Disaster area emergency communication method and system based on unmanned aerial vehicle edge computing

Abstract

The application provides a disaster area emergency communication method and system based on unmanned aerial vehicle edge computing, which comprises the following steps: based on the device positions of all target devices in the disaster area and the initial unmanned aerial vehicle deployment position, the data transmission rate between the target devices and the unmanned aerial vehicle is calculated; a task execution matrix for assisting the unmanned aerial vehicle to execute target tasks in the target devices is generated; the device information of the target devices and the task data volume of the target tasks are acquired; a time delay model and an energy consumption model of the target task execution process are constructed; a multi-objective optimization problem function is constructed; the multi-objective optimization problem function is solved to obtain an optimal unmanned aerial vehicle deployment position and an optimal task allocation strategy set; a communication connection between the unmanned aerial vehicle and the target devices is established, and the target tasks in all the target devices are processed based on the optimal task allocation strategy set. The application can obtain a Pareto optimal unmanned aerial vehicle deployment position set, thereby reducing the time delay and energy consumption of the entire emergency communication system.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of mobile edge computing, and particularly relates to a disaster area emergency communication method and system based on unmanned aerial vehicle edge computing. BACKGROUND

[0002] When natural disasters such as earthquakes, typhoons and floods occur, the power supply in the disaster area is often cut off, forcing the local servers and mobile communication base stations in the disaster area to shut down, resulting in poor communication in the disaster area and limited functions of intelligent devices. Both the reconstruction of local servers and the repair of circuits require a large amount of manpower and material resources, hindering the progress of disaster relief work. In view of this situation, mobile edge computing opens up a new way to solve the problem. Edge computing (MEC) can replace the centralized computing function of cloud servers by deploying edge servers at the network edge. Deploying edge servers in disaster areas can achieve the purpose of restoring computing power and maintaining the normal operation of intelligent devices in disaster areas, providing a large amount of computing resources. Unmanned aerial vehicles, with their flexibility, low cost and ease of deployment, are the preferred choice for carrying edge servers, enabling short-term large-scale service coverage. At the same time, compared with the original base station transmission, mobile edge computing realizes line-of-sight signal transmission to improve channel gain and reduce the energy consumption of user task offloading. The application of unmanned aerial vehicles also improves the flexibility and robustness of mobile edge computing.

[0003] Mobile edge computing is a cloud computing method that allocates local computing tasks of intelligent devices to edge nodes (unmanned aerial vehicles) for computation, and needs to consider indicators such as the task operation delay of the entire system, the overall energy consumption of the system, and data transmission security. In practical applications, the position of the unmanned aerial vehicle deployment, the motion trajectory of the unmanned aerial vehicle, the task offloading strategy, and the transmission channel allocation need to be considered. The main research directions are the optimization of average response time, the optimization of system energy consumption, and comprehensive optimization, and the methods used include convex optimization method, game theory method, intelligent optimization algorithm and reinforcement learning AI method.

[0004] In the existing technology of mobile edge computing, a computing offloading strategy based on edge-cloud cooperation is used to consider the dependency relationship and execution sequence between tasks, optimize the offloading efficiency, and minimize the weighted sum of system delay and energy consumption based on an improved hybrid particle swarm algorithm. In the above existing technology, the weighted sum of delay and energy consumption is used as the optimization target, although a relatively optimal task offloading strategy can be obtained in the end, the dimensional inconsistency of delay and energy consumption is not considered, and the result of the optimization weighted sum target function is relatively one-sided. In the actual application process, the position of the unmanned aerial vehicle needs to be deployed in advance according to the experience of the operator, and if the position of the unmanned aerial vehicle is deployed with a large deviation, the delay and energy consumption of the entire emergency communication system will be increased. SUMMARY

[0005] The application provides a disaster area emergency communication method and system based on unmanned aerial vehicle edge computing to solve the problem that the artificial deployment of unmanned aerial vehicles may increase the time delay and energy consumption of the entire emergency communication system.

[0006] In a first aspect, the application provides a disaster area emergency communication method based on unmanned aerial vehicle edge computing, which comprises the following steps:

[0007] Based on the device positions of all target devices in the target area and the initial unmanned aerial vehicle deployment position, the data transmission rate between the target devices and the unmanned aerial vehicle is calculated according to the free space path loss model, and the unmanned aerial vehicle is equipped with an edge server;

[0008] A task execution matrix is generated for the unmanned aerial vehicle to assist in executing target tasks in the target devices;

[0009] Device information of the target devices and task data volume of the target tasks are obtained;

[0010] A time delay model and an energy consumption model of the target task execution process are constructed in combination with the device information, the task data volume and the data transmission rate;

[0011] A multi-objective optimization problem function is constructed in combination with the time delay model, the energy consumption model and the task execution matrix, and the optimization target of the multi-objective optimization problem function is to make the response time of the target task execution process shortest and the total energy consumption least;

[0012] The multi-objective optimization problem function is solved by using a multi-objective genetic particle swarm nested greedy strategy algorithm to obtain an optimal unmanned aerial vehicle deployment position and an optimal task allocation matrix;

[0013] The unmanned aerial vehicle is deployed in the disaster area based on the optimal unmanned aerial vehicle deployment position;

[0014] A communication connection is established between the unmanned aerial vehicle and the target devices, and the target tasks in all the target devices are processed based on the optimal task allocation matrix.

[0015] Optionally, the data transmission rate between the target devices and the unmanned aerial vehicle is calculated based on the device positions of all target devices in the disaster area and the initial unmanned aerial vehicle deployment position according to the free space path loss model, which comprises the following steps:

[0016] The wireless channel link between all target devices in the disaster area and the unmanned aerial vehicle is evaluated according to the free space path loss model, and the power gain of the wireless channel link is calculated in combination with the device positions of the target devices and the initial unmanned aerial vehicle deployment position;

[0017] The data transmission rate between the target device and the UAV is calculated based on the power gain and according to Shannon theorem.

[0018] Optionally, the device information comprises device CPU frequency, CPU resource calculation period and data transmission power of the target device, the time delay model comprises a device time delay model when the target device locally executes the target task and a UAV time delay model when the UAV assists in executing the target task, and the energy consumption model comprises a device energy consumption model when the target device locally executes the target task and a UAV energy consumption model when the UAV assists in executing the target task.

