Perception-assisted unmanned aerial vehicle track and power distribution joint optimization method and system

Abstract

The invention discloses a perception-assisted unmanned aerial vehicle track and power distribution joint optimization method and system, and the method comprises the steps: firstly, extracting environment related features from high-dimensional CSI data, constructing a weighted graph, and calculating a shortest path distance matrix; and the CSI is mapped to a two-dimensional space through a twin neural network to maintain distance consistency, and user distribution is obtained by combining with known position correction. On the basis, an optimization model taking the minimization of the energy consumption of the unmanned aerial vehicle as a target and the user rate demand as a constraint is established, the optimization model is decomposed into a power distribution and trajectory optimization sub-problem, and the Lyapunov optimization and successive convex approximation are used for alternate solution. According to the method, high-precision positioning is realized through the channel map under the scene that the user position is unknown, the energy consumption of the unmanned aerial vehicle system is reduced by 27% or above through joint optimization, and meanwhile, 98% of user communication requirements are guaranteed.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of wireless communication, and particularly relates to a perception-assisted unmanned aerial vehicle trajectory and power allocation joint optimization method and system. BACKGROUND

[0002] Unmanned aerial vehicles (UAVs) have been widely used in many fields to replace humans to perform specific tasks due to their high controllability and strong maneuverability. In the field of wireless communication, UAVs have the potential to serve as aerial base stations or flying relays, which can provide fast and reliable communication connections for ground users. However, UAVs usually rely on battery power, and the energy required for communication and flight tasks is huge, so system energy efficiency becomes an important indicator that needs to be optimized.

[0003] In actual deployment, factors such as the flight trajectory and communication power of UAVs have a significant impact on system energy consumption. In the research of using UAVs as aerial base stations, some scholars have proposed deployment optimization strategies for aerial base stations to determine the optimal hovering position of UAVs, thereby improving the stability and quality of service of the connection. In the aspect of trajectory design, some existing works have adopted optimization methods such as convex optimization and successive convex approximation (SCA) to solve the optimal flight path of UAVs in dynamic scenarios. In addition, some research has introduced intelligent algorithms such as deep reinforcement learning, such as deep Q network (DQN), deep deterministic policy gradient (DDPG), and double deep Q network (DDQN), to improve the joint optimization performance of UAV trajectory planning and resource allocation by using their self-learning ability in high-dimensional dynamic decision-making scenarios.

[0004] However, most of the current research usually assumes that the user location is known, but in actual scenarios, user location information is often not directly available or has measurement errors, which will directly affect the reliability of the optimization results.

[0005] Channel Charting as a new wireless sensing technology can utilize the prior statistical characteristics of wireless channel, and realize the perception and estimation of user spatial distribution by learning the trend of change of Channel State Information (CSI) in space. Compared with traditional positioning technology, Channel Charting does not depend on external positioning system, and has stronger environmental adaptability and deployment flexibility. The typical Channel Charting construction method is: first, extract the channel features strongly related to the environment from the high-dimensional CSI data, then measure the similarity or dissimilarity of high-dimensional CSI samples as a distance measure between samples, and then reduce the dimension of high-dimensional CSI data to obtain low-dimensional coordinates through a low-dimensional mapping method. The mapped low-dimensional coordinates need to maintain the neighborhood relationship of high-dimensional CSI samples under the corresponding distance measure. In terms of dissimilarity measurement, the commonly used measurement methods include Euclidean distance, cosine correlation, etc.

[0006] However, due to the non-linear structure of the high-dimensional CSI data in the geometric space, simple Euclidean distance, cosine correlation and other measurement methods usually cannot make the dissimilarity well linear consistent with the actual distance, resulting in low positioning accuracy. SUMMARY

[0007] The technical problem to be solved by the present application is to provide a perception-assisted unmanned aerial vehicle trajectory and power distribution joint optimization method and system to solve the technical problem that the existing unmanned aerial vehicle trajectory optimization depends on known user positions, while in actual scenarios, position information is often missing or has errors, resulting in energy consumption optimization failure.

[0008] The present application adopts the following technical solutions: The perception-assisted unmanned aerial vehicle trajectory and power distribution joint optimization method comprises the following steps: scaling the autocorrelation matrix of high-dimensional CSI data to obtain a feature vector; using the Euclidean distance of the feature vector as the edge weight, generating a distance matrix representing channel differences through Dijkstra algorithm; using a twin neural network to map high-dimensional CSI data to a two-dimensional space through unsupervised learning, and the mapped two-dimensional coordinates maintain the relative distance relationship under high-dimensional CSI features; correcting the channel chart based on the known user position, and estimating the unknown user position by using k-nearest neighbor method; on the basis of the obtained user position, establishing an unmanned aerial vehicle energy consumption minimization model; The obtained UAV energy consumption minimization model is decomposed into a power allocation subproblem and a trajectory optimization subproblem, which are respectively converted into convex problems through Lyapunov optimization and first-order Taylor expansion; the two subproblems are iteratively optimized through alternation until convergence, and the optimal trajectory and power allocation scheme are output.

[0009] Preferably, the high-dimensional CSI data autocorrelation matrix is scaled to obtain the eigenvector, specifically: Each high-dimensional CSI data is represented by , and the autocorrelation matrix thereof is , which is scaled to , , is the feature extracted from the high-dimensional CSI data and strongly related to the environment.

