is a waveform matrix with N snapshots, is the total power of multi-user interference, beam matching model Among them, ω q ≥0 is the weight coefficient of the qth spatial angle, q=1,2,...,Q, Q is the total number of divided spatial angles, b q represents the qth desired transmit beam symbol, s H Ξ(θ q )s is the qth spatial angle θ q The received power at , the joint design factor ρ∈[0,1], s(i) represents the i-th symbol of waveform s, represents the reflection coefficient of the jth RIS element;
[0012] Step 2: Construct a parallel product complex circle manifold
[0013]
[0014] in, and are the feasible regions of waveform and RIS reflection coefficient, both of which belong to complex circular manifolds. and They are defined as:
[0015]
[0016] Step 3: Parallel optimization of the feasible reflection coefficient set on waveform s and RIS based on the parallel conjugate gradient method Get the ISAC waveform design optimization results.
[0017] Furthermore, in step 3, the parallel product complex circular manifold Convert to Euclidean space A Riemann submanifold on , defined using the complex Euclidean inner product On each point The metric of the parallel product complex circular manifold The Riemann gradient on .
[0018] Furthermore, step 3 specifically includes:
[0019] Step 301, calculate the waveform s and the reflection coefficient set based on the orthogonal projection from Euclidean space to tangent space Parallel Riemann gradient
[0020] Step 302, based on the update step size of the current iteration, in parallel multiply the complex circle manifold Update the waveform s and reflection coefficient set The feasible solution of , and the updated value of the next iteration is obtained Where k represents the number of iterations;
[0021] Step 303, determine whether the preset iteration convergence condition is met (the number of iterations reaches the upper limit or the objective function value converges). If not, use Armijo linear search to adaptively update the initial step size of the next iteration, and continue to execute steps 301 to 303; if satisfied, then based on the current update value The ISAC waveform design optimization results are obtained.
[0022] The technical solution provided by the present invention brings at least the following beneficial effects:
[0023] The present invention realizes simultaneous multi-target detection and communication under comprehensive performance based on the joint minimization of beam pattern matching error and RIS-assisted multi-user interference. Since the waveform and RIS phase shift are coupled together, this problem is difficult to solve; in addition, the constant modulus constraint imposed on both the RIS phase shift and the waveform sign further aggravates the difficulty. If the alternating optimization (AO) framework based on the conjugate gradient method is used to solve the beam optimization problem constructed by the present invention, it will lead to huge computational costs. The present invention is based on the parallel product complex circular manifold (P 2 C 2 M) framework, which uses the natural normal characteristics of waveforms and RIS to transform the beam optimization problem into a P 2 C 2 The unconstrained coupled quartic problem (UCQP) on M is used to derive a parallel conjugate gradient method to optimize the waveform and RIS in parallel to obtain the optimization result of the constructed beam optimization problem, thereby achieving the goal of improving the multi-target detection performance of the radar while providing services to all users of the ISAC system with the help of RIS. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0025] Figure 1 Convergence comparison diagram of the ISAC waveform design based on parallel product complex circle manifold RIS-assisted proposed in the embodiment of the present invention (Proposed method) and the alternating gradient descent (AGRM) algorithm based on Riemann manifold in different scenarios, where (1-a) corresponds to Case I scenario, and (1-b) corresponds to Case II;
[0026] Figure 2 The joint design performance comparison diagram of BUSM, AO, AGRM and the proposed method in the embodiment of the present invention in Case I scenario, where (2-a) corresponds to the beam pattern and (2-b) corresponds to the achievable total rate;
[0027] Figure 3 The joint design performance comparison diagram of BUSM, AO, AGRM and the proposed method in the embodiment of the present invention in the Case II scenario, where (3-a) corresponds to the beam pattern and (3-b) corresponds to the achievable total rate;
[0028] Figure 4 The performance comparison diagram of BUSM, AGRM and the proposed method of the embodiment of the present invention under different numbers of RIS reflection coefficients, wherein (4-a) corresponds to the normalized matching error (NME), and (4-b) corresponds to the achievable total rate under SNR = 25dB;
[0029] Figure 5 This is a comparison chart of the running time of the AGRM method and the proposed method in the embodiment of the present invention;
[0030] Figure 6 This is a graph showing the relationship between the objective function (parallel conjugate gradient algorithm PCG) and the running time in the Case II scenario. DETAILED DESCRIPTION
[0031] To make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be described in detail and completely in conjunction with the drawings in the implementation of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings can be arranged and designed using different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the present application for protection, but merely represents selected embodiments of the present invention.
