阅读说明：本技术 一种基于边缘计算的无人机目标追踪功耗优化方法及系统 (Unmanned aerial vehicle target tracking power consumption optimization method and system based on edge calculation ) 是由 邓晓衡 李君� 李博 于 2019-09-20 设计创作，主要内容包括：本发明公开了一种基于边缘计算的无人机目标追踪功耗优化方法及系统,包括步骤：S1、设置无人机追踪过程中视频处理模式；S2、计算任务分配,通过同时考虑无人机发射功率的调整和边缘节点的选择机制为无人机选择合适的发射功率和服务节点,最小化总开销。相比于一般的使用固定的最大的无人机发射功率进行数据传输,本发明提出的算法同时考虑无人机发射功率的调整和边缘节点的选择方案,以最小化能耗成本和时间成本之间的权衡值,在保证正常追踪的前提下,降低无人机的通信能耗。(the invention discloses an unmanned aerial vehicle target tracking power consumption optimization method and system based on edge calculation, which comprises the following steps: s1, setting a video processing mode in the unmanned aerial vehicle tracking process; and S2, calculating task allocation, and selecting appropriate transmitting power and service nodes for the unmanned aerial vehicle by simultaneously considering the adjustment of the transmitting power of the unmanned aerial vehicle and the selection mechanism of the edge nodes, thereby minimizing the total overhead. Compared with the common method of using the fixed maximum unmanned aerial vehicle transmitting power for data transmission, the algorithm provided by the invention considers the adjustment of the unmanned aerial vehicle transmitting power and the selection scheme of the edge node at the same time so as to minimize the balance value between the energy consumption cost and the time cost, and reduce the communication energy consumption of the unmanned aerial vehicle on the premise of ensuring normal tracking.)

1. An unmanned aerial vehicle target tracking power consumption optimization method based on edge calculation is characterized by comprising the following steps:

s1, setting a video processing mode in the unmanned aerial vehicle tracking process;

And S2, calculating task allocation, and selecting appropriate transmitting power and service nodes for the unmanned aerial vehicle by simultaneously considering the adjustment of the transmitting power of the unmanned aerial vehicle and the selection mechanism of the edge nodes, thereby minimizing the total overhead.

2. The method of claim 1, wherein the video processing modes include two modes:

The first method is that the unmanned aerial vehicle captures and stores the video, executes the task processing process locally until the unmanned aerial vehicle finishes the task and returns the result;

the second mode is to offload the video processing task to a ground server for processing.

3. The method for optimizing power consumption for target tracking of an unmanned aerial vehicle based on edge computing as claimed in claim 2, wherein the video processing mode is based on derivation of a functional relationship between a duration of the unmanned aerial vehicle and a weight of the unmanned aerial vehicle as a selection criterion, and specifically comprises:

By the estimation method of the endurance time of the unmanned aerial vehicle, the influence of the whole weight of the unmanned aerial vehicle on the endurance time is theoretically analyzed, and the relationship between the endurance time and the weight of the unmanned aerial vehicle body is deduced as follows:

Wherein, T_entIs the duration, G is the weight of the drone, W_bis the electric quantity of the battery, U_bIs the battery voltage, σ is the electrically regulated duty cycle, C_MIs the coefficient of propeller torque, D_PIs the diameter of the propeller, C_TIs the coefficient of tension of the propeller, K_TIs the motor torque constant, n_rIs the number of propellers, I_m0Is the nominal no-load current;

Through analyzing actual experimental data, the relation between the energy consumption (J/s) of the unmanned aerial vehicle per unit time and the weight (kg) of the unmanned aerial vehicle is simulated, and then the energy consumption under different video processing modes is compared.

4. The unmanned aerial vehicle target tracking power consumption optimization method based on edge computing according to claim 1, wherein the method for computing task allocation specifically comprises:

In the process of tracking by the unmanned aerial vehicle, the unmanned aerial vehicle unloads the calculation task to the ground edge node for cooperative processing, and receives a return result;

By usingthe total time delay of the task execution is represented by the following calculation formula:

With E_m,kthe energy consumption in the communication process is represented, and the calculation formula is as follows:

Wherein the content of the first and second substances,In order to delay the transmission of the task,Calculating time delays for tasks, S_mIs the amount of data in the time slot m,Is the channel capacity, beta₀indicates the reference distance d₀1m channel power, with p_m,kindicating that a drone is assigned to node n in slot m_m,kB denotes the channel bandwidth, σ²Is the power of the white gaussian noise,is a distance d₀Signal-to-noise ratio at 1m, r_m,kIs the edge node computing power.

