Ris-assisted distributed unmanned aerial vehicle multi-target regional power synthesis method

Abstract

The application relates to a distributed unmanned aerial vehicle multi-target area power synthesis method based on RIS assistance, which comprises the following steps of system modeling and parameter definition, generalized steering vector calculation, establishment of a received signal model, establishment of a multi-target optimization model, derivation of an upper limit of optimization, smoothing and approximation of a target function, effective solution by adopting an alternating optimization framework and a Riemann conjugate gradient algorithm. The RIS-assisted distributed power synthesis method can simultaneously provide differentiated power delivery for multiple target areas under an NLOS propagation environment, realizes cooperative optimization of the multiple target areas, and significantly improves the energy utilization efficiency and actual deployability of the system. The method has good flexibility and scalability, the number of target areas can be expanded by adjusting a set of weighting coefficients, and any combination configuration of the LOS nodes and the NLOS nodes is supported, so that the actual deployment environment can be flexibly adjusted.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology and relates to a RIS-assisted distributed UAV multi-target area power synthesis method. Background Technology

[0002] With the rapid development of technologies such as the Internet of Things (IoT), 6G communication, and smart cities, the demand for wireless power transfer (WPT) and spatial power control technologies in communication, power transfer, and other fields is increasing. Traditional wireless power transfer systems are mainly based on beamforming technology with a single transmitter. By adjusting the amplitude and phase weights of the antenna array, a directional beam is formed in two-dimensional angular space (azimuth and elevation angles) to deliver energy to the target area.

[0003] However, traditional beamforming methods have the following limitations. First, traditional methods can only control the distribution of electromagnetic energy in the angular dimension, lacking effective control over the distance dimension. When multiple targets are located at the same angle but different distances, traditional beamforming struggles to achieve precise distance-dimensional differentiation. Second, in scenarios requiring energy delivery to a specific three-dimensional volume region while avoiding electromagnetic interference with surrounding sensitive areas, the three-dimensional spatial selectivity of traditional methods needs improvement. Finally, traditional systems typically only provide a single power level output, making it difficult to simultaneously meet the needs of multiple target regions with different power requirements.

[0004] In recent years, three-dimensional spatial power combining technology based on distributed arrays has attracted much research attention. This technology draws on the principles of focused ultrasound surgery (FUS) in the medical field: by transmitting specifically designed waveforms through multiple spatially distributed transmitting nodes, energy is focused within a target three-dimensional volume region by utilizing the coherent superposition effect of electromagnetic waves in space, while simultaneously suppressing energy leakage within a protected region. Current research on three-dimensional spatial power combining using distributed arrays mainly focuses on aspects such as distributed array position optimization, derivation of positioning error boundaries, single-shot signal design, grating lobe suppression, and waveform design.

[0005] In real-world applications, especially in urban environments, indoor settings, or complex terrain, obstacles often obstruct the connection between the transmitting node and the target area, preventing the establishment of a direct line-of-sight (LOS) transmission link. These obstacles cause significant penetration loss, diffraction loss, and scattering loss. Some transmitting nodes, due to obstruction, cannot effectively contribute energy to the target area, reducing the optimizable variables and limiting performance limits. In more severe LOS scenarios, when all nodes are obstructed, the system completely fails, and due to the unpredictability of multipath propagation, unexpected energy hotspots may appear in the protected area.

[0006] Reconfigurable Smart Surface (RIS) technology offers a new approach to addressing the aforementioned bottlenecks. RIS is a reflective surface composed of numerous low-cost, electronically tunable reflective elements, whose reflection characteristics can be controlled by software, enabling flexible regulation of electromagnetic wave propagation. RIS-assisted wireless power transfer technology combines RIS with Wireless Power Communication Networks (WPCNs) to improve power transfer efficiency. RIS-assisted WPCN system resource allocation maximizes the system's weighted sum rate by optimizing RIS reflection beamforming, power transfer, and time slot allocation. UAV-RIS collaboratively optimizes UAV trajectory, RIS phase shift, and power allocation to improve system throughput. RIS-enabled MIMO-WPCN maximizes system spectral efficiency. Cellular-free massive MIMO employs a multi-UAV + RIS + distributed access point architecture to improve edge user performance and ensure fairness.

[0007] Existing technologies still have some drawbacks:

[0008] (1) Existing technologies only optimize power combining for a single target area. In practical application scenarios, there are often multiple target areas that require energy delivery. Single-target technologies can only select to serve one area, and there is room for improvement in the resource utilization efficiency of the system in multi-target scenarios.

[0009] (2) Different target regions may have different power requirements and priorities. Existing optimization objectives are mainly based on a single scalar function, which needs to be improved in modeling the differentiated requirements and priority relationships between multiple regions. This may lead to a mismatch between resource allocation and actual needs.

[0010] (3) Existing technologies are usually based on the assumption that all transmitting nodes propagate at LOS (line-of-sight) distances to the target area, or are deployed in controlled environments. In real-world environments such as urban environments, indoor scenes, or complex terrain, signals transmitted by nodes completely blocked by obstacles cannot reach the target area, resulting in limited functionality of some nodes. Although RIS technology provides a solution to the NLOS problem, further research is needed on how to organically combine RIS with multi-target power combining. Summary of the Invention

[0011] To address the problems existing in the above-mentioned traditional methods, this invention proposes a RIS-assisted distributed UAV multi-target area power synthesis method.