[0010] Preferably, the Euclidean distance of the eigenvector is taken as the edge weight, and the distance matrix representing the channel difference is generated through the Dijkstra algorithm, specifically: The Euclidean distance between the features of any two points is L points are regarded as nodes in the graph, a weighted undirected graph G=(V,E) is constructed, V is the node set, E is composed of edges between each node and its nearest n neighbors, and the weight of the edge is the Euclidean distance in the feature space. For the constructed graph G, the Dijkstra algorithm is used to calculate the shortest path length from each point to all other points in the graph, and the output is an LxL shortest path distance matrix D, , representing the channel difference between point i and the transmitter and point j and the transmitter.

[0011] Preferably, the loss function of the twin neural network is:

[0012] , wherein is the output of the neural network, represents the channel difference between point i and the transmitter and point j and the transmitter, and is the input of the neural network, i.e., the feature extracted from the high-dimensional CSI data.

[0013] Preferably, the UAV energy consumption minimization model is:

[0014] , wherein is the average energy consumption of the UAV in T time, is the bandwidth of the entire system, is the flight time of the UAV, is the standard deviation of the noise power, is the height of the UAV, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot,

[0015] P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot,

[0016] P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot,

[0017] P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, P(t) is the power transmitted by the drone to the ground users at the tth time slot, air density, For rotor realism, The rotor disk area, The channel gain is the result of the first-order Taylor expansion. Minimum data transfer rate for each user during this period. The coordinates of the drone's initial position projected onto the ground. The coordinates of the drone's termination position projected onto the ground. The maximum speed of the drone, For unit time slot length, The left and right boundaries of the region. This represents the upper and lower boundaries of the region.

[0018] Preferably, the power allocation subproblem and the trajectory optimization subproblem are as follows:

[0019] .

[0020] Preferably, the two subproblems are optimized alternately and iteratively until convergence, specifically as follows: Initialization: Input variables and initial value and Number of iterations Error accuracy Iterative computation; Output the UAV in all time slots Flight trajectory of the drone and the transmission power allocated to ground users .

[0021] Preferably, the iterative operation is as follows: fixed for Solve subproblem one to obtain the optimization variables. optimal solution ; fixed for Solve subproblem two to obtain the optimization variables. The optimal solution is ; renew = , = ; When satisfied hour, If the problem persists, return to the solution of subproblem one; otherwise, the iteration terminates.

[0022] Secondly, embodiments of the present invention provide a perception-assisted unmanned aerial vehicle (UAV) trajectory and power allocation joint optimization system, comprising: The feature module scales the autocorrelation matrix of high-dimensional CSI data to obtain feature vectors; using the Euclidean distance of the feature vectors as edge weights, a distance matrix representing channel differences is generated using the Dijkstra algorithm. The mapping module uses a Siamese neural network to map high-dimensional CSI data to a two-dimensional space through unsupervised learning. The mapped two-dimensional coordinates maintain the relative distance relationship under the high-dimensional CSI features. The estimation module corrects the channel map based on the known user locations and uses the k-nearest neighbor method to estimate the locations of unknown users. The module builds a model for minimizing drone energy consumption based on the obtained user location. The allocation module decomposes the obtained UAV energy consumption minimization model into a power allocation subproblem and a trajectory optimization subproblem, which are then transformed into convex problems through Lyapunov optimization and first-order Taylor expansion, respectively. The two subproblems are then iteratively optimized until convergence, and the optimal trajectory and power allocation scheme are output.

[0023] Thirdly, a computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the aforementioned perception-assisted UAV trajectory and power allocation joint optimization method.

[0024] Fourthly, embodiments of the present invention provide a computer-readable storage medium including a computer program, which, when executed by a processor, implements the steps of the aforementioned perception-assisted UAV trajectory and power allocation joint optimization method.

[0025] Fifthly, a chip includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the aforementioned perception-assisted UAV trajectory and power allocation joint optimization method.

[0026] In a sixth aspect, embodiments of the present invention provide an electronic device, including a computer program, wherein when the computer program is executed by the electronic device, it implements the steps of the above-described perception-assisted UAV trajectory and power allocation joint optimization method.

[0027] Compared with the prior art, the present invention has at least the following beneficial effects: The method comprises the following steps: a perception-assisted unmanned aerial vehicle trajectory and power distribution joint optimization method is used to force low-dimensional coordinates to maintain the original CSI distance relationship through a twin neural network, and a positioning distortion problem caused by high-dimensional data nonlinearity is solved; the positioning error is reduced by 32% compared with a traditional Euclidean distance, reliable position input is provided for trajectory optimization, a non-convex energy consumption model is decomposed into a power distribution and a trajectory optimization sub-problem, and iterative convergence is achieved. Compared with a single optimization strategy, the complexity is reduced by 50% and local optimization is avoided, energy saving of 27% is displayed in a 50-user scenario, and a target function simultaneously constrains user rate demand. The rate satisfaction rate is greater than 98%, and the optimization failure problem caused by position error in the background technology is solved.

[0028] Further, the scaling processing eliminates the influence of channel amplitude fluctuation, so that the feature vector only retains the spatial structure information strongly related to the environment. This operation improves the robustness of the feature to interference such as multipath and shielding, and provides a stable input for subsequent graph construction.

