Data collection and unloading method for distributed nodes in unmanned aerial vehicle assisted internet of things

Abstract

The invention provides a data collection and unloading method for distributed nodes in an unmanned aerial vehicle assisted internet of things, belongs to the technical field of unmanned aerial vehicle assisted mobile communication networks, and aims at an unmanned aerial vehicle assisted MEC system to realize energy efficiency maximization when solving joint data collection and calculation unloading tasks of large-scale internet of things nodes. According to the method, the problem is resolved into a nested mixed integer optimization problem, wherein an external problem is that a specific number of key INs are selected from potential INs for data transmission. The internal problem maximizes the energy efficiency of the unmanned aerial vehicle by optimizing the scheduling of the unmanned aerial vehicle, the three-dimensional flight trajectory, and the allocation of computing resources in order to perform data collection and compute offloading from the selected IN. An external problem is solved by converting selection of INs into a matrix eigenvalue optimization problem by using a method based on Gerre circle alignment, the external problem is divided into three sub-problems, and the model is convexly solved by using equivalent transformation, convex difference and continuous convex approximation.

Description

Technical Field

[0001] The present invention belongs to the technical field of drone-assisted mobile communication networks, and in particular relates to a method for collecting and unloading data from distributed nodes in a drone-assisted Internet of Things. Background Art

[0002] With the development of IoT technology, IoT terminals represented by wireless sensors are widely used in scenarios such as intelligent transportation, forest monitoring, and traffic safety. These sensor terminals do not require wired power connections and can be deployed more flexibly, thereby reducing installation and maintenance costs. In actual operation, these Internet of Things nodes (Ins) generate a large amount of various data, which need to be further transmitted to a central or distributed base station (BS) for further processing or analysis. However, in most cases, wireless connections make INs energy limited. The wide distribution and limited transmission power of INs make long-distance data communications unreliable.

[0003] At the same time, sensor nodes in the IoT usually have limited computing power and are unable to handle complex computationally intensive tasks independently. When processing latency-sensitive tasks, transmitting data to a centralized or distributed BS for further processing or analysis may result in high network latency, especially when the data transmission path is long or the network is congested. To meet the demand for high-performance computing, which far exceeds the inherent capabilities of IN, mobile edge computing (MEC) technology has been widely used in IoT applications. MEC allows IN to offload computing tasks to nearby edge servers (e.g., BS, access points) to achieve fast and energy-efficient task processing. However, in most cases, it is difficult and expensive to build a ground MEC infrastructure to serve a large number of INs distributed over a large geographical area.

[0004] The next generation of mobile networks, such as the fifth generation (5G), beyond 5G (B5G) and even the sixth generation (6G), have attracted more and more research efforts in recent years. Among them, unmanned aerial vehicle (UAV)-assisted mobile communication networks have attracted much attention recently. The advantages of UAVs, such as high mobility, low cost, easy deployment and line-of-sight link (LoS), enable them to be applied to different scenarios, thereby improving the performance of wireless networks. In addition, UAVs can not only collect data, but also act as edge computing nodes. During the flight, the data can be preprocessed or compressed on the UAV, thereby reducing the computational burden of IoT sensor nodes and cloud computing. The present invention comprehensively considers the energy efficiency, communication throughput and data offloading requirements of UAVs, and proposes a UAV three-dimensional trajectory and resource allocation strategy based on Gershgorin disk alignment and multivariable alternating iterative optimization to maximize the overall energy efficiency of UAV-assisted IoT systems. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a method for data collection and unloading of distributed nodes in a drone-assisted Internet of Things. For a scenario involving a single drone and multiple Internet of Things nodes, a system model is established to optimize energy efficiency and maximize the overall performance of the drone-assisted Internet of Things system.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A method for collecting and unloading data from distributed nodes in a drone-assisted Internet of Things, the method comprising the following steps:

[0008] Step 1: Design a drone-assisted mobile edge computing system, in which the drone needs to select a suitable ground IoT node from many distributed wireless ground IoT nodes to complete data collection and computation offloading tasks while ensuring data integrity. The system includes a drone 3D trajectory model, a node signal model, a fusion center signal reconstruction model, a probabilistic line-of-sight channel model, and a drone energy consumption and computation model.

[0009] Step 2: Propose a nested optimization problem; the inner problem selects a specific number of IoT nodes from a set of potential IoT nodes so that the data transmitted by the IoT nodes can reconstruct the global data with a certain accuracy, and the outer problem performs data collection and computation offloading from the selected IoT nodes by optimizing the scheduling, 3D flight trajectory, and computational resource allocation of the UAV to maximize the energy efficiency of the UAV.

[0010] Step 3: To solve the nested optimization problem P1, a two-stage algorithm is used to solve the problem.

[0011] The UAV-assisted mobile edge computing system described in step 1 consists of a UAV and IoT nodes, each of which can observe unknown parameters The observation results are sent to the fusion center, which estimates the parameters based on the received information. ,UAVs act as aerial collectors and mobile edge computing terminals,collecting data from ground IoT nodes and bringing the data back to the fusion center;

[0012] The collaborative relationship between the IoT nodes is represented by a connected undirected graph To represent; the undirected graph By triple Indicates that Represents an undirected graph In A collection of IoT nodes, Represent undirected graphs The nodes in Represents an undirected graph middle The collection of edges, is the edge weight A collection of Reflects two connected nodes and Correlation or similarity between A node in the IoT The coordinates of Indicates the height of the IoT node in the coordinates Negligible, approximately zero;

[0013] The UAV 3D trajectory model is as follows:

[0014] Assume that the starting and ending positions of the drone are predetermined, and the coordinates are expressed as , the real-time 3D trajectory of the UAV express;

[0015] Drone flight time Divided into Equal time slots , expressed as ; Each time slot are chosen to be short enough to assume that the drone is in the time slot The interior is approximately stationary; therefore, the trajectory of the drone Approximately a sequence of discrete points ; Due to limited energy, the flight time and maximum speed of the drone are and ; The maximum and minimum flight altitudes are and ,

[0016] Due to the maximum speed limit of the drone, the position between two time slots needs to meet the following constraints:

[0017] (1)

[0018] In formula (1), is the vertical ground trajectory of the UAV in the mth time slot;

[0019] (2)

[0020] In formula (2), D represents the displacement of the UAV, and:

[0021] (3)

[0022] In formula (3), is the speed of the drone in the mth time slot, is the acceleration of the UAV in the mth time slot, and:

[0023] (4)

[0024] and are the initial and final velocities of the drone, respectively;

[0025] (5)

[0026] (6)

[0027] The acceleration of unmanned aerial vehicles shall not exceed ;

[0028] (7)

[0029] Limit the speed of the drone to be greater than To keep flying;

[0030] (8)

[0031] Since the rotor UAV needs to maintain balance during flight, the limit of the UAV pitch angle is given by the following formula:

[0032] (9)

[0033] in, is the vertical ground speed of the UAV in the mth time slot, is a constant;

[0034] Considering the UAV using TDMA in the present invention, define the Boolean variable Describes the scheduling of all IoT nodes, where Indicates IoT nodes can be The transmission power required to meet the outage probability requirement is and transfer rate Transmit data to the drone, Indicates IoT nodes in the Time slots remain in sleep state;

[0035] Scheduling variables Described as:

[0036] and ;

[0037] The node signal model is as follows:

[0038] Undirected Graph IoT Node The signal value of the graph domain at , so the complete graph domain signal is expressed in the form of a vector, that is: ; Each IoT node Observation Noise observation is performed at all times, so the IoT node The real-valued observation Expressed as

[0039] ;

[0040] in Represents an IoT node The noise in the real-valued observations.