[0012] To achieve the above objectives, the embodiments of the present invention adopt the following technical solutions:

[0013] A RIS-assisted distributed UAV multi-target area power synthesis method is provided, including the following steps:

[0014] Step 1: Deployment The drones equipped with transmitters are divided into two groups based on their line-of-sight (LOS) and non-line-of-sight (NLOS) conditions with respect to the target area. The LOS group includes... Each drone can reach the target area directly; the NLOS group includes... The drone was obstructed by an obstacle. Deploy a carrier RIS with one reflection unit.

[0015] Step 2: Calculate the LOS node steering vector, RIS steering vector, and NLOS node to RIS steering vector matrix.

[0016] Step 3: Based on the LOS node steering vector, RIS steering vector, and NLOS node to RIS steering vector matrix, establish the received signal model and calculate the average synthesized power at the region of interest.

[0017] Step 4: With the minimum combined power in the set of all target regions and the maximum combined power in the set of protected regions as objectives, and considering constant mode constraints and RIS reflection coefficient constraints, construct a minimax multi-objective optimization model.

[0018] Step 5: Determine the upper bound of the objective function optimization based on the minima-maximum multi-objective optimization model.

[0019] Step 6: Smooth and approximate the objective function of the minima-maximum multi-objective optimization model to obtain the first multi-objective optimization model.

[0020] Step 7: The first multi-objective optimization model is optimized and solved using the alternating optimization framework and the Riemann conjugate gradient algorithm to obtain the combined power of the distributed UAV multi-objective region.

[0021] One of the above technical solutions has the following advantages and beneficial effects:

[0022] The aforementioned RIS-assisted distributed UAV multi-target area power combining method includes: system modeling and parameter definition, generalized steering vector calculation, establishing a received signal model, establishing a multi-objective optimization model, and deriving the optimization upper bound; smoothing and approximating the objective function; and using an alternating optimization framework and the Riemann conjugate gradient algorithm for effective solution. This RIS-assisted distributed power combining method can simultaneously provide differentiated power delivery to multiple target areas in an NLOS propagation environment, achieving collaborative optimization of multiple target areas and significantly improving the system's energy utilization efficiency and practical deployability. This method has good flexibility and scalability; the number of target areas can be expanded by adjusting the weighting coefficient set, while simultaneously supporting... LOS nodes and Any combination of NLOS nodes can be configured and flexibly adjusted according to the actual deployment environment. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of this application or the conventional technology, the drawings used in the description of the embodiments or the conventional technology will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart illustrating a RIS-assisted distributed UAV multi-target area power synthesis method in one embodiment.

[0025] Figure 2 This is a RIS-assisted distributed UAV multi-target region power synthesis spatial model in one embodiment;

[0026] Figure 3 This is a scene example diagram of two target regions considered in one embodiment;

[0027] Figure 4 This is a schematic diagram of the energy distribution in the region of interest under different weighting coefficients in one embodiment of the method, wherein... Figure 4 (a) is A schematic diagram of the energy distribution over the region of interest. Figure 4 (b) is A schematic diagram of the energy distribution over the region of interest. Figure 4 (c) is A schematic diagram of the energy distribution over the region of interest. Figure 4 (d) is A schematic diagram of the energy distribution over the region of interest;

[0028] Figure 5 This is a schematic diagram illustrating the change of the objective function of the method with the number of iterations under different weighting coefficients in one embodiment;

[0029] Figure 6 This is a schematic diagram illustrating the change of the combined power of target region 1 and target region 2 with the number of iterations under different weighting coefficients in one embodiment of the method.

[0030] Figure 7 This is a schematic diagram illustrating the changes in the minimum and maximum combined power of the target region 1 and the protected region 1 with the number of iterations under different weighting coefficients in one embodiment of the method. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0032] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the application.

[0033] It should be noted that, in this document, the reference to "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The presentation of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will understand that the embodiments described herein can be combined with other embodiments. The term "and / or" as used herein refers to any combination of one or more of the associated listed items, and all possible combinations, including such combinations.

[0034] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0035] In one embodiment, such as Figure 1 As shown, a RIS-assisted distributed UAV multi-target area power synthesis method is provided, which may include the following processing steps 1 to 7:

[0036] Step 1: Deployment The drones equipped with transmitters are divided into two groups based on their line-of-sight (LOS) and non-line-of-sight (NLOS) conditions with respect to the target area. The LOS group includes... Each drone can reach the target area directly; the NLOS group includes... The drone was obstructed by an obstacle. Deploy a carrier RIS with one reflection unit.

[0037] Specifically, for the NLOS propagation environment, a RIS-assisted multi-objective joint optimization method is proposed, which organically combines the RIS reflection mechanism with the multi-objective weighted optimization framework to realize multi-region power synthesis in the NLOS scenario.

[0038] RIS-assisted distributed UAV multi-target regional power synthesis spatial model, such as Figure 2 As shown.

