Dynamic task security acceptance method and system based on cloud edge collaboration

Abstract

The invention provides a dynamic task security acceptance method and system based on cloud edge collaboration. The method comprises the following steps: determining different task groups according to data information uploaded by an edge-end layer user; constructing a task acceptor model by utilizing a multi-objective function based on data attributes in the task group; based on the task acceptor model, task acceptance allocation is carried out through a multi-target edge server selection algorithm, and an optimal task acceptor is selected for a target task; and distributing the target task to a cloud server, an edge server or local equipment of an edge layer for task execution processing by utilizing the optimal task acceptor in combination with safety degree analysis. According to the method, the multi-objective optimization algorithm is utilized, and the safety of the system is guaranteed while the performance indexes such as energy consumption, delay and cost are met.

Description

Technical Field

[0001] The present invention relates to the field of cloud-edge computing and network security, and in particular to a dynamic task security acceptance method and system based on cloud-edge collaboration. Background Art

[0002] With the rapid development of the Internet of Things and mobile Internet, the demand for the generation and processing of massive data is increasing. Cloud computing and edge computing play an increasingly important role in data processing and service provision. Although the traditional cloud computing architecture can provide powerful computing and storage capabilities, its centralized structure leads to serious problems of data transmission delay and bandwidth consumption. The cloud-edge collaborative computing model can effectively reduce latency and improve data processing efficiency by distributing computing resources in the cloud and edge nodes. Data interaction between devices may face security risks, and the risks of tampering and leakage during data transmission cannot be ignored. Therefore, it is of great practical significance to propose a solution that can effectively coordinate cloud and edge computing resources, which can not only improve data processing efficiency but also enhance system security.

[0003] In the prior art, edge computing can significantly reduce the delay of data transmission and provide faster response due to its characteristics of being close to the data source. Now some researchers have considered task offloading in the cloud-edge collaborative environment, and achieved the maximum task offloading efficiency with the minimum resource consumption rate through the cloud and edge collaborative mechanism. For example, the published Chinese patents "Power task offloading method, device and computer equipment under cloud-edge collaboration" (publication number CN118093165A), "A cloud-edge resource collaborative scheduling method and system for computing network integration" (publication number CN117950860A), and "An edge computing offloading method based on the Internet of Things scenario" (publication number CN116827992A). However, on the one hand, the resources of edge computing nodes are limited, and edge computing alone cannot meet the needs of all tasks; on the other hand, cloud computing provides powerful computing and storage capabilities, but due to the long distance from the data source, there are certain delays and bandwidth limitations. Therefore, cloud-edge collaborative computing has become an effective solution to this problem. Through a reasonable task offloading strategy, tasks are dynamically allocated between cloud and edge nodes, which can fully utilize the advantages of both and achieve efficient and low-latency data processing.

[0004] Although some teams have studied the task scheduling strategy in cloud-edge collaboration, such as the published Chinese patent "A method for perceiving task resource requirements for cloud-edge collaboration scenarios" (publication number CN117135131A), the invention predicts resource requirements based on the type of task and task performance, and then makes quota decisions on the resources required for the task based on the resource status in the central cloud or edge cloud and the priority of the task. Finally, the decision plan is uploaded to the central cloud or edge cloud resource scheduling module to achieve the purpose of task processing. However, the security of the edge server is not considered in this process. If the security of the edge server is not guaranteed in this process, it may cause data and user privacy leakage. In practical applications, the task offloading decision not only involves the allocation of computing and communication resources, but also needs to consider the security of the system. Traditional optimization methods are difficult to achieve good results in complex industrial environments due to the lack of comprehensive consideration of security.

[0005] Therefore, how to ensure the security of the system while meeting performance indicators such as energy consumption, time and cost is an urgent problem to be solved. Summary of the invention

[0006] In order to solve the problem of how to ensure system security while meeting performance indicators such as energy consumption, delay and cost in the prior art, the present invention proposes a dynamic task security acceptance method based on cloud-edge collaboration, including:

[0007] Determine different task groups based on the data information uploaded by edge layer users;

[0008] Based on the data attributes in the task group, a task acceptor model is constructed using a multi-objective function;

[0009] Based on the task acceptor model, task acceptance allocation is performed through a multi-target edge server selection algorithm to select the best task acceptor for the target task;

[0010] Utilize the best task acceptor combined with safety analysis to distribute the target tasks to cloud servers, edge servers or local devices at the edge layer for task execution and processing.