[0041] If the signal of an IoT node is completely collected by the drone, we call the signal of the IoT node an effective observation signal. The fusion center can use the signal of the effective observation IoT node to reconstruct the complete signal. Then, all the observation signals that the drone can collect in one flight can be expressed as

[0042] ;

[0043] in represents the additive noise vector, Indicates the valid node data selectively collected by the drone during a flight; is a K×N matrix, Each row of represents a sampling vector with a modulus of 1, and each column corresponds to an IoT node. When a column is set to 1, the drone will completely collect data from the IoT node corresponding to the column. K is the number of nodes planned to be collected; Indicates time slot The set of nodes whose data has been completely collected;

[0044] Assume that the time series data collected by the IoT nodes are stored in their storage areas respectively; during a flight of the drone, the IoT nodes The amount of data that needs to be transferred is .

[0045] For a given undirected graph associated with IoT nodes , and its corresponding graph Laplacian matrix L can be expressed as

[0046] ;

[0047] in is a diagonal matrix, 1 is the vector of all 1s, Represents a diagonal matrix whose input vector elements are located on the main diagonal;

[0048] Signal Vector Smoothness The smoothness quantification model of the following formula is constructed on a given undirected weighted graph to quantify:

[0049] ,

[0050] The smaller the smoothness of a signal, the smoother the signal;

[0051] The fusion center signal reconstruction model is as follows:

[0052] Given a noisy graph signal , the goal of the fusion center is to restore the smooth graph signal; assuming that the fusion center adopts a biased graph signal reconstruction scheme based on graph Laplace regularization, through unconstrained Norm minimization realizes biased signal reconstruction. Specifically, graph Laplace regularization can be expressed as the optimization of the target signal, and its formula is as follows:

[0053] ;

[0054] in is the balanced graph Laplace regularization with Tradeoff parameter for normal data fidelity term; optimal solution for quadratic objective function It can be obtained by solving the following linear equation:

[0055] ;

[0056] For the The L constructed on , therefore, the above linear equation has a unique solution:

[0057] ;

[0058] Original signal With the reconstruction signal Minimizing the mean square error between

[0059] ;

[0060] let , and is a symmetric positive definite matrix, so given , and ,minimize The mean square error can be minimized;

[0061] The probabilistic line-of-sight channel model is as follows:

[0062] A quasi-static block fading channel model is used for the IoT node-UAV link, with The channel in each fading block remains constant and may change from block to block. The wireless communication link between The downlink channel power gain in is denoted as , and expressed as

[0063] ;

[0064] in, represents the large-scale attenuation coefficient, represents the small-scale Ricean attenuation coefficient, , represents the mean; and The calculation formula is

[0065] ;

[0066] ;

[0067] in, For IoT nodes The coordinates of is the path loss factor, The value is 2. It represents the power gain when the reference distance is 1m. represents the small-scale Ricean attenuation coefficient, represents the LoS channel part, , represents the random scattering part, ;

[0068] IoT Node In time slot The uplink transmission power for transmitting data to the UAV is , so the time slot No. The achievable transmission rate of attenuation blocks is calculated as follows:

[0069] ;

[0070] in, is the difference in signal-to-noise ratio between the actual modulation scheme and the theoretical Gaussian signal, is the channel bandwidth, represents the additive white Gaussian noise at the receiver. When the IoT node is awakened The probability of interruption with the drone is given by:

[0071] ;

[0072] in, represents the probability density function, It is a time slot hour The cumulative distribution function of is a non-decreasing function; when the maximum permissible interruption probability is hour, Satisfy the following formula

[0073] ;

[0074] in yes The inverse function of .

[0075] Therefore, distributed sensors are Total data transferred within It can be defined as

[0076] ;

[0077] The energy consumption and calculation model of the UAV are as follows:

[0078] The energy consumption of unmanned aerial vehicles generally consists of three main parts: communication-related energy consumption, propulsion energy consumption, and computing energy consumption. Among them, the communication-related energy consumption is much smaller than the propulsion energy consumption, so the communication-related energy consumption can be ignored.

[0079] For a rotorcraft UAV, its propulsion energy consumption for 3D flight can be expressed as

[0080] ;

[0081] in, and are two constants related to the wingspan efficiency, wing surface area, and mass of the drone; is the acceleration due to gravity, is the acceleration in the direction perpendicular to the ground, is the velocity in the x-axis direction of the Cartesian coordinate system, The velocity in the y-axis direction of the Cartesian coordinate system;

[0082] The computational energy of the drone in each time slot is as follows:

[0083] ;

[0084] in, is a constant representing the effective switched capacitance, Represents the drone in the time slot Assigned to IoT nodes computing resources;

[0085] In summary, the total energy consumption of the drone can be obtained as shown in the following formula:

[0086] ;

[0087] When the initial and final positions and speeds of the UAV are determined, the third and fourth terms in the above formula are constants, so they do not need to be considered in the optimization process;

[0088] In addition, considering the limited computing resources of the UAV, the maximum allocatable computing resources of the UAV are restricted in each time slot:

[0089] ;

[0090] The processing of all collected data should be completed within one flight cycle, which results in the following limitations:

[0091] ;

[0092] in, represents the number of CPU cycles required to calculate each data bit. This constraint ensures that the UAV The data collected in and after the time slot can be Completely unload after time slots;

[0093] The nested optimization problem proposed in step 2 is as follows:

[0094] By optimizing the trajectory , scheduling vector , calculate resource allocation vector , and a collection of IoT nodes that collect data , the energy efficiency maximization optimization problem P1 can be expressed as

[0095] ;

[0096] The optimization problem P1 is composed of an external problem and an internal problem. In the external problem, the goal is to find an optimal set of access IoT nodes so that the reconstruction error Satisfy a certain accuracy; at the same time, the goal of the internal problem is to determine the scheduling vector of the node , optimal trajectory , and the computational resource allocation given the optimal set of IoT nodes to access .

[0097] In step 3, a two-stage algorithm is used to solve the nested optimization problem P1; the specific method is as follows:

[0098] In the first stage, the selection of IoT nodes is transformed into a matrix eigenvalue optimization problem using a method based on Gale circle alignment to solve the internal problem in the optimization problem P1; the specific method is as follows:

[0099] The problem to be solved in the first stage is shown in question P1.1:

[0100]

[0101] because is a symmetric positive definite matrix that satisfies ; therefore, maximizing can minimize the upper bound of the mean square error; therefore, Problem P1.1 can be transformed into Problem P1.1a to maximize the matrix The minimum eigenvalue of is in the form of:

[0102]

[0103] The present invention converts the above matrix eigenvalue optimization problem into a scaling and translation operation of a Gale circle, and uses a method based on Gale circle alignment to solve problem P1.1a. The specific steps are as follows:

[0104] 1) Input IoT node undirected graph , reconstruction error and IN number budget K;

[0105] 2) Initialize the IoT node collection , N is the number of IoT nodes in the set;

[0106] 3) From node n=1 to node n=N, use the method based on the Gale circle theorem to estimate the covering subset of each Gale circle , and the resulting subset Merge to get a collection ;

[0107] 4) At the same time, if n≤K, select and set The subset with the largest intersection , subset Integration with IoT nodes After taking the intersection, find the set and the IoT node set The complement of the node set Reassign the value; then merge the current node n into the selected IoT node set , return to step 3) and take the next value of n;

[0108] 5) If n>K, then output the currently selected IoT node set , end the solution;

[0109] For different reconstruction errors Perform iterative solution to obtain the set of IoT nodes IoT node collection ;

[0110] In the second stage, the external problem is divided into three sub-problems and solved by convexifying the model using equivalent transformation, convex difference and continuous convex approximation; the specific method is as follows:

[0111] In the second stage of the problem, the collection and unloading of data sent by the selected IoT nodes of the external problem is completed by reasonably planning the flight path of the drone and allocating computing resources. The problems that need to be solved in the second stage are shown in Problem P1.2:

[0112]

[0113] Problem P1.2 is divided into three sub-problems, as follows:

[0114] a. IN node access sub-problem

[0115] Given a UAV flight trajectory and computing resource allocation In this case, the denominator of problem P1.2 is determined, and the sub-problem becomes the design of the access time slots of the drone data collection nodes, with the goal of maximizing the total amount of data collected.