[0039] deploy The drones equipped with transmitters are divided into two groups based on their line-of-sight (LOS) and non-line-of-sight (NLOS) conditions relative to the target area. The LOS group includes... Each node can directly reach the target area; the NLOS group includes... A node is obstructed by an obstacle and requires RIS assistance. A device equipped with RIS of a single reflection unit Represents the reflection coefficient of RIS, where , indicating the first The reflection coefficient of each reflecting element . For the first The node at the th The baseband signal sent by each snapshot indicates that... Representing waveform length, the waveform matrices transmitted by LOS UAV and NLOS UAV are... and It can be represented as:

[0040] (1)

[0041] (2)

[0042] in, They are the first, the second, and the third, respectively. The transmitted waveform vector of a LOS UAV. For the first d The LOS UAV in the first e The baseband signal sent by each snapshot , , They are the first, the second, and the third, respectively. The transmit waveform vector of each NLOS UAV For the first g The NLOS UAV in the first h The baseband signal sent by each snapshot , .

[0043] Step 2: Calculate the LOS node steering vector, RIS steering vector, and NLOS node to RIS steering vector matrix.

[0044] Step 3: Based on the LOS node steering vector, RIS steering vector, and NLOS node to RIS steering vector matrix, establish the received signal model and calculate the average synthesized power at the region of interest.

[0045] Specifically, the LOS node steering vector, RIS steering vector, and NLOS node-to-RIS steering vector matrix are then synthesized to obtain the result. The synthesized signal is then used to obtain the received signal model. Based on the received signal model, the signal at which the received signal is synthesized is determined. Within a discrete time interval, the region of interest is a point. The average combined power at that location.

[0046] Step 4: With the minimum combined power in the set of all target regions and the maximum combined power in the set of protected regions as objectives, and considering constant mode constraints and RIS reflection coefficient constraints, construct a minimax multi-objective optimization model.

[0047] Specifically, a multi-objective optimization framework is established, expanding the optimization objective from a single objective to multiple objectives, and a multi-objective regional power synthesis model is built. To ensure robust performance across multiple objective regions, the priority of these regions is considered. Introducing regularization parameters A Min-Max multi-objective optimization model was established.

[0048] A generalized steering vector for NLOS scenarios was derived, and an optimization model for NLOS scenarios was established. With the assistance of a UAV equipped with RIS, a reflection path can be provided for NLOS nodes obstructed by obstacles, enabling them to perform power synthesis on specific target areas, thereby compensating for the performance loss caused by obstacle obstruction.

[0049] A minima-maxima multi-objective optimization model was established for the minimum composite power in the target region set and the maximum composite power in the protection region set. The model was used to jointly optimize the minimum composite power in the target region and the maximum composite power in the protection region, which effectively improved the robustness of the composite power distribution.

[0050] Step 5: Determine the upper bound of the objective function optimization based on the minima-maximum multi-objective optimization model.

[0051] Specifically, based on the mathematical structure of the problem model, combined with the properties of semi-definite Hermitian matrices and unitary matrices, as well as the Rayleigh-Ritz theorem, the upper bound of the objective function was derived.

[0052] Step 6: Smooth and approximate the objective function of the minima-maximum multi-objective optimization model to obtain the first multi-objective optimization model.

[0053] Specifically, the objective function is smoothed and approximated. Using l p The approximate max function of the norm, and the objective function's By omitting the power constant, the non-smooth problem is transformed into a smooth problem.

[0054] Step 7: The first multi-objective optimization model is optimized and solved using the alternating optimization framework and the Riemann conjugate gradient algorithm to obtain the combined power of the distributed UAV multi-objective region.

[0055] Specifically, an alternating optimization framework and the Riemann conjugate gradient algorithm are used for efficient solution. The constant modulus constraint is treated as a complex circular manifold, transforming the constrained problem into an unconstrained problem, which is then solved efficiently using the Riemann conjugate gradient algorithm.

[0056] The aforementioned RIS-assisted distributed UAV multi-target area power combining method includes: system modeling and parameter definition, generalized steering vector calculation, establishing a received signal model, establishing a multi-objective optimization model, and deriving the optimization upper bound; smoothing and approximating the objective function; and using an alternating optimization framework and the Riemann conjugate gradient algorithm for effective solution. This RIS-assisted distributed power combining method can simultaneously provide differentiated power delivery to multiple target areas in an NLOS propagation environment, achieving collaborative optimization of multiple target areas and significantly improving the system's energy utilization efficiency and practical deployability. This method has good flexibility and scalability; the number of target areas can be expanded by adjusting the weighting coefficient set, while simultaneously supporting... LOS nodes and Any combination of NLOS nodes can be configured and flexibly adjusted according to the actual deployment environment.

[0057] In one embodiment, step 2 includes: the generalized steering vector of the LOS UAV transmitting waveform to the region of interest point is:

[0058] (3)

[0059] in For LOS node guide vectors, Represents the complex number symbol. Indicates the carrier frequency. Indicates the signal propagation to point Amplitude attenuation, Indicates the first m The signal emitted by the drone Spread to point The time delay, m =1,2,3,... , The total number of drones, Represents the speed of light. The coordinates of the points in the region of interest; For region of interest, , , These are the first drone, the second drone, and the third drone. The coordinates of the drone.

[0060] RIS is a... A uniform planar array of reflective units, and They represent their respective directions along shaft and The number of axial reflection units determines the distance from the RIS to the region of interest. The generalized steering vector is:

[0061] (4)

[0062] in, RIS to region of interest The generalized steering vector, For the (0,0)th reflection unit of the RIS to the region of interest point distance, , Let x be the x-coordinate of the (0,0)th reflection unit in the RIS. Here, are the heights of the (0,0)th reflector unit of the RIS. Let (the coordinates) be the coordinates of the (0,0)th reflection unit of the RIS. , These represent the (0,0)th reflection unit of the RIS and the region of interest, respectively. elevation and azimuth, , .in, These are the coordinates of the points in the region of interest.