5. The method for optimizing power consumption of unmanned aerial vehicle target tracking based on edge computing as claimed in claim 4, wherein to reduce the communication energy consumption and the weight of time delay, a total cost is used_m,kIt is shown that,

Where α (≧ 0) and β (≧ 0) represent relative weights for transmission energy consumption and total delay, which can be set and adjusted to meet different scenarios.

6. The method for optimizing power consumption of target tracking of unmanned aerial vehicle based on edge computing according to claim 5, wherein the objective of the computing task allocation is to jointly optimize the selection mechanism X of the transmission power P and the edge node EN of the unmanned aerial vehicle, so as to minimize the trade-off between the transmission energy consumption of the unmanned aerial vehicle and the total delay, and ensure normal tracking;

P and X are respectively expressed asAnd

The target function of the unmanned aerial vehicle target tracking system is as follows:

In order to ensure that the unmanned aerial vehicle can receive results and adjust in time to normally track the target, it needs to satisfy:

Wherein the content of the first and second substances,Is the task execution time, which is the sum of the task transmission time and the task processing time, δ_tIs a delay limitation of the computational task.

7. An unmanned aerial vehicle target tracking power consumption optimization system based on edge calculation is characterized by comprising:

The mode setting module is used for setting a video processing mode in the tracking process of the unmanned aerial vehicle;

And the edge calculation module is used for calculating task allocation, and selecting proper transmitting power and service nodes for the unmanned aerial vehicle by simultaneously considering the adjustment of the transmitting power of the unmanned aerial vehicle and the selection mechanism of the edge nodes, so that the total overhead is minimized.

8. The system of claim 7, wherein the video processing modes include two modes:

The first method is that the unmanned aerial vehicle captures and stores the video, executes the task processing process locally until the unmanned aerial vehicle finishes the task and returns the result;

The second mode is to offload the video processing task to a ground server for processing.

9. The system of claim 8, wherein the video processing mode is based on derivation of a functional relationship between a duration of the drone and a weight of the drone, and specifically comprises:

By the estimation method of the endurance time of the unmanned aerial vehicle, the influence of the whole weight of the unmanned aerial vehicle on the endurance time is theoretically analyzed, and the relationship between the endurance time and the weight of the unmanned aerial vehicle body is deduced as follows:

wherein, T_entIs the duration, G is the weight of the drone, W_bIs the electric quantity of the battery, U_bIs the battery voltage, σ is the electrically regulated duty cycle, C_Mis the coefficient of propeller torque, D_PIs the diameter of the propeller, C_TIs the coefficient of tension of the propeller, K_TIs the motor torque constant, n_ris a helixNumber of paddles I_m0is the nominal no-load current;

through analyzing actual experimental data, the relation between the energy consumption (J/s) of the unmanned aerial vehicle per unit time and the weight (kg) of the unmanned aerial vehicle is simulated, and then the energy consumption under different video processing modes is compared.

10. The system of claim 7, wherein the method for computing task allocation specifically comprises:

in the process of tracking by the unmanned aerial vehicle, the unmanned aerial vehicle unloads the calculation task to the ground edge node for cooperative processing, and receives a return result;

by usingThe total time delay of the task execution is represented by the following calculation formula:

With E_m,kthe energy consumption in the communication process is represented, and the calculation formula is as follows:

Wherein the content of the first and second substances,In order to delay the transmission of the task,Calculating time delays for tasks, S_mis the amount of data in the time slot m,is the channel capacity, beta₀indicates the reference distance d₀1m channel power, with p_m,kTo representallocation of drones to node n in time slot m_m，kb denotes the channel bandwidth, σ²Is the power of the white gaussian noise,Is a distance d₀Signal-to-noise ratio at 1m, r_m,kIs the edge node computing power.

Technical Field

the invention belongs to the field of unmanned aerial vehicle target tracking, and particularly relates to an unmanned aerial vehicle target tracking power consumption optimization method and system based on edge calculation.