[0029] Further, a weighted graph is constructed with the feature Euclidean distance as the edge weight, and a shortest path distance matrix is calculated through the Dijkstra algorithm. Compared with directly using the Euclidean distance, the matrix overcomes the "straight line distance distortion" problem in the nonlinear manifold, and the positioning accuracy is improved by 32%.

[0030] Further, the loss function explicitly constrains the two-dimensional Euclidean distance of the neural network output to be consistent with the original channel difference, and forces the low-dimensional mapping to maintain the topological structure of the high-dimensional CSI. This design solves the distortion problem of traditional dimension reduction methods on nonlinear manifolds, and makes the unknown user position estimation error less than 1.5 m.

[0031] Further, the target function includes aviation special parameters such as leaf type power and induced power, and accurately reflects the influence of flight speed on energy consumption; the constraint condition integrates the communication rate and the kinetic limit. The energy saving of the model is more than 27% compared with the scheme ignoring the flight power consumption.

[0032] Further, a virtual queue is introduced, a long-term rate constraint is converted into an instantaneous convex problem through Lyapunov optimization, the KKT condition is met to solve in real time, a first-order Taylor expansion is performed on a non-convex channel gain term, which is converted into a second-order cone programming, and the second-order cone programming is efficiently solved through CVX; the two are alternately iterated to ensure convergence.

[0033] Further, the power distribution sub-problem only optimizes the communication power consumption, and the trajectory optimization sub-problem only optimizes the flight energy consumption. After separation, the variable dimension of the sub-problem is reduced by 70%, and the calculation efficiency is improved.

[0034] It can be understood that the beneficial effects of the above-mentioned second aspect to sixth aspect can be referred to the related description in the first aspect, which will not be repeated here.

[0035] In summary, the present application realizes user distribution perception without external positioning through channel mapping technology, solves the problem of "optimization failure caused by unknown position" in the background technology, and the joint optimization framework saves energy by 27% in a 50-user complex scene, the rate satisfaction rate is >98%, the comprehensive energy efficiency is improved by 2.1 times compared with the traditional SCA method, the sub-problem alternating iteration algorithm has low complexity, and can be run in real time on an unmanned aerial vehicle embedded chip, thereby providing core technical support for 5G / 6G air base station deployment.

[0036] The technical solutions of the present application will be further described in detail below with the help of the accompanying drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 Flowchart of the present application Figure 2 Structure diagram of the twin neural network used in the present application Figure 3 Power comparison chart of the present application and other algorithms for rate requirement Figure 4 Comparison chart of the present application and other algorithms for meeting user rate requirement percentage for rate requirement Figure 5 Power comparison chart of the present application and other algorithms for user number 50 Figure 6 Comparison chart of the present application and other algorithms for meeting user rate requirement percentage for user number 50 Figure 7 Schematic diagram of a computer device provided by an embodiment of the present application Figure 8 Block diagram of a chip provided by an embodiment of the present application

[0038] Among them, 60. Computer device; 61. Processor; 62. Memory; 63. Computer program; 600. Electronic device; 610. Processing unit; 620. Storage unit; 6201. Random access storage unit; 6202. Cache storage unit; 6203. Read-only storage unit; 6204. Program / utility; 6205. Program module; 630. Bus; 640. Display unit; 650. Input / output interface; 660. Network adapter; 700. External device. DETAILED DESCRIPTION

[0039] The technical solutions in the embodiments of the present application will be described clearly and completely below with the help of the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0040] In the description of the present application, it should be understood that the terms "include" and "comprise" indicate the presence of the described features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0041] It should also be understood that the terms used in the present specification and the attached claims are only for the purpose of describing particular embodiments and do not intend to limit the present application. As used in the present specification and the attached claims, the singular forms "a," "an," and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0042] It should be further understood that the term "and / or" used in the present specification and the attached claims, means one or more of the associated listed items and all possible combinations of the items, and includes the combinations, for example, A and / or B can mean A alone, A and B together, and B alone. In addition, the character " / " in the present application generally represents an "or" relationship between the front and rear associated objects.

[0043] It should be understood that although the terms first, second, third, etc. can be used in the embodiments of the present application to describe a certain range, etc., these ranges should not be limited to these terms. These terms are only used to distinguish the ranges from each other. For example, the first range can also be referred to as the second range, and similarly, the second range can also be referred to as the first range, without departing from the scope of the embodiments of the present application.

[0044] Depending on the context, the word "if" as used herein can be interpreted to mean "when" or "while" or "in response to determining" or "in response to detecting." Similarly, the phrase "if determined" or "if detecting (a stated condition or event)" can be interpreted to mean "when determined" or "in response to determining" or "when detecting (a stated condition or event)" or "in response to detecting (a stated condition or event)," depending on the context.

[0045] Various structural diagrams according to the disclosed embodiments of the present application are shown in the accompanying drawings. These drawings are not drawn to scale, in which certain details are exaggerated for the purpose of clarity and certain details can be omitted. The shapes of various regions, layers, and their relative sizes and positional relationships shown in the drawings are only exemplary, and in actuality, they can be deviated due to manufacturing tolerances or technical limitations, and a person skilled in the art can additionally design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0046] The application provides a perception-assisted unmanned aerial vehicle trajectory and power distribution joint optimization method, first, environment-related features are extracted from a high-dimensional CSI data autocorrelation matrix, a weighted graph is constructed, and a channel difference distance matrix is generated through a Dijkstra algorithm; a twin neural network is used to map the CSI to a two-dimensional space, and user distribution is obtained by combining known position correction. An unmanned aerial vehicle energy minimization model is established, which is decomposed into a power distribution sub-problem (Lyapunov optimization) and a trajectory optimization sub-problem (first-order Taylor expansion convexity), and the optimal trajectory and power scheme are output through alternating iteration. In the scenario where the user position is unknown, the positioning error is less than 1.5 m, the joint optimization reduces the energy consumption by more than 27%, and the user rate satisfaction rate is greater than 98%.

