Unmanned aerial vehicle sensing resource allocation method and system for coverage path task

Abstract

This invention relates to a method and system for allocating UAV (Unmanned Aerial Vehicle) sensory computing resources for coverage path tasks, belonging to the field of wireless communication technology. The method includes: constructing a topology model based on the geographic information of the target area and deploying ground user terminals; constructing a UAV perception quality model, a communication rate model between the UAV and the ground terminals, a calculation model for the UAV data generation rate, and an energy consumption model for the total instantaneous power of the UAV; determining the decision variables to be optimized based on the energy consumption model, and setting multiple constraints based on the perception quality model, communication rate model, and calculation model; obtaining the instantaneous minimum power function based on the multiple constraints and the energy consumption model; obtaining the optimal speed of the UAV on the optimal path based on this function; and generating the optimal perception power, communication power, and calculation frequency based on the optimal speed to complete the allocation of UAV sensory computing resources. This invention improves the adaptability and robustness of UAV sensory computing resource allocation.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and in particular to a method and system for allocating sensory computing resources for UAVs to cover path tasks. Background Technology

[0002] With the rapid development of the low-altitude economy, unmanned aerial vehicles (UAVs) have become the core execution vehicle for regional coverage and periodic inspection tasks due to their advantages of high mobility, flexible deployment, and wide coverage. In applications such as smart agriculture, traffic network inspection, and pipeline and infrastructure monitoring, UAVs need to continuously cover linear or area target areas along preset or dynamically adjusted flight paths, while providing communication connections for ground terminals and performing real-time or near-real-time calculations on the collected sensing or communication data, forming an integrated service demand for sensing, communication, and computing.

[0003] To enhance the comprehensive service capabilities of unmanned aerial vehicle (UAV) systems, Integrated Sensing, Communication, and Computation (ISCC) technology has emerged. This technology integrates sensing, communication, and computing functions on a single UAV platform, relying on unified system design and resource scheduling to achieve multi-task collaborative execution. In an ISCC UAV system, flight requires simultaneous completion of comprehensive sensing, ground user communication services, and local / collaborative computing processing. Its operational performance is highly dependent on the coordinated matching of flight control strategies, ICC resource allocation methods, and path planning schemes; any disruption in the optimization of any of these components will lead to a decline in overall system performance.

[0004] Existing technologies for drone coverage path planning abstract it into a graph theory or geometric coverage problem, typically based on path modeling of the Traveling Salesman Problem (TSP) and the Chinese Postman Problem (CPP), with optimization objectives focused on minimizing flight distance or task completion time. However, these methods optimize only from the geometric features of the path, assuming that the drone has fixed or sufficient sensing, communication, and computing capabilities, without considering the dynamic impact of flight state changes such as flight speed and turning maneuvers on synesthetic computing performance. This leads to difficulties in matching the planned path with resource supply capacity during actual execution. On the other hand, most research on optimizing drone communication or computing capabilities assumes that the flight trajectory is known or fixed, focusing on resource scheduling issues such as communication power control, spectrum allocation, and computation offloading strategies. These methods ignore the coupling relationship between flight control variables and synesthetic computing resources, making it difficult for the designed resource allocation strategies to remain optimal during actual flight execution. Furthermore, some studies have attempted to jointly optimize flight trajectory and communication or sensing performance, but these typically only involve partial coupling between sensing and communication, or employ a phased, hierarchical decoupling strategy, i.e., first completing flight path planning and then allocating synergistic computing resources under a fixed path. This approach is prone to global performance degradation in real-world applications involving periodic coverage, multi-constraint collaboration, and real-time response, and lacks adaptability and robustness to dynamic environments.

[0005] In summary, existing technologies for ISCC UAV coverage path planning generally suffer from the following problems: First, the lack of a unified modeling and joint optimization framework for flight control, coverage path planning, and sensory computing resource allocation leads to insufficient synergy among the three. Second, existing models cannot accurately characterize the comprehensive impact mechanism of flight state changes on perception quality, communication performance, and computing latency. Third, the system optimization objectives are disconnected from engineering constraints such as UAV endurance, payload, and real-time performance, resulting in insufficient feasibility of the solution. Summary of the Invention

[0006] Therefore, the technical problem to be solved by the present invention is to overcome the problems of insufficient adaptability and robustness caused by the separation of UAV coverage path planning, flight control and synergistic computing resource allocation, lack of unified modeling and joint optimization in the prior art.

[0007] In a first aspect, to solve the above-mentioned technical problems, the present invention provides a method for allocating UAV sensory computing resources for covering path tasks, comprising:

[0008] S1. Based on the geographic information of the target area, construct a topology model of the target area, and deploy ground user terminals within the geographic area covered by the topology model;

[0009] S2. Construct a perception quality model for the UAV, a communication rate model between the UAV and the ground user terminal, a calculation model for the UAV data generation rate, and an energy consumption model for the total instantaneous power of the UAV.

[0010] S3. Based on the energy consumption model, construct a weighted cost function; use the weighted cost function as the optimization objective; determine the decision variables to be optimized based on the optimization objective; determine multiple constraints based on the decision variables to be optimized, the perceived quality model, the communication rate model, and the calculation model.

[0011] S4. Based on the topological model, obtain all nodes with odd degrees and construct a first set; based on the first set, construct a mixed-weight complete graph; calculate the weights of the edges in the mixed-weight complete graph; based on the mixed-weight complete graph and the weights, augment the undirected graph of the topological model to obtain an augmented undirected graph; based on the augmented undirected graph, obtain the optimal total path length.

[0012] S5. Based on the multiple constraints, obtain the optimal communication power closed-form solution, the optimal sensing power closed-form solution, and the optimal computation frequency closed-form solution; based on the optimal communication power closed-form solution, the optimal sensing power closed-form solution, the optimal computation frequency closed-form solution, and the energy consumption model, obtain the instantaneous minimum power function; based on the instantaneous minimum power function and the optimal total path length, calculate the optimal multiplier; based on the optimal multiplier, calculate the optimal speed of the UAV; based on the optimal speed, generate the optimal sensing power, the optimal communication power, and the optimal computation frequency, and complete the allocation of UAV communication and computing resources.