Background

Edge computing is used as a new computing mode, computing resources are pushed to the edge of the Internet close to a terminal user so as to meet the increasing requirements of applications such as the Internet of Things (IoT) on high computing capacity and low delay, and computing tasks are decomposed and migrated to edge nodes for processing by an edge computing model so as to reduce computing load of a cloud computing data center and reduce transmission delay, thereby achieving the purposes of reducing energy consumption and improving service time of mobile equipment.

Unmanned Aerial Vehicles (UAVs) are an important class of robots, and with the development of internet of things technology and the continuous improvement of Unmanned Aerial Vehicle performance, the Unmanned Aerial vehicles have been widely applied to various aspects including intelligence reconnaissance, communication relay, data collection, search and rescue, armed combat and the like. Practice has shown that UAVs are the best choice for performing boring, bad or Dangerous (Dull, Dirty, or Dangerous,3D) tasks.

compared with communication of fixed infrastructure, an unmanned aerial vehicle-assisted communication network is more flexible and can bring a lot of additional gains in mobility, so that more and more people recently apply the unmanned aerial vehicle as target tracking equipment to a lot of scenes, such as tracking high-mobility aerial targets in air-to-air operations, tracking and tracking ground moving targets by ground hitting tasks, tracking unexpected escape vehicles in urban accidents and the like. The method comprises the following steps that (1) moving target tracking is a complex and challenging problem, an unmanned aerial vehicle continuously shoots videos and tracks suspicious targets in the tracking process, due to the limitation of computing resources of the unmanned aerial vehicle, continuously generated video streams cannot be rapidly processed, and the video processing task is usually considered to be unloaded from the unmanned aerial vehicle to an edge node/server for cooperative processing; however, the unmanned aerial vehicle flight propulsion process and the unloading task process are accompanied by energy consumption, and according to unmanned aerial vehicle product data issued by the office network of the great Xinjiang, the maximum flight time of most unmanned aerial vehicles under ideal conditions is about 31 minutes, the working time is very limited, and because of the limitation of battery capacity, the energy consumption problem of the unmanned aerial vehicles is always a research hotspot.

Disclosure of Invention

The present invention considers a situation where in urban traffic a target vehicle is attempting to escape and a drone is used to track the target of the escape. In the tracking process, a camera on the unmanned aerial vehicle continuously captures a high-resolution video, and then the video needs to be analyzed and processed and subjected to target detection, so that the unmanned aerial vehicle can timely make adjustment to realize tracking. The task processing and feedback of the unmanned aerial vehicle can have two modes, the first mode is that the unmanned aerial vehicle captures and stores videos, and executes the task processing process locally until the unmanned aerial vehicle completes the task and returns the result. However, the hardware required for processing the video stream by using the image processing technology of the computation-and-memory-intensive DNN is heavy, and the energy consumption of the unmanned aerial vehicle during flight and the energy consumption of local computation due to the weight of the task processing module are not negligible. The second mode, in which video is offloaded from the UAV to the ground server for processing, reduces the energy consumption due to the weight of the mission processing module, but instead consumes energy during the transmission of video data.

According to the multi-rotor unmanned aerial vehicle knowledge manual, the functional relation between the endurance time (the endurance time reflects energy consumption) of the unmanned aerial vehicle and the weight of the unmanned aerial vehicle is deduced, and the second mode is proved to be superior to the first mode in terms of energy conservation.

However, there are two key issues with the second mode, i.e., the video is offloaded to the ground server for processing. The first is that due to time delay limitation, in the target tracking process of the unmanned aerial vehicle, continuously captured videos need to be transmitted and processed as soon as possible, so that the unmanned aerial vehicle can receive results in time and make corresponding adjustment to achieve successful tracking. The traditional way of communicating is for UAVs to upload real-time video to the cloud for processing, which is beneficial at task computing speeds, but in scenarios such as drone target tracking, the latency of the transmission may be unacceptable. The second problem is that the transmission energy consumption caused by real-time high-resolution video transmission is not negligible due to the limited battery energy of the unmanned aerial vehicle.

In order to solve the problems, the method provides energy-saving and efficient unmanned aerial vehicle target tracking by jointly considering the adjustment of the unmanned aerial vehicle transmitting power and the selection scheme of the ground edge node by using emerging edge calculation. The unmanned aerial vehicle offloads the computing task to a ground edge node instead of the cloud for cooperative processing, then receives the result, and adjusts to ensure successful tracking instead of losing the tracking target. Under the general fading channel model of the drone-edge node link, the invention considers both the transmit power of the drone and the selection scheme of the edge node to minimize the total cost, which is a weighted sum of the energy cost and the task execution time cost.