[0019] The time delay model and the energy consumption model of the target task execution process are constructed by combining the device information, the task data volume and the data transmission rate, and the construction comprises the following steps:

[0020] The device time delay model is constructed by combining the device CPU frequency, the CPU resource calculation period and the task data volume, and an expression formula of the device time delay model is as follows:

[0021]

[0022] In the formula, T denotes device execution time of the target task locally executed by the i-th target device, V i denotes the task data volume of the i-th target device, C i denotes the CPU resource calculation period of the i-th target device, denotes the device CPU frequency of the i-th target device;

[0023] The UAV time delay model is constructed by combining the CPU resource calculation period, the task data volume and the data transmission rate, and an expression formula of the UAV time delay model is as follows:

[0024]

[0025] In the formula, T denotes UAV total execution time of the target task of the i-th target device in the whole process assisted by the j-th UAV, denotes data transmission time between the j-th UAV and the i-th target device, denotes UAV execution time of the i-th target task executed by the j-th UAV, R i，j denotes the data transmission rate, denotes UAV CPU frequency of the j-th UAV;

[0026] The device energy consumption model is constructed by combining the device's CPU frequency and the device's latency model. The expression formula for the device energy consumption model is as follows:

[0027]

[0028] In the formula: The device execution power consumption of the i-th target device performing the target task locally is represented by κ, where κ represents the power consumption factor.

[0029] The UAV energy consumption model is constructed by combining the data transmission power and the UAV latency model. The formula for expressing the UAV energy consumption model is as follows:

[0030]

[0031] In the formula: This represents the total energy consumption of the j-th drone in the overall process of assisting the i-th target device in performing the target task. This represents the data transmission energy consumption from the i-th target device to the j-th UAV. P represents the energy consumption of the j-th UAV performing the ith target task. i This represents the data transmission power of the i-th target device.

[0032] Optionally, the expression for the task execution matrix is ​​as follows:

[0033]

[0034] In the formula: Q represents the task execution matrix, m represents the number of devices in the target device, n represents the number of drones, and q i，j ∈{0,1}, when the j-th UAV assists in performing the target task in the i-th target device, q i，j =1; when the j-th UAV does not assist in performing the target task in the ith target device, q i，j =0; when the i-th target device locally executes the target task, q i，n+1 =1; when the i-th target device does not execute the target task locally, q i，n+1 =0; n = n1 + n2 + n3, where n1, n2 and n3 represent the number of drones with different computing performance.

[0035] Optionally, the multi-objective optimization problem function can be expressed as follows:

[0036]

[0037] P1 represents a first optimization objective of minimizing a response time of the target task execution process, and P2 represents a second optimization objective of minimizing a total energy consumption of the target task execution process.

[0038] Optionally, the solving the multi-objective optimization problem function by using the multi-objective genetic particle swarm nested greedy strategy algorithm to obtain the optimal UAV deployment position and the optimal task allocation matrix comprises the following steps:

[0039] The UAV deployment position of the UAV is encoded as a target particle, and a plurality of target particles are collected as a target particle swarm;

[0040] A plurality of fitness functions are determined based on the optimization objectives of the multi-objective optimization problem function;

[0041] The speed and position of each target particle are updated by using a preset update strategy, and the individual optimal solution of the target particle and the non-dominated solution set in the target particle swarm are determined in combination with the plurality of fitness functions;

[0042] The population optimal solution of the target particle swarm is obtained based on the non-dominated solution set and by using an adaptive grid method;

[0043] The speed and position of the target particle are repeatedly updated, and the individual optimal solution, the population optimal solution and the non-dominated solution set are synchronously updated until the position change difference of the target particle is less than a preset difference threshold value;

[0044] The UAV deployment positions corresponding to all target particles in the non-dominated solution set after the last update iteration are taken as a Pareto optimal UAV deployment position set;

[0045] The optimal task allocation matrix is determined by using a greedy algorithm in combination with the Pareto optimal UAV deployment position set and the data transmission rate.

[0046] Optionally, the updating the speed and position of each target particle by using a preset update strategy and determining the individual optimal solution of the target particle and the non-dominated solution set in the target particle swarm in combination with the plurality of fitness functions comprises the following steps:

[0047] The particle speed of each target particle is updated by using a preset update strategy;

[0048] The particle position corresponding to the target particle is updated based on the particle speed;

[0049] The position change of each target particle is analyzed based on the plurality of fitness functions to obtain the individual optimal solution of each target particle;

[0050] Based on the plurality of fitness functions, all the individual optimal solutions of the target particles are compared to obtain a set of non-dominated solutions in the target particle swarm.

[0051] Optionally, the repeated updating iteration of the calculation formula of the velocity and position of the target particle is as follows:

[0052]

[0053]

[0054] In the formula: denotes the particle velocity of the kth target particle after the tth updating iteration, denotes the initial particle velocity of the kth target particle at the tth updating iteration, denotes the particle position of the kth target particle after the tth updating iteration, denotes the initial particle position of the kth target particle at the tth updating iteration, ω denotes an inertia factor, c1 denotes a local velocity factor, and c2 denotes a global velocity factor.

[0055] Optionally, the combining of the optimal UAV deployment position and the data transmission rate and the determination of the optimal task allocation strategy set through the greedy algorithm comprises the following steps:

[0056] According to the delay model and the energy consumption model, device execution delay and device execution energy consumption are respectively calculated when each target device locally executes the target task;

[0057] According to the delay model and the energy consumption model, device execution delay and device execution energy consumption are respectively calculated when each target device locally executes the target task;

[0058] According to the delay model and the energy consumption model, device execution delay and device execution energy consumption are respectively calculated when each target device locally executes the target task;

[0059] For any one group of target devices and corresponding optimal UAVs, it is judged whether the device execution delay is less than the UAV execution delay;

[0060] If the device execution delay is less than the UAV execution delay, it is judged whether the device execution energy consumption is less than the UAV execution energy consumption;

[0061] If the device execution energy consumption is less than the UAV execution energy consumption, it is determined that the optimal task allocation matrix between the target device and the corresponding optimal UAV is a first task allocation matrix, and the first task allocation matrix indicates that the target task is executed locally by the target device.