[0063] The NLOS node to RIS steering vector matrix is ​​as follows:

[0064] (5)

[0065] (6)

[0066] in, The NLOS node to RIS guiding vector matrix (i.e. (Generalized steering vector matrix from the drone to the smart reflector) The first in NLOS UAV One drone ( The generalized steering vector to the intelligent reflective surface. For the (0,0)th reflection unit of the RIS to the NLOS node, the... m The distance of the drone For the NLOS node of the 1st generation m The coordinates of the drone. , , , For the NLOS node of the 1st generation m A drone x Z-axis coordinates and z-axis coordinates Let (the height of) the (0,0)th reflector unit of the RIS be the height of the reflector unit. , These represent the (0,0)th reflection unit of the RIS and the th reflection unit in the NLOS node, respectively. m The elevation and azimuth angles of the drone.

[0067] In one embodiment, step 3 includes: establishing the received signal model based on the LOS node steering vector, the RIS steering vector, and the NLOS node-to-RIS steering vector matrix.

[0068] (7)

[0069] (8)

[0070] (9)

[0071] in, To find the region of interest The synthesized signal of the received signal, For region of interest, For LOS node guide vectors, RIS to region of interest The generalized steering vector, , These are the waveform matrices transmitted by LOS UAV and NLOS UAV, respectively. This represents the reflection coefficient matrix of RIS. This is the NLOS node-to-RIS guiding vector matrix.

[0072] exist Within a discrete time interval, the region of interest points The average combined power at that location is:

[0073] (10)

[0074] in, For region of interest Average synthesis power at, The waveform length.

[0075] In one embodiment, the minima-maxima multi-objective optimization model in step 4 is:

[0076] (11)

[0077] in, For the first l Priority of each target region , For regularization parameters, , For the center of the target area Average synthesis power at, For points in the target area, L The number of target areas. To protect the area The average combined power at point, To protect the points in the area, F To protect the area, For the first m The node at the th The baseband signal sent by each snapshot For complex fields, , Indicates the waveform length. Indicates the first n The reflection coefficient of each reflecting element This represents the number of drone nodes.

[0078] In one embodiment, step 5 includes: determining the upper bound of the objective function based on the minima-maximum multi-objective optimization model, combined with the properties of the semi-definite Hermitian matrix and the unitary matrix, and the Rayleigh-Ritz theorem:

[0079] (12)

[0080] in, For the target area center Average synthesis power at, , , , These are the waveform matrices transmitted by LOS UAV and NLOS UAV, respectively. Indicates the waveform length. for The largest eigenvalue, Guide vectors for LOS nodes in the target region. The generalized steering vector from RIS to the region of interest point in the target region. For points in the target area, This represents the reflection coefficient matrix of RIS. The NLOS node to RIS guiding vector matrix, This represents the number of drone nodes.

[0081] Specifically, by introducing weighting coefficients To achieve collaborative optimization of multiple target regions, the weighting coefficients are adjusted. This enables the system to allocate resources differently based on the actual characteristics of each target area.

[0082] A weighted multi-objective optimization framework was established, which can provide differentiated power delivery to two or more target regions simultaneously. By introducing weighting coefficients, the system can allocate resources differently according to the actual characteristics of each target region (such as power demand, priority, etc.).

[0083] In one embodiment, step 6 includes: using l p The norm approximates the max function, and the objective function of the minima-maxima multi-objective optimization model is... By omitting the power constant and transforming the non-smooth problem into a smooth problem, we obtain the first multi-objective optimization model as follows:

[0084] (13)

[0085] in, For the first l Priority of each target region , For regularization parameters, , For the target area The average combined power at point, For points in the target area, L The number of target areas. To protect the area The average combined power at point, To protect the points in the area, F To protect the area, for p Norm, For the first m The node at the th The baseband signal sent by each snapshot , , Indicates the waveform length. For complex fields, Indicates the first n The reflection coefficient of each reflecting unit. For the first n The reflection phase of each reflecting unit It is a constant.

[0086] In one embodiment, step 6 includes: decomposing the optimization problem into subproblem 1 and subproblem 2 according to the first multi-objective optimization model; wherein subproblem 1 is the reflection coefficient matrix of a fixed RIS. ,optimization Subproblem 2 is fixed. Optimize the reflection coefficient matrix of RIS The method of alternating optimization is used to solve subproblems 1 and 2 until the preset conditions are met, so as to obtain the combined power of the distributed UAV multi-target area; the Riemann conjugate gradient algorithm is used to solve subproblems 1 and 2.

[0087] Specifically, when solving multi-objective optimization models, L... p Norms approximate maxima problems in the solution. and When dealing with mutually coupled optimization problems, an alternating optimization method is used to solve the problem. For constant modulus constraints and RIS reflection phase constraints, the Riemann conjugate gradient algorithm is used to solve the problem, transforming the constrained problem into an unconstrained problem on a complex circular manifold, which significantly reduces the computational complexity of the algorithm.