[0013] In one embodiment of the present invention, in step S5, the optimal multiplier is calculated based on the instantaneous minimum power function and the optimal path total length; the optimal speed of the UAV is calculated based on the optimal multiplier as follows:

[0014] Construct a local Hamiltonian based on the instantaneous minimum power function;

[0015] Based on the local Hamiltonian, the total length of the optimal path, and the total capacity of the UAV's onboard battery, construct a dual expression;

[0016] The dual expression is iteratively searched using a bisection method. In each iteration, the instantaneous flight velocity at each path position is calculated until the global energy constraint is satisfied, and the optimal multiplier is obtained.

[0017] The optimal speed of the UAV is calculated based on the optimal multiplier.

[0018] In one embodiment of the present invention, in step S5, the expressions for the optimal communication power closed-form solution, the optimal sensing power closed-form solution, and the optimal computational frequency closed-form solution are respectively:

[0019] ;

[0020] ;

[0021] ;

[0022] in, This represents the closed-form solution for optimal communication power. This represents the channel gain for the bottleneck user within the dynamic coverage radius of the drone. Indicates the minimum communication rate threshold. Indicates the drone's location on the path. The user set within the coverage area Represents the threshold constant. This represents the energy efficiency constant of the UAV remote sensing camera system. Indicates bandwidth. Indicates noise power. This represents the number of CPU cycles required to compute a unit of data. Indicates instantaneous flight speed. Indicates the minimum perceived quality threshold. Indicates the speed at which remote sensing data is generated. Indicates pixel size, Indicates the principal distance. Indicates relative flight altitude. This indicates the width of the fixed ground strip covered by the camera sensor. This indicates the color bit depth of the image.

[0023] In one embodiment of the present invention, in step S2, the expression of the perceived quality model is:

[0024] ;

[0025] in, Indicates instantaneous perceived quality. This represents the energy efficiency constant of the UAV remote sensing camera system. Indicates relative flight altitude. Indicates the principal distance. Indicates pixel size, Indicates instantaneous sensed power. Indicates instantaneous flight speed. Indicates the path location label.

[0026] In one embodiment of the present invention, in step S2, the expression for the communication rate model between the UAV and the ground user terminal is:

[0027] ;

[0028] in, Represents the transmission efficiency factor. Indicates the drone's location on the path. Place and No. Instantaneous communication rate of a ground user terminal Indicates bandwidth. This represents the power of Gaussian white noise. Represents the Doppler coefficient. Indicates instantaneous flight speed. Indicates the transmission power. This represents the large-scale channel gain.

[0029] In one embodiment of the present invention, the expression for the calculation model of the UAV data generation rate in step S2 is as follows:

[0030] ;

[0031] in, Indicates the drone data generation rate. This indicates the width of the fixed ground strip covered by the camera sensor. Indicates the color bit depth of the image. Indicates instantaneous flight speed. It represents the quality perceived instantaneously.

[0032] In one embodiment of the present invention, in step S2, the expression for the energy consumption model of the total instantaneous power of the UAV is:

[0033] ;

[0034] in, Indicates the drone's location on the path. Total instantaneous power at the location, Indicates propulsion power. Indicates instantaneous flight speed. Indicates the sensed power. Indicates the transmission power. Indicates the chip's energy consumption coefficient. This indicates the frequency of calculation.

[0035] In one embodiment of the present invention, in step S3, the expression for the weighted cost function is constructed based on the energy consumption model as follows:

[0036] ;

[0037] in, This represents the weighted cost function. Indicates the total path length. This represents the time weighting coefficient. This represents the energy weighting coefficient. Indicates the drone's location on the path. Total instantaneous power at the location, Indicates instantaneous flight speed.

[0038] In one embodiment of the present invention, the plurality of constraints include perception quality constraints, communication rate constraints, computing power constraints, physical resource constraints, and global energy hard constraints.

[0039] Secondly, to solve the above-mentioned technical problems, the present invention provides a UAV sensory computing resource allocation system for covering path tasks, used to implement the above-mentioned UAV sensory computing resource allocation method for covering path tasks, including:

[0040] The deployment module is used to construct a topology model of the target area based on the geographic information of the target area, and to deploy ground user terminals within the geographic area covered by the topology model.

[0041] The model building module is used to build a perception quality model of the UAV, a communication rate model between the UAV and the ground user terminal, a calculation model of the UAV data generation rate, and an energy consumption model of the total instantaneous power of the UAV.

[0042] The optimization decision module is used to construct a weighted cost function based on the energy consumption model; use the weighted cost function as the optimization objective; determine the decision variables to be optimized based on the optimization objective; and determine multiple constraints based on the decision variables to be optimized, the perceived quality model, the communication rate model, and the calculation model.

[0043] The path planning module is used to: obtain all nodes with odd degrees according to the topology model and construct a first set; construct a mixed-weight complete graph according to the first set; calculate the weights of the edges in the mixed-weight complete graph; augment the undirected graph of the topology model according to the mixed-weight complete graph and the weights to obtain an augmented undirected graph; and obtain the optimal total path length according to the augmented undirected graph.

[0044] The sensory computing resource allocation module is used to obtain the optimal communication power closed-form solution, the optimal sensing power closed-form solution, and the optimal computing frequency closed-form solution based on the multiple constraints; to obtain the instantaneous minimum power function based on the optimal communication power closed-form solution, the optimal sensing power closed-form solution, the optimal computing frequency closed-form solution, and the energy consumption model; to calculate the optimal multiplier based on the instantaneous minimum power function and the optimal total path length; to calculate the optimal speed of the UAV based on the optimal multiplier; and to generate the optimal sensing power, optimal communication power, and optimal computing frequency based on the optimal speed, thereby completing the sensory computing resource allocation for the UAV.

