Edge computing information security optimization system and method
The intelligent reflective surface drone system addresses channel instability and security issues in wireless communication by optimizing communication environments for secure offloading, improving the secure computing capability of mobile edge computing systems.
Patent Information
- Application Number
- CN202411185722.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-27
- Publication Date
- 2024-12-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing wireless communication systems face challenges with channel instability, severe channel attenuation, and security issues due to obstacles and eavesdropping, limiting the effective and secure offloading of mobile edge computing.
An edge computing information security optimization system utilizing an intelligent reflective surface carried by a drone to optimize communication environments, enabling secure offloading of computing tasks between IoT devices and edge servers while avoiding eavesdroppers, by constructing a target optimization problem and iteratively solving for optimal phase, time slot allocation, transmission power, and UAV trajectory.
The system effectively reduces channel attenuation and prevents information leakage, enhancing the secure computing capability of mobile edge computing systems by optimizing communication environments and resource utilization.
Smart Images

Figure CN119095112A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of wireless communications, and in particular to an edge computing information security optimization system and method. Background Art
[0002] In Mobile Edge Computing (MEC) applications, wireless communication systems face challenges such as channel instability and security. Traditional wireless communication systems suffer from poor communication quality due to physical obstacles, interference, and multipath effects. When the offload link experiences severe channel attenuation due to congestion, the device suffers from low offload rates, which results in high latency and energy budget for computational offload. At the same time, due to the broadcast nature of wireless channels, data transmission in wireless channels is easily eavesdropped, and there is a risk of information leakage.
[0003] In recent years, the Intelligent Reconfigurable Surface (IRS) technology has attracted much attention due to its ability to intelligently reconfigure the wireless propagation environment through passive beamforming. By adjusting the phase shift of its units, IRS can enhance the signal strength to the legitimate receiving end and weaken the signal strength to the illegal eavesdropping end, thereby significantly improving the performance and security of the communication system. Since IRS does not require complex RF links and has the advantages of low power consumption, low cost and easy deployment, it has become one of the hot topics in the research of next-generation mobile communications. By utilizing IRS technology, the channel attenuation problem can be effectively alleviated, the signal coverage and transmission quality can be improved, and it provides a new way to solve the channel instability and security problems in existing wireless communication systems.
[0004] At present, there have been some studies on using IRS to assist in the safe offloading of mobile edge computing systems. However, the existing IRSs are deployed on the ground, and the deployment location of the IRS is not optimized. This fixed ground deployment method cannot guarantee that there are no obstacles between the IRS and the base station and IoT devices, so the channels between the IRS and the base station and IoT devices may still suffer from severe path attenuation. In addition, when the IoT device is located on the back of the IRS panel, the IRS cannot effectively reconfigure its channel environment, resulting in an inability to adapt to changing communication needs and environmental changes, which in turn leads to the inability of mobile edge computing to be effectively and safely offloaded, limiting the performance of the mobile edge computing system. Summary of the invention
[0005] In order to overcome the problem in existing wireless communication technologies that the IRS cannot effectively reconfigure its channel environment, resulting in the inability to adapt to changing communication needs and environmental changes, and further resulting in the inability to effectively and securely offload mobile edge computing, the present invention proposes an edge computing information security optimization system and method. The aerial IRS can configure an ideal channel environment for ground equipment, effectively reduce channel attenuation and prevent eavesdropping, thereby improving the secure computing capability of the mobile edge computing system.
[0006] The purpose of the present invention is achieved by the following technical solutions:
[0007] An edge computing information security optimization system, the system comprising an edge server, an eavesdropper, an intelligent reflective surface carried by a drone, and a plurality of Internet of Things devices;
[0008] Each of the IoT devices is used to perform local computation on a computing task or wirelessly transmit part of the computing task to an edge server;
[0009] The edge server is used to receive and process computing task data transmitted from the IoT device;
[0010] The eavesdropper is used to simulate eavesdropping on computing task data transmitted by IoT devices;
[0011] The intelligent reflective surface carried by the drone optimizes the communication environment between IoT devices and edge servers to prevent possible eavesdroppers.
[0012] In the above technical solution, an aerial intelligent reflective surface secure mobile edge computing system is constructed. The intelligent reflective surface carried by the drone can optimize the communication environment between the IoT device and the edge server in the air against eavesdroppers, thereby effectively assisting the ground IoT devices in unloading information to the edge server, while avoiding information leakage to eavesdroppers, and improving the secure computing capabilities of the mobile edge computing system.
[0013] An edge computing information security optimization method, the method comprising the following steps:
[0014] IoT devices implement partial offloading strategies, offloading some computing tasks to edge servers for processing;
[0015] The goal is to maximize the minimum safe computing rate of IoT devices and construct a target optimization problem.
[0016] Solving the target optimization problem to obtain a task safety offloading strategy;
[0017] According to the task security offloading strategy, the aerial intelligent reflective surface assists the IoT device to safely offload part of the computing tasks to the edge server for processing.
[0018] In the above technical scheme, the IoT device executes a partial offloading strategy, which can leave a part of the computing tasks for local processing and offload the other part of the tasks to the edge server for processing, so as to optimize the utilization of computing resources; the target optimization problem is solved, and the task safety offloading strategy obtained can enable the intelligent reflective surface in the air to effectively optimize the communication environment between the IoT device and the edge server according to the task safety offloading strategy. At the same time, it effectively configures an ideal channel environment for ground equipment, greatly reduces channel attenuation and interference, so as to adapt to different communication needs and environmental changes, and then assists the IoT device to safely offload part of the computing tasks to the edge server for processing, thereby improving the secure computing capability of the mobile edge computing system.
