A high-security offloading energy-efficient method using physical layer security technology in UAV-MEC environment

Abstract

A high-security offloading energy efficiency method using physical layer security technology in the UAV-MEC environment is proposed. A new drone edge offloading scenario is constructed. The goal is to maximize the ratio of the system's secure communication transmission rate to the energy consumption of legitimate drones. The drone maximum confidentiality energy efficiency model is established. Under the constraints of given latency, MEC server CPU computing frequency and offloading rate, the trajectory of legitimate drones, CPU computing frequency allocation and computing task offloading strategy are jointly optimized. At the same time, the secure communication in the UAV-MEC scenario is improved from the perspective of physical layer security. A global optimization algorithm based on block coordinate descent is constructed to solve the optimization problem. The optimization problem is decoupled into three sub-problems. The non-convex sub-problem is approximated as a local convex optimization problem by introducing slack variables. The convex problem is solved by multiple iterations using the CVX tool to obtain the approximate optimal solution of the system. This method fully considers the relationship between drone secure communication and drone energy consumption, and improves the overall Qos of the system.

Description

Technical Field

[0001] The present invention relates to the field of computer wireless communication technology, and in particular to a high-security unloading energy efficiency method using physical layer security technology in a UAV-MEC environment. Background Art

[0002] With the popularization of mobile terminal devices and the rapid development of mobile communication technology, computing-intensive services have placed high demands on the computing and storage capabilities of terminals. Therefore, many researchers have considered introducing the idea of ​​cloud computing into mobile communication networks, using mobile cloud computing to accelerate the operation and processing of computing-intensive services to improve user experience. In a typical mobile cloud computing network, cloud servers are generally deployed in the core network far away from users, and offloading services to cloud servers often faces large transmission delays. In view of this, the European Telecommunications Standards Institute ETSI Industry Specification Group ISG launched Mobile Edge Computing MEC as a means of extending intelligence to the edge of the network and higher processing and storage capabilities, achieving better computing and processing service effects.

[0003] Although the edge computing system assisted by UAV has great advantages, the communication problem between UAV and ground has to be considered. As we all know, due to the broadcast nature of wireless transmission, the communication between UAV and ground users can be easily heard by nearby malicious attackers.

[0004] From the perspective of ensuring physical layer security and saving energy consumption, the goal of the drone edge computing system is to maximize the confidentiality energy efficiency or security energy efficiency, that is, the ratio between the confidentiality capacity and the system energy consumption in the system. Confidentiality capacity, or security capacity, is defined as the difference between the channel capacity of the main link and the channel capacity of the eavesdropping link. If the confidentiality capability drops below zero, the transmission from the source to the destination becomes unsafe, and the eavesdropper can intercept the source transmission. In order to prevent eavesdroppers from stealing data during task offloading, the drone itself must have the ability to transmit interference signals and choose a better channel environment during data offloading. However, due to the limited power resources stored in the drone itself, there is a contradiction between the anti-eavesdropping power consumption and the amount of data offloading, that is, if the drone wants to obtain a large confidential communication offloading rate, it must consume more interference power. Therefore, it is of certain significance to study the problem of secure communication and energy efficiency optimization strategy in drone-assisted edge computing. Summary of the invention

[0005] In response to the problems existing in the above-mentioned background technology, the present invention proposes a high-security offloading energy efficiency method using physical layer security technology in a UAV-MEC environment. In an unmanned aerial vehicle-assisted edge computing environment where eavesdroppers are present, physical layer security technology is introduced to balance the secure communication transmission rate and the energy consumption of the unmanned aerial vehicle in the process of offloading ground terminal computing tasks to the edge server carried by the unmanned aerial vehicle, thereby achieving high-security offloading energy efficiency and improving the overall operating efficiency of the system.

[0006] A high-security offloading energy efficiency method using physical layer security technology in a UAV-MEC environment comprises the following steps:

[0007] Step 1, establish an edge unloading scenario with eavesdropping drones and legitimate drones;

[0008] Step 2, modeling the above edge offloading model as a formulated optimization problem through a mathematical model;

[0009] Step 3: The established optimization problem is divided into two modes according to whether the amount of ground offloaded user task data can meet the task delay requirements: local computing mode and MEC partial offload computing mode. If the ground terminal user can complete the computing task within the delay requirement, the local computing mode is selected; otherwise, the MEC partial offload mode is selected.