[0116] To solve this integer programming problem, we need to set the binary constraint Relax to .

[0117] The access time slot problem of the UAV data collection node is designed as follows:

[0118]

[0119]

[0120]

[0121]

[0122]

[0123] ;

[0124] b. UAV 3D trajectory optimization sub-problem

[0125] According to the given set of nodes to be visited , scheduling vector and resource allocation vector Optimizing drone trajectories and the associated speed and acceleration , the optimization problem P1.2 can be restated as:

[0126]

[0127] in and the numerator of the objective function Relative to All are non-concave;

[0128] Relative to is also non-convex, the denominator of the objective function is relative to and is non-convex.

[0129] are also non-convex; therefore, they need to be transformed into mathematically solvable problems;

[0130] (b1) The numerator of the objective function is relative to Convexification

[0131] because may be non-negative, so a slack variable is introduced , thus obtaining

[0132] ;

[0133] ;

[0134] ;

[0135] so, is converted to

[0136] .

[0137] (b2) The numerator of the objective function is relative to Convexification

[0138] In is a non-convex function, which is transformed into a convex function using a first-order Taylor expansion and applying a continuous convex approximation;

[0139] ;

[0140] Introducing auxiliary variables and This is to handle the non-convex denominator . In this way, the above formula becomes

[0141]

[0142] and The right side of the non-concave terms is bounded by the local point Performing a first-order Taylor expansion on the right side yields:

[0143] ;

[0144] ;

[0145] Through the above transformation, the formula The terms in can be transformed into a standard form consistent with convex optimization:

[0146] ;

[0147] (b3) Convexification

[0148] The Taylor expansion on the right also yields ;

[0149] and convexification of the numerator of the objective function

[0150] let represents the given trajectory in the rth iteration; by applying the first-order Taylor expansion, The lower bound of

[0151] ,in

[0152]

[0153] ;

[0154] .

[0155] Through the above operation, the numerator of the objective function becomes the following concave form

[0156] ;

[0157] Similarly, replace for

[0158] .

[0159] (b4) Convexification

[0160] At local point Department, about The first-order Taylor expansion of

[0161] ;

[0162] in and ;

[0163] ;

[0164] Will Substituting into

[0165] .

[0166] Through the above transformation, the optimization of the problem P1.2.2 related to the UAV trajectory can be transformed into the following form:

[0167]

[0168] c. Computational resource allocation optimization sub-problem

[0169] The computational resource offloading problem is formulated as maximizing the lower bound of the computational offloading data volume, i.e., Problem P1.2.3:

[0170]

[0171] d. A multivariable fixed alternating iteration method is used to optimize the second stage problem. The specific steps of this method are as follows:

[0172] d1) Initialize the number of iterations i and the initial trajectory , Initial resource allocation , initial speed , initial acceleration , intermediate variables and ;

[0173] d2) Using the given and Solve problem (P1.2.1) and get the optimal solution for the scheduling vector ;

[0174] d3) Using the given , , , , , and , solve problem P1.2.2a and obtain the optimal trajectory ,speed , acceleration , auxiliary variables and , intermediate variables and ;

[0175] d4) Based on the optimal trajectory obtained in step d3) and the scheduling vector obtained in step d2) Solve problem P1.2.3 and obtain the optimal resource allocation ;

[0176] d5) If the multivariable fixed alternating iterative algorithm converges, exit the iteration; otherwise, update the number of iterations i = i + 1 and use the trajectory , Resource Allocation ,speed , acceleration , intermediate variables and Bring it to step d1) for the next iteration.

[0177] The beneficial effects of the present invention are:

[0178] This paper designs a 3D trajectory optimization and resource allocation strategy based on drone assistance, and constructs a drone-assisted Internet of Things (IoT) data collection and computational offloading model. The key nodes are selected by the Gershgorin disk alignment algorithm to reduce redundant data transmission and reduce the total energy consumption of the system. The drone flies under the optimized 3D trajectory and collects and processes the data of the key nodes according to the allocated computing resources. After the task is completed, the drone will return to the ground base station or fusion center (FC). With the goal of maximizing the energy efficiency of the system, this paper proposes a joint algorithm for node selection and three-dimensional trajectory optimization, and at the same time improves the communication throughput and computational offloading performance of the drone through a multi-variable alternating iterative optimization method. Simulation results show that compared with other benchmark schemes, the algorithm based on joint 3D trajectory optimization and resource allocation proposed by us significantly improves the performance of data collection and computational offloading, and verifies the effectiveness of drone-related optimization and selective key node data collection in the IoT network. This shows that by optimizing the flight trajectory and computing resource allocation strategy of the drone, the data transmission efficiency and energy utilization rate of the drone-assisted IoT system can be greatly improved, which further proves the necessity and importance of this method in complex IoT environments.

[0179] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0180] Figure 1 Schematic diagram of the system scenario; in the figure: Computational Offloading, 3D Trajectory, Data Collection, Correlation between INs, The selected IN;

[0181] Figure 2 Random IN graph and graph signal on IN; (a) distribution of 50 random INs, (b) selected INs;

[0182] Figure 3 The influence of different flight durations T on the trajectory of the UAV;

[0183] Figure 4 The influence of different flight time T on the flight altitude of the UAV;

[0184] Figure 5The influence of different flight time T on the speed of UAV;

[0185] Figure 6 Node access time slot allocation under different data collection limits (a) T=100s, S k =9Mbit, (b) T=100s, S k =14Mbit.

[0186] Figure 7 Calculate the resource offloading allocation results for the UAV auxiliary system under different flight durations T (a) T=105s, (b) T=115s, (c) T=120s;

[0187] Figure 8 For different data collection lower limit S k Under the same conditions, the EE of the proposed method is compared with three baselines of unmanned aerial vehicles; in the figure: Energy Efficiency for the proposed scheme, Energy Efficiency for 2D scheme, Equal computational resource allocation, Fixed trajectory method.

[0188] Fig. 9 Collect lower limits S for different data k Comparison of the flight altitudes of unmanned aerial vehicles at different times.

[0189] Fig.10 The comparison between the proposed method and three baseline EEs under different flight durations T of UAV is shown in the figure. Figure 8 .

[0190] Fig.11 For different maximum computing resource constraints The EE of the proposed method is compared with three baselines of UAV under the same conditions. Figure 8 . DETAILED DESCRIPTION

[0191] The following will describe the technical solutions in the embodiments of the present invention in detail in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention.

[0192] In this embodiment, a method for collecting and unloading data from distributed nodes in a drone-assisted Internet of Things includes the following steps:

[0193] Step 1: Design a UAV-assisted MEC system, in which the UAV needs to select a suitable ground IN from many distributed wireless ground INs to complete the data collection and computation offloading tasks, including the UAV 3D trajectory model, node signal model, fusion center (FC) signal reconstruction model, probabilistic line-of-sight channel model, UAV energy consumption and computation model;

[0194] The drone-assisted MEC system consists of a drone and IN, each IN can observe unknown parameters The observations are sent to the Fusion Center (FC), which estimates the parameters based on the received information. ,The UAV acts as an aerial collector and mobile edge computing terminal, ,collecting data from the ground IN and bringing the data back to FC, as shown in Figure 1.