[0011] Optionally, determining different task groups according to data information uploaded by edge layer users includes:

[0012] According to the data information uploaded by edge layer users to edge layer sensors, different task groups are divided according to data types and processing requirements. Each task group consists of multiple subtasks. Suppose the user mobile device set is U = {u1, u2, ..., u n}, the task group is T = {T1, T2, T3, ..., T N}, subtask is T n =(s n, c n , d n ), where s n Represents subtask T n The size of c n Represents subtask T n The amount of calculation; d n Indicates completion of task T n The maximum acceptable delay.

[0013] Optionally, based on the data attributes in the task group, a task acceptor model is constructed using a multi-objective function, including:

[0014] Based on the data attributes of each task group, the execution energy consumption, transmission energy consumption, execution time, transmission time and transmission cost of the subtask are calculated;

[0015] Based on the execution energy consumption, transmission energy consumption, execution time, transmission time and transmission cost of the subtask, the total energy consumption, total delay and total cost objective functions are calculated in combination with the task offloading decision;

[0016] The task acceptor model is constructed according to the objective function and constraints of minimizing total energy consumption, total delay and total cost.

[0017] Optionally, the calculation of the execution energy consumption, transmission energy consumption, execution time, transmission time and transmission cost of the subtask based on the data attributes of each task group includes:

[0018] Calculate the energy consumption of the local execution of the subtask based on the computational load of the subtask, the computing power of the local device, and the coefficient of the device hardware architecture;

[0019] The transmission energy consumption of the subtask is calculated according to the transmission power of the subtask when unloading, the data size of the subtask, the transmission rate of the subtask, the interference ratio of the subtask during the transmission period, and the efficiency of the power amplifier of the local device;

[0020] Calculate the local execution time of the subtask according to the computation amount of the subtask, the computing power of the local device, and the resource utilization rate of the local device;

[0021] The time for the subtask to be executed non-locally is calculated based on the computational effort of the subtask and the computing power of the edge server;

[0022] The transmission time of the subtask is calculated according to the data size of the subtask, the transmission rate of the subtask and the interference ratio of the subtask during the transmission period;

[0023] The transmission cost of the subtask is calculated based on all the costs incurred by the subtask during the transmission process.

[0024] Optionally, the task acceptance allocation is performed based on the task acceptor model by using a multi-target edge server selection algorithm to select the best task acceptor for the target task, including:

[0025] Based on the task acceptor model, a multi-target edge server selection algorithm with congestion distance is added to select the best task acceptor for the target task according to the task load, transmission power and resource allocation of the task acceptor.

[0026] The multi-target edge server selection algorithm with crowding distance includes the following steps: encoding, initialization, non-dominated sorting, crowding distance calculation, elite retention strategy, selection, crossover and mutation.

[0027] Optionally, the crowding distance is calculated as follows:

[0028]

[0029] Among them, P[i] distance represents the crowding distance; P[i] E represents the crowding distance of energy consumption; P[i] TA represents the delayed crowding distance; P[i] C represents the crowding distance of the cost; E[i+1] and E[i-1] represent the values ​​of the two individuals adjacent to the i-th individual in the energy consumption objective function; TA[i+1] and TA[i-1] represent the values ​​of the two individuals adjacent to the i-th individual in the delay objective function; C[i+1] and C[i-1] represent the values ​​of the two individuals adjacent to the i-th individual in the cost objective function; F represents the number of all individuals in the layer where the current individual is located.

[0030] Optionally, after performing task acceptance allocation based on the task acceptor model and selecting the best task acceptor for the target task through a multi-target edge server selection algorithm, the method further comprises:

[0031] When the number of tasks processed by the task acceptor reaches the set maximum task processing number threshold, the task acceptor is reselected for the target task and optimized; otherwise, each time the task acceptor assigns a task, the task counter of the task acceptor is updated.