[0032] For the convenience of description, the following definitions are given first:
[0033] The transpose, conjugate, and conjugate transpose operations are respectively represented by (·) T ,(·) * ,(·) H express; Represents a complex domain; I N represents the N×N identity matrix; vec(·) represents the matrix column vector; diag(·) represents diagonalization; ||·|| F,|·|represent the matrix norm and the modulus of the complex number respectively; ⊙ represents the Hadamard product, represents the Kronecker product; represents the real part of a complex number; represents the expectation of a random variable; represents the Euclidean gradient of the variable X.
[0034] The embodiment of the present invention considers a RIS-assisted ISAC system consisting of a base station (BS) with M antennas, K single-antenna users, and L RIS elements. The system provides services for K users with the help of RIS, while improving the multi-target detection performance of the radar. All channels are assumed to be block-fading and remain constant within each frame. Under the assumption that the base station performs perfect estimation through pilot symbols, the received signal can be expressed as:
[0035]
[0036] in, and They represent the baseband channels from the base station to the user, from the reconfigurable intelligent surface (RIS) to the user, and from the base station to the RIS, respectively; represents the phase shift matrix of RIS, represents the set of feasible reflection coefficients on RIS, also called phase shift represents a waveform matrix with N snapshots, represents the noise matrix at the receiving end, and the noise of user k is 0 is the all-zero matrix, I is the identity matrix, σ 2 is the noise variance, and the expected symbol matrix at the user is expressed as The received signal can be rewritten as:
[0037]
[0038] in, This is called multi-user interference (MUI).
[0039] The total power of MUI is expressed as:
[0040]
[0041] This has a significant impact on the total rate achievable by the user. Specifically, the total rate achievable is expressed as:
[0042]
[0043] in
[0044]
[0045] Among them, γ k represents the received signal-to-interference-noise ratio of user k, x k,j represents the jth expected symbol of user k, yes The kth row of s j is the j-th column of S, E(·) is the expected operation, j=1,…,N.
[0046] Waveform diversity can be used to enhance radar detection capabilities by synthesizing specific transmit waveforms, which enables precise management of the spatial distribution of transmit power to improve performance. The spatial region of interest of the radar is denoted as Θ. Typically, a dot matrix is selected. Q is the total number of spatial angles, in θ q The power received at is:
[0047]
[0048] in is the waveform matrix, s = vec(S), a t (θ q ) is the steering vector, defined as:
[0049]
[0050] The desired transmit beam is b = [b1, b2, ..., b Q ] T Then, the beamforming matching model is as follows:
[0051]
[0052] Among them, ω q ≥0(q=1,2,...,Q) is the weight coefficient, and η is the proportional variable, that is, the scaling parameter.
[0053] Based on the above analysis, the modeling of the optimization problem can be expressed as follows:
[0054]
[0055] where ρ∈[0,1] is the joint design factor, and the two constraints are RIS and constant modulus constraint of the waveform; due to the waveform s, scaling parameter η and phase shift The coupling effect of the RIS reflection element and the waveform symbol makes this optimization problem difficult to solve. In addition, the constant modulus constraints on the RIS reflection element and the waveform symbol and the quartic programming form of the beam matching expression make the above optimization problem more difficult to solve. In the embodiment of the present invention, the parallel product complex circle manifold method is used to solve the optimization problem.