[0045] Compared with the prior art, the above-described technical solution of the present invention has the following advantages:

[0046] (1) The UAV sensing and computing resource allocation method and system for covering path tasks described in this invention unifies and optimizes the three types of tasks: sensing, communication and computing. This breaks through the limitations of the traditional single-task mode of UAVs and realizes the efficient reuse of airborne resources. While completing the task of full coverage monitoring of linear infrastructure, it can provide stable air communication access services for ground users, which greatly improves the comprehensive utilization efficiency of UAV platforms. By establishing quantitative constraints for sensing quality, communication rate and computing real-time performance, it effectively avoids problems such as motion blur, Doppler interference and computing overload that may occur during high-speed flight, ensuring that monitoring data is clear and usable, communication links are stable and reliable, and data processing is real-time and efficient, thereby enhancing the stability and robustness of task execution in complex scenarios.

[0047] (2) The present invention adopts a hierarchical alternating optimization solution method, which not only ensures the stability and convergence of the calculation process, but also reduces the complexity of algorithm implementation. It can be adapted to the hardware configuration of existing UAV platforms and is easy to deploy in engineering. The present invention is particularly suitable for UAV coverage inspection application scenarios with limited computing power and limited energy supply, and has good engineering feasibility and practicality. Attached Figure Description

[0048] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0049] Figure 1 This is a flowchart of a UAV sensory computing resource allocation method for a covered path task, according to a preferred embodiment of the present invention.

[0050] Figure 2 This is a schematic diagram of a path coverage application scenario for the integrated sensing and computing of unmanned aerial vehicles (UAVs) in a preferred embodiment of the present invention.

[0051] Figure 3This is a schematic diagram of the coverage path planning result and flight path in a preferred embodiment of the present invention;

[0052] Figure 4 This is a schematic diagram showing the comparison of penalty correction costs for different resource allocation strategies in a preferred embodiment of the present invention. Detailed Implementation

[0053] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0054] Example 1:

[0055] Reference Figure 1 As shown, the present invention provides a method for allocating UAV sensory computing resources for covering path tasks, including but not limited to the following steps:

[0056] S1. Based on the geographic information of the target area, construct a topology model of the target area and deploy ground user terminals within the geographic area covered by the topology model;

[0057] S2. Construct a perception quality model for the UAV, a communication rate model between the UAV and the ground user terminal, a calculation model for the UAV data generation rate, and an energy consumption model for the total instantaneous power of the UAV.

[0058] S3. Based on the energy consumption model, construct a weighted cost function; use the weighted cost function as the optimization objective; determine the decision variables to be optimized based on the optimization objective; determine multiple constraints based on the decision variables to be optimized, the perceived quality model, the communication rate model, and the calculation model.

[0059] S4. Based on the topological model, obtain all nodes with odd degrees and construct the first set; based on the first set, construct a mixed-weight complete graph; calculate the weights of the edges in the mixed-weight complete graph; based on the mixed-weight complete graph and the weights, augment the undirected graph of the topological model to obtain an augmented undirected graph; based on the augmented undirected graph, obtain the total length of the optimal path.

[0060] S5. Based on multiple constraints, obtain the closed-form solution for optimal communication power, optimal sensing power, and optimal computation frequency; based on the closed-form solution for optimal communication power, optimal sensing power, optimal computation frequency, and energy consumption model, obtain the instantaneous minimum power function; based on the instantaneous minimum power function and the optimal total path length, calculate the optimal multiplier; based on the optimal multiplier, calculate the optimal speed of the UAV; based on the optimal speed, generate the optimal sensing power, optimal communication power, and optimal computation frequency, and complete the allocation of UAV communication and computing resources.

[0061] This invention discloses a UAV-based sensory computing resource allocation method for coverage path tasks. This method effectively addresses the technical shortcomings of traditional UAV monitoring, such as single monitoring tasks, low resource utilization, insufficient task reliability, and unreasonable path planning, demonstrating significant technical advantages. Based on the geographic information of the target area, this method constructs a topology model and deploys ground user terminals, laying a precise scene foundation for subsequent path planning and sensory computing task collaboration. This ensures comprehensive coverage of the monitoring range and user communication needs, avoiding task deviations caused by inaccurate scene modeling. In step S2, a mathematical model is constructed relating UAV perception quality, communication rate, data generation rate, and total instantaneous power. This provides a scientific quantitative basis for subsequent optimization of target setting, constraint determination, and resource allocation, achieving accurate modeling of each sensory computing task and breaking the limitations of independent modeling and lack of collaboration in traditional monitoring. In step S3, a weighted cost function is constructed based on the energy consumption model as the optimization objective. Constraints are determined by combining the decision variables to be optimized and the models of each task, achieving coordinated optimization of task efficiency and energy consumption. This ensures that the optimization direction aligns with actual task requirements. Simultaneously, by clearly defining constraints, the basic performance of sensing, communication, and computing tasks is guaranteed, avoiding performance failures. Based on the topology model, singularity nodes are identified, a mixed-weight complete graph is constructed, and topology augmentation is performed to obtain the optimal total path length. This generates a fully covered, non-redundant optimal flight path, effectively shortening the path length, reducing energy waste from redundant flight, improving monitoring efficiency, and solving the problems of unreasonable and incomplete coverage in traditional path planning. Closed-form solutions for various resources are obtained based on constraints. The optimal speed of the UAV is calculated by combining the instantaneous minimum power function, the optimal total path length, and the optimal multiplier, ultimately achieving precise allocation of synesthetic computing resources. This realizes dynamic coordination between flight speed and synesthetic computing resources, ensuring the service quality of each task while minimizing UAV energy consumption and extending effective operating time. Furthermore, obtaining closed-form solutions improves the efficiency and accuracy of resource allocation, facilitating real-time engineering deployment. Overall, this step-by-step design achieves full-process collaboration in scenario modeling, task modeling, optimization decision-making, path planning, and resource allocation, improving the accuracy, efficiency, and reliability of UAV monitoring. It is applicable to various linear infrastructure monitoring scenarios and has strong versatility and scalability.

[0062] Specifically, refer to Figure 2As shown in the figure, this is an application scenario of path coverage assisted by integrated sensing and computing technologies from a UAV. The system in this scenario includes a UAV with radar sensing, wireless communication, and edge computing capabilities, multiple ground user terminals, and a linear target area to be monitored. Based on this scenario, this embodiment proposes a joint optimization method for UAV flight control and sensing / computing resources for path coverage planning tasks. By constructing a unified mathematical model between the coverage path, flight control variables, and sensing / computing resources, and employing a hierarchical alternating optimization approach to solve various decision variables step by step, a globally or near-optimal control and resource allocation method is obtained while satisfying system constraints.