[0019] Furthermore, the process of constructing the target optimization problem includes:
[0020] Based on the aerial intelligent reflective surface safe mobile edge computing system, a three-dimensional Cartesian coordinate system is constructed;
[0021] Based on the three-dimensional Cartesian coordinate system, communication constraints are set for the communication between the edge server, the eavesdropper, the intelligent reflective surface carried by the drone, and a number of IoT devices; and computing constraints are set for the IoT devices and edge servers in processing computing tasks respectively;
[0022] Based on communication constraints and computational constraints, a target optimization problem is constructed.
[0023] Furthermore, the process of setting communication constraints on the communication between the edge server, the eavesdropper, the intelligent reflective surface carried by the drone, and the plurality of IoT devices includes:
[0024] Given the initial and final positions of the drone, the flight time of the drone is divided into N time slots, and the trajectory constraints of the drone are set. The expression is:
[0025] ‖l[n+1]-l[n]‖≤D max ,n=1,2,...,N-1
[0026] l[1]=l0
[0027] l F = l[N];
[0028] Set the transmission power constraint of IoT devices during data transmission. The expression is:
[0029] 0≤p k [n]≤P k ;
[0030] The system uses time division multiple access to communicate and constrains the time slot allocation during the communication process. The expression is:
[0031]
[0032] The channel gain between the IoT device and the aerial intelligent reflector in time slot n is expressed as:
[0033]
[0034] The channel gains from the airborne intelligent reflection surface to the edge server and the eavesdropper in time slot n are expressed as:
[0035]
[0036] The channel gains from IoT devices to edge servers and eavesdroppers are expressed as:
[0037]
[0038] Based on the aforementioned constraints and channel gains, the safe offloading rate of IoT devices in the process of offloading part of the computing tasks to the edge server is obtained, which is expressed as:
[0039]
[0040] Where l[n] represents the horizontal coordinate of the drone at time slot n, n∈{1,...,N}; k represents the kth IoT device, p k [n] represents the transmission power of IoT device k in time slot n, P k represents the maximum transmission power of IoT device k; t k [n] = δa k [n] represents the time allocated to IoT device k in time slot n; ρ represents the path loss per unit distance under free space loss; d k,I represents the distance between IoT device k and the intelligent reflective surface in the air; d I,B represents the distance between the aerial intelligent reflective surface and the edge server; d k,B represents the distance between IoT device k and the edge server; d k,E represents the distance between IoT device k and the eavesdropper; denotes a random variable that follows a circularly symmetric complex Gaussian distribution with zero mean and unit variance; {x} + It means x=max{0,x}.
[0041] Furthermore, the process of setting computing constraints on IoT devices and edge servers in processing computing tasks includes:
[0042] Set the local computing amount of IoT device k, the expression is:
[0043]
[0044] Based on the local computing capacity, the computing capacity constraint of IoT device k is set, and the expression is:
[0045]
[0046] Set the computing capacity constraint of the edge server, the expression is:
[0047]
[0048] The energy consumption constraint of IoT device k in the entire computing cycle is expressed as:
[0049]
[0050] Among them, f k Calculate the CPU frequency for device k; is the maximum CPU frequency of the server; c B The number of CPU cycles required for the server to calculate one bit of data; represents the energy consumption of device k during local computing; E off,k =a k [n]δp k [n] indicates that the energy consumption during the computing task offloading process in time slot n is; R k,B [n] represents the task offloading rate between IoT device k and edge server in time slot n.
[0051] Furthermore, based on the communication constraints and computing constraints, the target optimization problem is constructed with the goal of maximizing the minimum secure computing rate of IoT devices. The secure computing rate expression of IoT device k is:
[0052]
[0053] Among them, R k,sec [n] represents the secure offloading rate of IoT device k in time slot n.
[0054] In the above technical scheme, by setting different constraints on each component in the system, constructing a target optimization problem with these constraints, and solving the target optimization problem, the task security offloading strategy obtained can enable the intelligent reflective surface in the air to optimize the communication environment between the IoT device and the edge server according to the task security offloading strategy, effectively assist the ground IoT device to unload information to the edge server, and avoid information leakage to eavesdroppers, thereby improving the secure computing capability of the mobile edge computing system.
[0055] Furthermore, the process of solving the target optimization problem includes:
[0056] The target optimization problem is divided into four optimization problems: phase optimization of the aerial intelligent reflector, time slot allocation optimization, IoT device transmission power and local computing frequency optimization, and UAV trajectory optimization.
[0057] The four optimization problems of phase optimization of the intelligent aerial reflector, time slot allocation optimization, IoT device transmission power and local computing frequency optimization, and UAV trajectory optimization are solved iteratively. When the target value reaches the convergence accuracy, the task security offloading strategy between the IoT device and the edge server is obtained.
[0058] According to the task security offloading strategy, the aerial intelligent reflective surface assists the IoT device to safely offload part of the computing tasks to the edge server for processing.
[0059] Furthermore, the process of dividing the target optimization problem into four optimization problems, namely, optimization of the phase of the aerial intelligent reflector, optimization of time slot allocation, optimization of the transmission power and local computing frequency of the IoT device, and optimization of the trajectory of the UAV, includes:
[0060] Fixed time slot allocation, UAV trajectory, transmission power and local calculation frequency, optimize the phase of the aerial intelligent reflector, the expression is:
[0061]
[0062] Fix the phase of the aerial intelligent reflector, the transmission power, the local calculation frequency and the trajectory of the UAV, and optimize the time slot allocation. The expression is:
[0063]
[0064] Fixed time slot allocation, airborne intelligent reflector phase and drone trajectory, jointly optimize the transmission power and local calculation frequency, the expression is:
[0065]
[0066] Fixed time slot allocation, phase of the airborne intelligent reflector, transmission power and local calculation frequency, optimize the trajectory of the UAV, the expression is:
[0067]
[0068] Furthermore, an alternating optimization method is adopted to iteratively solve the four optimization problems of phase optimization of aerial intelligent reflector surface, time slot allocation optimization, IoT device transmission power and local computing frequency optimization, and UAV trajectory optimization.