[0010] Step 4: The optimization problem in the MEC partial offloading mode is solved using the block coordinate algorithm, and it is decoupled into three sub-optimization problems: the non-convex sub-problem of legal UAV trajectory optimization, the sub-problem of UAV CPU computing frequency allocation, and the sub-problem of data offloading rate optimization.

[0011] Step 5, introduce slack variables into the non-convex sub-problem of UAV trajectory optimization to perform local convex approximation processing to solve the approximate optimal operation trajectory of the system;

[0012] Step 6: Transform the objective function, remove the constraints and variables irrelevant to the current sub-problem, and solve the sub-problem of the CPU calculation frequency allocation and the data offloading rate of the drone;

[0013] Step 7: Use the CVX tool to find local approximate solutions to the above three sub-problems respectively, and iterate the three local solutions to obtain the approximate optimal solution of the system as a whole, obtain the optimal flight path trajectory of the UAV, the optimal CPU computing frequency allocation of the UAV to each ground unloading user, and the computing task data unloading rate.

[0014] Furthermore, in step 1, the edge offloading scenario with eavesdropping drones and legitimate drones is established, including the following information: N ground terminal users, E eavesdropping drones, and a legitimate drone equipped with an edge server; the location of the i-th user is represented as A legal drone must be fixed at a height H above the ground. u , its horizontal position is express.

[0015] Further, in step 2, the established optimization problem specifically includes an objective function and constraint conditions;

[0016] The objective function is:

[0017]

[0018] In formula (1), η is defined as the maximum confidentiality energy efficiency of the system, and the system secure communication transmission rate is Defined as the data transmission rate r from ground user i to the legitimate UAV link iu Maximum data transmission rate of the link from user i to the eavesdropping drone The difference Where E i represents the total energy consumption of legitimate drone service user i, including the energy consumption of launching jamming and eavesdropping drones Communication transmission power consumption generated during receiving unloaded users and the computational energy consumed by offloading user data to legitimate drones

[0019] q and f are the flight trajectory of the drone and the calculation frequency assigned to each user, respectively. The variable ξ = [0, 1] is defined as the offloading coefficient for the user to establish an offloading connection with a legitimate drone;

[0020] The constraints include:

[0021] The latency requirement of user i in local computing mode should meet T i loc ≤T, where T i loc represents the local computing delay, and T is the maximum delay that the terminal task can tolerate;

[0022] User i chooses to partially uninstall the content Where, T i ul 、T i c They represent the offloading delay and the computation delay of the legitimate drone edge server, respectively. The delay transmitted from the drone to the user is negligible.

[0023] In edge offloading mode, the CPU computing frequency allocation of the drone must meet In the formula, Indicates the total CPU computing frequency of the drone.

[0024] Further, in step 4, the optimization problem in the MEC partial offloading mode is solved using a block coordinate algorithm, and is decoupled into three sub-optimization problems. The specific steps are as follows:

[0025] a. According to the block coordinate descent method, only the constraints related to the UAV trajectory are retained, and the UAV trajectory sub-optimization problem is expressed as:

[0026]

[0027] b. Only the constraints related to the UAV frequency allocation are retained, and the UAV frequency allocation sub-optimization problem is expressed as:

[0028]

[0029] c. According to the block coordinate descent method, only the constraints related to the drone unloading rate are retained. The drone unloading rate sub-problem is expressed as:

[0030]

[0031] Furthermore, in step 5, the slack variables are introduced to perform local convex approximation on the UAV trajectory optimization problem. The specific method is:

[0032] Introducing auxiliary variable φ iu and ε ie The variable r iu and The relaxation is the following inequality:

[0033]

[0034]

[0035] Auxiliary variable φ iu Yes iu The lower bound of ε ie yes The upper bound of .

[0036] Furthermore, in step 7, the CVX tool is used to find local approximate solutions to the three sub-problems, and the three local solutions are iterated to obtain the approximate optimal solution of the system as a whole. The specific method is:

[0037] When solving the UAV trajectory subproblem, the initial values ​​of each variable are given and substituted into the local approximate value of the i-th iteration of the trajectory. Then, the UAV trajectory value of the i-th iteration and other initial values ​​are substituted into the frequency optimization subproblem to solve the approximate solution of the frequency allocation subproblem under the i-th iteration. Then, the UAV trajectory solution and frequency solution of the i-th iteration are substituted into the approximate solution of the unloading rate subproblem. The number of iterations is continuously increased until the difference between the objective functions of the two iterations is within the range of [-0.01, 0.01]. A high unloading energy efficiency method can be obtained and the process is terminated.