[0063] Specifically, in step S1, the steps for constructing a topology model of the target area based on the geographic information of the target area, and deploying ground user terminals within the geographical area covered by the topology model are as follows:

[0064] S110. Construct a topological model of the target monitoring area. Obtain the geographic information of the linear infrastructure to be monitored (such as urban road networks and oil pipeline networks), and abstract it into a weighted undirected graph. .in, Represents the set of intersection nodes or endpoints. This represents the set of road segments that need to be covered. This represents the total number of intersection nodes or endpoints. This represents the total number of road segments that need to be covered. For an undirected graph... Each edge in Its weight Initialize to the actual physical length of the road segment. The drone needs to traverse the set. All edges in the equation must be encountered at least once.

[0065] S120. Deploy ground user terminals within the geographical area covered by the topology model. (In an undirected graph) Within the covered geographical area, it is generated based on the Poisson Point Process or other distribution models. There are 1 ground users, and their coordinate set is denoted as . ,in These users have a continuous need for communication access during drone coverage, and the drone needs to act as an aerial base station to provide them with wireless signal coverage.

[0066] Furthermore, in this embodiment, after deploying ground user terminals within the geographical area covered by the topology model, the method also includes initializing the UAV platform parameters and initializing the Quality of Service (QoS) indicators for the synaptic computing task.

[0067] S130. Initialize UAV platform parameters. In this embodiment, the number of UAVs performing the mission is set to 1, and the flight altitude is a fixed value. The total capacity of the onboard battery is Simultaneously configure the drone's dynamic parameters (such as drag coefficient and rotor area) and the upper limit of onboard resources, including maximum flight speed. Maximum calculation frequency Maximum sensing power and maximum transmission power .

[0068] S140. Initialize the service quality indicators for the synesthetic computing task, and set the minimum performance threshold that the coverage task must meet. Specifically, this includes setting a minimum perception quality threshold for the perception task. This is used to ensure the clarity of acquired images or radar data, preventing motion blur and unusability due to excessive speed; for communication tasks, a minimum communication rate threshold is set. This is used to ensure the real-time performance and reliability of control command issuance or monitoring data feedback; for computational tasks, data processing coefficients are set. This coefficient represents the number of CPU cycles required per unit of sensor data, measured in cycles / bit, to ensure that onboard computing can process the generated data stream in real time.

[0069] Specifically, in step S2, the specific steps for constructing the UAV's perception quality model, the communication rate model between the UAV and the ground user terminal, the calculation model of the UAV's data generation rate, and the energy consumption model of the UAV's total instantaneous power are as follows:

[0070] S210. Define the instantaneous perceived quality of the UAV at the corresponding location along the path, and establish a perceived quality model to address the motion blur effect caused by flight speed. The expression for the perceived quality model is:

[0071] ;

[0072] in, Indicates instantaneous perceived quality. This represents the energy efficiency constant of the UAV remote sensing camera system. Indicates pixel size, Indicates the principal distance (focal length). Indicates relative flight altitude. Indicates instantaneous sensed power. Indicates instantaneous flight speed. Indicates the path location label.

[0073] S220, defining drones and the first The instantaneous communication rate between ground users is considered. In this embodiment, the Doppler frequency shift introduced by the high-speed flight of the UAV disrupts the subcarrier orthogonality of the Orthogonal Frequency Division Multiplexing (OFDM) system, thus affecting communication performance. Based on this influencing factor, a communication rate model between the UAV and the ground user terminal is constructed. The expression for the communication rate model is:

[0074] ;

[0075] in, Represents the transmission efficiency factor. Indicates the drone's location on the path. Place and No. Instantaneous communication rate of a ground user terminal Indicates system bandwidth. This represents the power of Gaussian white noise. Represents the Doppler coefficient. Indicates the transmission power. This represents the large-scale channel gain (including path loss and shadow fading). This represents the inter-carrier interference power caused by the Doppler effect.

[0076] S230. Define the data generation rate of the UAV, which is positively correlated with the UAV's flight speed (i.e., scanning area rate) and sensing quality (i.e., sampling resolution). Therefore, the expression for the calculation model of the UAV data generation rate is:

[0077] ;

[0078] in, Indicates the drone data generation rate. This represents the width of the fixed ground strip covered by the camera sensor (where... (This refers to the sensor width) Indicates the color bit depth (bit / pixel) of the image. Indicates instantaneous flight speed. This represents the instantaneously perceived quality. In this embodiment, to meet real-time computing constraints, the CPU computing frequency... The required relation is: .

[0079] S240. Construct an energy consumption model for the total instantaneous power of the UAV. In this embodiment, the total instantaneous power of the UAV includes propulsion power and the power of the integrated sensor-computer interface payload. The energy consumption model is expressed as follows:

[0080] ;

[0081] in, Indicates the drone's location on the path. Total instantaneous power at the location, Indicates propulsion power. Indicates the sensed power. Indicates the transmission power. Indicates the chip's energy consumption coefficient. This indicates the frequency of calculation.

[0082] Furthermore, in this embodiment, the propulsion power Using a rotorcraft power model, the specific expression is as follows:

[0083] ;

[0084] in, and These represent the blade profile power and induced power in hovering mode, respectively. Indicates the speed at the tip of the blade. This represents the average rotor induced speed during hovering. Indicates the fuselage drag coefficient. Indicates air density, Indicates the area of ​​the rotor disk. Indicates rotor solidity.

[0085] Specifically, in step S3, the optimization problem is formally defined, including: determining the optimization objective, selecting optimization variables, establishing constraints, and decomposing the problem. The specific steps are as follows:

[0086] S310. Determine the optimization objective. Based on the energy consumption model constructed in step S2, construct a weighted cost function and use the weighted cost function as the optimization objective. In this embodiment, the purpose of constructing the weighted cost function is to minimize the task completion time and minimize the total energy consumption (i.e., maximize the task execution efficiency). The expression for the weighted cost function is:

[0087] ;

[0088] in, This represents the weighted cost function. Indicates the total path length. This represents the time weighting coefficient. This represents the energy weighting coefficient. Indicates the drone's location on the path. The total instantaneous power at the location.