[0069] In the above technical scheme, the target optimization problem is divided into four optimization problems, and iterative optimization is performed in an alternating optimization manner, which can effectively ensure that the optimization value can reach the required expected target optimization value. The target optimization value obtained is used as a task safety offloading strategy, which can enable the intelligent reflective surface in the air to effectively optimize the communication environment between the Internet of Things device and the edge server according to the task safety offloading strategy. At the same time, it effectively configures an ideal channel environment for ground equipment, effectively reduces channel attenuation and interference, and thus improves the system's safe computing capability.
[0070] An electronic device comprises a memory, a processor and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, the steps of an edge computing information security optimization method are implemented.
[0071] Compared with the prior art, the present invention has the following beneficial effects:
[0072] The present invention proposes an edge computing information security optimization system and method. First, the Internet of Things device executes a partial offloading strategy, which can leave a part of the computing tasks for local processing and offload the other part of the tasks to the edge server for processing, so as to optimize the utilization rate of computing resources; then, the target optimization problem is solved, and the obtained task security offloading strategy can enable the intelligent reflection surface in the air to effectively optimize the communication environment between the Internet of Things device and the edge server according to the task security offloading strategy, effectively reduce channel attenuation, and avoid information leakage to eavesdroppers, thereby improving the security computing capability of the mobile edge computing system. BRIEF DESCRIPTION OF THE DRAWINGS
[0073] Figure 1 A schematic diagram of the structure of an edge computing information security optimization system provided in an embodiment of the present application;
[0074] Figure 2 A flowchart of the steps of an edge computing information security optimization method provided in an embodiment of the present application;
[0075] Figure 3 A graph showing the convergence of the objective function value during the iterative solution of the target optimization problem provided in the embodiment of the present application;
[0076] Figure 4 A schematic diagram of the trajectory of a drone provided in an embodiment of the present application;
[0077] Figure 5 This is a performance comparison chart of the experimental process provided by the embodiment of the present application with the benchmark solution. DETAILED DESCRIPTION
[0078] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Preferred embodiments of the present invention are provided in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present invention more thorough and comprehensive.
[0079] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.
[0080] Embodiment 1:
[0081] This embodiment provides an edge computing information security optimization system. Figure 1 ,The system includes: an edge server, an eavesdropper, an intelligent reflective surface carried by a drone, and several IoT devices;
[0082] Each of the IoT devices is used to perform local computation on a computing task or wirelessly transmit part of the computing task to an edge server;
[0083] The edge server is used to receive and process computing task data transmitted from the IoT device;
[0084] The eavesdropper is used to simulate eavesdropping on computing task data transmitted by IoT devices;
[0085] The intelligent reflective surface carried by the drone optimizes the communication environment between IoT devices and edge servers to prevent possible eavesdroppers.
[0086] For example, see Figure 1 , build an airborne intelligent reconfigurable surface (IRS) secure mobile edge computing system. The airborne IRS assists ground IoT devices in unloading information to edge servers while preventing information leakage to eavesdroppers. The system consists of the following parts: K IoT devices, Each device has a certain amount of local computing power and wireless transmission capability; a single-antenna service base station (i.e., edge server) is responsible for receiving and processing data from IoT devices; a single-antenna eavesdropper simulates potential data eavesdropping threats; and an intelligent reflective surface carried by a drone, composed of M reflective elements, is responsible for optimizing the wireless channel environment.
[0087] In this embodiment, an aerial intelligent reflective surface secure mobile edge computing system is constructed. The intelligent reflective surface carried by the drone can optimize the communication environment between the IoT device and the edge server in the air against eavesdroppers, thereby effectively assisting the ground IoT device to unload information to the edge server, while avoiding information leakage to eavesdroppers, thereby improving the stability and security of the system communication environment.
[0088] Embodiment 2:
[0089] This embodiment provides an edge computing information security optimization method for the system described in Embodiment 1, see Figure 2 , the method comprises the following steps:
[0090] Step S1: The IoT device executes a partial offloading strategy and offloads part of the computing tasks to the edge server for processing;
[0091] Step S2: construct a target optimization problem with the goal of maximizing the minimum secure computing rate of IoT devices;
[0092] Step S3: Solve the target optimization problem to obtain a task safety offloading strategy;
[0093] Step S4: The aerial intelligent reflective surface assists the IoT device in safely offloading part of the computing tasks to the edge server for processing according to the task security offloading strategy.
[0094] In this embodiment, the IoT device executes a partial offloading strategy, which can leave a part of the computing tasks for local processing and offload the other part of the tasks to the edge server for processing, so as to optimize the utilization of computing resources; the target optimization problem is solved, and the task security offloading strategy obtained can enable the intelligent reflective surface in the air to optimize the communication environment between the IoT device and the edge server according to the task security offloading strategy, effectively reducing channel attenuation, while avoiding information leakage to eavesdroppers, and improving the system's secure computing capabilities.