[0195] There is a collaborative relationship between INs, which is connected through an undirected graph To represent. By triple Definition, where Represents an undirected graph In A collection of INs, Represent undirected graphs A node in Representative figure The collection of edges, is the edge weight A collection of Reflects two connected nodes and The correlation or similarity between them. The coordinates of Compared with the height of the drone, the height of the IN It can be ignored and will not affect the conclusion of the present invention.

[0196] 1. UAV 3D trajectory model

[0197] Assume that the starting and ending positions of the drone are predetermined, and the coordinates are expressed as . Real-time 3D trajectory of drones express.

[0198] Among them, due to limited energy, the flight time and maximum speed of the drone are and The maximum and minimum flight altitudes are and ,satisfy (1),

[0199] This is for safety reasons (such as avoiding buildings).

[0200] In the continuous time case, the position variables of the UAV are infinite and difficult to obtain. For the sake of explanation, we will Divide into equal time slots, represented by ,in is the time slot length. Each time slot is chosen to be short enough so that the UAV is assumed to be approximately stationary within the time slot. In view of this, the trajectory of the UAV Can be approximated as a discrete point sequence Due to the maximum speed limit of the drone, the position between two time slots needs to meet the following constraints:

[0201] (2)

[0202] in (3)

[0203] (4)

[0204] and are the initial and final velocities of the drone, respectively;

[0205] (5)

[0206] (6) The acceleration of unmanned aerial vehicles shall not exceed ;

[0207] (7) The speed of the unmanned aerial vehicle must be greater than To keep flying;

[0208] (8)

[0209] Since the rotor UAV needs to maintain balance during flight, the limit of the UAV pitch angle is given by the following formula:

[0210] (9) Among them is a constant;

[0211] Consider a drone using time division multiple access and define a Boolean variable Describes the schedule of all INs, where Indicates The IN can be The transmission power required to meet the outage probability requirement is and transfer rate Transmit data to the drone, Indicates IN in the The scheduling variable Can be described as

[0212] and .

[0213] 2. Node signal model

[0214] The present invention is used Representation diagram IN The complete graph domain signal can be written in the form of a vector, namely: Each IN A deterministic parameter (e.g. temperature) to make noise observations. The real-valued observation Modeled as

[0215] ;

[0216] in Indicates IN The noise in the real-valued observations.

[0217] If the signal of an IN is completely collected by the drone, we call the signal of the IN an effective observation signal, and FC can reconstruct the complete signal using the signal of the effective observation IN. Then, all the observation signals that the drone can collect in one flight can be expressed as

[0218] ;

[0219] in It is a K The matrix of represents the additive noise vector, Indicates the valid node data selectively collected by the drone during a flight. Each row of represents a sampling vector with a modulus of 1, and each column corresponds to an IN. When a column is set to 1, the drone will completely collect the data of the IN corresponding to the column, and K is the number of nodes planned to be collected. We use To indicate the time slot The set of nodes whose data has been completely collected.

[0220] Assume that the time series data collected by IN are stored in their storage areas respectively. During a flight of the drone, IN The amount of data that needs to be transferred is .

[0221] For a given graph associated with IN , and its corresponding graph Laplacian matrix It can be expressed as

[0222] ;

[0223] in is a diagonal matrix. 1 is the vector of all 1s, Represents a diagonal matrix with the input vector elements located on the main diagonal. The signal vector The smoothness of can be quantified by constructing a smoothness quantification model of the following formula on a given undirected weighted graph:

[0224] , the smaller the smoothness of the signal, the smoother the signal.

[0225] 3. FC signal reconstruction model

[0226] Given a noisy graph signal ,The goal of FC is to recover the smooth graph signal.

[0227] Assuming that FC adopts a biased graph signal reconstruction scheme based on graph Laplace regularization (GLR), The norm minimization realizes the reconstruction of biased signals. Specifically, GLR can be expressed as the optimization of the target signal, and its formula is as follows:

[0228] ;

[0229] in is to balance GLR and The tradeoff parameter for the normal data fidelity term. The optimal solution for the quadratic objective function It can be obtained by solving the following linear equation:

[0230] ;

[0231] For a connected graph Built on , there is a positive . Therefore, the above equation has a unique solution:

[0232] ;

[0233] Original signal With the reconstruction signal Minimizing the MSE between

[0234] ;

[0235] let , and is a symmetric positive definite matrix. Therefore, given , and ,minimize The MSE can be minimized.

[0236] 4. Probabilistic Line-of-Sight Channel Model

[0237] A quasi-static block fading channel model is used for the IN-UAV link, with The channel remains constant within each fading block and may change from block to block. The wireless communication link between The downlink channel power gain in is denoted as , and expressed as

[0238] ;

[0239] in, represents the large-scale attenuation coefficient, represents the small-scale Ricean attenuation coefficient, . and The calculation formula is

[0240] ;

[0241] ;

[0242] in, is the path loss factor, represents the power gain when the reference distance is 1 m. The small-scale Rice attenuation coefficient is In the present invention, The value of is 2. Indicates the LoS channel portion ( ), represents the random scattering part (a zero-mean unit-variance circularly symmetric complex Gaussian random variable connected to the NloS channel part). Therefore, we have .

[0243] IN In time slot The uplink transmission power for transmitting data to the UAV is Therefore, the time slot No. The achievable transmission rate of attenuation blocks is calculated as follows:

[0244] ;

[0245] in, is the difference in signal-to-noise ratio between the actual modulation scheme and the theoretical Gaussian signal, is the channel bandwidth, represents the additive white Gaussian noise at the receiver. When IN is awakened The probability of interruption with the UAV is given by

[0246] ;

[0247] in, It is a time slot hour The cumulative distribution function (CDF) of is a non-decreasing function. When the maximum permissible interruption probability is hour, Satisfy the following formula

[0248] ;

[0249] in yes The inverse function of .

[0250] Therefore, distributed sensors are Total data transferred within It can be defined as

[0251] .

[0252] 5. UAV energy consumption and calculation model

[0253] The energy consumption of unmanned aerial vehicles generally consists of three main parts: communication-related energy consumption, propulsion energy consumption, and computing energy consumption. For a rotary-wing UAV, its 3D flight propulsion energy consumption can be expressed as

[0254] ;

[0255] in, and are two constants related to the wingspan efficiency, wing surface area, and mass of the drone; is the acceleration due to gravity.

[0256] The computational energy of the drone in each time slot is as follows:

[0257] ;

[0258] in, is a constant representing the effective switched capacitance, Represents the drone in the time slot Assigned to IN computing resources.

[0259] IN is limited by maximum energy , which can be determined according to the actual IN type. Considering that the communication-related energy consumption is generally within a few joules, which is much smaller than the propulsion energy consumption, the present invention ignores it here and does not affect the subsequent conclusions. Based on the above description, we can get the total energy consumption of the drone, as shown in the following formula.

[0260] ;

[0261] When the initial and final positions and speed of the UAV are determined, the third and fourth terms in the above formula are constants and can be ignored during the optimization process. Considering the limited computing resources of the UAV, the maximum allocatable computing resources of the UAV are restricted in each time slot:

[0262] ;

[0263] The processing of all collected data should be completed within one flight cycle. This results in the following limitations:

[0264] ;

[0265] in, represents the number of CPU cycles required to calculate each data bit. This constraint ensures that the UAV The data collected in and after the time slot can be After a time slot, it is completely unloaded.