[0088] In one embodiment, the solution process for subproblem 1 includes: for any given... Subproblem 1 is:

[0089] (14)

[0090] in, Let be the objective function. For the first l Priority of each target region , For regularization parameters, , For the target area The average combined power at point, For points in the target area, L The number of target areas. To protect the area The average combined power at point, To protect the points in the area, F To protect the area, for p Norm, , , Indicates the waveform length. It is a constant. , , These are the waveform matrices transmitted by LOS UAV and NLOS UAV, respectively. for , Guide vectors for LOS nodes in the target region. The generalized steering vector from RIS to the region of interest point in the target region. for , To protect the LOS node guide vectors in the region, The generalized steering vector from the RIS to the region of interest point in the protected area. This represents the reflection coefficient matrix of RIS. The NLOS node to RIS guiding vector matrix;

[0091] Treating the constant modulus constraint as a complex circular manifold transforms the constrained problem into an unconstrained problem; the objective function of the unconstrained problem is:

[0092] (15)

[0093] in, Let the objective function be the unconstrained problem. , , , , , It is a K-dimensional identity matrix. It is a complex field.

[0094] Calculating the Euclidean gradient of the objective function of the unconstrained problem yields the following Euclidean gradient:

[0095] (16)

[0096] in, The Euclidean gradient of the objective function for an unconstrained problem. This is the vectorized waveform vector at the j-th iteration. , .

[0097] The Euclidean gradient in the tangent space The Riemann gradient is obtained by orthogonal projection onto the surface.

[0098] Based on the RCG algorithm and the Riemann gradient, determine the first... The gradient descent direction in the next iteration is:

[0099] (17)

[0100] in, , They were respectively in the second j Second and third j The gradient descent direction at -1 iterations, The parameters are calculated using the Polak-Ribiere rule. For the Riemann gradient, For Hadamah accumulation;

[0101] The updated solution extends beyond the manifold. The tangent vector outside the projection is projected onto the manifold by the pullback operator. superior:

[0102] ;

[0103] in, This is the vectorized waveform vector at the (j+1)th iteration. Indicates the pull-back operator, This indicates the update step size, which is solved using the Armijo linear search criterion;

[0104] Continuously updated The solution to subproblem 1 is obtained when the preset convergence condition is met.

[0105] In one embodiment, the solution process for subproblem 2 includes: For any given X, subproblem 2 is:

[0106] (18)

[0107] in, , , Let be the objective function of subproblem 2.

[0108] Subproblem 2 has the same structure as subproblem 1. Subproblem 2 is solved using the same RCG algorithm as subproblem 1, with the only difference being the method of calculating the Euclidean gradient. The Euclidean gradient is calculated using Wirtinger's fundamental calculus theory as follows:

[0109] (19)

[0110] in, For the objective function Euclidean gradient, For the reflection phase, For the target area The average combined power at point, To protect the area The average combined power at point, The set of points in the region of interest within the target region. , , The NLOS node to RIS guiding vector matrix;

[0111] ;

[0112] ;

[0113] ;

[0114] ;

[0115] when hour, A is a transformation matrix.

[0116] In a specific embodiment, consider a scene instance with two target regions (e.g.) Figure 3 ), target area and and protected areas All located in Within the plane. The swarm of drones carrying transmitters contains a total of 1 node A single drone forms a LOS node. A number of drones form an NLOS node, of which the first... One drone The coordinates of the location are Furthermore, there exists a location at coordinate position. The drone carries with A RIS (reconfigurable intelligent surface) with a reflective element. Represents the reflection coefficient of RIS, where Indicates the first The reflection coefficient of each reflecting element. The node at the th The baseband signal sent by each snapshot indicates Let K represent the waveform length. Then, the waveform matrices transmitted by LOS UAV and NLOS UAV... and It can be represented as:

[0117] (20)

[0118] (twenty one)

[0119] Step 10: Assuming these platforms are fully calibrated, and disregarding phase synchronization and position errors, the LOS UAV transmits the waveform to the region of interest. Generalized steering vector As shown in formula (3).

[0120] Assuming the intelligent reflective surface is a... A uniform planar array of reflective units, and They represent their respective directions along shaft and The number of axis elements corresponds to the distance from the intelligent reflective surface to the region of interest. Generalized steering vector As shown in formula (4).

[0121] So, in NLOS interference UAV, the first... One drone ( Generalized steering vector to the intelligent reflective surface As shown in formula (6).

[0122] Therefore, the NLOS node to RIS steering vector matrix As shown in formula (5).

[0123] Therefore, in The synthesized signal (i.e. the received signal model) is shown in formula (7).

[0124] exist K Within a discrete time interval, point The average combined power at the point is shown in formula (10).

[0125] Step 20: To ensure the robustness of power combining, the optimization model should consider the minimum combined power in the two target region sets and the maximum combined power in the protection region set, and then consider the waveform constant mode constraint and RIS reflection coefficient constraint, combined with the priority of the two target regions. Introducing regularization parameters A Min-Max multi-objective optimization problem can be established as shown in formula (11).

[0126] Step 30: Introducing a constant When handling negative values, the following conditions must be met: .

[0127] make ,but ,because It is a positive semi-definite Hermitian matrix, and its largest eigenvalue is denoted as . According to the Rayleigh-Ritz theorem:

[0128] (twenty two)

[0129] under constant modulus constraint Under the given conditions, we have: .therefore:

[0130] (twenty three)

[0131] Expand The expressions are:

[0132] (twenty four)

[0133] make ,Depend on unitary properties ( ):

[0134] (25)

[0135] so:

[0136] (26)

[0137] but: Therefore, the upper bound of the objective function is shown in formula (12).