The invention originally provides two key points as follows:

(1) Power consumption optimization model in unmanned aerial vehicle target tracking scene based on edge calculation

the research on the power consumption optimization problem of the unmanned aerial vehicle has a lot of achievements in the academic world, but most of the power consumption optimization problem is researched in the flight propulsion process of the unmanned aerial vehicle when the unmanned aerial vehicle is applied to scenes such as data collection and the like.

(2) Design energy-saving and efficient unmanned aerial vehicle task allocation (EUTD) algorithm

The traditional power consumption optimization algorithm of the unmanned aerial vehicle mainly reduces energy consumption by optimizing the flight path of the unmanned aerial vehicle, but the invention researches the problem of energy consumption optimization in the unloading process of the video task of the unmanned aerial vehicle and fully considers the computing power (CPU and memory) and the position (GPS) of the edge node to determine the task allocation scheme of the unmanned aerial vehicle. Compared with the common method of using the fixed maximum unmanned aerial vehicle transmitting power for data transmission, the algorithm provided by the invention considers the adjustment of the unmanned aerial vehicle transmitting power and the selection scheme of the edge node at the same time so as to minimize the balance value between the energy consumption cost and the time cost, and reduce the communication energy consumption of the unmanned aerial vehicle on the premise of ensuring normal tracking.

Specifically, the invention provides an unmanned aerial vehicle target tracking power consumption optimization method based on edge calculation, which is characterized by comprising the following steps:

S1, setting a video processing mode in the unmanned aerial vehicle tracking process;

And S2, calculating task allocation, and selecting appropriate transmitting power and service nodes for the unmanned aerial vehicle by simultaneously considering the adjustment of the transmitting power of the unmanned aerial vehicle and the selection mechanism of the edge nodes, thereby minimizing the total overhead.

further, the video processing modes include two types:

the first method is that the unmanned aerial vehicle captures and stores the video, executes the task processing process locally until the unmanned aerial vehicle finishes the task and returns the result;

The second mode is to offload the video processing task to a ground server for processing.

Further, the video processing mode is based on derivation of a functional relationship between the duration of the unmanned aerial vehicle and the weight of the unmanned aerial vehicle as a selection basis, and specifically comprises:

By the estimation method of the endurance time of the unmanned aerial vehicle, the influence of the whole weight of the unmanned aerial vehicle on the endurance time is theoretically analyzed, and the relationship between the endurance time and the weight of the unmanned aerial vehicle body is deduced as follows:

wherein, T_entIs the duration, G is the weight of the drone, W_bis the electric quantity of the battery, U_bIs the battery voltage, σ is the electrically regulated duty cycle, C_Mis the coefficient of propeller torque, D_PIs the diameter of the propeller, C_TIs the coefficient of tension of the propeller, K_TIs the motor torque constant, n_rIs the number of propellers, I_m0Is the nominal no-load current;

Through analyzing actual experimental data, the relation between the energy consumption (J/s) of the unmanned aerial vehicle per unit time and the weight (kg) of the unmanned aerial vehicle is simulated, and then the energy consumption under different video processing modes is compared.

Further, the method for computing task allocation specifically includes:

in the process of tracking by the unmanned aerial vehicle, the unmanned aerial vehicle unloads the calculation task to the ground edge node for cooperative processing, and receives a return result;

By usingthe total time delay of the task execution is represented by the following calculation formula:

Wherein the content of the first and second substances,In order to delay the transmission of the task,calculating time delays for tasks, S_mis the amount of data in the time slot m,Is the channel capacity, beta₀Indicates the reference distance d₀1m channel power, with p_m,kIndicating that a drone is assigned to node n in slot m_m,kB denotes the channel bandwidth, σ²Is the power of the white gaussian noise,Is a distance d₀signal-to-noise ratio at 1m, r_m,kIs the edge node computing power.

With E_m,kthe energy consumption in the communication process is represented, and the calculation formula is as follows:

In order to reduce the weight of communication energy consumption and time delay, the total cost is used_m,kIt is shown that,

Where α (≧ 0) and β (≧ 0) represent relative weights for transmission energy consumption and total delay, which can be set and adjusted to meet different scenarios.