[0089] S320. Determine the optimization variables based on the optimization objective. In this embodiment, the decision variables to be optimized include: coverage path. and instantaneous flight speed along the path Sensing power Communication transmission power and calculate frequency .

[0090] S330. Based on the aforementioned decision variables to be optimized, the perceived quality model, the communication rate model, and the computational model, several constraints are determined. These constraints include perceived quality constraints, communication rate constraints, computational capability constraints, physical resource constraints, and global energy hard constraints.

[0091] Furthermore, the expression for the perceived quality constraint is:

[0092] .

[0093] Furthermore, the expression for the communication rate constraint is:

[0094] .

[0095] Furthermore, the expression for the computational capability constraint is:

[0096] .

[0097] Furthermore, the expression for physical resource constraints is:

[0098] ;

[0099] in, This indicates the minimum flight speed.

[0100] Furthermore, the expression for the global energy hard constraint is:

[0101] .

[0102] S340. A hierarchical decoupling strategy is adopted to decompose the original optimization problem into two sub-problems. Sub-problem one is the coverage path planning problem (i.e., determining the total path length). and coverage path Subproblem two is the joint flight control and resource allocation problem (i.e., determining instantaneous flight speed). Sensing power Communication transmission power and calculate frequency ).

[0103] Specifically, in step S4, drone coverage path planning is performed for the target area, and the specific steps are as follows:

[0104] S410. Based on the topological model, obtain all nodes with odd degrees and construct the first set; based on the first set, construct a mixed-weight complete graph; calculate the weights of the edges in the mixed-weight complete graph. The specific steps are as follows:

[0105] S411, Traversing an Undirected Graph Identify all nodes with odd degrees and form the first set. .

[0106] S412, For the first set Construct a hybrid weighted complete graph. For any node pair... Calculate the weights of its edges. The weight is taken as the shortest path distance on the graph. Straight-line flight distance The smaller value of is expressed as follows:

[0107] ;

[0108] in, This indicates the additional cost of switching flight modes (such as steering). and Let u and v represent the coordinate vectors of nodes u and v in the mixed-weight complete graph, respectively.

[0109] S420, Based on the mixed weight complete graph and weights For undirected graphs of topological models The process involves augmentation to obtain an augmented undirected graph. Specifically, the Minimum Weight Perfect Matching (MWPM) algorithm is performed on the mixed-weight complete graph to obtain the optimal matching set. Based on the optimal matching set For the original undirected graph To augment, if the weight For the corresponding air distance, add virtual flight path edges; otherwise, add duplicate edges of the original graph path to obtain an augmented undirected graph.

[0110] S430. Based on the augmented undirected graph, the optimal total path length is obtained. Specifically, the augmented graph structure satisfies the criteria for an Eulerian graph and possesses the topological foundation for the existence of closed-loop traversal paths. Based on this graph structure, the Hierholzer algorithm is used to solve for and generate closed-loop traversal paths, thereby obtaining closed-loop covering paths that meet the full coverage requirement and have no redundant cycles. And further determine the total length of the optimal path corresponding to the optimal coverage path. Among them, closed-loop coverage path It is an Euler circuit.

[0111] Specifically, in step S5, instantaneous resources are allocated. This process includes obtaining the optimal communication power, optimizing the configuration of sensing and computing resources, calculating the instantaneous total power function, executing global speed control, and making joint optimization decisions based on the above characteristics, thereby achieving coordinated optimization and globally optimal allocation of communication, sensing, and computing resources. The specific steps are as follows:

[0112] S510. According to the communication rate model, at a given instantaneous speed Under the given conditions, derive the thresholds that satisfy the communication interruption probability and the minimum communication rate. The minimum transmit power closed-form solution (i.e., the optimal communication power closed-form solution) is derived based on the sensing quality model and constraints; the sensing quality is satisfied. The minimum sensing power closed-form solution (i.e., the optimal sensing power closed-form solution) is obtained; based on the real-time data generation rate, the optimal CPU frequency is determined, and the optimal computation frequency closed-form solution is obtained.

[0113] Furthermore, the expressions for the above-mentioned closed-form solution of optimal communication power, optimal sensing power, and optimal computational frequency are as follows:

[0114] ;

[0115] ;

[0116] ;

[0117] in, This represents the closed-form solution for optimal communication power. This represents the channel gain for the bottleneck user within the dynamic coverage radius of the drone. Indicates the minimum communication rate threshold. Indicates the drone's location on the path. The user set within the coverage area Represents the empty set; Represents the threshold constant. Indicates the transmission efficiency factor; This represents the energy efficiency constant of the UAV remote sensing camera system. Indicates bandwidth. Indicates noise power. This represents the number of CPU cycles (CPU cycles / bit) required to compute a unit of data. Indicates the minimum perceived quality threshold. This indicates the speed at which remote sensing data is generated.

[0118] S520, Solve the optimal communication power closed-form solution Optimal sensing power closed-form solution And the optimal calculation frequency closed-form solution Substitute into the energy consumption model , obtain only information about speed instantaneous minimum power function .

[0119] S530. Calculate the optimal multiplier based on the instantaneous minimum power function and the optimal total path length; the specific steps for calculating the optimal speed of the UAV based on the optimal multiplier are as follows:

[0120] S531. Construct a local Hamiltonian based on the instantaneous minimum power function. In this embodiment, Lagrange multipliers are introduced. Relax the global energy constraint and define the local Hamiltonian, with the following expression:

[0121] .

[0122] S532, Based on local Hamiltonian Total length of the optimal path and the total capacity of the drone's onboard battery Construct the dual expression. This step transforms the original problem into the dual problem. , where the dual expression for:

[0123] .