[0056] Although existing methods can solve the optimization problem shown in formula (9), this usually requires huge computational costs and is easy to fall into a local minimum solution. To address this problem, the embodiment of the present invention proposes a parallel product complex circular manifold (P 2 C 2 M) framework, using the inherent constant mode characteristics of waveforms and RIS. First, construct (P 2 C 2 M) space as the solution space to accommodate the waveform and RIS reflection coefficient. 2 C 2 M), the problem is converted into (P 2 C 2 M), and derive a parallel conjugate gradient (PCG) algorithm to optimize the waveform and RIS in parallel.
[0057] First build P 2 C 2 In general, the constrained optimization problem can be transformed into an unconstrained optimization problem on a specially designed manifold space that contains all the constraints. The feasible region of the waveform and the RIS with constant modulus constraints in the optimization problem shown in formula (9) is equivalent to P 2 C 2 M space, which consists of the product of two complex circular manifolds. The dimension of this space is MN+L and can be expressed as:
[0058]
[0059] in, and are the feasible regions of waveform and RIS reflection coefficient respectively. and All belong to complex circular manifolds, and they can be defined as:
[0060]
[0061]
[0062] Where s(i) represents the i-th symbol of the waveform and ψ(j) represents the reflection coefficient of the j-th RIS element.
[0063] In order to 2 C 2 An efficient gradient-based algorithm is derived in M space, which uses the property that smooth manifolds can be locally linearized around each point. The linearization at Tangent space is defined as follows:
[0064]
[0065] Points on any smooth curve m on the manifold The tangent vector Indicates the direction in which it can move. Cut Space It can be decomposed into the product of two tangent spaces:
[0066]
[0067] in and The tangent spaces at are:
[0068]
[0069] Among them, ξ s and They are s and The tangent vector at MN and 0 L is the zero vector; from this we can get the point on the manifold The tangent space can be expressed as:
[0070]
[0071] To obtain The gradient of the cost function on needs to be chosen The embodiment of the present invention chooses to assign an induced Riemann metric to it. Convert to Euclidean space is a Riemann submanifold on ; therefore, the conventional complex Euclidean inner product is used to define each point on the manifold The measure of is given by:
[0072]
[0073] in, Based on the metric formula (18), P 2 C 2 Manifolds in M-space Transformed into Euclidean space Riemann submanifold on .
[0074] Then, the problem is solved using a parallel conjugate gradient algorithm, which consists of three main steps: 1) computing parallel Riemann gradients, 2) updating feasible solutions, and 3) adaptively updating the step size using Armijo linear search.
[0075] In P 2 C 2In M space, the optimization problem shown in formula (9) has been transformed into an unconstrained coupled quartic problem (UCQP), and the problem can be rewritten as follows:
[0076]
[0077] According to (19), when the waveform s is given, the scaling parameter η has a closed-form solution, expressed as:
[0078]
[0079] Where v = [v1,...,v Q ] T And v q =s H Ξ(θ q )s.
[0080] Derivation of s and The steps are as follows: First, calculate the parallel Riemann gradient. Given η, formula (19) can be transformed into:
[0081]
[0082] Where x = vec(X), The Riemann gradient Defined as tangent space An element on , which satisfies:
[0083]
[0084] in yes At the point Chu Yan Directional derivative of direction; because P 2 C 2 M is the Euclidean space The directional derivative can be expressed as:
[0085]
[0086] in, yes The Euclidean gradient of It means arrive Orthogonal projection of .