[0095] As a preferred embodiment, in step S2, the process of constructing the target optimization problem includes:
[0096] S21: constructing a three-dimensional Cartesian coordinate system based on the aerial intelligent reflective surface safe mobile edge computing system;
[0097] S22: Based on the three-dimensional Cartesian coordinate system, communication constraints are set for the communication between the edge server, the eavesdropper, the intelligent reflective surface carried by the drone, and a plurality of IoT devices; and computing constraints are set for the IoT devices and edge servers in processing computing tasks respectively;
[0098] S23: Construct the target optimization problem based on communication constraints and computational constraints.
[0099] In step S21, see Figure 1 , a three-dimensional Cartesian coordinate system is established. The coordinates of the service base station, eavesdropper, and IoT device k are [w B T ,0] T , [w E T ,0] T , [w k T ,0] T , k∈{1,...,K}. Among them, w B =[x B ,y B ] T 、w E =[x E ,y E ] T 、w k =[x k ,y k ] T are the horizontal coordinates of the serving base station, the eavesdropper, and the device k. The flight time T of the drone is divided into N equal-length sub-slots, and the slot length δ = T / N. In time slot n, the position of the drone is [l[n] T ,H] T , n∈{1,...,N}. Among them, l[n]=[x U [n],y U [n]] T , H is the given flight altitude of the UAV.
[0100] In step S22, the process of setting communication constraints on the communication between the edge server, the eavesdropper, the intelligent reflective surface carried by the drone, and the plurality of IoT devices includes:
[0101] The maximum flight speed of the drone is V max , the maximum moving distance D in a time slot max =V max δ, the trajectory of the UAV should satisfy constraint (1):
[0102]
[0103] Define the transmission power of device k in time slot n as p k [n],p k [n] should satisfy the constraints of formula (2), where P kis the maximum transmission power of device k. To avoid mutual interference between devices during data unloading, in this system, time division multiple access is used for unloading communication. The time slot allocation should satisfy the constraints of formula (3). In time slot n, the time t allocated to device k is k [n] = δa k [n].
[0104] 0≤p k [n]≤P k (2)
[0105]
[0106] In this system, the IRS is a uniform planar array (UPA) mounted horizontally on the drone to achieve panoramic reflection. The number of reflective elements of the IRS is M = m x m y , the reflection phase shift matrix of IRS is expressed as θ (p,q) [n] represents the phase shift of the (p,q)th IRS element in the nth time slot, where 0≤p≤m x -1,0≤q≤m y -1.
[0107] θ (p,q) [n]=[0,2π) (4)
[0108] In this system, the passive eavesdropping scenario is considered, that is, the eavesdropper also belongs to the service base station, but not the target service base station. The IoT device calculates confidential information and does not want base stations other than the target service base station to receive this information. Therefore, the service base station and the eavesdropper have perfect channel state information. The flight altitude of the drone is high enough, and the air channel can be assumed to be a line-of-sight channel. The channel gain between device k and IRS in time slot n is:
[0109]
[0110] Where ρ represents the path loss per unit distance under free space loss, d k,I represents the distance between device k and IRS, χ and λ represent the antenna spacing and carrier wavelength, respectively; φ k,I [n] and They represent the azimuth and elevation angles from device k to IRS respectively.
[0111] In time slot n, the channel gain from IRS to base station is h I,B [n], the channel gain from IRS to the eavesdropper is h I,E [n], their expressions are similar to h k,I [n] Similar to.
[0112]
[0113] in, φ I,E [n] and are the azimuth and elevation angles from the IRS to the eavesdropper. φ I,B [n] and They represent the azimuth and elevation from IRS to the base station respectively.
[0114] In this system, consider that there are obstacles between the device and the base station (eavesdropper), the path attenuation is serious, and the ground channel is assumed to be a Rayleigh channel. The channel gains from device k to the base station and to the eavesdropper are h k,B and h k,E , as shown in equations (11) and (12). Where d k,B d k,E are the distances from device k to the base station and from device k to the eavesdropper, respectively. is a circularly symmetric complex Gaussian distributed random variable with zero mean and unit variance.
[0115]
[0116] The system bandwidth is B, R k,B [n] is the unloading rate from device k to the base station in the nth time slot, R k,E [n] is the eavesdropping offloading rate from device k to the eavesdropper in the nth time slot. Then the security offloading rate can be expressed as:
[0117]
[0118] Where l[n] represents the horizontal coordinate of the drone at time slot n, n∈{1,...,N}; k represents the kth IoT device, p k [n] represents the transmission power of IoT device k in time slot n, P k represents the maximum transmission power of IoT device k; t k [n] = δa k [n] represents the time allocated to IoT device k in time slot n; ρ represents the path loss per unit distance under free space loss; d k,I represents the distance between IoT device k and the intelligent reflective surface in the air; d I,B represents the distance between the aerial intelligent reflective surface and the edge server; d k,B represents the distance between IoT device k and the edge server; d k,E represents the distance between IoT device k and the eavesdropper; denotes a random variable that follows a circularly symmetric complex Gaussian distribution with zero mean and unit variance; {x} +It means x=max{0,x}.