[0266] Step 2: A nested optimization problem is proposed; the inner problem selects a specific number of INs from a set of potential INs so that the data transmitted by the INs can reconstruct the global data with a certain accuracy, and the outer problem performs data collection and computation offloading from the selected INs to maximize the energy efficiency of the UAV by optimizing the scheduling, 3D flight trajectory, and computational resource allocation of the UAV;

[0267] The proposed optimization problem is as follows:

[0268] By optimizing the trajectory , scheduling vector , calculate resource allocation vector , and the IN set of collected data , the energy efficiency (EE) maximization optimization problem can be expressed as

[0269] ;

[0270] question It is a nested optimization problem, which consists of an external problem and an internal problem. In the external problem, the goal is to find an optimal set of visits IN so that the reconstruction error At the same time, the goal of the internal problem is to determine the scheduling vector of the node , optimal trajectory , and the computational resource allocation given the optimal access IN set .

[0271] This nested structure makes It becomes a non-convex problem of hybrid combinatorial optimization, and the optimization variables are highly coupled, which brings great challenges to finding a solution.

[0272] Step 3: To solve the non-convex problem, a two-stage algorithm is proposed to solve the above nested optimization problem;

[0273] In the first stage, the selection of Ins is transformed into a matrix eigenvalue optimization problem using a method based on Gale circle alignment (GDA) to solve the internal problem; the specific method is as follows:

[0274] The first stage mainly solves the external combinatorial optimization problem, namely the node selection subproblem, which aims to select a certain number of ground nodes from the IN set to transmit data to the UAV so that the reconstruction error of FC using these data to reconstruct the global data meets the requirements. shown.

[0275]

[0276] because is a symmetric positive definite matrix that satisfies Therefore, maximizing The upper bound of MSE can be minimized. Therefore, the problem Can be transformed into a problem Maximize Matrix The form of the minimum eigenvalue of .

[0277]

[0278] The present invention solves the problem (P1.1a) by converting the above matrix eigenvalue optimization problem into the scaling and translation operations of the Gershgorin disk using a GDA-based method. The steps of Algorithm 1 are as follows:

[0279] 1) Input IN undirected graph , Error margin and IN number budget K;

[0280] 2) Initialize the IN collection , N is the number of sets IN;

[0281] 3) From node n=1 to node n=N, use the method based on the Gershgorin disk theorem to estimate the covering subset of each Gershgorin disk , and the resulting subset Merge to get a collection ;

[0282] 4) At the same time, if n≤K, select and set The subset with the largest intersection , subset With IN collection After taking the intersection, find the set and the IN set The complement of the node set Reassign; then merge the current node n into the selected IN set , return to step 3) and take the next value of n;

[0283] 5) If n>K, then output the currently selected IN set , the solution is finished.

[0284] Algorithm 1 For different Perform iterative solution and use binary search to obtain the optimal boundary. Iterative solution can obtain the set of IN and the corresponding .

[0285] In the second stage, the external problem is divided into three sub-problems and solved by convexifying the model using equivalent transformation, convex difference and continuous convex approximation (SCA); the specific methods are as follows:

[0286] In the second stage of the problem, the drone aims to solve the internal problem. The external problem is completed by rationally planning the flight path of the drone and allocating computing resources to collect and offload the data sent by the selected IN. The problem in the second stage is as follows shown.

[0287]

[0288] question The second stage is still a mixed integer non-convex problem, and it is generally difficult to get an optimal solution. Therefore, the goal of the solution is to obtain An effective suboptimal solution.

[0289] question It is divided into three sub-problems, as follows:

[0290] a. IN node access sub-problem

[0291] Given a UAV flight trajectory and computing resource allocation In this case, the problem is identified The denominator of , the sub-problem becomes the design of the access time slot of the drone data collection node, with the goal of maximizing the total amount of data collected.

[0292] To solve this integer programming problem, we need to transform the binary constraints Relax to .

[0293] The access time slot problem of the UAV data collection node is designed as follows:

[0294]

[0295]

[0296]

[0297]

[0298]

[0299]

[0300] This problem is a standard linear programming (LP) problem and is easy to solve.

[0301] b. Optimization of UAV 3D trajectory

[0302] We use the given set of nodes to be visited , scheduling vector and resource allocation vector Optimizing drone trajectories and the associated speed and acceleration The optimization problem can be reformulated as

[0303]

[0304] in and the numerator of the objective function Relative to All are non-concave.

[0305] Relative to is also non-convex, the denominator of the objective function is relative to and is non-convex.

[0306] are also non-convex. Therefore, we need to transform them into mathematically solvable problems.

[0307] (b1) The numerator of the objective function is relative to Convexification

[0308] because may be non-negative, so a slack variable is introduced , thus obtaining

[0309] ;

[0310] ;

[0311] ;

[0312] so, can be converted to

[0313] .

[0314] (b2) The numerator of the objective function is relative to Convexification

[0315] In is a non-convex function. Use first-order Taylor expansion to convert it into a convex function and apply SCA.

[0316] ;

[0317] Introducing auxiliary variables and This is to handle the non-convex denominator . In this way, the above formula becomes

[0318]

[0319] and The right side of the non-concave terms is bounded by the local point Performing a first-order Taylor expansion on the right side yields:

[0320] ;

[0321] ;

[0322] Through the above transformation, the formula The terms in can be transformed into a standard form consistent with convex optimization:

[0323] ;

[0324] (b3) Convexification

[0325] The Taylor expansion on the right also yields .

[0326] and convexification of the numerator of the objective function

[0327] let represents the given trajectory in the rth iteration. By applying the first-order Taylor expansion, The lower bound of

[0328] ,in

[0329]

[0330] ;

[0331] .

[0332] Through the above operations, the numerator of the objective function can be transformed into the following concave form

[0333] ;

[0334] Similarly, we can replace for

[0335] .

[0336] (b4) Convexification

[0337] At local point Department, about The first-order Taylor expansion of

[0338] ;

[0339] in and .

[0340] ;

[0341] Will Substituting into

[0342] .

[0343] Through the above transformation, the problems related to the trajectory of the drone The optimization can be transformed into the following problem

[0344]

[0345] c. Optimization of computing resource allocation

[0346] Through the above analysis and optimization, we can get the , , and the scheduling vector The propulsion energy consumption and the total data collected during the flight of the UAV can be determined. The computational energy consumption is much smaller than the propulsion energy consumption and the amount of data collected, and has little impact on energy efficiency, so we ignore the computational energy consumption. In the computational resource allocation task, the computational resource offloading capacity of the UAV is maximized so that the UAV can collect more data, thereby improving EE.

[0347] The computational resource offloading problem is formulated as maximizing the lower bound of the computational offloading data volume, that is, .

[0348]

[0349] The first two constraints in the problem seek to maximize the lower bound of the offloaded data, and the remaining constraints ensure that the computational resources allocated to each IN will offload the collected data. This problem is a convex optimization problem and can be solved using standard convex optimization techniques.

[0350] Based on the above analysis and derivation, we proposed a multi-variable fixed alternating iteration algorithm to optimize the second-stage problem.

[0351] d. Algorithm 2 is used to solve (P1.2). The optimal trajectory q, velocity v, acceleration a, scheduling vector t and computing resource allocation f can be obtained by iteratively solving (P1.2.1), (P1.2.2a) and (P1.2.3). The specific steps are as follows:

[0352] d1) Initialize the number of iterations i and the initial trajectory , Initial resource allocation , initial speed , initial acceleration , intermediate variables and ;

[0353] d2) Using the given and Solve problem (P1.2.1) and get the optimal solution for the scheduling vector ;

[0354] d3) Using the given , , , , , and , solve problem (P1.2.2a) and get the optimal trajectory ,speed , acceleration , auxiliary variables and , intermediate variables and ;

[0355] d4) Based on the optimal trajectory obtained in step d3) and the scheduling vector obtained in step 2) Solve the problem (P1.2.3) to obtain the optimal resource allocation ;

[0356] d5) If the multivariable fixed alternating iterative algorithm converges, exit the iteration; otherwise, update the number of iterations i = i + 1 and use the trajectory , Resource Allocation ,speed , acceleration , intermediate variables and Bring it to step d1) for the next iteration.