[0038] The beneficial effects of the present invention are: jointly optimizing the drone trajectory planning, drone CPU computing frequency allocation and unloading rate allocation strategy in drone-assisted edge computing in an eavesdropping environment, balancing the relationship between the secure communication transmission rate and energy consumption in the system, so that the system as a whole has the maximum secure communication energy efficiency ratio; compared to simply pursuing the maximum secure communication transmission rate during the unloading process, this method takes into account the system energy consumption. From the perspective of the system as a whole, it can improve unnecessary system energy waste while balancing a certain secure communication transmission rate, thereby improving the overall operation efficiency of the system. In addition, compared with the solution that simply pursues the minimum system energy consumption, considering the secure unloading data rate will further improve the security in the system. In summary, this method fully considers the relationship between drone energy consumption and secure communication transmission rate, and has the effect of improving the overall Qos of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a system model diagram of the high-security unloading energy efficiency method in an embodiment of the present invention.

[0040] Figure 2 Schematic diagram of a high-security unloading energy efficiency method in an embodiment of the present invention DETAILED DESCRIPTION

[0041] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings.

[0042] As shown in the figure, the high-security offloading energy-efficiency method using physical layer security technology in the UAV-MEC environment includes the following specific steps:

[0043] Step 1: Establish an edge unloading scenario with eavesdropping drones and legitimate drones.

[0044] Step 2: Model the above edge offloading model as a formulated optimization problem through a mathematical model.

[0045] Step 3: The established optimization problem is divided into two modes according to whether the amount of ground offloaded user task data can meet the task delay requirements: local computing mode and MEC partial offload computing mode. If the ground terminal user can complete the computing task within the delay requirement, the local computing mode is selected. Otherwise, the MEC partial offload mode is selected.

[0046] Step 4: The optimization problem in the MEC partial offloading mode is solved using the block coordinate algorithm, and it is decoupled into three sub-optimization problems: the legal UAV trajectory planning sub-problem, the UAV frequency allocation sub-problem, and the offloading rate optimization sub-problem.

[0047] Step 5: Introduce slack variables into the non-convex sub-problem of UAV trajectory optimization for local convex approximation processing to solve the system's approximately optimal operation trajectory.

[0048] Step 6: Transform the objective function, remove the constraints and variables that are not related to the current sub-problem, and solve the drone CPU computing frequency allocation sub-problem and data offloading rate sub-problem.

[0049] Step 7: Use the CVX tool to find local approximate solutions to the above three sub-problems respectively, and iterate the three local solutions to obtain the approximate optimal solution of the system as a whole, obtain the optimal flight path trajectory of the UAV, the optimal CPU frequency allocation of the UAV to each ground unloading user, and calculate the task data unloading rate.

[0050] like Figure 1 As mentioned above, for step 1: Figure 1The system model of the present invention is presented, including N ground terminal users, E eavesdropping drones and a legitimate drone equipped with an edge server. Each user terminal has a certain computing power to meet the local requirements for performing some simple tasks. The drone equipped with the edge server has strong computing power. The ground terminal user will unload data to the legitimate drone, and the data in the process will be eavesdropped by the nearby eavesdropping drone. The sets of ground users and eavesdropping drones are defined as N and E respectively. Here we assume that the legitimate drone knows the exact location of the ground user and the eavesdropping drone, and there is no communication between different eavesdropping drones. Here each user has a set of tasks to process. Due to limited local computing power and latency requirements, users can process their tasks locally or offload part of the tasks to the legitimate drone for processing. In the presence of an eavesdropping drone, the legitimate drone operating in full-duplex mode is equipped with two antennas, one of which is used to receive the unloading signal from the unloading user, and the other is used to send interference signals to the eavesdropping drone. The ground user and the eavesdropping drone are equipped with one antenna for transmission and eavesdropping respectively. Here a multi-access channel is considered, and all ground users can use the same channel to send their signals at the same time. When a legitimate drone emits electromagnetic waves to interfere with nearby eavesdroppers, the electromagnetic waves will also interfere with themselves, generating self-interference signals.