[0119] Furthermore, the process of setting computing constraints on IoT devices and edge servers in processing computing tasks includes:
[0120] This system adopts a partial offloading strategy, that is, the IoT device calculates part of the task locally and offloads the other part of the task to the server base station for calculation. k The CPU cycles required to calculate 1 bit of data for IoT device k, f k is the CPU calculation frequency (cycles / second) of device k, and the local calculation amount can be expressed as Since the local computing capability of device k is limited by its own maximum computing capability, f k The following constraints should be met:
[0121]
[0122] Definition c B The number of CPU cycles required for the server to calculate one bit of data, is the maximum CPU frequency of the server. The number of secure computing bits at the server cannot exceed its computing power, so the computing power of the server needs to meet the following constraints:
[0123]
[0124] Define Q k is the minimum computing requirement of device k. In order to meet the basic computing requirements of all devices, we can get:
[0125]
[0126] The energy consumption of device k during local calculation is where k k It is expressed as the effective capacitance coefficient of device k. In the nth time slot, the energy consumption during unloading is E off,k =a k [n]δp k [n]. The energy of device k is limited in the entire computing cycle, and its energy consumption constraint is:
[0127]
[0128] Among them, f k Calculate the CPU frequency for device k; is the maximum CPU frequency of the server; c B The number of CPU cycles required for the server to calculate one bit of data; represents the energy consumption of device k during local computing; E off,k =a k [n]δpk [n] indicates that the energy consumption during the computing task offloading process in time slot n is; R k,B [n] represents the task offloading rate between IoT device k and edge server in time slot n.
[0129] In step S23, based on the communication constraints and computational constraints, a target optimization problem is constructed with the goal of maximizing the minimum secure computational rate of IoT devices. The average secure computational rate of device k is defined as the ratio of the secure offload amount plus the local computation amount to the computational time T, expressed as:
[0130]
[0131] In the optimization problem (P1), constraint (21a) is the transmission power constraint of device k; (21b) is the UAV trajectory constraint; (21c) is the time slot allocation constraint; (21d) is the IRS phase constraint; (21e) is the device k calculation frequency constraint; (21f) is the constraint to ensure that the offloaded task can complete the calculation at the server base station; (21g) is the constraint to ensure that the minimum computing task can be completed; (21h) is the total energy finite constraint used by device k for transmitting signals and local calculations.
[0132] It can be understood that by setting different constraints on each component in the system, constructing the target optimization problem with these constraints, and solving the target optimization problem, the task safety offloading strategy obtained can enable the intelligent reflection surface in the air to effectively optimize the communication environment between the IoT device and the edge server according to the task safety offloading strategy, greatly reducing channel attenuation, and avoiding information leakage to eavesdroppers, thereby improving the system's secure computing capabilities.
[0133] As a preferred embodiment, in step S3, the process of solving the target optimization problem includes:
[0134] S31: The target optimization problem is divided into four optimization problems: phase optimization of the aerial intelligent reflector, time slot allocation optimization, IoT device transmission power and local computing frequency optimization, and UAV trajectory optimization;
[0135] S32: Iteratively solve the four optimization problems of phase optimization of the intelligent air reflector, time slot allocation optimization, IoT device transmission power and local computing frequency optimization, and drone trajectory optimization. When the target value reaches the convergence accuracy, the task security offloading strategy between the IoT device and the edge server is obtained;
[0136] S33: The aerial intelligent reflective surface assists the IoT device in safely offloading part of the computing tasks to the edge server for processing according to the task security offloading strategy.
[0137] In step S31, the process of dividing the target optimization problem into four optimization problems of airborne intelligent reflector phase optimization, time slot allocation optimization, IoT device transmission power and local computing frequency optimization, and drone trajectory optimization includes:
[0138] The problem of discontinuous optimization objectives needs to be solved. The optimization objective of problem (P1) contains the symbol {.} + , which results in the optimization objective being a discontinuous function, which is difficult to solve. k [n], can guarantee R k,sec [n]>=0, so the symbol {.} in (P1) optimization objective + It can be removed without affecting the optimization results.
[0139] Next, we need to deal with the problem of optimization variable coupling. The optimization variables of the optimization problem (P1) are highly coupled. In order to effectively solve the proposed non-convex nonlinear optimization problem, the block coordinate descent method is used to decouple the optimization variables in problem (P1), and problem (P1) is divided into four sub-problems: 1) Given the UAV trajectory, time slot allocation, transmit power and local calculation frequency, optimize the IRS phase. 2) Given the IRS phase, transmit power, local calculation frequency and UAV trajectory, optimize the time slot allocation. 3) Given the time slot allocation, IRS phase and UAV trajectory, sub-problem 3 jointly optimizes the transmit power and local calculation frequency. 4) Given the time slot allocation, IRS phase, transmit power and local calculation frequency, optimize the UAV trajectory. The four sub-problems are solved alternately until the target optimization value reaches the convergence accuracy.
[0140] 1) Optimize IRS phase
[0141] First, the time slot allocation, drone trajectory, transmission power and local calculation frequency are fixed to optimize the IRS phase. In the nth time slot, a corresponding optimal IRS phase is optimized for each IoT device. The IRS phases of different time slots and different IoT devices are independent of each other and do not affect each other, so the optimization of the IRS phase can be restated as problem (P2):
[0142]
[0143] Introducing vectors in, By definition We can get:
[0144]
[0145] in, H k [n] = h k [n]h k H[n],G k [n] = g k [n]g k H [n]. Introducing slack variables The optimization problem (P2) can be rewritten as a semi-definite programming (SDP) problem (P3):
[0146]
[0147] The optimization objective (25) of (P3) is to maximize the form of concave function minus concave function, which is non-convex. We use Successive Convex Approximation (SCA) to approximate (25). Applying the property that the first-order Taylor expansion of a convex function at any point is a global lower bound estimate of its original function, we can obtain:
[0148]
[0149] Where r is the number of iterations.
[0150] According to the continuous convex approximation, the lower bound of (26) is used to approximate the original function, and (25) can be rewritten as:
[0151]
[0152] The rank-one constraint (25f) is difficult to handle. Here we first remove the rank-one constraint (25f) and then use Gaussian randomization to restore it. The optimization problem can finally be transformed into (P4):
[0153]
[0154] The optimization objectives and constraints of the above problems are all in standard convex form and can be solved using optimization toolkits, such as the Matlab software for disciplined Convex programming (CVX).