[0032] Optionally, reselecting a task acceptor for the target task and optimizing it includes:

[0033] Selecting a new task acceptor for the target task based on a multi-target edge server selection algorithm, and resetting the task counter of the task acceptor;

[0034] Evaluate the new task acceptor based on a multi-objective function, and dynamically adjust the task processing quantity threshold of the new task acceptor according to the evaluation result;

[0035] The minimization multi-objective function value of the new task acceptor is monitored and optimized in real time.

[0036] Optionally, the method of utilizing the best task acceptor in combination with safety analysis to allocate the target task to a cloud server, an edge server or a local device at the edge layer for task execution processing includes:

[0037] Based on the best task acceptor, prioritize the evaluation of whether the comprehensive security of local equipment meets security standards;

[0038] When the comprehensive security level of the local device meets the security standards, the target task is offloaded to the local device for task execution processing;

[0039] When the comprehensive security level of the local device does not meet the security standards, the edge cloud fuzzy reasoning method is used to allocate the target task to the cloud server or edge server based on the judgment factors.

[0040] Optionally, the evaluating the comprehensive security of the local device includes:

[0041] Total security value TV of local equipment l t , calculated as follows:

[0042]

[0043] Among them, TV comp Indicates the computing power safety value; TV sys Indicates the system integrity security value; TV data Indicates the data protection capability security value.

[0044] Optionally, the computing capacity safety value TV comp , calculated as follows:

[0045]

[0046] in, Indicates the computing power of the local device; c n represents the computational effort of the task; σ com Indicates the standard deviation of computing power.

[0047] Optionally, the system integrity security value TV sys , calculated as follows:

[0048] TV sys =min(S os , Spatch , S smalware )

[0049] Among them, S os Indicates the security score of the operating system; S patch Indicates the timeliness score of the patch; S smalware Indicates the malware protection score.

[0050] Optionally, the data protection capability security value TV data , calculated as follows:

[0051]

[0052] Among them, S enc Indicates the encryption strength score; c enc represents the mean value of encryption strength; σ enc Indicates the standard deviation of encryption strength.

[0053] Optionally, the edge cloud fuzzy inference method is used to allocate the target task to the cloud server or the edge server according to the determination factor, including:

[0054] The resource utilization, transmission power and task delay of the edge server are used as the input elements of the edge cloud fuzzy reasoning method, and edge cloud fuzzification, edge cloud fuzzy reasoning and edge cloud defuzzification are performed in sequence to obtain the decision of assigning the target task to the cloud server or edge server for execution and processing.

[0055] Optionally, the step of sequentially performing edge cloud fuzzification, edge cloud fuzzy reasoning, and defuzzification of edge cloud includes:

[0056] The membership function and non-membership function of the input and output variables of the resource utilization, transmission power and task delay of the edge server are established by using trapezoidal membership function and triangular membership function to perform edge cloud fuzzification.

[0057] Using fuzzy inference system to infer the membership and non-membership functions of input variables;

[0058] The center of gravity method of edge cloud defuzzification is used to transform the fuzzy result obtained by the fuzzy reasoning system into a clear decision value, and the edge cloud fuzzy reasoning result of the target task is output.

[0059] Optionally, obtaining a decision of allocating the target task to a cloud server or an edge server includes:

[0060] If the edge cloud fuzzy inference result of the target task is greater than the threshold r1, the target task is assigned to the cloud server for processing; otherwise, the target task is assigned to the edge server for processing;

[0061] If the edge cloud fuzzy inference result of the target task is greater than the threshold r2, the target task is assigned to the adjacent edge server for processing; otherwise, the target task is assigned to the local edge server for processing.