[0051] For step 2, a mathematical model is used to model it into a formula optimization problem. Based on the above description, the following mathematical model can be established first:

[0052] (a) Communication model

[0053] For the air-to-ground channel, assuming that the Doppler shift in communication can be compensated by the receiver, the channel quality depends on the link between the drone and the user. Since the drone links are all Loss (line-of-sight links), that is, the channel gain obeys the free space loss model, the channel power gain between the i-th user and the legitimate user can be expressed as shown in the following formula:

[0054]

[0055] Among them, q is the flight trajectory of the UAV, d i Refers to the Euclidean distance between user terminal k and the drone in the nth frame. β1 refers to the received power when the reference distance is 1m and the transmission power is 1W. Here β1 is expressed as: Here t and g r They are the transmission gain of the ground user antenna and the receiving gain of the drone antenna respectively. λ represents the wavelength of the transmission signal. Assume that the e-th eavesdropper drone is located at a fixed height H e , horizontal coordinate It is completely known on the legitimate drone. Then the channel gain from the i-th user to the e-th eavesdropping drone can be expressed as:

[0056]

[0057] Similarly, in the air-to-air channel between the legitimate drone and the eavesdropper drone, the channel power gain between the eavesdropper drone and the legitimate drone can be expressed as:

[0058]

[0059] in, β2 refers to the channel gain between the legitimate UAV and the eavesdropping UAV when the reference distance is 1m.

[0060] Assuming that the tasks of the user terminal can be divided, the computing tasks can be assigned to the user's local and legal drone MEC servers for execution. The drone cannot perform data computing tasks at the same time during the data offloading process. Since different ground users need to calculate different amounts of data, some users need to offload tasks to the MEC carried by legal drones, while other users can perform computing tasks locally while meeting local computing latency requirements. It is defined as the offloading coefficient for the user to establish an offloading connection with a legitimate drone. The drone and the user can jointly determine the size of the offloading coefficient based on their own computing resources. When the user's own computing task meets the local computing delay, the user can offload or perform the calculation locally. This embodiment defaults to users who perform all local calculations (i.e., ξ i =0) There is no problem of information being stolen, that is, the information security problem only occurs during the offloading process. When the i-th user offloads the computing task to the legitimate UAV, the data transmission rate of the uplink between the user and the UAV can be expressed as:

[0061]

[0062] Here i is the transmission power of the i-th terminal user. 2 is the noise power, γ is the self-interference coefficient, p jam The legitimate drone emits electromagnetic waves to interfere with the power of the eavesdropping drone, which will cause the legitimate drone to generate γp itself. jam The self-interference power.

[0063] Similarly, the data transmission rate between the user and the eavesdropping drone can be expressed as:

[0064]

[0065] Therefore, the system confidentiality capacity problem generated during data offloading can be expressed as: in It represents the maximum data transmission rate of the eavesdropping drone eavesdropping on the i-th user. Here we define

[0066] (b) Delay model

[0067] The amount of data required to perform the task by the i-th user is defined as D i , computing task D i The number of CPU cycles required for each bit of data is defined as C i , it is assumed here that the computing task of each user needs to be completed within T time, the transmission bandwidth of each user in the channel is B, and users who cannot meet the local computing delay must offload the task to the legal drone. The entire computing process of the terminal user data can be divided into two modes: local computing and partial offloading. Therefore, the amount of data offloaded by the user to the legal drone is ξ i D i , the amount of task data calculated locally by the user is (1-ξ i )D i Since the amount of data returned to the user after calculation by the legitimate drone is very small compared to the amount of unloaded input data, the delay of the data transmitted from the drone to the user in the present invention can be ignored.

[0068] Based on the above analysis, the latency of local computing by the end user can be expressed as:

[0069]

[0070] f0 refers to the user's local CPU computing frequency. Here, by default, the CPU computing frequency of each ground terminal user is f0.

[0071] The delay of offloading the ground end-user tasks associated with the UAV to the UAV can be expressed as:

[0072]

[0073] The delay caused by the calculation of the drone performing the user task is:

[0074]

[0075] f iu The CPU computing frequency assigned to user i by the drone. Since the computing power is proportional to the cube of the CPU frequency, the drone computing power can be expressed as: Among them, ò is the power consumption coefficient, and its value depends on the chip structure of the drone.