[0062] if the device execution energy consumption is greater than or equal to the UAV execution energy consumption and / or the device execution time delay is greater than or equal to the UAV execution time delay, the target task is unloaded to the optimal UAV;

[0063] the number of tasks in the optimal UAV is counted, and it is determined whether the number of tasks is greater than the maximum number of tasks that the optimal UAV can run;

[0064] if the number of tasks is less than or equal to the maximum number of tasks that the optimal UAV can run, the optimal task allocation matrix between the target device and the corresponding optimal UAV is determined as a second task allocation matrix, which indicates that the optimal UAV assists in executing the target task;

[0065] if the number of tasks is greater than the maximum number of tasks that the optimal UAV can run, the optimal task allocation matrix between the target device and the corresponding optimal UAV is determined as a third task allocation matrix, which indicates that the task with the minimum data transmission rate in the target task is executed locally on the target device, and the optimal UAV assists in executing other tasks in the target task.

[0066] In a second aspect, the present application also provides a disaster area emergency communication system based on UAV edge computing, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the method as described in the first aspect when executing the computer program.

[0067] The present application has the following advantages:

[0068] First, the data transmission rate between the target device and the UAV in the disaster area is calculated, and a task execution matrix of the UAV assisting in executing the target task in the target device is generated. Then, a time delay model and an energy consumption model of the target task execution process in the target device are constructed, and a multi-objective optimization problem function is further constructed, taking the shortest response time and the least total energy consumption of the target task execution process as the optimization goal. The multi-objective genetic particle swarm nested greedy strategy algorithm is used to solve the multi-objective optimization problem function, and the optimal UAV deployment position and the optimal task allocation matrix are obtained. Finally, the UAV is deployed in the disaster area based on the optimal UAV deployment position, the communication connection between the UAV and the target device is established, and the target task in all the target devices is processed based on the optimal task allocation matrix. Compared with manually determining the deployment position of the UAV, the performance of the UAV position deployment is higher and the effect is better, so that the increase of the time delay and the energy consumption of the entire emergency communication system caused by the deviation of the deployment position of the UAV can be avoided as much as possible. BRIEF DESCRIPTION OF DRAWINGS

[0069] Figure 1 The flowchart of the disaster area emergency communication method based on the unmanned aerial vehicle edge computing in the application.

[0070] Figure 2 The model diagram of the emergency communication based on the unmanned aerial vehicle mobile edge computing in the application. DETAILED DESCRIPTION

[0071] The application discloses a disaster area emergency communication method based on unmanned aerial vehicle edge computing.

[0072] Reference Figure 1 The disaster area emergency communication method based on the unmanned aerial vehicle edge computing specifically comprises the following steps:

[0073] S101. Based on the device positions of all target devices in the target area and the initial unmanned aerial vehicle deployment position, the data transmission rate between the target devices and the unmanned aerial vehicle is calculated according to the free space path loss model.

[0074] The target area refers to an area where the power supply is cut off due to natural disasters or other special circumstances, and the local server and mobile communication base station are closed. The communication in the target area is not smooth, and the functions of intelligent devices are limited. The intelligent devices in the target area, such as mobile phones, computers, and Internet of Things devices, are target devices in the target area. Rebuilding the local server and repairing the circuit in the target area requires a lot of manpower and material resources. Mobile edge computing opens up a new way to solve the problem. Edge computing (MEC) can replace the centralized computing function of the cloud server by deploying edge servers at the network edge. Deploying edge servers in the target area can achieve the purpose of restoring computing power and maintaining the normal operation of target devices in the target area. A large amount of computing resources are provided. The unmanned aerial vehicle carries the edge server to achieve short-term large-scale service coverage. Compared with the original base station transmission, mobile edge computing realizes line-of-sight signal transmission to improve channel gain and reduce the energy consumption of user computing task offloading. The application of unmanned aerial vehicles also improves the flexibility and robustness of mobile edge computing.

[0075] Reference Figure 2 The coordinates of the target devices (MD1, MD2,..., MD n ) and the unmanned aerial vehicles (UAV1, UAV2,..., UAV n ) are defined using a three-dimensional Cartesian coordinate system, without considering the altitude of the target devices, and assuming that the unmanned aerial vehicles fly at the same horizontal height. Since the signal between MD and UAV adopts line-of-sight (LoS) propagation, the free space path loss (FSPL) model can be used to evaluate the wireless channel between the unmanned aerial vehicle and the intelligent device.

[0076] S102. Generating a task execution matrix of assisting the target devices to execute the target tasks by the UAVs.

[0077] wherein, it is assumed that all target tasks in the target devices are offloadable, thus the target task execution has two ways of completion. The first way is to execute the task by the computing power of the target device itself, which is called local execution. The rate of local execution is limited, and simply relying on local execution will lead to slow task completion, resulting in time delay and affecting the quality of service. Therefore, in the case of base station damage, the task is generally executed by relying on the UAV edge computing server to assist, which is the second execution way: UAV assisted execution. The UAV assistance process involves data transmission and processing process.

[0078] In this embodiment, a task execution matrix Q∈{0,1} m×(n+1) is generated to represent the task execution of the target tasks, and if it is UAV assisted execution, it means that all target tasks in the target device are completely offloaded to the UAV. The specific expression of the task execution matrix is as follows:

[0079]

[0080] In the formula: Q represents the task execution matrix, m represents the number of target devices, n represents the number of UAVs, q i，j ∈{0,1}, when the jth UAV assists the execution of the target task in the ith target device, q i，j =1; when the jth UAV does not assist the execution of the target task in the ith target device, q i，j =0; when the ith target device executes the target task locally, q i，n+1 =1; when the ith target device does not execute the target task locally, q i，n+1 =0; n=n1+n2+n3, n1, n2 and n3 respectively represent the number of UAVs of different computing performance, specifically, n1 represents the number of A-level UAVs, n2 represents the number of B-level UAVs, and n3 represents the number of C-level UAVs. The higher the level of the UAV, the stronger the computing performance of the UAV.

[0081] S103. Obtaining device information of the target devices and task data volume of the target tasks.

[0082] wherein, the device information includes device CPU frequency, CPU resource calculation period, data transmission power, device power, etc. of the target devices.

[0083] S104. Constructing a delay model and an energy consumption model of the target task execution process in combination with the device information, the task data volume and the data transmission rate.

[0084] The time delay of the task execution can be calculated by a time delay model. The time required for the target task execution is related to the task data volume of the task itself, the CPU resource calculation period in the execution device, and the CPU calculation frequency of the execution device. The calculation of the energy consumption of the task execution needs to consider the power of the execution device in addition to the calculation time. Since the target task execution mode includes local execution and UAV-assisted execution, two time delay models and two energy consumption models need to be constructed according to the execution device.