[0062] Optionally, the edge cloud fuzzy inference method is used to allocate the target task to the cloud server or the edge server according to the determination factor, including:

[0063] Based on the edge server selected by the target task, a security value evaluation is performed on the message receiving rate, message transmission rate and message tampering rate of the edge server node, and whether the security value of the current edge server node period meets the final security value is determined according to the final security value evaluation result;

[0064] If the security value of the edge server node meets the final security value, the security value evaluation result of the edge server node is used as the historical final security value evaluation result of the next cycle, and the target task is deployed to the edge server node, and the security value of the edge server node is updated at the same time;

[0065] Otherwise, the edge server node is isolated, and a security value evaluation is performed on the message acceptance rate, message transmission rate, and message tampering rate of the adjacent edge server nodes until an edge server node that meets the final security value evaluation result is found.

[0066] Optionally, the final safety value assessment result of the message reception rate of the edge server node is obtained according to the safety value assessment result of the message reception rate of the edge server node in the current cycle, the final safety value assessment result of the message reception rate in the previous cycle, the influencing factor of the network environment change, the influencing factor of the task urgency, and the weight of each assessment result;

[0067] The final security value evaluation result of the message transmission rate of the edge server node is calculated based on the security value evaluation result of the message transmission rate of the edge server node in the current cycle, the final security value evaluation result of the message transmission rate in the previous cycle, the user's behavior influencing factor, the influencing factor of the network congestion situation, and the weight of each evaluation result to obtain the final security value evaluation result of the message transmission rate of the edge server node;

[0068] The final security value assessment result of the message tampering rate of the edge server node is obtained based on the security value assessment result of the message tampering rate of the edge server node in the current cycle, the final security value assessment result of the message tampering rate in the previous cycle, the influencing factor of historical tampering behavior, the influencing factor of the security event correlation and the weight of each assessment result.

[0069] Based on the same inventive concept, the present invention also provides a dynamic task security acceptance system based on cloud-edge collaboration, including:

[0070] The task grouping module is used to determine different task groups based on the data information uploaded by edge layer users;

[0071] A model building module, used for building a task acceptor model based on data attributes in the task group using a multi-objective function;

[0072] An acceptor selection module, used to perform task acceptance allocation based on the task acceptor model through a multi-target edge server selection algorithm, and select the best task acceptor for the target task;

[0073] The task allocation module is used to use the best task acceptor combined with safety analysis to allocate the target task to the cloud server, edge server or local device at the edge layer for task execution processing.

[0074] Optionally, the task grouping module is specifically used to:

[0075] According to the data information uploaded by edge layer users to edge layer sensors, different task groups are divided according to data types and processing requirements. Each task group consists of multiple subtasks. Suppose the user mobile device set is U = {u1, u2, ..., u n}, the task group is T = {T1, T2, T3, ..., T N}, subtask is T n =(s n , c n , d n ), where s n Represents subtask T n The size of c n Represents subtask T n The amount of calculation; d n Indicates completion of task T n The maximum acceptable delay.

[0076] Optionally, the model building module includes:

[0077] A calculation unit, for calculating the execution energy consumption, transmission energy consumption, execution time, transmission time and transmission cost of the subtasks based on the data attributes of each task group; and calculating the total energy consumption, total delay and total cost objective functions in combination with the task offloading decision based on the execution energy consumption, transmission energy consumption, execution time, transmission time and transmission cost of the subtasks;

[0078] The construction unit is used to construct a task acceptor model according to the minimization objective function and constraint conditions of total energy consumption, total delay and total cost.

[0079] Optionally, the receiver selection module includes:

[0080] An algorithm unit, for selecting a multi-target edge server based on a task acceptor model by adding a crowding distance;

[0081] A selection unit, used for selecting an optimal task acceptor for a target task according to the task load, transmission power and resource allocation of the task acceptor;

[0082] The multi-target edge server selection algorithm with crowding distance includes the following steps: encoding, initialization, non-dominated sorting, crowding distance calculation, elite retention strategy, selection, crossover and mutation.

[0083] Optional, including:

[0084] The acceptor optimization module is used to reselect and optimize the task acceptor for the target task when the task processing quantity of the task acceptor reaches the set maximum task processing quantity threshold;

[0085] The timer update module is used to update the task counter of the task acceptor each time the task acceptor assigns a task when the task processing quantity of the task acceptor does not reach the set maximum task processing quantity threshold.