[0076] (c) Energy consumption model

[0077] From the above delay and power model analysis, it can be seen that when executing user offload data, the energy consumption generated by the legal drone mainly includes three aspects: the energy consumption generated by emitting interference electromagnetic waves to interfere with eavesdroppers, the communication energy consumption generated in the process of receiving user offload data, and the computing energy consumption generated by executing user offload data. Since legal drones only emit interference signals when receiving offload data, the interference energy consumption generated by the electromagnetic waves emitted by legal drones can be expressed as:

[0078]

[0079] Let p u is the power of the legitimate UAV receiving the unloading signal, then the communication transmission power consumption generated by the legitimate UAV receiving the unloading user can be expressed as:

[0080]

[0081] Similarly, the computational energy consumption of a legitimate drone executing user offloaded data can be expressed as:

[0082]

[0083] The total power generated by the UAV to process ground users is:

[0084] Based on the above analysis, the confidentiality energy efficiency is defined as the ratio of the secure communication transmission rate to the energy consumption of the drone. Then the optimization problem in the present invention can be modeled as problem P1:

[0085]

[0086] In the above problem, C1 represents the offloading coefficient, C2 ensures the time constraint that the task needs to meet in the local computing mode, C3 indicates that the user task must be fully guaranteed to be completed within the total time T in the MEC offloading mode, C4 indicates that the total allocated frequency should be guaranteed to be within the total frequency range of the drone CPU, and C5 indicates that the frequency allocated to the legal drone should be non-negative.

[0087] The present invention mentioned in step 4 uses the block coordinate descent method for solving; when solving the optimization problem, the block coordinate descent global optimization algorithm only updates one or several variable blocks at a time, and iteratively solves the problem by alternately optimizing the three sub-problems to obtain an approximate solution to the original problem.

[0088] Step 5, the non-convex sub-problem of drone trajectory optimization is treated as an approximate convex sub-optimization problem by introducing auxiliary variables and adopting a first-order Taylor expansion. The specific processing flow is as follows:

[0089] For this non-convex subproblem:

[0090]

[0091] because is not always less than or equal to 0, so the target problem is non-convex. The main reason for the non-convexity is that the variable r iu and Therefore, an auxiliary variable φ is introduced here iu and ε ie The variable r iu and The relaxation is the following inequality: Here, the auxiliary variable φ iu Yes iu The lower bound of ε ie yes The upper bound of , so the target problem can be expressed as follows:

[0092]

[0093] Due to the introduction of auxiliary variables, the objective function becomes a convex function, but constraints C2 and C3 are still non-convex. The main reason for non-convexity is r iu and The legal drone trajectory q is non-convex, but r iu About ||qw i || 2 is convex, About ||qv e || 2 It is convex, so here we use SCA technology to re-express constraints C2 and C3:

[0094]

[0095] Here L1 is the first-order Taylor expansion lower bound expression of the legal drone position q[m] in the mth iteration:

[0096]

[0097] Similarly, to locally convexify ε ie :

[0098]

[0099] Among them, L2 also uses SCA technology to express it as the first-order Taylor expansion lower bound of the legal drone position q[m] in the mth iteration:

[0100]

[0101] Similarly, in order to convexify L3, an auxiliary variable λ is introduced e ,

[0102]

[0103] where λ e satisfy:

[0104]

[0105] Therefore, the UAV trajectory planning problem can be transformed into solving the following convex problem:

[0106]

[0107] Since this problem is a convex problem, it can be solved using the CVX tool and standard convex optimization techniques to obtain the local optimal solution of the UAV trajectory q[m] at the mth iteration.

[0108] The sub-problem of the CPU frequency allocation and the sub-problem of the data offloading rate in step 6 can be further transformed through the following process:

[0109] The sub-problem of the drone CPU calculation frequency allocation is:

[0110]

[0111] Since both the objective function and the constraints are convex functions, the CVX tool can also be used to solve them using the standard convex optimization algorithm to obtain the local optimal solution for the CPU computing frequency allocation of the drone under the current number of iterations.