[0357] The total computational complexity of Algorithm 2 is .

[0358] Numerical simulation results

[0359] Next, we simulated and verified the performance of the designed three-dimensional trajectory optimization and resource allocation joint algorithm for distributed data acquisition and offloading in UAV-assisted wireless networks. We generated a random sensor graph with 50 INs according to “gsp_sensors” in GSPBOX. Therefore, INs are randomly distributed in the geographical area of ​​[0-2000 m]×[0-2000 m], and their distribution is shown in Figure 2(a). Assume that the signal at IN is an artificial graph signal with a bandwidth of K = 5, satisfying the smoothness assumption. In graph signal theory, a necessary condition for a graph signal with a bandwidth of K to be perfectly recovered from its noise-free samples is that the number of selected sensors is greater than or equal to the signal bandwidth K. Specifically, if ,and is with The Laplacian matrix corresponding to the nodes The minimum mean square error solution provides a perfect reconstruction from the sampled image signal. In the first stage, by solving the problem The obtained location of the IN used to transmit data to the UAV is shown in Figure 2(b).

[0360] The starting and ending positions of the drone are set as and (Unit: meter). The selection of other simulation parameters is shown in Table 1.

[0361] Table 1

[0362] We evaluate the performance of the proposed algorithm through the following three benchmarks:

[0363] Two-dimensional trajectory planning scheme: enables the drone to plan its trajectory at a fixed altitude. The 2D trajectory planning is replaced, and the node access subproblem and the computing resource allocation subproblem adopt the method proposed in the present invention.

[0364] Fixed trajectory method: when solving the UAV trajectory optimization problem, the trajectory of the UAV is fixed as a parabola and optimized alternately. and , thus obtaining the optimal solution for this example.

[0365] Equivalent computing resource allocation: When solving the optimization problem, all INs have the same computing resources and are optimized alternately. and To obtain the optimal solution.

[0366] Figure 3 The effect of different flight durations $T$ on the flight trajectory of the UAV is shown. It can be found that when the flight duration T increases, the UAV covers a wider range during the entire flight. The UAV always flies as close to Ins as possible, because being close to Ins can improve the data transmission rate and communication quality, thereby ensuring the completion of the lower limit data collection task for each node. In addition, more time periods allow the UAV to offload more data computing tasks, thereby getting closer to IN and collecting more data.

[0367] Figure 4 shows the trend of the flight altitude of the UAV under different flight times T. When the flight time is 105s, the flight altitude of the UAV is relatively low. As the flight time increases, the average flight altitude of the UAV also increases. There are two reasons for this. On the one hand, the UAV has more time to collect data, so the UAV can get relatively close to the IN less frequently. On the other hand, the computing power of the UAV is limited, and the flight altitude needs to be increased to reduce the amount of data collected so that the collected data can meet the computing offloading capacity. Figure 5 shows the speed change trend of the UAV with different flight times T at different time periods. It can be seen from the figure that the speed of the UAV is fluctuating. When the UAV approaches the IN that is collecting data, its speed will decrease in order to collect more data. When the UAV is on the way to the next node, the UAV will increase its speed to approach the next node in order to reach the destination faster. In addition, as the flight time T continues to increase, the UAV has more time to collect data from the IN, so it can reduce its speed to obtain more EE.

[0368] Figure 6(a) and Figure 6(b) show the data collection time slot allocation of the UAV under different data collection limits when the flight duration T = 100s. It can be seen from the figure that the UAV visits IN 3, IN 4, IN 5, IN2 and IN 1 in sequence along the flight trajectory. Compared with the data collection limit of 14 Mbit, when the data collection limit is 9 Mbit, the UAV reserves less time for IN 5 and IN 2. This is because the UAV chooses to collect more data near IN 4 to obtain higher EE, while the lower data collection limit allows the UAV to meet the data transmission requirements of IN 5 and IN 2 in a short time.

[0369] Figure 7 shows the computational resource allocation results for computational offloading of the UAV-assisted system. Combined with the trajectory shown in Figure 3, it can be seen that during the data collection process of each IN, the UAV tends to allocate more computational resources to the current IN to process the data of that IN. In addition, we can also notice that as the flight time of the UAV increases, the color of the allocated computational resource heat map will also become lighter. The increase in duration gives the UAV more time to offload the collected data, so the computational time allocated to each node increases, and the size of the allocated computational resources also decreases.

[0370] Figure 8 compares the data collection thresholds S at different IN kThe proposed 3D-EE scheme and the EE of the three benchmarks under the condition. It is not difficult to find from the figure that the proposed strategy has the largest EE compared with the three benchmarks. Specifically, compared with the 2D trajectory planning scheme, the equal computing resource allocation method and the fixed trajectory method, the average performance improvement is 34.76%, 60.65% and 71.65% respectively, which fully demonstrates the effectiveness of the proposed strategy. In addition, the EE of the proposed scheme increases with S k As shown in Figure 9, this is because the proposed scheme can appropriately adjust the flight altitude of the UAV under the higher data collection lower limit requirement. When the data collection lower limit is increased, the flight altitude of the UAV is reduced, thereby increasing the EE.

[0371] Figure 10 shows the impact of the proposed method on the UAV EE at different flight durations T. Figure 10 compares the proposed scheme with three baselines. The minimum data collection amount S for all four schemes k are all set to 7 Mbit. From the results, it can be seen that the proposed scheme can achieve better EE performance regardless of the value of the flight duration T. In addition, as the flight duration T increases, the EE of the proposed scheme has a peak value, which is located near T=105-110s in the figure. The peak value of the equal computing resource allocation benchmark is located near T=120s. This is due to the introduction of 3D UAV trajectory planning and UAV attitude changes, which have an impact on data collection. The UAV can flexibly adjust the flight position to achieve the EE peak within the appropriate flight time.

[0372] Figure 11 shows the maximum computing resource constraints under different The EE of the proposed method is compared with that of the three benchmarks. With the increase of , the EE of the proposed scheme increases significantly and shows the best EE. Under this condition, the EE of the proposed scheme still increases rapidly, indicating that the increase in computing energy consumption in the proposed scheme does not significantly affect the change in EE, which further confirms the rationality of ignoring computing energy consumption in the optimization of the third sub-problem in the second stage of the algorithm. The increase of the fixed flight trajectory method and the two-dimensional scheme also shows an upward trend, while the computational energy consumption of the two strategies increases with the maximum computing resource limit. This is because although the efficiency values ​​of the three are close, the fixed flight trajectory method and the two-dimensional scheme are two-dimensional flight strategies, and the amount of data collected is significantly less than that of the three-dimensional flight strategy. The changes will have a certain impact on EE.

[0373] In summary, this paper proposes an optimization strategy for UAV-assisted MEC systems to maximize energy efficiency when solving the joint data collection and computation offloading tasks of large-scale IoT nodes. We propose a nested mixed integer optimization problem, where the external problem is to select a specific number of key INs from potential INs for data transmission. The internal problem maximizes the energy efficiency of the UAV by optimizing the scheduling, three-dimensional flight trajectory, and computational resource allocation of the UAV to perform data collection and computational offloading from the selected INs. First, the external IN selection problem is solved using the eigenvalue optimization technique based on Gale circle alignment. Second, a multivariable fixed alternating iterative algorithm is proposed to optimize the scheduling and energy efficiency of the UAV by decomposing the external problem and combining convex difference techniques and continuous convex approximation methods. Simulation results show that the optimization strategy proposed in this paper performs well in improving the energy efficiency of the UAV, and can significantly improve the energy efficiency and system performance of the UAV compared with the two-dimensional trajectory optimization and computational resource equalization algorithms. Future research can further explore optimization problems in more complex environments, such as optimization problems under the collaborative work of multiple UAVs or dynamic environmental changes. In addition, considering the continuous expansion of wireless networks in the future, how to improve the real-time and scalability of the algorithm will also be an important research direction.