[0086] Optionally, the receiver optimization module includes:

[0087] A reset unit, configured to select a new task acceptor for the target task based on a multi-target edge server selection algorithm, and reset a task counter of the task acceptor;

[0088] An evaluation unit, used for evaluating the new task acceptor based on a multi-objective function;

[0089] The adjustment unit is used to dynamically adjust the task processing quantity threshold of the new task acceptor according to the evaluation result, and to monitor and optimize the minimized multi-objective function value of the new task acceptor in real time.

[0090] Optionally, the task allocation module includes:

[0091] A local assessment unit, used to assess the comprehensive security of local equipment;

[0092] A local judgment unit, used to judge whether the comprehensive security of the local device meets the security standards;

[0093] A local allocation unit, used for offloading the target task to the local device for task execution processing when the comprehensive safety degree of the local device meets the safety standard;

[0094] The non-local allocation unit is used to allocate the target task to the cloud server or edge server according to the judgment factors by using the edge cloud fuzzy reasoning method when the comprehensive security degree of the local device does not meet the security standards.

[0095] Optionally, the non-local allocation unit includes:

[0096] The reasoning subunit is used to use the resource utilization, transmission power and task delay of the edge server as input elements of the edge cloud fuzzy reasoning method, and sequentially perform edge cloud fuzzification, edge cloud fuzzy reasoning and defuzzification of the edge cloud;

[0097] The decision subunit is used to generate a decision on assigning the target task to a cloud server or edge server for execution and processing.

[0098] Optionally, the reasoning subunit is specifically used for:

[0099] The membership function and non-membership function of the input and output variables of the resource utilization, transmission power and task delay of the edge server are established by using trapezoidal membership function and triangular membership function to perform edge cloud fuzzification.

[0100] Using fuzzy inference system to infer the membership and non-membership functions of input variables;

[0101] The center of gravity method of edge cloud defuzzification is used to transform the fuzzy result obtained by the fuzzy reasoning system into a clear decision value, and the edge cloud fuzzy reasoning result of the target task is output.

[0102] Optionally, the decision subunit is specifically used to:

[0103] If the edge cloud fuzzy inference result of the target task is greater than the threshold r1, the target task is assigned to the cloud server for processing; otherwise, the target task is assigned to the edge server for processing;

[0104] If the edge cloud fuzzy inference result of the target task is greater than the threshold r2, the target task is assigned to the adjacent edge server for processing; otherwise, the target task is assigned to the local edge server for processing.

[0105] Optional, including:

[0106] An edge evaluation subunit, configured to perform security value evaluation on a message receiving rate, a message transmission rate, and a message tampering rate of an edge server node based on the edge server selected by the target task;

[0107] The edge judgment subunit is used to determine whether the security value of the current edge server node cycle meets the final security value based on the final security value evaluation result; if the security value of the edge server node meets the final security value, the security value evaluation result of the edge server node is used as the historical final security value evaluation result of the next cycle, and the target task is deployed to the edge server node, and the security value of the edge server node is updated at the same time; otherwise, the edge server node is isolated, and the message acceptance rate, message transmission rate and message tampering rate of the adjacent edge server nodes are evaluated for security values ​​until an edge server node that meets the final security value evaluation result is found.

[0108] In another aspect, the present application further provides an electronic device, comprising: at least one processor and a memory; the memory and the processor are connected via a bus;

[0109] The memory is used to store one or more programs;

[0110] When the one or more programs are executed by the at least one processor, a dynamic task security acceptance method based on cloud-edge collaboration as described above is implemented.

[0111] On the other hand, the present application also provides a readable storage medium on which an execution program is stored. When the execution program is executed, a dynamic task security acceptance method based on cloud-edge collaboration as described above is implemented.