[0112] Based on the local optimal solution of the drone trajectory obtained from the above sub-problems and the allocation of the drone CPU computing frequency, for any given unloading rate ξ, problem P1 can be transformed into the following form:

[0113]

[0114] When other variables are fixed, solve the problem to find the maximum value, that is, to find the unloading rate ξ i Minimum problem, and ξ i It is also determined by the maximum system delay. According to constraint C1, we can get:

[0115]

[0116] The process of using the CVX tool to find approximate solutions to the above three sub-problems described in step 7 is as follows: first, give the initial values ​​of each variable, substitute them into the solution of the drone trajectory optimization problem to obtain the local approximate value of the i-th iteration, then substitute the drone trajectory value of the i-th iteration and other initial values ​​into the CPU calculation frequency optimization sub-problem, solve the approximate solution of the CPU calculation frequency allocation sub-problem under the i-th iteration, then substitute the drone trajectory solution and CPU calculation frequency solution of the i-th iteration into the approximate solution of the offloading rate sub-problem, and continuously increase the number of iterations until the difference between the objective functions of the two iterations is within the range of [-0.01, 0.01]. Then, the high offloading energy efficiency method of the system can be obtained, that is, the optimal flight path trajectory, the optimal frequency allocation of the drone to each ground offloading user, and the calculation task data offloading rate.

[0117] In summary, the present invention simultaneously considers the relationship between the maximum secure offload data rate and the energy consumption of the drone in the drone-assisted edge computing system under an eavesdropping environment. A maximum confidentiality energy efficiency model for drones is established. Under the constraints of given latency, MEC server CPU computing frequency and offload rate, the legal drone trajectory, frequency allocation and computing task offload strategy are jointly optimized, while improving the secure communication in the UAV-MEC scenario from the perspective of physical layer security. The method adopts a global optimization algorithm based on the block coordinate descent method, decouples the optimization problem into three sub-problems, and locally approximates the non-convex sub-optimization problem to a local convex optimization problem by referencing slack variables. The convex problem is solved by multiple iterations using the CVX tool to obtain the approximate optimal solution of the system. The invention fully considers the relationship between the safety communication of drones and the energy consumption of drones, and has the purpose of improving the overall Qos of the system.

[0118] The above description is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiment. Any equivalent modifications or changes made by ordinary technicians in this field based on the contents disclosed by the present invention should be included in the protection scope recorded in the claims.

Claims

1. A high-security offloading energy efficiency method using physical layer security technology in a UAV-MEC environment, characterized by: The method comprises the following steps: Step 1, establish an edge unloading scenario with eavesdropping drones and legitimate drones; Step 2, modeling the above edge offloading model as a formulated optimization problem through a mathematical model; In step 2, the following mathematical model is established: Communication model: including the channel power gain between the ith user and the legitimate user, the channel gain from the ith user to the eavesdropping UAV, and the channel power gain between the eavesdropping UAV and the legitimate UAV; when the ith user offloads the computing task to the legitimate UAV, the data transmission rate of the uplink between the user and the UAV, and the data transmission rate between the user and the eavesdropping UAV are obtained; Delay model: The delay transmitted from the drone to the user is negligible, and the delay of local calculation by the end user, the delay of offloading the ground end user task associated with the drone to the drone, and the delay caused by the calculation of the drone executing the user task are obtained; Energy consumption model: When executing user offload data, the energy consumption generated by the legitimate drone includes the energy consumption generated by emitting interference electromagnetic waves to interfere with eavesdroppers, the communication energy consumption generated in the process of receiving user offload data, and the computing energy consumption generated by executing user offload data; The confidentiality energy efficiency is defined as the ratio of the secure communication transmission rate to the energy consumption of the drone. The optimization problem is then modeled as an optimization problem that maximizes the confidentiality energy efficiency, and the corresponding constraints are given; the constraints include the offloading coefficient, the time constraints that tasks need to meet in the local computing mode, the user tasks need to be fully guaranteed to be completed within the total time T in the MEC offloading mode, the total allocated frequency should be guaranteed to be within the total frequency range of the drone CPU, and the frequency allocated to the legal drone should be non-negative; Step 3: The established optimization problem is divided into two modes according to whether the amount of ground offloaded user task data can meet the task delay requirements: local computing mode and MEC partial offload computing mode. If the ground terminal user can complete the computing task within the delay requirement, the local computing mode is selected; otherwise, the MEC partial offload mode is selected. Step 4: The optimization problem in the MEC partial offloading mode is solved using the block coordinate algorithm, and it is decoupled into three sub-optimization problems: the non-convex sub-problem of legal UAV trajectory optimization, the sub-problem of UAV CPU computing frequency allocation, and the sub-problem of data offloading rate optimization. Step 5, introduce slack variables into the non-convex sub-problem of UAV trajectory optimization to perform local convex approximation processing to solve the approximate optimal operation trajectory of the system; Step 6: Transform the objective function, remove the constraints and variables irrelevant to the current sub-problem, and solve the sub-problem of the CPU calculation frequency allocation and the data offloading rate of the drone; Step 7: Use the CVX tool to find local approximate solutions to the above three sub-problems respectively, and iterate the three local solutions to obtain the approximate optimal solution for the system as a whole, obtain the optimal flight path trajectory of the drone, the optimal CPU computing frequency allocation of the drone to each ground unloading user, and the computing task data unloading rate, balance the relationship between the security communication transmission rate and energy consumption in the system, so that the overall system has the maximum security communication energy efficiency ratio.