[0155] 2) Optimize time slot allocation
[0156] Given the IRS phase, transmit power, local calculation frequency, and drone trajectory, optimize the time slot allocation. The sub-problem of optimizing time slot allocation can be expressed as:
[0157]
[0158] Problem (P5) is a standard linear programming problem and can be solved using CVX.
[0159] 3) Jointly optimize transmission power and local computing frequency
[0160] definition Given the IRS phase, time slot allocation, and UAV trajectory, the subproblem of jointly optimizing the transmit power and local computation frequency can be formulated as:
[0161]
[0162] Among them, R k,B [n] = log2(1 + x k [n]p k [n]), R k,E [n] = log2(1 + y k [n]p k [n]).
[0163] Due to the non-convexity of constraint (30a), the optimization problem (P6) is a non-convex optimization problem. Using SCA, we can k,E [n] Performing a first-order Taylor expansion yields:
[0164]
[0165] Where r represents the rth iteration.
[0166] definition The optimization problem (P6) can be transformed into:
[0167]
[0168] The optimization objectives and constraints of the above problems are all in standard convex form and can be processed using CVX.
[0169] 4) Optimize drone trajectory
[0170] Given the IRS phase, transmit power, local computation frequency, and time slot allocation, the subproblem of optimizing the UAV trajectory can be formulated as Problem (P8):
[0171]
[0172] From formulas (6), (8), and (10), we can see that the steering vector a k [n]、a B [n]、a E [n] is complexly coupled with the trajectory and is nonlinear, which is difficult to handle. Referring to some existing work, the trajectory is optimized using the rth iteration Approximate processing.
[0173] To process R k,B [n] and R k,E[n] Complex expressions, defined by You can get:
[0174]
[0175] Among them, the coefficient A 1,k [n], B 1,k [n], C 1,k [n], A 2,k [n], B 2,k [n], C 2,k [n] is:
[0176]
[0177] Introducing slack variables The optimization problem (P8) can be transformed into the optimization problem (P9):
[0178]
[0179] Among them, the optimization variables R k,B [n], R k,E The expression of [n] is (37) and (38), D k [n], E k The expressions of [n] are equation (39) and equation (40).
[0180] R k,B [n] = log2(1 + u 1,k [n]) (37)
[0181] R k,E [n] = log2(1 + u 2,k [n]) (38)
[0182]
[0183] In problem (P9), due to the non-convexity of constraints (36a), (36b), and (36c), (P9) is a non-convex optimization problem.
[0184] a) Processing constraints (36a)
[0185] The first-order Taylor expansion of a convex function is a property of its global lower bound, for -log2(1+u 2,k [n]) is expanded by the first order Taylor, we can get:
[0186]
[0187] b) Processing constraints (36b)
[0188] Before processing constraint (36b), determine the constant A 1,k [n], B 1,k [n], C 1,k [n] is positive or negative. According to expression (35), we can get: So A 1,k [n], C 1,k [n] is non-negative; B 1,k [n] cannot be determined by the expression, it can be positive or negative. Here, we introduce a binary vector k={1,...,K}. When B 1,k When [n]≥0, I k [n] = 0; B 1,k [n]<0,I k [n] = 1. We can get:
[0189]
[0190] In order to handle constraint (36b), a slack variable is introduced You can get:
[0191]
[0192] The first and third terms on the right hand side of the inequality sign in equation (44) are concave, while the second and fourth terms are convex. In order to convert (44) into a convex form, the SCA technique is used to perform a first-order Taylor expansion on the second and fourth terms, and we obtain:
[0193]
[0194] Among them, v1 r [n]、v 2,k r [n] is the value of the rth iteration.
[0195] Taking the squares on both sides of the inequalities in equations (45)-(48), we can obtain the following through inequality transformation:
[0196]
[0197] It can be proved that the formula ||l[n]-w|| 2 、e -2v[n] is convex, where w∈{w B ,w k},v∈{v1,v 2,k}. (50) and (51) are forms where convex functions are less than or equal to convex functions. Using SCA technology, we can perform first-order Taylor expansion on the right hand side of the inequality sign. (52) and (53) are forms where convex functions are greater than or equal to convex functions. Using SCA technology, we can perform first-order Taylor expansion on the left hand side of the inequality sign. We can get:
[0198]
[0199] c) Processing constraints (36c)
[0200] According to formula (35), we can determine A 2,k [n], C 2,k [n] is non-negative, B 2,k [n] can be positive or negative. Define a binary vector When B 2,k When [n]≥0, L k [n] = 0; B 2,k [n]<0,L k [n] = 1. We can get:
[0201]
[0202] Introducing slack variables You can get:
[0203]
[0204] Similar to the treatment of constraint (36b), we can obtain:
[0205]
[0206]
[0207] In summary, the optimization problem can be transformed into (P10):
[0208]
[0209] The optimization objectives and constraints of the above problems are all in standard convex form and can be processed using CVX.
[0210] Furthermore, an alternating optimization method is adopted to iteratively solve the four optimization problems of phase optimization of aerial intelligent reflector surface, time slot allocation optimization, IoT device transmission power and local computing frequency optimization, and UAV trajectory optimization.