[0085] S105. Construct a multi-objective optimization problem function in combination with the time delay model, the energy consumption model, and the task execution matrix.

[0086] The optimization objective of the multi-objective optimization problem function is to minimize the response time and the total energy consumption of the target task execution process.

[0087] S106. Solve the multi-objective optimization problem function by using a multi-objective genetic particle swarm nested greedy strategy algorithm to obtain the optimal UAV deployment position and the optimal task allocation matrix.

[0088] Since the multi-objective optimization problem function needs to consider two variables, the UAV deployment position and the optimal task allocation strategy, the solution space of the problem will become very large if the two variables are optimized at the same time, and it is difficult to obtain the Pareto optimal solution by iteration. Therefore, the multi-objective genetic particle swarm nested greedy strategy (MOGAPSO-G) algorithm can be used to solve the multi-objective optimization problem function. In the solving process of the multi-objective genetic particle swarm nested greedy strategy algorithm, the outer layer is the multi-objective genetic particle swarm algorithm, which is used to solve the UAV deployment position. The inner layer is the greedy algorithm, which is used to obtain the optimal task allocation matrix.

[0089] S107. Deploy the UAV in the disaster area based on the optimal UAV deployment position.

[0090] S108. Establish a communication connection between the UAV and the target device, and process the target tasks in all target devices based on the optimal task allocation matrix.

[0091] The optimal task allocation matrix contains the optimal execution mode of the target tasks in all target devices, so the optimal task allocation matrix can be traversed by the UAV and the target device to make each target task in each target device have an optimal execution mode.

[0092] The implementation principle of the embodiment is as follows:

[0093] First, the data transmission rate between the target device and the unmanned aerial vehicle in the disaster area is calculated, and a task execution matrix of the unmanned aerial vehicle assisting in executing the target task in the target device is generated. Then, a time delay model and an energy consumption model of the target task execution process in the target device are constructed, and a multi-objective optimization problem function with the shortest response time and the least total energy consumption of the target task execution process as the optimization objectives is further constructed. The multi-objective genetic particle swarm nested greedy strategy algorithm is used to solve the multi-objective optimization problem function, and the optimal unmanned aerial vehicle deployment position and the optimal task allocation matrix are obtained. Finally, the unmanned aerial vehicle is deployed in the disaster area based on the optimal unmanned aerial vehicle deployment position, the communication connection between the unmanned aerial vehicle and the target device is established, and the target task in all the target devices is processed based on the optimal task allocation matrix. Compared with manually determining the deployment position of the unmanned aerial vehicle, the performance of the unmanned aerial vehicle position deployment is higher and the effect is better, so that the increase of the time delay and the energy consumption of the entire emergency communication system caused by the deviation of the deployment position of the unmanned aerial vehicle can be avoided as much as possible.

[0094] In one embodiment, the step S101 of calculating the data transmission rate between the target device and the unmanned aerial vehicle based on the device positions of all the target devices in the disaster area and the initial unmanned aerial vehicle deployment position according to the free space path loss model specifically includes the following steps:

[0095] The wireless channel link between all the target devices and the unmanned aerial vehicle in the disaster area is evaluated according to the free space path loss model, and the power gain of the wireless channel link is calculated in combination with the device positions of the target devices and the initial unmanned aerial vehicle deployment position.

[0096] The data transmission rate between the target device and the unmanned aerial vehicle is calculated based on the power gain and according to the Shannon theorem.

[0097] In this embodiment, the coordinates of the target device (MD1, MD2,..., MD n ) and the unmanned aerial vehicle (UAV1, UAV2,..., UAV n ) are defined by using a three-dimensional Cartesian coordinate system, without considering the altitude of the target device, and assuming that the unmanned aerial vehicle flies at the same horizontal height. The position coordinates of the i-th MD are The position coordinates of the j-th UAV are Since the signal between the MD and the UAV adopts line-of-sight (LoS) propagation, the free space path loss (FSPL) model is used to evaluate the wireless channel between the unmanned aerial vehicle and the intelligent device. The calculation formula of the power gain of the wireless channel link is:

[0098]

[0099] In the formula, h i，jrepresents the power gain, G0is the channel power gain of unit space distance, representing the transmission capability characteristics of the wireless channel link itself, independent of input and output, which varies with time or frequency. i，j represents the Euclidean three-dimensional space distance between the ith MD and the jth UAV.

[0100] In order to complete the data transmission as soon as possible to improve the signal-to-noise ratio of the wireless channel link, it is assumed that the data transmission power of the ith MD is constant as P i Therefore, according to the Shannon theorem, the calculation formula of the data transmission rate between the ith MD and the jth UAV is as follows:

[0101]

[0102] In the formula, R i，j represents the data transmission rate between the ith MD and the jth UAV, B represents the channel bandwidth (unit: MHz), σ 2 represents the Gaussian white noise power in the wireless channel link.

[0103] In one embodiment, the device information includes the device CPU frequency, CPU resource calculation period and data transmission power of the target device, the latency model includes a device latency model when the target device locally executes the target task and a UAV latency model when the UAV assists in executing the target task, and the energy consumption model includes a device energy consumption model when the target device locally executes the target task and a UAV energy consumption model when the UAV assists in executing the target task.

[0104] In this embodiment, step S104, i.e., constructing the latency model and the energy consumption model of the target task execution process in combination with the device information, the task data volume and the data transmission rate, specifically includes the following steps:

[0105] The device latency model is constructed in combination with the device CPU frequency, the CPU resource calculation period and the task data volume, and the expression formula of the device latency model is as follows:

[0106]

[0107] In the formula, T represents the device execution time of the ith target device locally executing the target task, V i represents the task data volume of the ith target device, C i represents the CPU resource calculation period of the ith target device, represents the device CPU frequency of the ith target device;

[0108] The process assisted by the UAV is as follows: the target device transmits the target task to the UAV through a wireless channel link, and the edge computing server carried by the UAV executes the task and transmits the result data back to the target device. Therefore, the time required for the UAV assisted execution is divided into three parts: data transmission time, UAV execution time and data transmission back time. Since the amount of data transmitted back is extremely small and can be compressed, the time and energy consumption required for this part are extremely low, and therefore this process can be ignored. In summary, the UAV delay model can be constructed by combining the CPU resource calculation period, task data volume and data transmission rate. The expression formula of the UAV delay model is as follows:

[0109]

[0110] In the formula: represents the total execution time of the jth UAV for the overall process of the target task in the ith target device, represents the data transmission time between the jth UAV and the ith target device, represents the UAV execution time of the jth UAV for the ith target task, R i，j represents the data transmission rate, represents the UAV CPU frequency of the jth UAV;

[0111] The power consumption of the CPU with frequency f can be approximated as κf 3 , and the power consumption factor κ is determined by the structure of the CPU. Therefore, the device energy consumption model is constructed by combining the device CPU frequency and the device delay model. The expression formula of the device energy consumption model is as follows:

[0112]

[0113] In the formula: represents the device execution energy consumption of the ith target device for executing the target task, and κ represents the power consumption factor;

[0114] The UAV energy consumption model is constructed by combining the data transmission power and the UAV delay model. The expression formula of the UAV energy consumption model is as follows:

[0115]

[0116] In the formula: represents the total energy consumption of the jth UAV for the overall process of the target task in the ith target device, represents the data transmission energy consumption of the target task in the ith target device to the jth UAV, represents the UAV execution energy consumption of the jth UAV for the ith target task, P i represents the data transmission power of the ith target device.

[0117] In the embodiment, the expression formula of the multi-objective optimization problem function is as follows:

[0118]

[0119] In the formula, P1 represents a first optimization objective of minimizing the response time of the target task execution process, and P2 represents a second optimization objective of minimizing the total energy consumption of the target task execution process.

[0120] The constraint conditions of the multi-objective optimization problem function are as follows:

[0121]

[0122] In the formula, C1, C2, C3, C4, and C5 are constraint conditions, wherein the constraint condition C1 and the constraint condition C2 are position constraints of the UAV, x max and y max respectively represent the maximum horizontal and vertical coordinates in which the UAV can be deployed. The constraint condition C3 indicates that the task execution matrix Q is a 0-1 matrix, and there are only two cases of execution and non-execution. The constraint condition C4 restricts the execution of the tasks, indicating that each task can be executed only once and must be executed. The constraint condition C5 restricts the execution of the UAV, and the number of tasks assisted by each UAV for execution cannot exceed the maximum number of concurrent tasks N max of the UAV.

[0123] In one embodiment, the step S106, i.e., the multi-objective genetic particle swarm nested greedy strategy algorithm is used to solve the multi-objective optimization problem function to obtain the optimal UAV deployment position and the optimal task allocation matrix, and specifically includes the following steps:

[0124] The UAV deployment position of the UAV is encoded as a target particle, and a plurality of target particles are collected as a target particle swarm;

[0125] A plurality of fitness functions are determined based on the optimization objectives of the multi-objective optimization problem function;

[0126] The speed and position of each target particle are updated by using a preset update strategy, and the individual optimal solution of the target particle and the non-dominated solution set in the target particle swarm are determined in combination with the plurality of fitness functions;

[0127] The population optimal solution of the target particle swarm is obtained based on the non-dominated solution set and by using the self-adaptive grid method;

[0128] The speed and position of the target particle are repeatedly updated, and the individual optimal solution, the population optimal solution, and the non-dominated solution set are synchronously updated until the position change difference of the target particle is less than a preset difference threshold value;

[0129] The deployment position of each unmanned aerial vehicle corresponding to all target particles in the non-dominated solution set after the last update iteration is taken as a set of Pareto optimal unmanned aerial vehicle deployment positions;

[0130] The optimal task allocation strategy set is determined by combining the set of optimal Pareto optimal unmanned aerial vehicle deployment positions and the data transmission rate and through a greedy algorithm.

[0131] In the embodiment, the outer algorithm of the multi-objective genetic particle swarm nested greedy strategy algorithm is a multi-objective genetic particle swarm algorithm, which is used to solve the unmanned aerial vehicle deployment position, and the inner algorithm is a greedy algorithm, which is used to obtain the optimal task allocation matrix. The multi-objective genetic particle swarm algorithm is a swarm intelligence algorithm simulating the foraging of birds, and is an embodiment of biological individual intelligence and swarm intelligence. The algorithm updates the speed and position of the particles through three aspects of information, i.e., individual inertia, self-cognition and social cognition, so that the particle swarm eventually converges to the optimal solution. The particle swarm algorithm searches a continuous space, and is suitable for solving continuous optimization problems. The multi-objective genetic particle swarm algorithm needs to construct an external storage area to retain all non-dominated solutions, and the elite individuals in the external storage area guide the entire particle swarm to constantly approach the Pareto frontier.

[0132] Referring to Table 1, in order to obtain the Pareto optimal solution of the unmanned aerial vehicle deployment coordinates, the position coordinates of the unmanned aerial vehicles are encoded into a target particle. The actual coordinates of each unmanned aerial vehicle are (x i , y i , z0), and each particle represents a deployment strategy of the unmanned aerial vehicle. The number of the unmanned aerial vehicle also corresponds to the heterogeneous unmanned aerial vehicles (three levels of A, B and C) with different computing capabilities, so as to realize multi-level computing.

[0133] Table 1 is a coding table for encoding the position coordinates of the unmanned aerial vehicles into particles

[0134]

[0135] The multi-objective optimization problem function contains multiple fitness functions. In the embodiment, the two fitness functions are the time delay and the energy consumption in the model, which are two optimization objective function values. When the time delay and the energy consumption of a target particle are not higher than those of other particles, the target particle is said to be not dominated by other target particles.

[0136] In one embodiment, the step of updating the speed and position of each target particle through a preset update strategy and determining the individual optimal solution of the target particle and the non-dominated solution set in the target particle swarm in combination with multiple fitness functions includes the following steps:

[0137] The particle speed of each target particle is updated by using a preset update strategy;

[0138] The particle position of the corresponding target particle is updated based on the particle speed.

[0139] The position changes of each target particle are analyzed based on multiple fitness functions to obtain the individual optimal solution for each target particle;

[0140] By comparing the individual optimal solutions of all target particles using multiple fitness functions, the non-dominated solution set in the target particle swarm is obtained.

[0141] In this embodiment, due to the existence of two fitness functions, the individual optimal solution pBest is not simply a single target particle, but rather a set containing multiple target particles. The target particles in the set of individual optimal solutions are not dominated by the target particles formed by the historical positions traversed by that target particle, and the number of target particles in the set of individual optimal solutions is p.