[0124] S533, utilizing total energy consumption with Lagrange multipliers Taking advantage of the monotonically decreasing property, a bisection method is used to iteratively search the dual expression. In each iteration, the instantaneous flight velocity at each path position is calculated until the global energy constraint is satisfied. Furthermore, the error converges, yielding the optimal multiplier. The formula for calculating instantaneous flight speed is as follows: .

[0125] S534, Based on the optimal multiplier Calculate the optimal velocity profile of the UAV along its entire path. The speed automatically decreases when the channel is poor or the sensing requirements are high, and automatically increases when the environment is good.

[0126] S540, Optimal Speed Substituting the solutions into the optimal communication power closed-form solution, the optimal sensing power closed-form solution, and the optimal calculation frequency closed-form solution respectively, the optimal sensing power is generated. Optimal communication power and optimal calculation frequency This allows for the allocation of computing resources for drone sensing.

[0127] Through the above steps, a final joint control strategy for UAV flight control and synaptic computing resource allocation was finally achieved under the premise of satisfying all instantaneous quality of service (QoS) constraints and global energy hard constraints, providing technical support for the efficient and stable execution of UAV path coverage missions.

[0128] Furthermore, the specific process by which the UAV performs coverage flight and synesthetic computing tasks according to the joint optimization results is as follows:

[0129] Step 1: The drone is pre-loaded with the flight trajectory and resource scheduling instruction table generated in step S5, and then follows the planned path. Fly, and at each location point along the path. Execute the corresponding speed control and power allocation commands to ensure that the sensing computing task and flight coverage are carried out synchronously and efficiently.

[0130] Step 2: Real-time acquisition of system operating status and feedback to the optimization module. During flight execution, the UAV monitors its actual remaining battery power, sudden environmental interference, and severe channel fluctuations in real time. If the detected status deviation exceeds a preset threshold, the path and resource replanning mechanism is triggered, and the optimization algorithm described in step S5 is invoked to quickly re-optimize the flight parameters and resource scheduling scheme of the remaining path, ensuring the safe and successful completion of the UAV path coverage mission.

[0131] The method described in this invention unifies and jointly optimizes three types of tasks: perception, communication, and computation. This solves the problem of fragmented flight control decisions, perception task execution, communication service support, and computational resource allocation in existing UAV coverage path planning and integrated sensing-computing systems, achieving efficient reuse of airborne resources. While completing the task of full-coverage monitoring of linear infrastructure, this method provides stable airborne communication access services to ground users, significantly improving the overall utilization efficiency of the UAV platform. By establishing quantitative constraints for perception quality, communication rate, and computational real-time performance, it effectively avoids problems such as motion blur, Doppler interference, and computational overload that may occur during high-speed flight, ensuring clear and usable monitoring data, stable and reliable communication links, and real-time and efficient data processing, thus enhancing the stability and robustness of task execution in complex scenarios. Simultaneously, this method constructs a weighted optimization with the goal of minimizing task completion time and total energy consumption. Under the premise of satisfying the physical resources and global energy constraints of the UAV, it achieves dynamic coordination between flight control and resource allocation, effectively reducing UAV energy consumption, extending effective operating time, and improving the continuous monitoring capability of large-scale linear infrastructure. In terms of path planning, a hierarchical strategy employing singularity node identification, mixed weight perfect matching, graph augmentation, and the Hierholzer algorithm to generate Eulerian circuits automatically generates fully covered, non-redundant, and optimal closed-loop flight paths. This avoids redundant flight in traditional traversal methods, shortens the total path length, and further improves overall monitoring efficiency. Furthermore, this method decomposes the complex joint optimization problem into two sub-problems—path planning and resource allocation—through a hierarchical decoupling strategy, significantly reducing the difficulty of solving the problem. This facilitates engineering implementation and real-time deployment, making it applicable to various linear infrastructure monitoring scenarios such as urban road networks, oil pipelines, and power transmission lines. It demonstrates strong versatility, scalability, and engineering practicality.

[0132] To verify the effectiveness and robustness of the UAV sensory computing resource and flight control joint optimization method (TAHDA) proposed in this embodiment of the invention, a simulation experimental platform was built based on real geographic information system data, and comparative tests were conducted with various existing technical solutions such as Standard Coverage Path Planning (StdCPP), Fixed Velocity Strategy (FixVel), Myopic Strategy, and MaxPerf Strategy (MaxPerf).

[0133] In terms of experimental setup, a portion of the actual road network in the East Taihu Lake area of ​​Suzhou was selected as the monitoring object. First, the raw map data exported from OpenStreetMap was preprocessed (including densification, simplification, and topology repair), abstracting the complex road network into a weighted undirected graph model. , where nodes Represents a road intersection, edge This represents the linear road segments that need to be covered. It is set within the map boundaries. The locations of several ground users follow a Poisson point process. The main simulation parameters are set as follows: UAV communication system bandwidth... =1MHz, carrier frequency =2GHz, maximum transmit power =2W, minimum perceived quality threshold =10 (m) -1 Total battery capacity J, time weighting coefficient =1.5, energy weighting coefficient =0.01.

[0134] Furthermore, under the aforementioned pre-set experimental configuration conditions, the results obtained in this experiment, specifically the actual effect of the coverage path planning, are explained and analyzed in detail, clarifying the performance of this path planning scheme in terms of coverage efficiency and resource consumption. (Refer to...) Figure 3 As shown in the figure, the drone coverage flight trajectory generated using the method described in this embodiment of the invention is illustrated. The color intensity of the trajectory represents the flight progress. Experimental results show that the path generated by the hybrid metric path planning strategy used in this embodiment of the invention... As a closed Eulerian circuit, this ensures that the UAV can traverse all target road segments and ultimately return safely to the starting point. Compared with the traditional standard CPP algorithm based on ground distance, this invention can intelligently identify odd-degree nodes in the graph and utilize the UAV's aerial maneuverability to establish aerial shortcuts between nodes, thereby avoiding detours limited by ground topology and significantly shortening the total mission range.