Claims

1. A method for collecting and unloading data from distributed nodes in a drone-assisted Internet of Things, characterized in that: The method comprises the following steps: Step 1: Design a drone-assisted mobile edge computing system, in which the drone needs to select a suitable ground IoT node from many distributed wireless ground IoT nodes to complete data collection and computing offloading tasks while ensuring data integrity; Step 2: Propose a nested optimization problem; the inner problem selects a specific number of IoT nodes from a set of potential IoT nodes so that the data transmitted by the IoT nodes can reconstruct the global data with a certain accuracy, and the outer problem performs data collection and computation offloading from the selected IoT nodes by optimizing the scheduling, 3D flight trajectory, and computational resource allocation of the UAV to maximize the energy efficiency of the UAV. Step 3: To solve the nested optimization problem, a two-stage algorithm is used to solve the problem; 1) In the first stage, the selection of IoT nodes is transformed into a matrix eigenvalue optimization problem using a method based on Gale circle alignment to solve the internal problem in the optimization problem; 2) In the second stage, the external problem is divided into three sub-problems and solved by convexifying the model using equivalent transformation, convex difference and continuous convex approximation.

2. The method for collecting and unloading data from distributed nodes in the UAV-assisted Internet of Things according to claim 1 is characterized in that: The UAV-assisted mobile edge computing system described in step 1 includes a UAV 3D trajectory model, a node signal model, a fusion center signal reconstruction model, a probabilistic line-of-sight channel model, and an energy consumption and computing model of the UAV; The UAV-assisted mobile edge computing system consists of a UAV and IoT nodes, each of which can observe unknown parameters The observation results are sent to the fusion center, which estimates the parameters based on the received information. ,UAVs act as aerial collectors and mobile edge computing terminals,,collecting data from ground IoT nodes and bringing the data back to the,fusion center; The collaborative relationship between the IoT nodes is represented by a connected undirected graph To represent; the undirected graph By triple Indicates that Represents an undirected graph In A collection of IoT nodes, Represent undirected graphs The nodes in Represents an undirected graph middle The collection of edges, is the edge weight A collection of Reflects two connected nodes and Correlation or similarity between A node in the IoT The coordinates of Indicates the height of the IoT node in the coordinates Negligible, approximately zero.

3. The method for collecting and unloading data from distributed nodes in the UAV-assisted Internet of Things according to claim 2 is characterized in that: The UAV 3D trajectory model is as follows: Assume that the starting and ending positions of the drone are predetermined, and the coordinates are expressed as , the real-time 3D trajectory of the UAV is used express; Drone flight time Divided into Equal time slots , expressed as ; Each time slot are chosen to be short enough to assume that the drone is in the time slot The interior is approximately stationary; therefore, the trajectory of the drone Approximately a sequence of discrete points ; Due to limited energy, the flight time and maximum speed of the drone are and ; The maximum and minimum flight altitudes are and , Due to the maximum speed limit of the drone, the position between two time slots needs to meet the following constraints: （1） In formula (1), is the vertical ground trajectory of the UAV in the mth time slot; （2） In formula (2), D represents the displacement of the UAV, and: （3） In formula (3), is the speed of the drone in the mth time slot, is the acceleration of the UAV in the mth time slot, and: （4） and are the initial and final velocities of the drone, respectively; （5） （6） The acceleration of unmanned aerial vehicles shall not exceed ; （7） Limit the speed of the drone to be greater than To keep flying; （8） Since the rotor UAV needs to maintain balance during flight, the pitch angle limit of the UAV is given by the following formula: （9） in, is the vertical ground speed of the UAV in the mth time slot, is a constant; Considering that drones use time division multiple access, define a Boolean variable Describes the scheduling of all IoT nodes, where Indicates IoT nodes can be The transmission power required to meet the outage probability requirement is and transfer rate Transmit data to the drone, Indicates IoT nodes in the Time slots remain in sleep state; Scheduling variables Described as: and .

4. The method for collecting and unloading data from distributed nodes in the UAV-assisted Internet of Things according to claim 3 is characterized in that: The node signal model is as follows: Undirected Graph IoT Node The signal value of the graph domain at , so the complete graph domain signal is expressed in the form of a vector, that is: ; Each IoT node Observation Noise observation is performed at all times, so the IoT node The real-valued observation Expressed as ； in Represents an IoT node Noise in the real-valued observations of ; If the signal of an IoT node is completely collected by the drone, we call the signal of the IoT node an effective observation signal. The fusion center can use the signal of the effective observation IoT node to reconstruct the complete signal; then, all the observation signals that the drone can collect in one flight can be expressed as ； in represents the additive noise vector, Indicates the valid node data selectively collected by the drone during a flight; is a K×N matrix, Each row of represents a sampling vector with a modulus of 1, and each column corresponds to an IoT node. When a column is set to 1, the drone will completely collect data from the IoT node corresponding to the column. K is the number of nodes planned to be collected; Indicates time slot The set of nodes whose data has been completely collected; Assume that the time series data collected by the IoT nodes are stored in their storage areas respectively; during a flight of the drone, the IoT nodes The amount of data that needs to be transferred is ; For a given undirected graph associated with IoT nodes , and its corresponding graph Laplacian matrix L can be expressed as ； in is a diagonal matrix, 1 is the vector of all 1s, Represents a diagonal matrix whose input vector elements are located on the main diagonal; Signal Vector Smoothness The smoothness quantification model of the following formula is constructed on a given undirected weighted graph to quantify: ， The smaller the smoothness of a signal, the smoother the signal.

5. The method for collecting and unloading data from distributed nodes in the UAV-assisted Internet of Things according to claim 4 is characterized in that: The fusion center signal reconstruction model is as follows: Given a noisy graph signal , the goal of the fusion center is to restore the smooth graph signal; assuming that the fusion center adopts a biased graph signal reconstruction scheme based on graph Laplace regularization, through unconstrained Norm minimization realizes biased signal reconstruction; specifically, graph Laplace regularization can be expressed as the optimization of the target signal, and its formula is as follows: ； in is the balanced graph Laplace regularization with Tradeoff parameter for normal data fidelity term; optimal solution for quadratic objective function It can be obtained by solving the following linear equation: ； For the The L constructed on , therefore, the above linear equation has a unique solution: ； Original signal With the reconstruction signal Minimizing the mean square error between ； let , and is a symmetric positive definite matrix, so given , and ,minimize The mean square error can be minimized.