[0211] For example, the IRS phase, time slot allocation, transmission power, local calculation frequency and drone trajectory are jointly optimized to maximize the minimum average safe calculation rate of the device. Combining the previous theoretical analysis and the processing of the four sub-problems, the specific algorithm steps of the loop iteration are as follows:
[0212] (1) Set system parameters, such as the drone flight time T, the number of time slots N, and the device energy consumption budget. Parameters such as system bandwidth;
[0213] (2) Set the loop number r = 0 and initialize the optimization target Convergence accuracy ε=10 -3 , initialize time slot allocation a k (r) [n], local calculation frequency f k (r) [n], the signal transmission power p of the device k (r) [n], drone position l (r) [n];
[0214] (3) Substitution Solve problem (P4) and get the local optimal solution Θ k (r+1) [n];
[0215] (4) Substitution Solve problem (P5) and get the optimal solution a k (r+1) [n];
[0216] (5) Substitution Solve problem (P7) and get the local optimal solution f k (r+1) 、p k (r+1) [n];
[0217] (6) Substitution calculate If That is, the algorithm converges to the preset threshold range and the algorithm solution is completed. Otherwise, execute step (7).
[0218] (7) Substitution Solve problem (P10) and get l (r+1) [n]. Update the number of iterations r = r + 1 and repeat steps (3) to (6).
[0219] It can be understood that dividing the target optimization problem into four sub-optimization problems and performing iterative optimization in an alternating optimization manner can effectively ensure that the optimization value can reach the required expected target optimization value. The target optimization value obtained is used as a task safety offloading strategy, which can enable the intelligent reflective surface in the air to effectively optimize the communication environment between the IoT device and the edge server according to the task safety offloading strategy, greatly reducing channel attenuation and avoiding information leakage to eavesdroppers, thereby improving the system's secure computing capabilities.
[0220] In this embodiment, IRS is used to assist secure mobile edge computing without changing the framework of the existing communication network, so that the system can be directly applied to the existing communication network. As a passive device, IRS does not increase the energy budget of the system. Through intelligent beamforming, it enhances the signal to the receiving end and effectively offsets the signal at the eavesdropping end. IRS is installed on the drone, which gives the system greater flexibility and maneuverability. By reasonably planning the flight trajectory of the drone, the position of IRS can be optimized, thereby improving the overall performance of the system. This system is not only suitable for the safe unloading of IoT devices, but can also be widely used in other data transmission scenarios that require high security and high efficiency.
[0221] Embodiment three:
[0222] This embodiment provides corresponding experimental data description for the method mentioned in Embodiment 1, which is as follows:
[0223] See also Figure 3 , to set the maximum transmission power P = 20dBm, the UAV flight time (total calculation cycle) T = 20s, the number of time slots N = 20, and the equipment energy consumption budget Precision threshold ε = 1e -3 , computer simulation results of the convergence of the proposed algorithm in the alternating iteration process. Figure 3 It can be seen that in the first three iterations, the objective function value increases rapidly; after three iterations, the growth of the objective value becomes more and more gentle; after the eighth iteration, the objective function value tends to converge. This shows that the proposed algorithm can reach convergence in a smaller number of iterations.
[0224] like Figure 4 As shown in the figure, the optimized trajectory of the drone is shown, where the base station (edge server) is located at (0,0), the locations of the IoT devices are (160,60), (160,0) and (160,-60), and the eavesdropper is located at (10,0). The drone equipped with the smart reflective surface flies from the initial position (80,80) to the final position (80,-80). Figure 4As can be seen in the figure, the drone's trajectory first flies close to the base station, then tries to stay near the base station, and finally flies back to the end point. This flight trend becomes more obvious when the flight time increases. Benefiting from the passive beamforming of IRS, the drone can weaken the signal from the device to the eavesdropper, so the IRS can get closer to the eavesdropper, effectively improving the security performance of the system.
[0225] like Figure 5 The figure shows the performance comparison between the proposed algorithm and the baseline scheme. It can be seen that the proposed algorithm outperforms all the baseline schemes. This shows that the system performance can be significantly improved by jointly optimizing time slot allocation, device transmit power, local calculation frequency, IRS passive beamforming, and drone trajectory.
[0226] Compared with the proposed algorithm, the performance of the unoptimized UAV trajectory (IRS position) is significantly reduced, indicating that the optimization of UAV trajectory plays a key role in improving system performance. Especially when the transmission power is high, the effect of trajectory optimization is more significant; when there is no IRS, the system's security calculation rate is the lowest, indicating that IRS plays an important role in improving the channel environment and offsetting eavesdropping signals; when the maximum transmission power is small, the proposed algorithm is similar to the comparison scheme of unoptimized time slot allocation and unoptimized calculation frequency, but our scheme is still better than the comparison scheme. In the comparison scheme of unoptimized transmission power, the performance is poor because the transmission power is the initial value, but it is still better than the scheme without IRS.
[0227] Embodiment 4:
[0228] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, steps of an edge computing information security optimization method are implemented.
[0229] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. An edge computing information security optimization system, characterized in that: The system comprises: Edge servers, eavesdroppers, smart reflective surfaces carried by drones, and several IoT devices; Each of the IoT devices is used to perform local computation on a computing task or wirelessly transmit part of the computing task to an edge server; The edge server is used to receive and process computing task data transmitted from the IoT device; The eavesdropper is used to simulate eavesdropping on computing task data transmitted by IoT devices; The intelligent reflective surface carried by the drone optimizes the communication environment between IoT devices and edge servers to prevent possible eavesdroppers.
2. An edge computing information security optimization method, the method is used to implement the function of the system described in claim 1, characterized in that: The method comprises the following steps: IoT devices implement partial offloading strategies, offloading some computing tasks to edge servers for processing; The goal is to maximize the minimum safe computing rate of IoT devices and construct a target optimization problem. Solving the target optimization problem to obtain a task safety offloading strategy; According to the task security offloading strategy, the aerial intelligent reflective surface assists the IoT device to safely offload part of the computing tasks to the edge server for processing.