[0087] It follows that in the product manifold Parallel Riemann gradient on It can be expressed as:
[0088]
[0089] in, and They represent s and Based on formula (21), the Euclidean gradient of s can be calculated as follows:
[0090]
[0091] where t q =Ξ(θ q )s, similar, remove Irrelevant items, The Euclidean gradient of can be calculated as follows:
[0092]
[0093] in u=diag(U),
[0094] From this we can get the Riemann gradient and It can be obtained by the orthogonal projection from Euclidean space to tangent space, and the formula is expressed as:
[0095]
[0096] Generally speaking, the steepest gradient descent method can be used to solve the manifold The update criterion for the unconstrained optimization problem on is:
[0097] e k+1 =e k+1 +α k d k , (28)
[0098] where d k = -gradf(e k ) is the fastest gradient descent direction, α k >0 is the descending step length. However, in practical applications, the convergence speed of the steepest gradient descent algorithm is relatively slow, and it is easy to fall into the local minimum solution; unlike the steepest gradient descent algorithm, the embodiment of the present invention notes that the quaternary optimization process is equivalent to the movement of particles on a high-dimensional plane, which indicates that the actual descending direction is a composite direction of the steepest gradient descent direction and the initial momentum direction. Based on this feature, the embodiment of the present invention derives the parallel conjugate gradient (PCG) algorithm to efficiently solve the optimization problem given by equation (21).
[0099] Descent direction in PCG algorithm and The expressions are as follows:
[0100]
[0101] where β k ≥0 is the conjugation parameter, which can be calculated by the following formula:
[0102]
[0103] in and
[0104]
[0105] The projection transformation is introduced in formula (31) to solve the calculation relationship between tangent vectors in different tangent spaces. from Project to another point The calculation formula is:
[0106]
[0107] in
[0108]
[0109] Based on the above analysis, in the manifold superior arrive The update process can be reformulated as:
[0110]
[0111] where α k is the step size of the adaptive update of the Armijo linear search algorithm given in the next step. The retraction operation can be expressed as:
[0112]
[0113] Right now
[0114] Where R s (ξ s )and They are all based on the retraction operation on the complex circular manifold, and this operation can be defined as:
[0115]
[0116] So R s (ξ s )and They can be expressed as:
[0117]
[0118] In addition, the overall descent direction is defined as To simplify the subsequent calculation process; using the Armijo linear search strategy to update the step size, the search step size can be adaptively adjusted according to the change of momentum, thereby ensuring the non-incremental nature of the algorithm and improving the convergence speed of the algorithm. The adaptive update strategy can be expressed as:
[0119]
[0120] in Used to characterize the initial step size of the kth iteration, τ2 is a coefficient greater than 0 and less than 1, and n is the number of linear searches.
[0121] In order to further speed up the convergence speed, the initial step size α of the next iteration can be appropriately adjusted according to the change of the number of linear searches n (i.e. ). If n = 1, it means that the step size is too small, and the initial step size of the next iteration should be increased by α = ξ1α k ,ξ1>1; if n=2, it means that the initial step size is reasonable, and the initial step size of the next iteration should be kept as α=α k ; If n>2, it means that the step size of the current point is too small, and the initial step size of the next iteration should be increased to α=ξ2α k ,ξ2>1. In summary, the update rule for the initial step size α of the next iteration is as follows:
[0122]
[0123] In summary, the specific implementation of the Armijo linear search algorithm used in the embodiment of the present invention is shown in Table 1, which is used to solve the overall P of the ISAC waveform design optimization problem (9). 2 C 2 The specific implementation of the M algorithm is shown in Table 2.
[0124] Table 1
[0125]
[0126] Table 2
[0127]
[0128]
[0129] In order to verify the performance of the waveform designed by the RIS-assisted ISAC waveform design based on parallel product complex circular manifolds proposed in the embodiment of the present invention, the waveform design performance is compared between the existing scheme and the scheme proposed in the embodiment of the present invention, wherein the existing scheme includes a RIS-free waveform design method and a RIS-assisted waveform design method.