[0142] For a single particle, in each iteration, the updated particle position needs to be compared with the previous optimal solution pBest0. If the updated particle position dominates the previous optimal solution pBest0, then the optimal solution for this iteration is the updated particle position. If the previous optimal solution pBest0 dominates the updated particle position, then the optimal solution remains pBest0. If the updated particle position and pBest0 do not dominate each other, then each is assigned a 50% selection probability, and either the updated particle position or pBest0 is randomly selected as the optimal solution for this iteration based on the selection probability. Furthermore, in each iteration, the optimal solutions of all target particles are compared to obtain the set of non-dominated solutions.

[0143] In one implementation, the velocity and position of the target particle are repeatedly updated and iterated in the step, and the individual optimal solution, the population optimal solution, and the set of non-dominated solutions are updated simultaneously until the difference in the position change of the target particle is less than a preset difference threshold. The calculation formula for repeatedly updating and iterating the velocity and position of the target particle is as follows:

[0144]

[0145]

[0146] In the formula: This represents the particle velocity of the k-th target particle after the t-th update iteration. This represents the initial particle velocity of the k-th target particle during the t-th update iteration. This represents the position of the k-th target particle after the t-th update iteration. Let ω represent the initial particle position of the k-th target particle in the t-th update iteration, ω represent the inertia factor, c1 represent the local velocity factor, and c2 represent the global velocity factor.

[0147] In the embodiment, the target particles contained in the set of population optimal solution gBest are not dominated by the target particles constituted by all the historical positions experienced by the entire target particle population, and the number of target particles in the set of population optimal solution gBest is q. The multi-objective genetic particle swarm algorithm cannot compare the optimality of the target particles in the set, and needs to select the leaders in the individual optimal solution pBest and the population optimal solution gBest at each population update.

[0148] The set of Pareto optimal solutions currently searched by the target particle population is archived in the external storage area, and the population optimal solution gBest can be obtained by the adaptive grid method. The particle space is 2n-dimensional, and the space is reduced to the particle space in the external storage area and is divided into LxL layers. That is, the maximum value max and the minimum value min of the target particles in the external storage area are found in each dimension, and the coordinate axis is evenly divided into L segments in [min, max]. In the L 2n In the particle space in the L layer, the particle crowding density is the number of particles, and the smaller the number represents the less crowded. The population optimal solution gBest is selected from the external storage area by the roulette wheel method.

[0149] In one embodiment, if a target particle in the target particle population obtained in each round of update is not dominated by all the target particles in the external storage area, it is supplemented into the external storage area. In order to solve the overflow problem of the external storage area, the external storage area needs to be sheared. When the number of target particles in the external storage area exceeds the number limit, the crowding density of each target particle in the external storage area needs to be calculated according to the adaptive grid method, a target particle in the grid with the largest crowding density is randomly deleted and the crowding density is updated, and the external storage area is repeatedly sheared until the size requirement is met. In this way, while maintaining the archive size of the external storage area, the search of the area with low crowding degree can be increased. After continuous update iteration, all particle populations will eventually converge to several different positions, and these particles are not dominated by other particles in the searched space, which are called the Pareto optimal solution set of the problem.

[0150] In one embodiment, due to the large number of Pareto optimal solutions of the multi-objective optimization problem function, some Pareto optimal solutions may be missed in the search process of the multi-objective genetic particle swarm algorithm. In order to increase the diversity of the algorithm search, a crossover mutation operator is introduced: a target particle in the population optimal solution gBest is randomly selected as a parent particle, a point position of the parent particle is randomly selected, the parent particle and the corresponding target particle in the target particle group are replaced according to the point position, and the value of the replaced target particle in the target particle group is mutated based on a preset mutation probability. The value of the value mutation is a random number between the minimum value and the maximum value of the target particle.

[0151] In one embodiment, the step of combining the optimal UAV deployment position and the data transmission rate and determining the optimal task allocation matrix by the greedy algorithm comprises the following steps:

[0152] According to the time delay model and the energy consumption model, the device execution time delay and the device execution energy consumption of each target device when executing the target task locally are calculated respectively;

[0153] According to the time delay model and the energy consumption model, the device execution time delay and the device execution energy consumption of each target device when executing the target task locally are calculated respectively;

[0154] According to the time delay model and the energy consumption model, the device execution time delay and the device execution energy consumption of each target device when executing the target task locally are calculated respectively;

[0155] For any group of target devices and corresponding optimal UAVs, it is determined whether the device execution time delay is less than the UAV execution time delay;

[0156] If the device execution time delay is less than the UAV execution time delay, it is determined whether the device execution energy consumption is less than the UAV execution energy consumption;

[0157] If the device execution energy consumption is less than the UAV execution energy consumption, the optimal task allocation matrix between the target device and the corresponding optimal UAV is determined as a first task allocation matrix, and the first task allocation matrix indicates that the target task is executed locally at the target device;

[0158] If the device execution energy consumption is greater than or equal to the UAV execution energy consumption and / or the device execution time delay is greater than or equal to the UAV execution time delay, the target task is unloaded to the optimal UAV;

[0159] The number of tasks of the existing tasks in the optimal UAV is counted, and it is determined whether the number of tasks is greater than the maximum number of executable tasks preset by the optimal UAV;

[0160] If the number of tasks is less than or equal to the maximum number of executable tasks, the optimal task allocation matrix between the target device and the corresponding optimal UAV is determined as a second task allocation matrix, and the second task allocation matrix indicates that the optimal UAV assists in executing the target task;

[0161] If the number of tasks is greater than the maximum number of executable tasks preset by the optimal UAV, the optimal task allocation matrix between the target device and the corresponding optimal UAV is determined as a third task allocation matrix, and the third task allocation matrix indicates that the task with the minimum data transmission rate in the target task is executed locally at the target device, and the optimal UAV assists in executing other tasks in the target task.