3. The edge computing information security optimization method according to claim 2 is characterized in that: The process of constructing the target optimization problem includes: Based on the aerial intelligent reflective surface safe mobile edge computing system, a three-dimensional Cartesian coordinate system is constructed; Based on the three-dimensional Cartesian coordinate system, communication constraints are set for the communication between the edge server, the eavesdropper, the intelligent reflective surface carried by the drone, and a number of IoT devices; and computing constraints are set for the IoT devices and edge servers in processing computing tasks respectively; Based on communication constraints and computational constraints, a target optimization problem is constructed.
4. The edge computing information security optimization method according to claim 3 is characterized in that: The process of setting communication constraints on the communication between the edge server, the eavesdropper, the smart reflective surface carried by the drone, and several IoT devices includes: Given the initial and final positions of the drone, the flight time of the drone is divided into N time slots, and the trajectory constraints of the drone are set. The expression is: ‖l[n+1]-l[n]‖≤D max ,n=1,2,...,N-1 l[1]=l0 l F =l[N]; Set the transmission power constraint of IoT devices during data transmission. The expression is: 0≤p k [n]≤P k ; In order to prevent mutual interference of information transmission of IoT devices, the system uses time division multiple access to communicate, and the expression is: The channel gain between the IoT device and the aerial intelligent reflector in time slot n is expressed as: The channel gains from the airborne intelligent reflection surface to the edge server and the eavesdropper in time slot n are expressed as: The channel gains from IoT devices to edge servers and eavesdroppers are expressed as: Based on the aforementioned constraints and channel gains, the safe offloading rate of the IoT device offloading part of the computing tasks to the edge server is obtained, which is expressed as: Where l[n] represents the horizontal coordinate of the drone at time slot n, n∈{1,...,N}; k represents the kth IoT device, p k [n] represents the transmission power of IoT device k in time slot n, P k represents the maximum transmission power of IoT device k; t k [n] = δa k [n] represents the time allocated to IoT device k in time slot n; ρ represents the path loss per unit distance under free space loss; d k,I represents the distance between IoT device k and the intelligent reflective surface in the air; d I,B represents the distance between the aerial intelligent reflective surface and the edge server; d k,B represents the distance between IoT device k and the edge server; d k,E represents the distance between IoT device k and the eavesdropper; denotes a random variable with circularly symmetric complex Gaussian distribution with zero mean and unit variance; {x} + It means x=max{0,x}.
5. The edge computing information security optimization method according to claim 4, characterized in that: The process of setting computational constraints on IoT devices and edge servers involved in processing computational tasks includes: Set the local computing amount of IoT device k, the expression is: Set the computation frequency constraint of IoT device k, the expression is: Set the computing capacity constraint of the edge server, the expression is: Set the energy consumption constraint of IoT device k in the entire computing cycle, the expression is: Among them, f k Calculate the CPU frequency for device k; Calculates the maximum CPU frequency for the server; c B The number of CPU cycles required for the server to calculate one bit of data; represents the energy consumption of device k during local computing; E off,k =a k [n]δp k [n] represents the energy consumption during the computing task offloading process in time slot n; R k,B [n] represents the task offloading rate between IoT device k and edge server in time slot n.
6. The edge computing information security optimization method according to claim 5, characterized in that: Based on communication constraints and computational constraints, the target optimization problem is constructed with the goal of maximizing the minimum average secure computation rate of IoT devices. The secure computation rate expression of IoT device k is: Among them, R k,sec [n] represents the secure offloading rate of IoT device k in time slot n.
7. The edge computing information security optimization method according to claim 6, characterized in that: The process of solving the target optimization problem includes: The target optimization problem is divided into four optimization problems: phase optimization of the aerial intelligent reflector, time slot allocation optimization, IoT device transmission power and local computing optimization, and UAV trajectory optimization. The four optimization problems of phase optimization of the aerial intelligent reflector, time slot allocation optimization, IoT device transmission power and local computing optimization, and drone trajectory optimization are solved iteratively. When the target value reaches the convergence accuracy, the task security offloading strategy between the IoT device and the edge server is obtained. According to the task security offloading strategy, the aerial intelligent reflective surface assists the IoT device to safely offload part of the computing tasks to the edge server for processing.
8. The edge computing information security optimization method according to claim 7, characterized in that: The target optimization problem is divided into four optimization sub-problems: phase optimization of the intelligent reflector in the air, time slot allocation optimization, IoT device transmission power and local computing frequency optimization, and drone trajectory optimization. The process includes: Fixed time slot allocation, UAV trajectory, transmission power and local calculation frequency, optimize the phase of the aerial intelligent reflector, the expression is: Fix the phase of the aerial intelligent reflector, the transmission power, the local calculation frequency and the trajectory of the UAV, and optimize the time slot allocation. The expression is: Fixed time slot allocation, airborne intelligent reflector phase and drone trajectory, jointly optimize the transmission power and local calculation frequency, the expression is: Fixed time slot allocation, phase of the airborne intelligent reflector, transmission power and local calculation frequency, optimize the trajectory of the UAV, the expression is:
9. The edge computing information security optimization method according to claim 7, characterized in that: An alternating optimization method is adopted to iteratively solve the four optimization problems of phase optimization of aerial intelligent reflector, time slot allocation optimization, IoT device transmission power and local computing frequency optimization, and UAV trajectory optimization.
10. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, the steps of the method according to any one of claims 2 to 9 are implemented.