[0130] The simulation experiment conditions are set as follows: consider a RIS-assisted uniform linear array synaesthesia system, the number of antennas M = 6, the array spacing is half a wavelength, and it is assumed that N = 4 samples are sampled within a pulse interval; in the downlink, there are K = 4 users in the synaesthesia system, the channel parameter H obeys the complex Gaussian distribution, the carrier is f0 = 3GHz, and the expected symbol matrix X is generated according to the unit power QPSK symbol table. For radar detection, the spatial region of interest is φ = [-90°, 90°] and we set The Q=181 and the interval is 1°.
[0131] The scenarios considered include the following two cases.
[0132] Case I: The weight coefficient is fixed to ρ = 0.5, the number of different RIS reflection coefficients is L = 45, and the expected waveform has only one main lobe centered at The width is Δθ = 60°, and the calculation formula of normalized matching error (NME) is NME = J R (η,s) / η 2 ;
[0133] Case II: The weight coefficient is ρ = 0.5, the number of different RIS reflection coefficients is L = 45, and the expected waveform has three main lobes, located at The width of each main lobe is Δθ=20°. More specifically, the desired waveform d(θ p ) is expressed as
[0134]
[0135] Among them, j=1,2,3.
[0136] For fair comparison, the initialization of all methods used for comparison is the same, the initialization of the RIS phase shifter is a unit amplitude complex random vector, and the initialization waveform is set to an orthogonal linear frequency modulation (LFM) waveform.
[0137] The existing solution adopted in this embodiment is:
[0138] (1) BUSM algorithm: Reference: S. Shi, Z. Wang, Z. He, and Z. Cheng, “Constrained waveform design for dual-functional MIMO radar-communication system,” Signal Processing, vol. 171, p. 107530, 2020.
[0139] (2) AO algorithm: Reference "X.Wang, Z.Fei, Z.Zheng, and J.Guo, "Joint waveformdesign and passive beamforming for RIS-assisted dual-functional radar-communication system," IEEE Transactions on Vehicular Technology, vol.70, no.5, pp.5131–5136, 2021.";
[0140] (3) AGRM algorithm: Reference: D.An, J.Hu, and C.Huang, “Joint design of transmitwaveform and passive beamforming for RIS-assisted ISAC system,” SignalProcessing, vol. 204, p. 108854, 2023.
[0141] Figure 1 The convergence comparison of the RIS-assisted ISAC waveform design based on parallel product complex circular manifold and the AGRM algorithm in different scenarios is shown, with the weight coefficient set to 0.5. Figure 1(1-a) and (1-b) in the figure respectively show the relationship between the cost function and the number of iterations in the Case I and Case II scenarios. It can be seen from Figure (1-a) that in the Case I scenario, the solution proposed in the embodiment of the present invention converges to 0.4306 after 351 iterations, while the AGRM method converges to 0.4818 after 1793 iterations; it can be seen from Figure (1-b) that the solution proposed in the embodiment of the present invention converges faster than the AGRM algorithm. Specifically, the objective function of the solution proposed in the embodiment of the present invention converges to 1.159 after 351 iterations, while the objective function of the AGRM algorithm converges to 1.229 after 1811 iterations. In addition, it can be clearly seen that after 600 iterations, the objective function of the AGRM method is reduced by a small value, which means that the performance is significantly reduced.
[0142] Figure 2 The joint design performance comparison of BUSM, AO, AGRM and the solution proposed in the embodiment of the present invention in the Case I scenario is shown, where the experimental results are the average of 50 Monte Carlo runs. Figure 2 (2-a) in the figure shows the beam pattern matching comparison. In addition, it can be seen from (2-a) that the scheme proposed in the embodiment of the present invention obtains lower radar beam matching error and sidelobe, while the BUSM method without RIS and the AO method with RIS have poor performance. Generally speaking, the NME obtained by the scheme proposed in the embodiment of the present invention is 17.72dB (0.0169), which is about 1.61dB lower than AO, about 1.38dB lower than AGRM, and about 3.59dB lower than BSUM. Figure 2 (2-b) in FIG. 1 shows the relationship between the achievable sum rate and the signal-to-noise ratio. As can be seen from the figure, compared with the BUSM, AO and AGRM methods, the scheme proposed in the embodiment of the present invention achieves better performance in terms of the achievable sum rate of communication. Generally, when SNR = 25dB, the achievable sum rate of the scheme proposed in the embodiment of the present invention is about 0.77 bits / second / Hz higher than the AGRM method, 14.47 bits / second / Hz higher than the AO method, and about 24 bits / second / Hz higher than the BUSM method without RIS.