[0138] Therefore, constant The closed expression can be .

[0139] For two target regions, the constant The closed expression is:

[0140] (27)

[0141] Step 40: From the objective function and the constant modulus and reflection coefficient constraints, it can be seen that this multi-objective optimization problem is a non-convex, non-smooth, and NP-hard (Non-deterministic Polynomial Hard) problem. Furthermore, the presence of a large number of discrete points in the spatial domain makes the problem difficult to handle. Therefore, the following approach is adopted... l p The norm is used for smoothing and approximation:

[0142] (28)

[0143] when When large enough, l p The norm will tend to l ∞ This is a typical multi-objective nonconvex optimization problem, because p The effect of the exponent on the objective function may lead to the existence of many saddle points in the solution space of this optimization problem, thus affecting the objective function. p While exponentiation provides smoothness, it fails to alter the inherent computational complexity of the problem. Furthermore, because... and The problem is difficult to solve directly due to the mutual coupling between the two elements. Therefore, the Alternating Optimization (AO) method is used to solve the problem.

[0144] Step 50: Fix ,optimization For any given The optimization problem can be expressed as:

[0145] (29)

[0146] Since the optimization variables satisfy the constant modulus constraint, this problem is non-convex and NP-hard. To solve this problem efficiently, we can approach it from the perspective of Riemannian space geometry, treating the constant modulus constraint as a complex circular manifold, thereby quickly finding a high-quality solution.

[0147] First, for ease of subsequent derivation, the relevant variables are vectorized:

[0148] (30)

[0149] (31)

[0150] (32)

[0151] Therefore, the above optimization problem can be regarded as a complex circular manifold. Unconstrained problems in

[0152] (33)

[0153] manifold It can be represented as:

[0154] (34)

[0155] make , The objective function can be simplified to:

[0156] (35)

[0157] First, we calculate the Euclidean gradient. The Euclidean gradient of the objective function can be expressed as:

[0158] (36)

[0159] in: .

[0160] At point The tangent space at a point can be represented as: .

[0161] Next, the Riemann gradient is calculated. It is its Euclidean gradient In tangent space Orthographic projection on:

[0162] (37)

[0163] According to the RCG algorithm, in the... The gradient descent direction at the next iteration can be expressed as:

[0164] (38)

[0165] Where parameters Calculated using the Polak-Ribiere rule:

[0166] (39)

[0167] Finally, the updated solution is delimited from the manifold. The tangent vector outside the projection is projected onto the manifold by the pullback operator. superior:

[0168] (40)

[0169] in, Indicates the pullback operator:

[0170] (41)

[0171] The update step size, solved using the Armijo linear search criterion, can be expressed as:

[0172] (42)

[0173] Where usually taken , , It is the smallest negative integer that satisfies equation (b) above. Based on the above equation, we can obtain... Therefore, the RCG optimization framework can make the objective function of the problem monotonically decrease during the iteration process.

[0174] Based on the above algorithm, continuously update Until the preset convergence condition is met. The solution to this subproblem can then be obtained. In this algorithm, the algorithm tends to converge when the objective value no longer changes significantly with the number of iterations. The convergence condition can be set as follows:

[0175] (43)

[0176] Step 60: Fix ,optimization For any given The optimization problem can be expressed as:

[0177] (44)

[0178] Clearly, subproblem 2 has the same structure as subproblem 1, so it is solved using the same RCG algorithm, and the process is the same as for subproblem 1. The difference lies in the calculation of the Euclidean gradient.

[0179] set up ,Will Divided into blocks Then the power term can be expressed as:

[0180] (45)

[0181] Ignore and For irrelevant terms, the power term simplifies to:

[0182] (46)

[0183] in , .

[0184] because Therefore, the power term can be defined as:

[0185] (47)

[0186] in , .

[0187] Therefore, the objective function can be simplified to:

[0188] (48)

[0189] Calculate the Euclidean gradient using Wirtinger's fundamental theory of calculus:

[0190] (49)

[0191] in, For the objective function Euclidean gradient, For the reflection phase, For the target area The average combined power at point, To protect the area The average combined power at point, , , The NLOS node to RIS guiding vector matrix;

[0192] ;

[0193] ;

[0194] ;

[0195] ;

[0196] when hour, .

[0197] In summary, based on the analysis of the two sub-problems mentioned above, alternating optimization is employed. and The solution to the above optimization problem is obtained through iteration.

[0198] In a verification implementation, multiple sets of experiments were conducted on this method. The energy distribution in the region of interest under different weighting coefficients is shown below. Figure 4 As shown, where Figure 4 (a) is A schematic diagram of the energy distribution over the region of interest. Figure 4 (b) is A schematic diagram of the energy distribution over the region of interest. Figure 4 (c) is A schematic diagram of the energy distribution over the region of interest. Figure 4 (d) is A schematic diagram of the energy distribution over the region of interest. Figure 4 It can be seen that this method can achieve differentiated power synthesis of two target regions by adjusting different weighting coefficients, while controlling the region that needs to be protected to maintain low power. This shows that the algorithm effectively suppresses energy leakage while optimizing the target power.