[0162] In the embodiment, the optimal unmanned aerial vehicle deployment position has been obtained by the outer algorithm of the multi-objective genetic particle swarm nested greedy strategy algorithm, so the optimal task allocation matrix of the target task execution process in the target device can be further determined by the inner algorithm (greedy algorithm). First, the optimal unmanned aerial vehicle with the maximum data transmission rate is found for each target device, and then the delay and energy consumption of the optimal unmanned aerial vehicle for assisting in executing the target task offloading are calculated after the target task in the target device is allocated to the optimal unmanned aerial vehicle. If the delay and energy consumption of the target device for locally executing the target task are both less than the delay and energy consumption of the optimal unmanned aerial vehicle for assisting in executing the target task, the optimal task allocation matrix of the target task in the target device and the optimal unmanned aerial vehicle indicates that the target task is executed locally in the target device. In addition, since the task processing capacity of the unmanned aerial vehicle is limited, if the number of tasks of the unmanned aerial vehicle for the target device to perform task offloading is greater than the maximum number of executable tasks, the target task with the minimum data transmission rate among the tasks offloaded to the unmanned aerial vehicle is found, and the target task is executed locally in the target device without being offloaded to the unmanned aerial vehicle.

[0163] The application further discloses an emergency communication system in a disaster area based on unmanned aerial vehicle edge computing, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the emergency communication method in the disaster area based on unmanned aerial vehicle edge computing in any of the above embodiments when executing the computer program.

[0164] The implementation principle of the embodiment is as follows:

[0165] Through the calling of the program, the data transmission rate between the target device and the unmanned aerial vehicle in the disaster area is first calculated, and a task execution matrix of the unmanned aerial vehicle for assisting in executing the target task in the target device is generated. Then, a delay model and an energy consumption model of the target task execution process in the target device are constructed, and a multi-objective optimization problem function with the shortest response time and the least total energy consumption of the target task execution process as the optimization goal is further constructed. The multi-objective genetic particle swarm nested greedy strategy algorithm is used to solve the multi-objective optimization problem function, and the optimal unmanned aerial vehicle deployment position and the optimal task allocation strategy set are obtained. Finally, the unmanned aerial vehicle is deployed in the disaster area based on the optimal unmanned aerial vehicle deployment position, the communication connection between the unmanned aerial vehicle and the target device is established, and the target task in all target devices is processed based on the optimal task allocation strategy set. Compared with manually determining the deployment position of the unmanned aerial vehicle, the accuracy of the deployment position of the unmanned aerial vehicle is higher, so that the increase of the delay and energy consumption of the entire emergency communication system caused by the deviation of the deployment position of the unmanned aerial vehicle can be avoided as much as possible.

[0166] Those skilled in the art should understand that the above discussion of any of the embodiments is merely exemplary and is not intended to be limiting of the scope of protection as set forth in the appended claims; the above embodiments or technical features among different embodiments can also be combined, steps can be implemented in any order, and there are many other changes, such as the above-mentioned different aspects of one or more embodiments of the present application, which are not provided in detail for the sake of brevity.

[0167] One or more embodiments of the present application are intended to cover all such alternatives, modifications, and variations as fall within the broad scope of the application. Accordingly, any and all such alternatives, modifications, equivalents, improvements and the like as are deemed by those of ordinary skill in the art to fall within the spirit and scope of one or more embodiments of the present application are intended to be included within the scope of the present application.