The invention also provides an unmanned aerial vehicle target tracking power consumption optimization system based on edge calculation, which is characterized by comprising the following steps:

The mode setting module is used for setting a video processing mode in the tracking process of the unmanned aerial vehicle;

and the edge calculation module is used for calculating task allocation, and selecting proper transmitting power and service nodes for the unmanned aerial vehicle by simultaneously considering the adjustment of the transmitting power of the unmanned aerial vehicle and the selection mechanism of the edge nodes, so that the total overhead is minimized.

Compared with the prior art, the invention has the following beneficial effects:

1) The invention theoretically explains the influence of the weight of the unmanned aerial vehicle on the endurance time of the unmanned aerial vehicle. By analyzing the actually measured data, the functional relation between the energy consumption per unit time (J/s) and the weight (kg) of the unmanned aerial vehicle is simulated. In addition, we focus on energy consumption comparison and analysis of the drone in video local processing and video offload processing.

2) the invention researches the problem of energy-saving computing task allocation based on edge computing, namely, an unmanned aerial vehicle unloads a computing task to a ground edge node instead of a cloud for processing. This is an offload service at the edge of the network, and the basic link for mobile devices to communicate with edge nodes can avoid high latency and additional energy consumption due to long transmission distances and server congestion compared to traditional communication links through the cloud.

3) Unlike most tasks of offloading the mission of the drone using a fixed maximum transmission power, the present invention provides a suitable way to adjust the transmission power and select the appropriate serving edge node, considering that the transmission power and the distance from the drone to the edge node affect the transmission energy consumption and transmission time. Furthermore, different edge nodes have different processing capabilities, which means that task computation time should also be taken into account, since the processing results should be fed back as fast as possible to ensure tracking. Based on the above description, the invention designs an energy-saving and efficient unmanned aerial vehicle task allocation (EUTD) algorithm for adjusting the transmission power and selecting a proper edge node in the unmanned aerial vehicle target tracking process.

4) In addition, the invention verifies that different relative weights of energy cost and time cost correspond to different transmission power and edge node selection schemes through simulation experiments, and proves that the EUTD algorithm can realize more reliable and energy-saving target tracking compared with other algorithms.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.

FIG. 1 is a graph of the relationship between energy consumption per unit time and weight of the unmanned aerial vehicle of the present invention;

FIG. 2 is a graph comparing the energy consumption of two modes of the present invention;

FIG. 3 is a diagram of the unmanned aerial vehicle target tracking scene based on edge calculation according to the present invention;

FIG. 4 is a diagram of the unmanned aerial vehicle mission Allocation (EUTD) algorithm of the present invention;

FIG. 5 is a graph of the comparative effect of the present invention on giving different weights to the energy consumption and delay of the drone;

FIG. 6 is a graph showing the effect of the proposed solution of the present invention compared to two reference solutions;

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments that can be derived by one of ordinary skill in the art from the embodiments given herein are intended to be within the scope of the present invention.

Video processing mode in unmanned aerial vehicle tracking process

unmanned aerial vehicle is carrying out the in-process of target tracking, and unmanned aerial vehicle carries out target identification through shooting the video and realizes tracking, and the video stream of constantly shooing needs real-time analysis and processing to give unmanned aerial vehicle with the processing result and guarantee normal pursuit and be unlikely to lose the target. Because unmanned aerial vehicle has certain load capacity, a feasible mode is to mount a processor on unmanned aerial vehicle, video analysis and processing work just can be accomplished locally like this, unmanned aerial vehicle just can obtain the feedback result along with video processing in real time, then in time make action adjustment according to the feedback result and guarantee normal pursuit, whole pursuit process just can be accomplished alone by unmanned aerial vehicle, higher automation and low latency have, but what come with is the weight of load to the influence of unmanned aerial vehicle flight in-process energy consumption, and because video analysis and processing are the work that high intensive and high resource occupy, video local processing also can produce some energy consumption. The other method is to unload the video processing task to a ground server for processing, and the method can remove a task processing module on the unmanned aerial vehicle and reduce the weight of the unmanned aerial vehicle body, thereby reducing the energy consumption in the flight process, but the energy consumption and the wireless bandwidth consumption of the unmanned aerial vehicle caused by the transmission of a large amount of video data are not negligible, compared with the local processing of the unmanned aerial vehicle, the method has the following problems that firstly, the delay is increased; second, mobile connections have instability in performance and reliability due to the mobility of the drone.