[0199] The objective function of this method changes with the number of iterations under different weighting coefficients, as shown below. Figure 5 As shown, from Figure 5 It can be seen that the method proposed in this invention has good stability. Under different weight configurations, the algorithm can converge within 50-60 iterations.

[0200] The variation of the combined power of target region 1 and target region 2 with the number of iterations under different weighting coefficients is as follows: Figure 6 As shown, from Figure 6 It can be seen that the weights are biased towards effectiveness, as... As the value increases, the power in region 1 increases monotonically, while the power in region 2 decreases monotonically, and the power difference gradually widens, verifying the effectiveness of the weighting coefficient.

[0201] The changes in the minimum and maximum combined power of the target region 1 and the protected region 1 with the number of iterations under different weighting coefficients are as follows: Figure 7 As shown, from Figure 7 It can be seen that the multi-objective optimization method proposed in this application can effectively suppress energy leakage in the protected area while increasing the power in the target area, thus achieving the synergistic achievement of the dual optimization objectives.

[0202] It is worth noting that: (1) When designing the weights for multi-objective regions, a scheme based on region and power requirements can be considered to reflect the priority differences of different regions. At the same time, dynamic weights based on real-time feedback can also be considered, and the weight coefficients can be dynamically adjusted according to the actual power level of each target region after each iteration. (2) When establishing a multi-objective optimization problem, the weighted MSE criterion can be used to replace the Min-Max criterion. Although it will generate undesirable space-frequency composite power in the target and protection areas, and some grid points may have extreme cases of excessively low or high power, it can improve the accuracy of interference averaging. (3) When solving multi-objective optimization problems, the alternating direction multiplier method can be used to solve the problem based on the Lp norm approximation. Although the complexity is higher than the RCG method, the ADMM framework is flexible and can handle various constraint types. (4) In terms of system architecture, multiple RIS can be deployed for collaborative optimization. At the same time, active RIS with power amplifiers can be used to replace passive RIS to overcome the dual path loss problem of passive RIS and significantly improve the performance of long-distance scenarios.

[0203] It should be understood that, although the above Figure 1 The steps are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise explicitly stated in this document, there is no strict order in which these steps are executed; they can be performed in other orders. Furthermore, the above... Figure 1 At least some of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0204] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0205] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of protection of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and all such modifications and improvements fall within the scope of protection of this application.

Claims

1. A RIS-assisted distributed unmanned aerial vehicle multi-target regional power synthesis method, characterized by, The method comprises the steps of: Step 1: Deployment The UAVs with transmitters are divided into LOS group and NLOS group according to the line-of-sight conditions with the target area, wherein the LOS group includes The UAVs in the LOS group can directly reach the target area, and the NLOS group includes The UAVs in the NLOS group are blocked by obstacles, A RIS with A reflection unit is deployed; Step 2: calculating the LOS node steering vector, the RIS steering vector and the NLOS node-to-RIS steering vector matrix; Step 3: establishing a received signal model according to the LOS node steering vector, the RIS steering vector and the NLOS node-to-RIS steering vector matrix, and calculating the average combined success rate at a point in a region of interest; Step 4: taking the minimum combined success rate in all target region sets and the maximum combined success rate in a protection region set as targets, considering the constant modulus constraint and the RIS reflection coefficient constraint, and constructing a minimax multi-objective optimization model as follows: ； wherein, is a priority of the l th target region, , is a regularization parameter, , is an average joint success rate at a point in the target region, is a point in the target region, L is a number of target regions, is an average joint success rate at a point in the protection region, is a point in the protection region, F is a protection region, is a baseband signal transmitted by the m th node at the th snapshot, , , denotes a waveform length, is a number of UAV nodes, is a complex field, denotes a reflection coefficient of the n th reflection unit, is a reflection phase of the n th reflection unit; Step 5: determining an upper limit of optimization of a target function according to the minimax multi-objective optimization model; specifically, determining the upper limit of optimization of the target function as follows according to the minimax multi-objective optimization model, in combination with the properties of a semi-positive Hermitian matrix and a unitary matrix and the Rayleigh-Ritz theorem: ； wherein, , , , are the waveform matrices transmitted by the LOS UAV and the NLOS UAV, respectively, is the maximum eigenvalue of is the LOS node steering vector in the target region, is the generalized steering vector from the RIS to the point of interest in the target region, denotes the reflection coefficient matrix of the RIS, is the NLOS node-to-RIS steering vector matrix;​ Step 6: smoothing and approximating the target function of the minimax multi-objective optimization model to obtain a first multi-objective optimization model; Step 7: using an alternating optimization framework and a Riemann conjugate gradient algorithm to optimize and solve the first multi-objective optimization model to obtain a distributed unmanned aerial vehicle multi-target region combined success rate.