[0047] Please refer to Figure 1 The application provides a perception-assisted unmanned aerial vehicle trajectory and power distribution joint optimization method, comprising the following steps: S1, extracting environment-related channel features from high-dimensional CSI data; Each CSI data is represented by The autocorrelation matrix of the transmitter with B antennas is The autocorrelation matrix is scaled , wherein . That is, the environment-related features extracted from the high-dimensional CSI data.

[0048] S2, measuring the similarity or difference of high-dimensional CSI samples as a distance measure between samples; The Euclidean distance between the features of any two points is These points (assuming a total of L points) are regarded as nodes in the graph, and a weighted undirected graph G=(V,E) is constructed, wherein V is a node set, E is composed of edges between each node and its nearest n neighbors, and the weight of the edge is the Euclidean distance in the feature space.

[0049] For the constructed graph G, the Dijkstra algorithm is used to calculate the shortest path length from each point to all other points in the graph, and an LxL shortest path distance matrix D is output, wherein The shortest path distance from node i to node j, that is, the channel difference between point i and the transmitter and point j and the transmitter.

[0050] S3, using a twin neural network to map high-dimensional CSI data to a two-dimensional space through unsupervised learning, and the mapped two-dimensional coordinates need to maintain the relative distance relationship under the high-dimensional CSI features; Please refer to Figure 2 , a twin neural network is constructed, and the loss function of the neural network is:

[0051] wherein is the output of the neural network.

[0052] Only need to input high-dimensional CSI data into the neural network, and the channel map is obtained.

[0053] S4, correct the relative position in the channel map using the known real position information of part of the users to obtain the real position information of all users; Use the k-nearest neighbor method to find the neighbors of the unknown point, and average the known part of the user position, that is, the estimated actual coordinates of the unknown point.

[0054] S5, on the basis of the user position obtained in step S4, model the unmanned aerial vehicle flight trajectory design and power allocation optimization problem. The optimization problem includes optimization objectives, optimization variables, and constraint conditions; The model of the unmanned aerial vehicle flight trajectory design and power allocation problem is as follows:

[0055] wherein, . and are the blade power and induced power in the hovering state, respectively, represents the tip speed of the rotor blade, is the average rotor induced speed in the hovering state, and are the fuselage drag ratio and rotor solidity, respectively, and represent the air density and rotor disc area. is a set of ground users, each user has coordinates , and . is a set of time slots, is the length of a unit time slot.

[0056] The height of the unmanned aerial vehicle is , and the position of the unmanned aerial vehicle in the th time slot is , wherein .

[0057] The coordinates of the initial position and the final position of the unmanned aerial vehicle projected onto the ground are and , the speed of the unmanned aerial vehicle in the th time slot is , and the maximum flight speed of the unmanned aerial vehicle is .

[0058] In the first time slot, the power transmitted by the UAV to the ground users is , the maximum total transmit power of the UAV is . The bandwidth of the entire system is , the channel power gain per unit distance is , the noise power is , and the minimum data transmission rate of each user in this period of time is .

[0059] S6, decompose the original problem of step S5 into two sub-problems; fix the flight trajectory of the UAV, and update the optimization problem of the UAV transmit power allocation to model sub-problem one; fix the UAV transmit power allocation, and update the optimization problem of the UAV flight trajectory to model sub-problem two. Sub-problem one: The problem of fixing the flight trajectory of the UAV and updating the UAV transmit power allocation is modeled as follows:

[0060] Introduce a virtual queue , and use the Lyapunov optimization method to convert the problem into the following convex problem:

[0061] The above problem satisfies the KKT condition and can be solved directly.

[0062] Sub-problem two: The problem of fixing the UAV transmit power allocation and updating the UAV flight trajectory is modeled as follows:

[0063] Using first-order Taylor expansion, given the UAV trajectory of the th iteration, the channel gain of the th iteration is expanded as , and the problem can be converted into a convex problem:

[0064] This problem can be solved using CVX.

[0065] S7, alternately iterate and optimize the two sub-problems of step S6, and use a joint algorithm to jointly optimize the flight trajectory and transmit power allocation of the UAV.

[0066] The joint optimization algorithm of the flight trajectory and transmit power allocation of the UAV is implemented through the following steps: S701, initialization: input the initialization values of the variables and ​​ and iteration number error precision ; S702, iteration operation; In this step, the following operations are performed in turn: S7021, fixing for , solving subproblem one, obtaining the optimal solution of the optimization variable ; ; S7022, fixing for , solving subproblem two, obtaining the optimal solution of the optimization variable ; ; S7023, updating = , = ; S7024, when , , return to step S7021, otherwise iteration is terminated.

[0067] S703, output: output the flight trajectory of the unmanned aerial vehicle on all time slots and the transmission power allocated to the ground user .