6. The method for collecting and unloading data from distributed nodes in the UAV-assisted Internet of Things according to claim 5 is characterized in that: The probabilistic line-of-sight channel model is as follows: A quasi-static block fading channel model is used for the IoT node-UAV link, with fading blocks, where the channel within each fading block remains constant and may change from block to block; connect the drone to the IoT node The wireless communication link between The downlink channel power gain in is denoted as , and expressed as ； in, represents the large-scale attenuation coefficient, represents the small-scale Ricean attenuation coefficient, , represents the mean function; and The calculation formula is ； ； in, For IoT nodes The coordinates of is the path loss factor, The value is 2. It represents the power gain when the reference distance is 1m. represents the small-scale Ricean attenuation coefficient, represents the LoS channel part, , represents the random scattering part, ; IoT Node In time slot The uplink transmission power for transmitting data to the UAV is , so the time slot No. The achievable transmission rate of attenuation blocks is calculated as follows: ； in, is the difference in signal-to-noise ratio between the actual modulation scheme and the theoretical Gaussian signal, is the channel bandwidth, represents the additive white Gaussian noise at the receiver; in the time slot When the IoT node is awakened The probability of interruption with the drone is given by: ； in, represents the probability density function, It is a time slot hour The cumulative distribution function of is a non-decreasing function; when the maximum permissible interruption probability is hour, Satisfy the following formula ； in yes The inverse function of Therefore, distributed sensors are Total data transferred within It can be defined as 。 7. The method for collecting and unloading data from distributed nodes in a drone-assisted Internet of Things according to claim 6, characterized in that: The energy consumption and calculation model of the UAV are as follows: The energy consumption of unmanned aerial vehicles generally consists of three main parts: communication-related energy consumption, propulsion energy consumption, and computing energy consumption. Among them, the communication-related energy consumption is much smaller than the propulsion energy consumption, so the communication-related energy consumption can be ignored. For a rotorcraft UAV, its propulsion energy consumption for 3D flight can be expressed as ； in, and are two constants related to the wingspan efficiency, wing surface area, and mass of the drone; is the acceleration due to gravity, is the acceleration in the direction perpendicular to the ground, is the velocity in the x-axis direction of the Cartesian coordinate system, The velocity in the y-axis direction of the Cartesian coordinate system; The computational energy of the drone in each time slot is as follows: ； in, is a constant representing the effective switched capacitance, Represents the drone in the time slot Assigned to IoT nodes computing resources; In summary, the total energy consumption of the drone can be obtained as shown in the following formula: ； When the initial and final positions and speeds of the UAV are determined, the third and fourth terms in the above formula are constants, so they do not need to be considered in the optimization process; In addition, considering the limited computing resources of the UAV, the maximum allocatable computing resources of the UAV are restricted in each time slot: ； The processing of all collected data should be completed within one flight cycle, which results in the following limitations: ； in, represents the number of CPU cycles required to calculate each data bit. This constraint ensures that the UAV The data collected in and after the time slot can be After a time slot, it is completely unloaded.

8. The method for collecting and unloading data from distributed nodes in the UAV-assisted Internet of Things according to claim 7 is characterized in that: The nested optimization problem proposed in step 2 is as follows: By optimizing the trajectory , scheduling vector , calculate resource allocation vector , and a collection of IoT nodes that collect data , the energy efficiency maximization optimization problem P1 can be expressed as ； The optimization problem P1 is composed of an external problem and an internal problem. In the external problem, the goal is to find an optimal set of access IoT nodes so that the reconstruction error Satisfy a certain accuracy; at the same time, the goal of the internal problem is to determine the scheduling vector of the node , optimal trajectory , and the computational resource allocation given the optimal set of IoT nodes to access .

9. The method for collecting and unloading data from distributed nodes in the UAV-assisted Internet of Things according to claim 8, characterized in that: In the first stage of step 1) of step 3, the selection of IoT nodes is transformed into a matrix eigenvalue optimization problem using a method based on Gale circle alignment to solve the internal problem in the optimization problem P1; the specific method is as follows: The problem to be solved in the first stage is shown in question P1.1: ； because is a symmetric positive definite matrix that satisfies ; therefore, maximizing can minimize the upper bound of the mean square error; therefore, Problem P1.1 can be transformed into Problem P1.1a to maximize the matrix The minimum eigenvalue of is in the form of: ； By converting the above matrix eigenvalue optimization problem into the scaling and translation operation of the Gale circle, the method based on the Gale circle alignment is used to solve the problem P1.1a. The specific steps are as follows: 1) Input IoT node undirected graph , reconstruction error and IN number budget K; 2) Initialize the IoT node collection , N is the number of IoT nodes in the set; 3) From node n=1 to node n=N, use the method based on the Gale circle theorem to estimate the covering subset of each Gale circle , and the resulting subset Merge to get a collection ; 4) At the same time, if n≤K, select and set The subset with the largest intersection , subset Integration with IoT nodes After taking the intersection, find the set and the IoT node set The complement of the node set Reassignment; Then merge the current node n into the selected IoT node set , return to step 3) and take the next value of n; 5) If n>K, then output the currently selected IoT node set , end the solution; For different reconstruction errors Perform iterative solution to obtain the set of IoT nodes IoT node collection .

10. The method for collecting and unloading data from distributed nodes in the UAV-assisted Internet of Things according to claim 9, characterized in that: Step 2 of Step 3 In the second stage, the external problem is divided into three sub-problems and solved by convexifying the model using equivalent transformations, convex differences, and continuous convex approximations; the specific methods are as follows: In the second stage of the problem, the collection and unloading of data sent by the selected IoT nodes of the external problem is completed by reasonably planning the flight path of the drone and allocating computing resources. The problems that need to be solved in the second stage are shown in Problem P1.2: ； Problem P1.2 is divided into three sub-problems, as follows: a. IN node access sub-problem Given a UAV flight trajectory and computing resource allocation In this case, the denominator of problem P1.2 is determined, and the sub-problem becomes the design of the access time slot of the drone data collection node, with the goal of maximizing the total amount of data collection; To solve this integer programming problem, we need to set the binary constraint Relax to ; The access time slot problem of the UAV data collection node is designed as follows: ； ； ； ； ； ； b. UAV 3D trajectory optimization sub-problem According to the given set of nodes to be visited , scheduling vector and resource allocation vector Optimizing drone trajectories and the associated speed and acceleration , the optimization problem P1.2 can be restated as: ； in and the numerator of the objective function Relative to All are non-concave; Relative to is also non-convex, the denominator of the objective function is relative to and is non-convex; are also non-convex; therefore, they need to be transformed into mathematically solvable problems; b1, the numerator of the objective function relative to Convexification of: because may be non-negative, so a slack variable is introduced , thus obtaining ； ； ； so, is converted to ； b2, the numerator of the objective function relative to Convexification of: In is a non-convex function, which is transformed into a convex function using a first-order Taylor expansion and applying a continuous convex approximation; ； Introducing auxiliary variables and This is to handle the non-convex denominator ; In this way, the above formula becomes ； and The right side of the non-concave terms is bounded by the local point Performing a first-order Taylor expansion on the right side yields: ； ； Through the above transformation, the formula The terms in can be transformed into a standard form consistent with convex optimization: ； b3. Convexification of: The Taylor expansion on the right also yields ; and convexification of the numerator of the objective function let represents the given trajectory in the rth iteration; by applying the first-order Taylor expansion, The lower bound of ,in ； ； ； Through the above operation, the numerator of the objective function becomes the following concave form ； Similarly, replace for ； b4. Convexification of: At local point Department, about The first-order Taylor expansion of ； in and ; ； Will Substituting into ； Through the above transformation, the optimization of the problem P1.2.2 related to the UAV trajectory can be transformed into the following form: ； c. Computational resource allocation optimization sub-problem The computational resource offloading problem is formulated as maximizing the lower bound of the computational offloading data volume, i.e., Problem P1.2.3: ； d. A multivariable fixed alternating iteration method is used to optimize the second stage problem. The specific steps of this method are as follows: d1) Initialize the number of iterations i and the initial trajectory , Initial resource allocation , initial velocity , initial acceleration , intermediate variables and ; d2) Using the given and Solve problem P1.2.1 and obtain the optimal solution for the scheduling vector ; d3) Using the given , , , , , and , solve problem P1.2.2a and obtain the optimal trajectory ,speed , acceleration , auxiliary variables and , intermediate variables and ; d4) Based on the optimal trajectory obtained in step 3) and the scheduling vector obtained in step d2) Solve problem P1.2.3 and obtain the optimal resource allocation ; d5) If the multivariable fixed alternating iterative algorithm converges, exit the iteration; otherwise, update the number of iterations i = i + 1 and use the trajectory , Resource Allocation ,speed , acceleration , intermediate variables and Bring it to step d1) for the next iteration.

Images