Claims

1. A method for emergency communication in disaster area based on UAV edge computing, characterized in that, The method comprises the following steps: Based on the device location of all target devices in the target area and the initial UAV deployment location, and according to the free space path loss model, the data transmission rate between the target device and the UAV is calculated, and the UAV is equipped with an edge server; A task execution matrix is generated for the UAV to assist in executing target tasks in the target device; Device information of the target device and task data volume of the target task are obtained; A time delay model and an energy consumption model of the target task execution process are constructed in combination with the device information, the task data volume and the data transmission rate; A multi-objective optimization problem function is constructed in combination with the time delay model, the energy consumption model and the task execution matrix, and the optimization target of the multi-objective optimization problem function is to make the response time of the target task execution process shortest and the total energy consumption least; The multi-objective optimization problem function is solved by using a multi-objective genetic particle swarm nested greedy strategy algorithm to obtain an optimal UAV deployment location and an optimal task allocation matrix; The UAV is deployed in the disaster area based on the optimal UAV deployment location; A communication connection is established between the UAV and the target device, and the target tasks in all the target devices are processed based on the optimal task allocation matrix. 2.The UAV-based edge computing enabled emergency communication method in a disaster area according to claim 1, wherein, Based on the device location of all target devices in the target area and the initial UAV deployment location, and according to the free space path loss model, the data transmission rate between the target device and the UAV is calculated, which comprises the following steps: According to the free space path loss model, the wireless channel link between all target devices in the disaster area and the UAV is evaluated, and the power gain of the wireless channel link is calculated in combination with the device location of the target device and the initial UAV deployment location; Based on the power gain and according to Shannon's theorem, the data transmission rate between the target device and the UAV is calculated. 3.The UAV-based edge computing enabled emergency communication method in a disaster area according to claim 1, wherein, The device information comprises device CPU frequency, CPU resource calculation period and data transmission power of the target device, the time delay model comprises a device time delay model when the target device executes the target task locally and a UAV time delay model when the UAV assists in executing the target task, and the energy consumption model comprises a device energy consumption model when the target device executes the target task locally and a UAV energy consumption model when the UAV assists in executing the target task; The time delay model and the energy consumption model of the target task execution process are constructed in combination with the device information, the task data volume and the data transmission rate, which comprises the following steps: The device time delay model is constructed in combination with the device CPU frequency, the CPU resource calculation period and the task data volume, and the expression formula of the device time delay model is as follows: In the formula: represents the device execution time of the i-th target device locally executing the target task, represents the task data volume of the i-th target device, represents the CPU resource calculation period of the i-th target device, represents the device CPU frequency of the i-th target device; The UAV time delay model is constructed in combination with the CPU resource calculation period, the task data volume and the data transmission rate, and the expression formula of the UAV time delay model is as follows: In the formula: , , represents the total execution time of the jth unmanned aerial vehicle in assisting the execution of the ith target task in the target device as a whole, represents the data transmission time between the jth unmanned aerial vehicle and the ith target device, represents the unmanned aerial vehicle execution time of the jth unmanned aerial vehicle in executing the ith target task, represents the data transmission rate, represents the unmanned aerial vehicle CPU frequency of the jth unmanned aerial vehicle; The device energy consumption model is constructed in combination with the device CPU frequency and the device time delay model, and the expression formula of the device energy consumption model is as follows: In the formulae: represents the device execution energy consumption of the ith target device locally performing the target task, represents the power consumption factor; The UAV energy consumption model is constructed in combination with the data transmission power and the UAV delay model, and an expression formula of the UAV energy consumption model is as follows: In the formula: , , represents the total energy consumption of the jth unmanned aerial vehicle in the overall process of assisting the ith target device to perform the target task, represents the data transmission energy consumption of the ith target device to the jth unmanned aerial vehicle, represents the unmanned aerial vehicle execution energy consumption of the jth unmanned aerial vehicle in performing the ith target task, represents the data transmission power of the ith target device. 4.The UAV edge computing based emergency communication method in disaster area according to claim 3, characterized in that, An expression of the task execution matrix is as follows: In the formula, Q represents the task execution matrix, m represents the number of target devices, n represents the number of UAVs, When the jth UAV assists in executing the target task in the ith target device, ; when the jth unmanned aerial vehicle does not assist in performing the target task in the ith target device, ; when the ith target device locally executes the target task, ; when the ith target device does not locally execute the target task, respectively represent the number of the UAVs of the UAVs with different computing performance.​​​​ 5.The UAV edge computing based emergency communication method in disaster area according to claim 4, characterized in that, An expression formula of the multi-objective optimization problem function is as follows: In the formulae: represents a first optimization objective of minimizing the response time of the target task execution process, represents a second optimization objective of minimizing the total energy consumption of the target task execution process. 6.The UAV edge computing based emergency communication method in a disaster area according to claim 1, wherein, The multi-objective optimization problem function is solved by using the multi-objective genetic particle swarm nested greedy strategy algorithm to obtain the optimal UAV deployment position and the optimal task allocation matrix, and the method comprises the following steps: The UAV deployment position of the UAV is encoded into a target particle, and a plurality of target particles are collected into a target particle swarm; A plurality of fitness functions are determined based on the optimization objectives of the multi-objective optimization problem function; The speed and position of each target particle are updated iteratively by using a preset update strategy, and the individual optimal solution of the target particle and the non-dominated solution set in the target particle swarm are determined in combination with the plurality of fitness functions; The population optimal solution of the target particle swarm is obtained based on the non-dominated solution set and by using an adaptive grid method; The speed and position of the target particle are repeatedly updated iteratively, and the individual optimal solution, the population optimal solution and the non-dominated solution set are synchronously updated until the position change difference of the target particle is less than a preset difference threshold value; The UAV deployment positions corresponding to all target particles in the non-dominated solution set after the last update iteration are taken as a Pareto optimal UAV deployment position set; An optimal task allocation matrix is determined in combination with the Pareto optimal UAV deployment position set and the data transmission rate by using a greedy algorithm. 7.The UAV edge computing based emergency communication method in disaster area according to claim 6, characterized in that, The speed and position of each target particle are updated iteratively by using a preset update strategy, and the individual optimal solution of the target particle and the non-dominated solution set in the target particle swarm are determined in combination with the plurality of fitness functions, and the method comprises the following steps: The particle speed of each target particle is updated by using a preset update strategy; The particle position of the target particle is updated based on the particle speed; The position change of each target particle is analyzed based on the plurality of fitness functions to obtain the individual optimal solution of each target particle; The individual optimal solutions of all target particles are compared based on the plurality of fitness functions to obtain the non-dominated solution set in the target particle swarm. 8.The UAV edge computing based emergency communication method in disaster area according to claim 6, characterized in that, The calculation formula of repeatedly updating the speed and position of the target particle is as follows: wherein: represents the particle velocity of the kth target particle after the tth update iteration, represents the initial particle velocity of the kth target particle at the tth update iteration, represents the particle position of the kth target particle after the tth update iteration, represents the initial particle position of the kth target particle at the tth update iteration, represents an inertial factor, represents a local velocity factor, represents a global velocity factor; The is based on a plurality of fitness functions respectively analyzing the position changes of each target particle to obtain an individual optimal solution of each target particle; the individual optimal solution is not simply one target particle, but a set containing a plurality of target particles; the target particles contained in the set of individual optimal solutions are not dominated by the target particles constituted by the historical positions experienced by the target particles, and the number of target particles in the set of individual optimal solutions is p; The is based on non-dominated solution set and obtains the population optimal solution of the target particle group by adaptive grid method; the population optimal solution in the set is not dominated by the target particles formed by all historical positions experienced by the entire target particle group, and the population optimal solution in the set is q. 9.The UAV edge computing based emergency communication method in disaster area according to claim 6, wherein, An optimal task allocation matrix is determined in combination with the optimal UAV deployment position and the data transmission rate by using a greedy algorithm, and the method comprises the following steps: The device execution delay and the device execution energy consumption when each target device locally executes the target task are calculated according to the delay model and the energy consumption model; The optimal UAV corresponding to each target device is screened out in combination with the optimal UAV deployment position and the data transmission rate; The UAV execution delay and the UAV execution energy consumption when each optimal UAV assists in executing the target task of the corresponding target device are calculated according to the delay model and the energy consumption model; For any group of target devices and corresponding optimal UAVs, it is judged whether the device execution delay is less than the UAV execution delay; if the device execution latency is less than the UAV execution latency, determining whether device execution energy consumption is less than UAV execution energy consumption; if the device execution energy consumption is less than the UAV execution energy consumption, determining that an optimal task allocation matrix between the target device and the optimal UAV is a first task allocation matrix, the first task allocation matrix indicating that the target task is executed locally on the target device; if the device execution energy consumption is greater than or equal to the UAV execution energy consumption and / or the device execution latency is greater than or equal to the UAV execution latency, offloading the target task to the optimal UAV; counting a number of tasks of tasks already in the optimal UAV, and determining whether the number of tasks is greater than a maximum number of tasks that the optimal UAV can run; if the number of tasks is less than or equal to the maximum number of tasks that the optimal UAV can run, determining that the optimal task allocation matrix between the target device and the optimal UAV is a second task allocation matrix, the second task allocation matrix indicating that the optimal UAV assists in executing the target task; if the number of tasks is greater than the maximum number of tasks that the optimal UAV can run, determining that the optimal task allocation matrix between the target device and the optimal UAV is a third task allocation matrix, the third task allocation matrix indicating that a task with the smallest data transmission rate in the target task is executed locally on the target device, and the optimal UAV assists in executing other tasks in the target task.

10. A disaster area emergency communication system based on UAV edge computing, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that, The processor, when executing the computer program, implements the method of any one of claims 1-9.

Images