[0135] Furthermore, to verify the superiority of the method described in the embodiments of the present invention, it is compared and analyzed with existing technologies, specifically including Alternating Optimization (AO), Standard Coverage Path Planning (StdCPP), Fixed Resource Strategy (FixRes), Fixed Velocity Strategy (FixVel), Myopic Strategy, and MaxPerf Strategy. The performance differences of each strategy are compared and explained from two core dimensions: overall cost and constraint satisfaction capability. (Refer to...) Figure 4 As shown in the figure, the comparison results between the present invention (TAHDA) and various benchmark algorithms in terms of penalty correction cost are illustrated. The correction cost metric introduces a penalty term for energy overconsumption to comprehensively evaluate the economy and feasibility of the algorithm. Specifically, penalty correction cost refers to the introduction of a penalty term for battery energy overconsumption on top of a weighted objective function that includes time and energy consumption, used to comprehensively quantify the task execution cost of the algorithm under the premise of strictly satisfying the energy hard constraint. Furthermore, Figure 4 The "√" mark indicates that the algorithm can meet the energy constraint, otherwise it is marked as "×".

[0136] Experimental data show that among all testing methods, only the present invention (TAHDA), alternating optimization (AO), and fixed-rate strategy (FixVel) can strictly satisfy the global battery energy hard constraint (i.e., Other benchmark algorithms (such as Myopic and MaxPerf) suffer from severe energy waste due to their aggressive strategies, making them unsuitable for practical, physically constrained systems. While ensuring feasibility, this invention achieves a more effective balance between mission time and energy consumption compared to conservative fixed-speed strategies through dynamic joint optimization of flight speed, sensing power, and communication resources. Experimental results show that the correction cost of this invention is reduced by up to 49.72% compared to benchmark methods, demonstrating that this method can find the globally optimal control strategy within a strictly physically constrained feasible region.

[0137] This invention proposes a joint optimization method for UAV flight control and sensory computing resources for coverage path planning. By performing unified mathematical modeling and collaborative optimization design of UAV flight control parameters and three types of resources—sensing, communication, and computing—it effectively achieves a dynamic balance between coverage quality, resource utilization efficiency, and UAV energy consumption. Compared to existing technologies that separate path planning and resource allocation, this invention significantly reduces UAV onboard battery energy consumption while ensuring coverage integrity in the target area, and simultaneously improves the overall performance of sensory computing tasks, overcoming the technical deficiency of existing methods in simultaneously addressing multiple objectives.

[0138] Furthermore, this invention employs a hierarchical alternating optimization solution method, which not only ensures the stability and convergence of the computation process but also reduces the complexity of algorithm implementation. It is adaptable to the hardware configurations of existing UAV platforms and easy to deploy in engineering. This method is particularly suitable for UAV-based coverage and inspection applications with limited computing power and energy supply, demonstrating good engineering feasibility, practicality, and broad application and promotion value.

[0139] Example 2:

[0140] Based on the same inventive concept, this embodiment provides a UAV synesthetic computing resource allocation system for covering path tasks. The principle of solving the problem is similar to that of the UAV synesthetic computing resource allocation method for covering path tasks provided in Embodiment 1, and the repeated parts will not be described again.

[0141] This embodiment provides a UAV sensory computing resource allocation system for covering path tasks, used to implement the UAV sensory computing resource allocation method for covering path tasks described in Embodiment 1, including:

[0142] The deployment module is used to construct a topology model of the target area based on the geographic information of the target area, and to deploy ground user terminals within the geographic area covered by the topology model.

[0143] The model building module is used to build a perception quality model for the UAV, a communication rate model between the UAV and the ground user terminal, a calculation model for the UAV data generation rate, and an energy consumption model for the total instantaneous power of the UAV.

[0144] The optimization decision module is used to construct a weighted cost function based on the energy consumption model; use the weighted cost function as the optimization objective; determine the decision variables to be optimized based on the optimization objective; and determine multiple constraints based on the decision variables to be optimized, the perceived quality model, the communication rate model, and the computation model.

[0145] The path planning module is used to obtain all nodes with odd degrees based on the topology model and construct a first set; construct a mixed-weight complete graph based on the first set; calculate the weights of the edges in the mixed-weight complete graph; augment the undirected graph of the topology model based on the mixed-weight complete graph and the weights to obtain an augmented undirected graph; and obtain the optimal total path length based on the augmented undirected graph.

[0146] The sensory computing resource allocation module is used to obtain the optimal communication power closed-form solution, the optimal sensing power closed-form solution, and the optimal computing frequency closed-form solution based on multiple constraints; obtain the instantaneous minimum power function based on the optimal communication power closed-form solution, the optimal sensing power closed-form solution, the optimal computing frequency closed-form solution, and the energy consumption model; calculate the optimal multiplier based on the instantaneous minimum power function and the optimal total path length; calculate the optimal speed of the UAV based on the optimal multiplier; and generate the optimal sensing power, optimal communication power, and optimal computing frequency based on the optimal speed, thus completing the sensory computing resource allocation for the UAV.

[0147] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0148] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0149] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0150] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0151] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for allocating sensory computing resources for UAVs in a path-covering task, characterized in that, include: S1. Based on the geographic information of the target area, construct a topology model of the target area, and deploy ground user terminals within the geographic area covered by the topology model; S2. Construct a perception quality model for the UAV, a communication rate model between the UAV and the ground user terminal, a calculation model for the UAV data generation rate, and an energy consumption model for the total instantaneous power of the UAV. S3. Construct a weighted cost function based on the energy consumption model; use the weighted cost function as the optimization objective; determine the decision variables to be optimized based on the optimization objective; Based on the decision variables to be optimized, the perceived quality model, the communication rate model, and the computational model, multiple constraints are determined. S4. Based on the topological model, obtain all nodes with odd degrees and construct a first set; based on the first set, construct a mixed-weight complete graph; calculate the weights of the edges in the mixed-weight complete graph; based on the mixed-weight complete graph and the weights, augment the undirected graph of the topological model to obtain an augmented undirected graph; based on the augmented undirected graph, obtain the optimal total path length. S5. Based on the multiple constraints, obtain the optimal communication power closed-form solution, the optimal sensing power closed-form solution, and the optimal computation frequency closed-form solution; based on the optimal communication power closed-form solution, the optimal sensing power closed-form solution, the optimal computation frequency closed-form solution, and the energy consumption model, obtain the instantaneous minimum power function; based on the instantaneous minimum power function and the optimal total path length, calculate the optimal multiplier; based on the optimal multiplier, calculate the optimal speed of the UAV; based on the optimal speed, generate the optimal sensing power, the optimal communication power, and the optimal computation frequency, and complete the allocation of UAV communication and computing resources. The steps for calculating the optimal speed of the drone are as follows: Construct a local Hamiltonian based on the instantaneous minimum power function; Based on the local Hamiltonian, the total length of the optimal path, and the total capacity of the UAV's onboard battery, construct a dual expression; The dual expression is iteratively searched using a bisection method. In each iteration, the instantaneous flight velocity at each path position is calculated until the global energy constraint is satisfied, and the optimal multiplier is obtained. The optimal speed of the UAV is calculated based on the optimal multiplier.