[0068] In another embodiment of the present application, a perception-aided unmanned aerial vehicle trajectory and power allocation joint optimization system is provided, which can be used to implement the above-mentioned perception-aided unmanned aerial vehicle trajectory and power allocation joint optimization method. Specifically, the perception-aided unmanned aerial vehicle trajectory and power allocation joint optimization system includes a feature module, a mapping module, an estimation module, a construction module, and a distribution module.

[0069] The feature module performs scaling processing on the high-dimensional CSI data autocorrelation matrix to obtain a feature vector; and generates a distance matrix representing channel differences by Dijkstra algorithm, taking the Euclidean distance of the feature vector as the edge weight. The mapping module uses a twin neural network to map high-dimensional CSI data into a two-dimensional space through unsupervised learning, and the mapped two-dimensional coordinates maintain the relative distance relationship under high-dimensional CSI features. The estimation module corrects the channel atlas based on the known user position and estimates the unknown user position by using the k-nearest neighbor method. The construction module establishes an unmanned aerial vehicle energy consumption minimization model based on the obtained user position. ​The distribution module decomposes the obtained UAV energy consumption minimization model into a power distribution sub-problem and a trajectory optimization sub-problem, and converts them into convex problems through Lyapunov optimization and first-order Taylor expansion, respectively; the two sub-problems are iteratively optimized until convergence, and the optimal trajectory and power distribution scheme are output.

[0070] The application provides a terminal device, which comprises a processor and a memory, the memory is used for storing a computer program, the computer program comprises program instructions, and the processor is used for executing the program instructions stored in the computer storage medium. The processor can be a central processing unit (CPU), and can also be other general-purpose processors, graphics processing units (GPUs), tensor processing units (TPUs), digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components and the like, which are the computing core and control core of the terminal, and are suitable for implementing one or more instructions, and are specifically suitable for loading and executing one or more instructions to implement a corresponding method flow or a corresponding function; the processor in the embodiments of the application can be used for the operation of the perception-assisted UAV trajectory and power distribution joint optimization method, which comprises the following steps: The high-dimensional CSI data autocorrelation matrix is scaled to obtain an eigenvector; a distance matrix representing channel differences is generated by Dijkstra algorithm with the Euclidean distance of the eigenvector as an edge weight; the high-dimensional CSI data is mapped into a two-dimensional space by unsupervised learning using a twin neural network, and the mapped two-dimensional coordinates maintain the relative distance relationship under the high-dimensional CSI feature; the channel atlas is corrected based on the known user position, and the k-nearest neighbor method is used to estimate the unknown user position; a UAV energy consumption minimization model is established based on the obtained user position; the obtained UAV energy consumption minimization model is decomposed into a power distribution sub-problem and a trajectory optimization sub-problem, and is converted into a convex problem through Lyapunov optimization and first-order Taylor expansion, respectively; the two sub-problems are iteratively optimized until convergence, and the optimal trajectory and power distribution scheme are output.

[0071] Please refer to Figure 7The terminal device is a computer device. The computer device 60 in this embodiment includes a processor 61, a memory 62, and a computer program 63 stored in the memory 62 and capable of running on the processor 61. When the computer program 63 is executed by the processor 61, the method for estimating the concentration of radioactive iodine species in the post-accident containment in the embodiment is implemented. To avoid repetition, details are not described here. Alternatively, when the computer program 63 is executed by the processor 61, the functions of various models / units in the system for jointly optimizing the trajectory and power allocation of the UAV with the aid of perception in the embodiment are implemented. To avoid repetition, details are not described here.

[0072] The computer device 60 can be a desktop computer, a notebook, a palm computer, a cloud server, and the like. The computer device 60 can include, but is not limited to, the processor 61 and the memory 62. Those skilled in the art can understand that the computer device 60 can include more or fewer components, or some components are combined, or different components, for example, the computer device can also include an input / output device, a network access device, a bus, and the like. Figure 7 The computer device 60 is only an example and does not constitute a limitation on the computer device 60, and can include more or fewer components than those shown, or some components are combined, or different components, for example, the computer device can also include an input / output device, a network access device, a bus, and the like.

[0073] The processor 61 can be a central processing unit (CPU), and can also be other general-purpose processors, graphics processing units (GPUs), tensor processing units (TPUs), digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, and the like. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0074] The memory 62 can be an internal storage unit of the computer device 60, such as a hard disk or a memory of the computer device 60. The memory 62 can also be an external storage device of the computer device 60, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, and the like.

[0075] Further, the storage 62 can include both an internal storage unit of the computer device 60 and an external storage device. The storage 62 is used to store computer programs and other programs and data required by the computer device. The storage 62 can also be used to temporarily store data that has been output or will be output.

[0076] Referring to Figure 8 , the terminal device is an electronic device 600 in the form of a general computing device. Components of the electronic device can include, but are not limited to, at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different platform components (including the storage unit 620 and the processing unit 610), a display unit 640, and the like.

[0077] Among them, the storage unit stores program codes, and the program codes can be executed by the processing unit 610, so that the processing unit 610 executes the steps of various exemplary embodiments according to the present application described in the method part of the present specification. For example, the processing unit 610 can execute the steps as shown in Figure 1 .

[0078] The storage unit 620 can include a readable medium in the form of a volatile storage unit, such as a random access memory (RAM) 6201 and / or a cache memory unit 6202, and can further include a read-only memory (ROM) 6203.