The invention researches the factors influencing the endurance time of the unmanned aerial vehicle, focuses on the influence of the weight of the unmanned aerial vehicle, theoretically analyzes the influence of the whole weight of the unmanned aerial vehicle on the endurance time by the estimation method of the endurance time of the unmanned aerial vehicle, and deduces the relation between the endurance time and the weight of the unmanned aerial vehicle asWherein, T_entis the duration, G is the weight of the drone, W_bis the electric quantity of the battery, U_bIs the battery voltage, σ is the electrically regulated duty cycle, C_MIs the coefficient of propeller torque, D_PIs the diameter of the propeller, C_TIs the coefficient of tension of the propeller, K_TIs the motor torque constant, n_rIs the number of propellers, I_m0Is the nominal idling current, it can be seen from this derivation that the endurance T of the drone is inversely proportional to the weight G.

By analyzing the actual experimental data, the relationship between the energy consumption (J/s) of the unmanned aerial vehicle per unit time and the weight (kg) of the unmanned aerial vehicle is simulated, as shown in fig. 1. Then, the invention compares the energy consumption of the unmanned aerial vehicle under the two modes of video local processing and video unloading processing, as shown in fig. 2. From the results, we can see the difference between the energy consumption of the unmanned aerial vehicle in the two modes of the video local processing and the video unloading processing, and although the video unloading processing has instability in performance and reliability, the video unloading processing is continuously improved along with the progress of the technology, and in contrast, the video processing module carried by the video local processing of the unmanned aerial vehicle causes more energy consumption.

Second, system model

an unmanned aerial vehicle target tracking scene graph based on edge calculation is shown in fig. 3.

Assuming that the unmanned aerial vehicle unloads the video task to the ground edge node EN within the time T, assuming that the unmanned aerial vehicle reaches the flight height H (m), the flight trajectory of the unmanned aerial vehicle is projected to the ground and is represented as q (T) epsilon R^2×1T is 0 ≦ T, and the time T is discretized into M time slots, i.e., T ≦ M δ_t，δ_tindicating the length of each slot, the position of the drone is considered constant in a slot, so the trajectory of the drone can be approximately discretized into a set q_m,1≤m≤M},q_mIndicating the position of the drone in slot m. Suppose that the amount of data generated by the drone in slot m is S_mAnd bits, in two continuous time slots, the unmanned plane transmits the data generated in the last time slot in the next time slot.

Suppose there are Z edge nodes EN in a region of a city, using n_z1 ≦ Z ≦ Z } indicates that, in general, each edge node has limited computational resources, and different edge nodes EN have different computational resources, if edge node n is_zIs not powerful enough, n_zIt cannot serve as a service node, and as the drone moves, the set of edge nodes that can serve the drone is not fixed, and there may be different sets of service nodes in different time slots. Assume that at time slot m, the drone is able to offload a computing task into a set of children nodes with K nodes, denoted as n_m,k,1≤k≤K,1≤m≤M}，{n_m,kIs { n }_zsubset of { n }, n_m,kIs denoted by w_m,k∈R^2×1，r_m,kIndicating the processing power of the kth node in time slot m.

The invention designs an edge node selection mechanism by simultaneously considering the computing power of edge nodes and the distance from the nodes to the unmanned aerial vehicle, supposes that only one edge node is selected as a service node in a time slot, and uses x_m,kIndicates a node selection variable if n_m,kIs selected as the service node, then x_m,k1, otherwise x_m,k0, so we have the limitation

If n is_m,kis selected as a compute node, then the drone and node n_m,kDistance between d_m,kExpressed as shown in equation (1).

We assume that the communication link from the drone to the ground EN is a quasi-static block fading channel, where the channel remains constant within each time slot but changes over time slots. For ease of illustration, it is assumed that the communication link is a line-of-sight (LoS) channel. Indeed, the drone ground channel is more likely to have a LoS link than the ground channel. Thus, the quasi-static block fading channel follows a free space path loss model, which can be expressed as

β₀indicates the reference distance d₀1m channel power, with p_m,kIndicating that a drone is assigned to node n in slot m_m,kThen, the signal communication rate is expressed as equation (3) in bps.