[0143] Figure 3 The joint design performance comparison of BUSM, AO, AGRM and the scheme proposed in the embodiment of the present invention in Case II is shown, and the experimental results are the average of 50 Monte Carlo runs. Figure 3(3-a) in Figure 1 shows the beam pattern matching comparison. As shown in the figure, the scheme proposed in the embodiment of the present invention obtains lower beam pattern matching error and radar sidelobe. Under normal circumstances, the sidelobe null value obtained by the scheme proposed in the embodiment of the present invention is -14.84dB (0.03282), which is 1.77dB lower than AO, 4.77dB lower than AGRM, and 17.1dB lower than BUSM. Figure 3 (3-b) in Figure 3 shows the relationship between the achievable sum rate and the signal-to-noise ratio. It can be seen that compared with the BUSM, AO and AGRM methods, the scheme proposed in the embodiment of the present invention achieves better performance in terms of the achievable sum rate of communication. Generally, when SNR = 25dB, the achievable sum rate of the scheme proposed in the embodiment of the present invention is about 15.37 bits / s / Hz higher than the AO method, 0.74 bits / s / Hz higher than the AGRM method, and 24 bits / s / Hz higher than the BUSM method without RIS.
[0144] Figure 4 The performance comparison of BUSM, AGRM and the scheme proposed in the embodiment of the present invention is shown, in which different numbers of RIS reflection coefficients are considered. The experimental results are the average of 50 Monte Carlo runs. Figure 4 (4-a) in Figure 4 shows the NME of BUSM, AGRM and the scheme proposed in the embodiment of the present invention under different numbers of RIS reflection coefficients. It can be seen from the figure that as the number of RIS reflection coefficients increases, the NME of beam pattern matching decreases. RIS-assisted integrated waveform design achieves better performance, while the BUSM method without RIS performs poorly. In Case II, when L=30, the NME obtained by the scheme proposed in the embodiment of the present invention is 0.0515, which is 0.0096 higher than that of the AGRM method. Figure 4 (4-b) in the figure shows the achievable sum rate for different numbers of RIS reflection coefficients obtained by BUSM, AGRM and the scheme proposed in the embodiment of the present invention when the signal-to-noise ratio is 25 dB. It can be seen that the achievable sum rate of communication increases with the increase in the number of RIS reflection coefficients. RIS-assisted integrated waveform design achieves better performance, while the BUSM method without RIS performs poorly. Generally, when L=30, the achievable sum rate obtained by the scheme proposed in the embodiment of the present invention is 2.29 bits / s / Hz higher than that of the AGRM method and about 22.96 bits / s / Hz higher than that of the BUSM method without RIS.
[0145] Figure 5 The running time comparison of the AGRM method and the solution proposed in the embodiment of the present invention is shown, where the experimental results are the average of 50 Monte Carlo runs. As shown in the figure, the running time required by the solution proposed in the embodiment of the present invention is about half of that of the AGRM method.
[0146] Figure 6 The relationship between the objective function and the running time of the embodiment of the present invention in Case II is shown. The curve in the figure is the result obtained under 50 different inputs. It can be seen that the performance of the scheme proposed in the embodiment of the present invention at 0.5 seconds is slightly better than the performance of the AGRM method at 4.5 seconds. The calculation time of the scheme proposed in the embodiment of the present invention is about one order of magnitude lower than that of the AGRM method, while the performance is almost the same.