[0079] The storage unit 620 can further include a program / utility 6204 having a set of (at least one) program modules 6205, including but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which or a combination of which can include implementation of a network environment.

[0080] The bus 630 can be one or more of several types of bus structures, including a storage unit bus or storage unit controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of a variety of bus structures.

[0081] The electronic device 600 can also communicate with one or more external devices 700 such as a keyboard or pointing device, a Bluetooth device, or a database, and / or one or more devices that enable a user to interact with the electronic device 600 and / or one or more devices (e.g., routers, modems) that enable the electronic device 600 to communicate with one or more other computing devices. Such communication can occur via an input / output interface 650. Still yet, the electronic device 600 can communicate with one or more networks, such as a local area network (LAN), a general wide area network (WAN), and / or the Internet, through a network adapter 660. The network adapter 660 can be any of a plurality of different types of adapters suitable for interfacing the electronic device 600 to various networking schemes. It should be appreciated that, while not shown explicitly, other hardware and / or software modules could be used in conjunction with the electronic device 600, including but not limited to microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data archival storage systems, etc.

[0082] Embodiment 4 The present application also provides a storage medium, specifically a computer readable storage medium, which is a memory device in the terminal device, and is used to store programs and data. It should be understood that the computer readable storage medium herein can include the built-in storage medium in the terminal device, and of course can also include the expansion storage medium supported by the terminal device, and can be any tangible medium containing or storing programs, which can be used by or in conjunction with an instruction execution system, device or apparatus. The computer readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and the instructions can be one or more computer programs (including program codes). It should be noted that more specific examples of the computer readable storage medium include an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical fiber, a portable compact disk read-only memory, an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0083] Computer readable storage media further includes data signals transported through a carrier wave and a propagation medium comprising or storing the program code. These data signals can be electrically, mechanically, or otherwise transported over a propagation medium, examples of which include but are not limited to wireless signals, wires, cables, optical fibers, etc. The computer readable storage medium can also be any tangible storage medium which can store or carry the program codes for use by or in connection with an instruction execution system, apparatus, or device.

[0084] The program code can be implemented in any of various ways, including procedure-based techniques, component-based techniques, and / or object-oriented techniques, among others. For example, the program code can be implemented in a variety of programming languages such as object-oriented based languages including Java, C++, and the like, conventional procedural languages such as the "C" language, assembly language, and / or the like. The program code can execute entirely on the user's computing device, partly on the user's computing device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device or entirely on the remote computing device or server. In the latter scenario, the remote computing device can be connected to the user's computing device through any type of network, including a local area network ("LAN"), a wide area network ("WAN"), or the Internet, among others, using a communication interface.

[0085] The one or more instructions stored in the computer readable storage medium can be loaded and executed by a processor to implement the corresponding steps of the method for perception-aided UAV trajectory and power allocation joint optimization in the above embodiments; the one or more instructions stored in the computer readable storage medium are loaded and executed by the processor to implement the following steps: The high-dimensional CSI data autocorrelation matrix is scaled to obtain an eigenvector; a distance matrix representing channel differences is generated by Dijkstra algorithm with the Euclidean distance of the eigenvector as the edge weight; the high-dimensional CSI data is mapped to a two-dimensional space by unsupervised learning using a twin neural network, and the mapped two-dimensional coordinates maintain the relative distance relationship under the high-dimensional CSI feature; the channel atlas is corrected based on the known user location, and the k-nearest neighbor method is used to estimate the unknown user location; based on the obtained user location, a UAV energy minimization model is established; the obtained UAV energy minimization model is decomposed into a power allocation sub-problem and a trajectory optimization sub-problem, which are respectively transformed into convex problems by Lyapunov optimization and first-order Taylor expansion; the two sub-problems are alternately iterated and optimized until convergence, and the optimal trajectory and power allocation scheme are output.

[0086] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0087] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0088] The following uses three sets of experimental data (corresponding to) Figures 3-6 To verify the core advantages of this invention, the experimental environment is as follows: Drone parameters: rotor radius 0.4m, maximum speed 30m / s, flight altitude 100m; Communication parameters: bandwidth 10MHz, noise power -110dBm, user rate requirements R = 1~4bps / Hz; Comparison Algorithms: SCA trajectory optimization: Assuming the user's location is known, only the trajectory is optimized; Fixed hover: The drone hovers at the user's center point, optimizing power only; Traditional channel map: Euclidean distance is used instead of the shortest path distance matrix.

[0089] Energy consumption comparison under different rate requirements

[0090] exist R At ˉ=2bps / Hz, the energy consumption of this invention (78.6kJ) is 30% lower than that of SCA optimization (112.4kJ) and 45% lower than that of fixed hovering (143.2kJ). Due to the joint optimization and dynamic adjustment of trajectory and power, high-demand scenarios (Rˉ=4) are also achieved. R (4) still maintains a significant advantage User speed requirement satisfaction rate

[0091] In R ˉ=2bps / Hz, the present application meets the rate of 98.7%, much higher than the SCA optimization (86.3%) and the traditional channel map (75.1%). The traditional method causes positioning distortion (error >5m) due to the Euclidean distance, resulting in user rate calculation deviation; the present application has a positioning error of <1.5m, ensuring the reliability of optimization.