2. The RIS-aided distributed UAV multi-target regional power combining method based on claim 1, characterized in that, Step 2 comprises: The generalized steering vector of a LOS unmanned aerial vehicle transmitting a waveform to a point in a region of interest is: ； wherein is the LOS node orientation vector, denotes a complex symbol, denotes a carrier frequency, denotes the amplitude attenuation of the signal propagating from to the point denotes the time delay of the signal transmitted by the m th UAV propagating from to the point , m = 1,2,3,..., , is the total number of UAVs, denotes the speed of light, is the point of interest region coordinate; is the point of interest region, , , are the coordinates of the first UAV, the second UAV and the M th UAV, respectively; RIS is a... A uniform planar array of reflective units, and They represent their respective directions along shaft and The number of axis reflection units determines the distance from the RIS to the region of interest. The generalized steering vector is: ； in, RIS to region of interest The generalized steering vector, For the (0,0)th reflection unit of the RIS to the region of interest point distance, Let x be the x-coordinate of the (0,0)th reflection unit in the RIS. Here, are the heights of the (0,0)th reflector unit of the RIS. Let (the coordinates) be the coordinates of the (0,0)th reflection unit of the RIS. , These represent the (0,0)th reflection unit of the RIS and the region of interest, respectively. The elevation and azimuth angles; The NLOS node-to-RIS steering vector matrix is: ； ； wherein, is a NLOS node to RIS steering vector matrix, is a generalized steering vector from the m th UAV in the NLOS UAV to the smart reflector, is a distance from the (0, 0)th reflecting element of the RIS to the m th UAV in the NLOS node, is a coordinate of the m th UAV in the NLOS node, , are an elevation angle and an azimuth angle of the (0, 0)th reflecting element of the RIS and the m th UAV in the NLOS node, respectively.​​ 3. The RIS-aided distributed UAV multi-target regional power combining method based on claim 1, characterized in that, Step 3 comprises: The received signal model is established according to the LOS node steering vector, the RIS steering vector and the NLOS node-to-RIS steering vector matrix as follows: ； wherein, is a composite signal received at a point of interest region , is a point of interest region, is a LOS node steering vector, is a generalized steering vector of the RIS to a point of interest region , , denotes a reflection coefficient matrix of the RIS, is a NLOS node to RIS steering vector matrix; At a discrete time instant, the average combined power at a point of interest region is: ； wherein, is the average combined power at the region of interest point .

4. The RIS-aided distributed UAV multi-target regional power combining method based on claim 1, characterized in that, Step 6 comprises: The l p norm approximation max function, and the minimum-maximum multi-objective optimization model of the objective function of power constant omission, the non-smooth problem is converted into a smooth problem, and the first multi-objective optimization model is obtained as follows: ； in, For the first l Priority of each target region , For regularization parameters, , for p Norm, It is a constant.

5. The RIS-aided distributed UAV multi-target regional power combining method based on claim 1, characterized in that, Step 7 comprises: According to the first multi-objective optimization model, the optimization solving problem is decomposed into subproblem 1 and subproblem 2; wherein subproblem 1 is to fix the reflection coefficient matrix of the RIS , optimize ; subproblem 2 is to fix , optimize the reflection coefficient matrix of the RIS ; The sub-problems 1 and 2 are solved by using an alternating optimization method until a preset condition is met, and the distributed unmanned aerial vehicle multi-target region combined success rate is obtained; wherein the sub-problems 1 and 2 are solved by using a Riemann conjugate gradient algorithm.

6. The RIS-aided distributed UAV multi-target regional power combining method based on claim 5, characterized in that, The solving process of the sub-problem 1 comprises: For any given Sub-problem 1 is: ； wherein, is an objective function, is a priority of the l th target region, , is a regularization parameter, , is an norm, is a constant, is an , is an , is a LOS node steering vector in the protection region, is a generalized steering vector from the RIS to a point of interest in the protection region; The constant modulus constraint condition is regarded as a complex circle manifold, and the constraint problem is converted into an unconstrained problem; wherein the target function of the unconstrained problem is as follows: ； wherein is the objective function of the unconstrained problem, , , , , , is the K-dimensional identity matrix; The Euclidean gradient of the target function of the unconstrained problem is calculated to obtain the Euclidean gradient of the target function of the unconstrained problem as follows: ； wherein, the Euclidean gradient of the objective function of the unconstrained problem, , , is the vectorized waveform vector at the jth iteration; orthogonal projection of the Euclidean gradient on the tangent space yields the Riemannian gradient; According to the RCG algorithm and the Riemannian gradient, the gradient descent direction at the first iteration is determined as ； wherein, , are the gradient descent directions at the 1st j and 1st j -1th iteration, respectively, is a parameter computed by the Polak-Ribiere rule, is the Riemannian gradient, is the Hadamard product; The updated solution is projected by the pullback operator to the manifold outside the manifold above: ； where is the vectorized waveform vector at the j+1th iteration, denotes the pullback operator, denotes the update step, solved by the Armijo linear search criterion; constantly updated The solution of subproblem 1 is obtained until the preset convergence condition is satisfied.

7. The RIS-aided distributed UAV multi-target regional power combining method based on claim 6, characterized in that, The solving process of the sub-problem 2 comprises: For any given X, the sub-problem 2 is as follows: ； wherein , , is the objective function for subproblem 2; The sub-problem 2 has the same structure as the sub-problem 1, and the sub-problem 2 is solved by using the same RCG algorithm as the sub-problem 1, and the solving process is only different in the calculation method of the Euclidean gradient; wherein the Euclidean gradient is calculated by using the Wirtinger calculus basic theory as follows: ； wherein, is the Euclidean gradient of the objective function , is the reflection phase, is the average combined power at a point in the target region, , is the average combined power at a point in the protection region, , is the set of points of interest in the target region; ； ； ； ； ； ； When , , A is a transformation matrix.

Images