Where B denotes the channel bandwidth, σ²is the power of the white gaussian noise,Is a distance d₀signal to noise ratio at 1 m.

in the process of tracking by the unmanned aerial vehicle, the unmanned aerial vehicle unloads the calculation task to the ground edge node for cooperative processing, and receives a return result. The task transmission delay is expressed asNamely, it isThe task computation time delay is expressed asassuming that the resulting feedback delay is a fixed, very small value, ignored here, the task execution delay isandIs expressed by the formula (4).

In time slot m, the calculation task has a delay limit delta_tthat is, the transmission delay and the computation delay of the task should not exceed δ_tso that the unmanned aerial vehicle can timely receive the feedback result to adjust and guarantee tracking, namelywith E_m,kRepresents the energy consumption during communication as follows:

The objective of the subject is to reduce the weight of communication energy consumption and time delay, and use total cost_m,kAnd (4) showing.

Alpha (0) and beta (0) represent relative weights of transmission energy consumption and total delay, which can be set and adjusted to meet different situations. Generally speaking, we attach more importance to the energy consumption to prolong unmanned aerial vehicle's life.

wherein the content of the first and second substances,In order to delay the transmission of the task,Calculating time delays for tasks, S_mIs the amount of data in the time slot m,Is the channel capacity, beta₀indicates the reference distance d₀1m channel power, with p_m,kIndicating that a drone is assigned to node n in slot m_m,kB denotes the channel bandwidth, σ²is the power of the white gaussian noise,Is a distance d₀Signal-to-noise ratio at 1m, r_m,kis the edge node computing power.

p and X are respectively expressed asandour goal is to jointly optimize the selection of the transmission powers P and EN of the dronesmechanism X in order to minimize the trade-off between drone transmission energy consumption and total delay, while ensuring normal tracking. Our problems are expressed as follows

equation (7) is the objective function of the drone target tracking system, and constraint (8) ensures that the drone can receive the results and adjust in time to track the target normally.

Third, algorithm

the invention designs an energy-saving and efficient unmanned aerial vehicle task allocation (EUTD) algorithm, as shown in fig. 4, which considers the transmission power of the unmanned aerial vehicle and the edge node selection scheme at the same time to minimize the weight between the energy consumption cost and the time cost.

Fourth, experimental results

Fig. 5 shows the total cost of the drone, the transmission energy consumption, the transmission power, the delay, and the edge node selected per timeslot when different weights are given to the drone energy consumption and the delay (α ═ 0.9, β ═ 0.1), (α ═ 0.8, β ═ 0.2), (α ═ 0.7, β ═ 0.3), and it can be seen from the figure that the larger the value of α (i.e., the larger the weight given to the drone energy consumption), the lower the total cost of the drone and the transmission energy consumption. When α is 0.9 and β is 0.1, the transmit power fluctuates between 18 and 30mw and when α is 0.7 and β is 0.3, the transmit power fluctuates between 60 and 90mw, with greater weighting of the energy consumption implying longer delays, but still within an acceptable range. Therefore, we propose to attach more importance to energy consumption.

Fig. 6 is a comparison between the scheme proposed by the present invention and two reference schemes, the scheme of the present invention is to adjust the transmission power of the drone, and select a service node by comprehensively considering the calculation capability of the edge node and the distance between the edge node and the drone, the first reference scheme is to select the nearest edge node as the service node while the transmission power remains the maximum value of 0.1W, and the second reference scheme is to select the edge node as the service node randomly while the transmission power remains the same.

In conclusion, the method analyzes the energy consumption of the unmanned aerial vehicle in the video local processing and the video unloading processing by simulating the influence of the weight of the unmanned aerial vehicle on the endurance time of the unmanned aerial vehicle, and proves that the video unloading processing is superior to the video local processing in the aspect of energy saving. Then, the problem of efficient and reliable calculation task allocation in the target tracking process of the unmanned aerial vehicle is researched, the calculation task of the unmanned aerial vehicle is unloaded to a ground edge node instead of a cloud server, a task allocation algorithm is designed, appropriate transmission power and service nodes are selected for the unmanned aerial vehicle by considering adjustment of transmission power of the unmanned aerial vehicle and a selection mechanism of the edge node, total cost is minimized, and the total cost is balance between energy consumption cost and time cost. Designing a simulation experiment, wherein the experimental result shows that different relative weights of energy cost and time cost correspond to different transmission power and edge node selection schemes; the performance of the algorithm provided by the invention is superior to that of other reference schemes.

The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.