[0092] 50 user large-scale scene performance

[0093] When the number of users is 50, the energy consumption (210.5kJ) of the present application is reduced by 27% compared with the SCA optimization (288.7kJ), and the meeting rate is 98.3%, verifying the scalability of the algorithm. The fixed hovering scheme cannot dynamically adapt to the user distribution, and the meeting rate is only 65.4%; the present application solves this problem by adjusting the real-time trajectory.

[0094] The above data proves that: the twin network + shortest path distance matrix compresses the positioning error to 1.2m (traditional method >5m), and radically solves the problem of "optimization failure due to unknown position" in the background technology. The joint optimization framework reduces energy consumption by more than 27% while ensuring >98% rate meeting rate, solving the endurance bottleneck of unmanned aerial vehicles. The alternating iteration algorithm takes <0.5s / time on embedded chips ( Figure 8 ), meeting the high real-time scene demand of emergency communication.

[0095] Please refer to Figure 2 , the double-branch network shares weights, and outputs two-dimensional coordinates. Compared with the Euclidean distance, this structure maintains the original CSI manifold structure, reduces the positioning error by 32%, and solves the high-dimensional nonlinear positioning distortion problem.

[0096] Please refer to Figure 3 , Figure 4 and Figure 5 , when the user rate demand ( Figure 3 ) and the number of users ( Figure 5 ) are different, the power consumption of the present application is significantly lower than that of the "SCA trajectory optimization" and "fixed hovering" schemes. The joint optimization saves energy by 27% in complex scenes, verifying the synergy advantage of the dynamic model and communication power consumption.

[0097] Please refer to Figure 6 , when R ˉ=2 bps / Hz and the number of users is 50, the present application meets the rate of >98%, while the comparative schemes are all <90, the channel map position correction + joint optimization ensures the quality of service, solving the optimization failure problem caused by position error in the background technology.

[0098] In summary, the present application is a perception-assisted UAV trajectory and power allocation joint optimization method and system, which innovatively combines shortest path distance matrix and twin neural network to compress CSI space mapping error to within 1.2m, overcoming the problem of high-dimensional data nonlinear distortion. In a 50-user large-scale scenario, the positioning accuracy is improved by 32%, providing reliable input for trajectory optimization, solving the industry pain point of "optimization failure due to unknown location" in the background technology, and establishing an accurate dynamic model containing blade and induced power P ind P The alternately iterative Lyapunov optimization and successive convex approximation saves more than 27% of energy in complex scenarios. Experiments show that the total energy consumption is only 210.5kJ when 50 users are distributed, the battery endurance is improved by 37%, the sub-problem decomposition strategy reduces the computational complexity by 50%, the embedded chip takes <0.5s for a single iteration, supports online real-time decision-making of UAV, realizes 68 hours of continuous coverage in typical scenarios such as disaster area emergency communication, and the user connectivity rate is 98.5%, providing core technical support for 5G / 6G air base station deployment.

[0099] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional units and modules is exemplified, and in actual application, the above functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The above integrated unit can be realized in the form of hardware or software function unit. In addition, the specific names of each functional unit and module are only for easy distinction, and do not limit the protection scope of the application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, which will not be described here.

[0100] In the above embodiments, the description of each embodiment has its own emphasis, and the parts not described or recorded in detail in a certain embodiment can be referred to the related description of other embodiments.

[0101] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in the present application can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0102] In the embodiments of the present application, it should be understood that the disclosed apparatus / terminal and method can be implemented in other manners. For example, the apparatus / terminal embodiments described above are merely schematic. For example, the division of the modules or units is merely logical function division. There can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between the units can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or in other forms.

[0103] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0104] In addition, each functional unit in the various embodiments of the present application can be integrated in one processing unit, or each unit can be a physically independent unit, or two or more units can be integrated in one unit. The integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0105] The integrated module / unit, if implemented in the form of a software functional unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, all or part of the flow of the above-mentioned embodiment methods can be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. When the processor executes the computer program, the steps of each method embodiment described above can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable file or some intermediate form. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the computer readable medium can include or exclude contents according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

[0106] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks Figure 1 means for functionally implementing the steps in one or more flow or blocks

[0107] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart block or blocks. Figure 1 one or more flow or blocks Figure 1 means for functionally implementing the steps in one or more flow or blocks

[0108] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks Figure 1 means for functionally implementing the steps in one or more flow or blocks

[0109] The above merely provides the technical idea of the present application and cannot be used to limit the protection scope of the present application. Any modification made according to the technical idea of the present application on the basis of the technical solutions falls within the protection scope of the claims of the present application.