2. According to claim 1, a high-security offloading energy efficiency method using physical layer security technology in a UAV-MEC environment is characterized by: In step 1, the establishment of an edge unloading scenario with eavesdropping drones and legitimate drones includes the following information: N ground terminal users, E eavesdropping drones and a legitimate drone equipped with an edge server; the location of the i-th user is represented by The legal drone is fixed at a height Hu above the ground, and its horizontal position is express.

3. According to claim 1, a high-security offloading energy efficiency method using physical layer security technology in a UAV-MEC environment is characterized by: In step 2, the established optimization problem specifically includes an objective function and constraint conditions; The objective function is: In formula (1), η is defined as the maximum confidentiality energy efficiency of the system, and the system secure communication transmission rate is Defined as the data transmission rate r from ground user i to the legitimate UAV link iu Maximum data transmission rate of the link from user i to the eavesdropping drone The difference Where E i represents the total energy consumption of legitimate drone service user i, including the energy consumption of launching jamming and eavesdropping drones Communication transmission power consumption generated during receiving unloaded users and the computational energy consumed by offloading user data to legitimate drones q and f are the flight trajectory of the drone and the calculation frequency assigned to each user, respectively. The variable ξ = [0, 1] is defined as the offloading coefficient of the user to establish an offloading connection with the legitimate drone; The constraints include: The latency requirement of user i in local computing mode should meet T i loc ≤T, where T i loc represents the local computing delay, and T is the maximum delay that the terminal task can tolerate; User i chooses to partially uninstall the content Where, T i ul , T i c They represent the offloading delay and the computation delay of the legitimate drone edge server, respectively. The delay transmitted from the drone to the user is negligible. In edge offloading mode, the CPU computing frequency allocation of the drone must meet In the formula, Indicates the total CPU computing frequency of the drone.

4. According to claim 1, a high-security offloading energy efficiency method using physical layer security technology in a UAV-MEC environment is characterized in that: In step 4, the optimization problem in the MEC partial offloading mode is solved using the block coordinate algorithm, and is decoupled into three sub-optimization problems. The specific steps are as follows: a. According to the block coordinate descent method, only the constraints related to the UAV trajectory are retained. The UAV trajectory sub-optimization problem is expressed as: b. Only the constraints related to the UAV frequency allocation are retained, and the UAV frequency allocation sub-optimization problem is expressed as: c. According to the block coordinate descent method, only the constraints related to the drone unloading rate are retained. The drone unloading rate sub-problem is expressed as: 。 5. According to claim 1, a high-security offloading energy efficiency method using physical layer security technology in a UAV-MEC environment is characterized by: In step 5, the slack variables are introduced to perform local convex approximation on the UAV trajectory optimization problem. The specific method is: Introducing auxiliary variable φ iu and ε ie The variable r iu and The relaxation is the following inequality: Auxiliary variable φ iu Yes iu The lower bound of ε ie yes The upper bound of .

6. According to claim 1, a high-security offloading energy efficiency method using physical layer security technology in a UAV-MEC environment is characterized by: In step 7, the CVX tool is used to find local approximate solutions to the three sub-problems, and the three local solutions are iterated to obtain the approximate optimal solution of the system as a whole. The specific method is: When solving the UAV trajectory subproblem, the initial values ​​of each variable are given and substituted into the local approximate value of the i-th iteration of the trajectory solution. Then, the UAV trajectory value of the i-th iteration and other initial values ​​are substituted into the frequency optimization subproblem to solve the approximate solution of the frequency allocation subproblem under the i-th iteration. Then, the UAV trajectory solution and frequency solution of the i-th iteration are substituted into the approximate solution of the unloading rate subproblem. The number of iterations is continuously increased until the difference between the objective functions of the two iterations is within the range of [-0.01, 0.01], that is, a high unloading energy efficiency method is obtained and the process ends.

Images