[0147] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
[0148] The above are only some embodiments of the present invention. For those skilled in the art, several modifications and improvements can be made without departing from the creative concept of the present invention, which all belong to the protection scope of the present invention.
Claims
1. RIS-assisted ISAC waveform design based on parallel product complex circular manifolds, characterized by: The following steps are performed in a RIS-assisted ISAC system including a base station with M antennas, K single-antenna users and L RIS elements: Step 1: Construct the optimization problem of RIS-assisted ISAC waveform design: in, is the set of feasible reflection coefficients on the waveform s=vec(S), the scaling parameter η and RIS The objective function is is a waveform matrix with N snapshots, is the total power of multi-user interference, beam matching model Among them, ω q ≥0 is the weight coefficient of the qth spatial angle, q=1,2,...,Q, Q is the total number of divided spatial angles, b q represents the qth desired transmit beam symbol, s H Ξ(θ q )s is the qth spatial angle θ q The received power at , the joint design factor ρ∈[0,1], s(i) represents the i-th symbol of waveform s, represents the reflection coefficient of the jth RIS element; Step 2: Construct a parallel product complex circle manifold in, and are the feasible regions of waveform and RIS reflection coefficient, both of which belong to complex circular manifolds. and They are defined as: Step 3: Parallel optimization of the feasible reflection coefficient set on waveform s and RIS based on the parallel conjugate gradient method Get the ISAC waveform design optimization results.
2. The RIS-assisted ISAC waveform design based on parallel product complex circular manifolds as claimed in claim 1, characterized in that: In step 3, parallel product complex circular manifold Convert to Euclidean space A Riemann submanifold on , defined using the complex Euclidean inner product On each point The metric of the parallel product complex circular manifold The Riemann gradient on .
3. The RIS-assisted ISAC waveform design based on parallel product complex circular manifolds as claimed in claim 2, characterized in that: Step 3 specifically includes: Step 301, calculate the waveform s and the reflection coefficient set based on the orthogonal projection from Euclidean space to tangent space Parallel Riemann gradient Step 302, based on the update step size of the current iteration, in parallel multiply the complex circle manifold Update the waveform s and reflection coefficient set The feasible solution of , and the updated value of the next iteration is obtained Where k represents the number of iterations; Step 303, determine whether the preset iterative convergence condition is met. If not, use Armijo linear search to adaptively update the initial step size of the next iteration, and continue to execute steps 301 to 303; if satisfied, then based on the current update value The ISAC waveform design optimization results are obtained.
4. The RIS-assisted ISAC waveform design based on parallel product complex circular manifolds as claimed in claim 3, characterized in that: The iterative convergence condition is that the number of iterations reaches the upper limit or the objective function value converges.
5. The RIS-assisted ISAC waveform design based on parallel product complex circular manifolds as claimed in claim 3, characterized in that: The scaling parameter η is set to: Where, the desired transmit beam b = [b1, b2, ..., b Q ] T , auxiliary quantity v=[v1,...,v Q ] T , and v q =s H Ξ(θ q )s,q=1,2,...,Q.
6. The RIS-assisted ISAC waveform design based on parallel product complex circular manifolds according to any one of claims 3 to 5, characterized in that: In step 3, s and s are optimized in parallel based on the parallel conjugate gradient method. When , the objective function is transformed into: Among them, the expected symbol vector x = vec(X), the expected symbol matrix at the user Auxiliary amount I N Represents the N×N identity matrix, the phase shift matrix of RIS and represent the baseband channels from the base station to the user, from the RIS to the user, and from the base station to the RIS, respectively, ||·|| F is the symbol for the matrix bi-norm.
Citation Information
Cited By
RIS-assisted MIMO radar transmission beam forming method based on beam pattern matching
CN119986621A
RIS-Assisted MIMO Radar Transmit Beamforming Method Based on Beam Pattern Matching
CN119986621B