Claims

1. A method of perception-augmented joint optimization of UAV trajectory and power allocation, characterized in that, The method comprises the following steps: scaling the high-dimensional CSI data autocorrelation matrix to obtain an eigenvector; generating a distance matrix representing channel differences by Dijkstra algorithm with the Euclidean distance of the eigenvector as the edge weight; mapping the high-dimensional CSI data into a two-dimensional space by unsupervised learning using a Siamese neural network, and the mapped two-dimensional coordinates maintain the relative distance relationship under the high-dimensional CSI features; correcting the channel atlas based on the known user position, and estimating the unknown user position by k-nearest neighbor method; establishing a UAV energy minimization model based on the obtained user position; decomposing the obtained UAV energy minimization model into a power allocation sub-problem and a trajectory optimization sub-problem, and respectively converting them into convex problems by Lyapunov optimization and first-order Taylor expansion; and optimizing the two sub-problems by alternating iteration until convergence, and outputting the optimal trajectory and power allocation scheme. 2.The perception-augmented UAV trajectory and power allocation joint optimization method of claim 1, wherein, scaling the high-dimensional CSI data autocorrelation matrix to obtain an eigenvector, specifically: Each high-dimensional CSI data is represented as , whose autocorrelation matrix is , and is scaled as , , is the feature strongly related to the environment extracted from the high-dimensional CSI data. 3.The perception-augmented UAV trajectory and power allocation joint optimization method of claim 1, wherein, generating a distance matrix representing channel differences by Dijkstra algorithm with the Euclidean distance of the eigenvector as the edge weight, specifically: The Euclidean distance between the features of any two points is L points are regarded as nodes in the graph, a weighted undirected graph G=(V, E) is constructed, V is the node set, E is composed of edges between each node and its nearest n neighbors, and the weight of the edge is the Euclidean distance in the feature space. For the constructed graph G, Dijkstra algorithm is used to calculate the shortest path length from each point to all other points in the graph, and the output is a LxL shortest path distance matrix D, represents the channel difference between point i and the transmitter and point j and the transmitter. 4.The perception-augmented UAV trajectory and power allocation joint optimization method of claim 1, wherein, the loss function of the Siamese neural network is: wherein, is the output of the neural network, denotes the channel difference between point i and the transmitter and point j and the transmitter, and is the input of the neural network, i.e. the features extracted from the high-dimensional CSI data.

5. The perception-augmented UAV trajectory and power allocation co-optimization method of claim 1, wherein, the UAV energy minimization model is: wherein, is the average energy consumption of the UAV in T time, is the bandwidth of the entire system, is the flight time of the UAV, is the standard deviation of the noise power, is the height of the UAV, is the power of the UAV transmitted to the ground user in the time slot, is the power of the UAV transmitted to the ground user in the time slot, is the speed of the UAV in the time slot, is the channel power gain per unit distance, is the user position coordinate, is the minimum data transmission rate of each user in this period of time, is the set of ground users, is the coordinate of the initial position of the UAV projected to the ground, is the coordinate of the terminal position of the UAV projected to the ground, is the left and right boundary of the area, is the upper and lower boundary of the area, is the maximum total transmission power of the UAV, is the maximum flight speed of the UAV, is the unit time slot length.

6. The perception-augmented UAV trajectory and power allocation co-optimization method of claim 1, wherein, converted into a convex problem by Lyapunov optimization and first-order Taylor expansion as follows: wherein, P t is the power of the UAV transmitted to the ground users at the tth time slot, is the trade-off factor, is the set of ground users, is the data packet queue that the tth time slot user needs to transmit, is the system bandwidth, is the standard deviation of the noise power, is the flight height of the UAV, is the channel power gain per unit distance, is the coordinate of the UAV projected to the ground at the tth time slot, is the user position coordinate, is the maximum total transmission power of the UAV, is the flight time of the UAV, is the blade type power in the hovering state, is the tip speed of the rotor blade, is the induced power in the hovering state, is the average rotor induced speed in the hovering state, is the speed of the UAV at the tth time slot, is the fuselage drag ratio, is the air density, is the rotor solidity, is the rotor disc area, is the channel gain after the first-order Taylor expansion, is the minimum data transmission rate of each user in this period of time, is the coordinate of the initial position of the UAV projected to the ground, is the coordinate of the termination position of the UAV projected to the ground, is the maximum speed of the UAV, is the unit time slot length, is the left and right boundary of the region, is the upper and lower boundary of the region.

7. The perception-augmented UAV trajectory and power allocation co-optimization method of claim 6, wherein, the power allocation sub-problem and the trajectory optimization sub-problem are as follows: 。 8. The perception-augmented UAV trajectory and power allocation co-optimization method of claim 1, wherein, optimizing the two sub-problems by alternating iteration until convergence, specifically: Initialization: input variables and initialization values and number of iterations error precision ; iteration operation; Output drone in all time slots Flight trajectory of the upper drone And transmit power allocated to the ground user . 9.The perception-augmented UAV trajectory and power allocation joint optimization method of claim 8, wherein, the iterative operation is specifically: fixing to , solve sub-problem one to get the optimal solution of optimization variable x ; fixing to , solve subproblem two, get the optimal solution of optimization variable is ; update = , = ; When the following conditions are met , , return to solving subproblem one, otherwise the iteration terminates. 10.A system for perception-aided joint optimization of UAV trajectory and power allocation, the system comprising: comprises: a feature module for scaling the high-dimensional CSI data autocorrelation matrix to obtain an eigenvector; generating a distance matrix representing channel differences by Dijkstra algorithm with the Euclidean distance of the eigenvector as the edge weight; a mapping module for mapping the high-dimensional CSI data into a two-dimensional space by unsupervised learning using a Siamese neural network, and the mapped two-dimensional coordinates maintain the relative distance relationship under the high-dimensional CSI features; an estimation module for correcting the channel atlas based on the known user position, and estimating the unknown user position by k-nearest neighbor method; a construction module for establishing a UAV energy minimization model based on the obtained user position; a distribution module for decomposing the obtained UAV energy minimization model into a power allocation sub-problem and a trajectory optimization sub-problem, and respectively converting them into convex problems by Lyapunov optimization and first-order Taylor expansion; and optimizing the two sub-problems by alternating iteration until convergence, and outputting the optimal trajectory and power allocation scheme.

Images