2. The UAV sensory computing resource allocation method for covering path tasks according to claim 1, characterized in that, In step S5, the expressions for the optimal communication power closed-form solution, the optimal sensing power closed-form solution, and the optimal calculation frequency closed-form solution are respectively: ； ； ； ； in, This represents the closed-form solution for optimal communication power. This represents the channel gain for the bottleneck user within the dynamic coverage radius of the drone. Indicates the minimum communication rate threshold. Indicates the drone's location on the path. The user set within the coverage area Represents the threshold constant. Represents the transmission efficiency factor. This represents the energy efficiency constant of the UAV remote sensing camera system. Indicates bandwidth. Indicates noise power. This represents the number of CPU cycles required to compute a unit of data. Indicates instantaneous flight speed. Indicates the minimum perceived quality threshold. Indicates the speed at which remote sensing data is generated. Indicates pixel size, Indicates the principal distance. Indicates relative flight altitude. This indicates the width of the fixed ground strip covered by the camera sensor. This indicates the color bit depth of the image.

3. The UAV sensory computing resource allocation method for covering path tasks according to claim 1, characterized in that, In S2, the expression for the perceived quality model is: ； in, Indicates instantaneous perceived quality. This represents the energy efficiency constant of the UAV remote sensing camera system. Indicates relative flight altitude. Indicates the principal distance. Indicates pixel size, Indicates instantaneous sensed power. Indicates instantaneous flight speed. Indicates the path location label.

4. The UAV sensory computing resource allocation method for covering path tasks according to claim 1, characterized in that, In step S2, the expression for the communication rate model between the UAV and the ground user terminal is: ； in, Represents the transmission efficiency factor. Indicates the drone's location on the path. Place and No. Instantaneous communication rate of a ground user terminal Indicates bandwidth. This represents the power of Gaussian white noise. Represents the Doppler coefficient. Indicates instantaneous flight speed. Indicates the transmission power. This represents the large-scale channel gain.

5. The UAV sensory computing resource allocation method for covering path tasks according to claim 1, characterized in that, In S2, the expression for the calculation model of the UAV data generation rate is: ； in, Indicates the drone data generation rate. This indicates the width of the fixed ground strip covered by the camera sensor. Indicates the color bit depth of the image. Indicates instantaneous flight speed. It represents the quality perceived instantaneously.

6. The UAV sensory computing resource allocation method for covering path tasks according to claim 1, characterized in that, In S2, the energy consumption model for the total instantaneous power of the UAV is expressed as follows: ； in, Indicates the drone's location on the path. Total instantaneous power at the location, Indicates propulsion power. Indicates instantaneous flight speed. Indicates the sensed power. Indicates the transmission power. Indicates the chip's energy consumption coefficient. This indicates the frequency of calculation.

7. The UAV sensory computing resource allocation method for covering path tasks according to claim 1, characterized in that, In step S3, based on the energy consumption model, the expression for the weighted cost function is constructed as follows: ； in, This represents the weighted cost function. Indicates the total path length. This represents the time weighting coefficient. This represents the energy weighting coefficient. Indicates the drone's location on the path. Total instantaneous power at the location, Indicates instantaneous flight speed.

8. The UAV sensory computing resource allocation method for covering path tasks according to claim 1, characterized in that, The constraints include perception quality constraints, communication rate constraints, computing power constraints, physical resource constraints, and global energy hard constraints.

9. A UAV sensory computing resource allocation system for covering path tasks, used to implement the UAV sensory computing resource allocation method for covering path tasks as described in any one of claims 1 to 8, characterized in that, include: The deployment module is used to construct a topology model of the target area based on the geographic information of the target area, and to deploy ground user terminals within the geographic area covered by the topology model. The model building module is used to build a perception quality model of the UAV, a communication rate model between the UAV and the ground user terminal, a calculation model of the UAV data generation rate, and an energy consumption model of the total instantaneous power of the UAV. The optimization decision module is used to construct a weighted cost function based on the energy consumption model; use the weighted cost function as the optimization objective; and determine the decision variables to be optimized based on the optimization objective. Based on the decision variables to be optimized, the perceived quality model, the communication rate model, and the computational model, multiple constraints are determined. The path planning module is used to: obtain all nodes with odd degrees according to the topology model and construct a first set; construct a mixed-weight complete graph according to the first set; calculate the weights of the edges in the mixed-weight complete graph; augment the undirected graph of the topology model according to the mixed-weight complete graph and the weights to obtain an augmented undirected graph; and obtain the optimal total path length according to the augmented undirected graph. The sensory computing resource allocation module is used to obtain the optimal communication power closed-form solution, the optimal sensing power closed-form solution, and the optimal computing frequency closed-form solution based on the multiple constraints; to obtain the instantaneous minimum power function based on the optimal communication power closed-form solution, the optimal sensing power closed-form solution, the optimal computing frequency closed-form solution, and the energy consumption model; to calculate the optimal multiplier based on the instantaneous minimum power function and the optimal total path length; to calculate the optimal speed of the UAV based on the optimal multiplier; and to generate the optimal sensing power, optimal communication power, and optimal computing frequency based on the optimal speed, thereby completing the sensory computing resource allocation for the UAV.

Images