[0112] Compared with the prior art, the present invention has the following beneficial effects:

[0113] The present invention provides a method for dynamic task security acceptance based on cloud-edge collaboration, including: determining different task groups according to data information uploaded by edge layer users; constructing a task acceptor model based on data attributes in the task group using a multi-objective function; performing task acceptance assignment based on the task acceptor model using a multi-objective edge server selection algorithm to select the best task acceptor for the target task; using the best task acceptor combined with security analysis to assign the target task to a cloud server, edge server or local device at the edge layer for task execution processing. The present invention ensures the security of the system while meeting performance indicators such as energy consumption, delay and cost by using a multi-objective optimization algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0114] Figure 1 A flow chart of a dynamic task security acceptance method based on cloud-edge collaboration provided by the present invention;

[0115] Figure 2 A schematic diagram of the overall process of a dynamic task security acceptance method based on cloud-edge collaboration provided by the present invention;

[0116] Figure 3 A schematic diagram of a process flow for selecting the best task acceptance provided by the present invention;

[0117] Figure 4 A schematic diagram of the process flow of updating and optimizing the acceptor provided by the present invention;

[0118] Figure 5 A schematic diagram of the process of edge server security value node judgment provided by the present invention;

[0119] Figure 6 A schematic diagram of the overall structure of a dynamic task security acceptance system based on cloud-edge collaboration provided by the present invention;

[0120] Figure 7 A schematic diagram of an electronic device for a dynamic task security acceptance method based on cloud-edge collaboration provided by the present invention. DETAILED DESCRIPTION

[0121] The present invention proposes a dynamic task security acceptance method based on cloud-edge collaboration, which uses a multi-objective edge service selection algorithm, uses an edge-cloud fuzzy reasoning method to handle the uncertainty of edge services and cloud servers, and uses security analysis to determine the security of local devices and edge servers, thereby achieving dynamic task allocation and security assurance. The specific embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings.

[0122] Embodiment 1:

[0123] A dynamic task security acceptance method based on cloud-edge collaboration, such as Figure 1 As shown, including:

[0124] Step 1: Determine different task groups based on the data information uploaded by edge layer users;

[0125] Step 2: Based on the data attributes in the task group, a task acceptor model is constructed using a multi-objective function;

[0126] Step 3: Based on the task acceptor model, the task acceptance allocation is performed through a multi-target edge server selection algorithm to select the best task acceptor for the target task;

[0127] Step 4: Utilize the best task receiver combined with safety analysis to assign the target task to the cloud server, edge server or local device at the edge layer for task execution and processing.

[0128] Below Figure 2 Take this as an example to explain this implementation in detail:

[0129] Step 1: Determining different task groups according to the data information uploaded by edge layer users includes:

[0130] According to the data information uploaded by edge layer users to edge layer sensors, different task groups are divided according to data types and processing requirements. Each task group consists of multiple subtasks. Suppose the user mobile device set is U = {u1, u2, ..., u n}, the task group is T = {T1, T2, T3, ..., T N}, subtask is T n =(s n , c n , d n ), where s n Indicates the size of subtask n; c n represents the computational effort of subtask n; d n represents the maximum acceptable delay in completing task n.

[0131] Step 2: Based on the data attributes in the task group, a task acceptor model is constructed using a multi-objective function, including:

[0132] Based on the data attributes of each task group, the execution energy consumption, transmission energy consumption, execution time, transmission time and transmission cost of the subtask are calculated;

[0133] Based on the execution energy consumption, transmission energy consumption, execution time, transmission time and transmission cost of the subtask, the total energy consumption, total delay and total cost objective functions are calculated in combination with the task offloading decision;

[0134] The task acceptor model is constructed according to the objective function and constraints of minimizing total energy consumption, total delay and total cost.

[0135] A task acceptor model is constructed based on the data attributes of each task group and the task acceptor selection problem. The task acceptor selection problem is defined as a multi-objective optimization problem, and the objective functions of total energy consumption, total delay and total cost are set.

[0136] The objective function E of the total energy consumption is calculated as follows:

[0137]

[0138] Where N represents the total number of tasks; x n A binary decision variable indicating whether the nth task is offloaded to the edge node for processing, x n =1 means the task is offloaded, x n =0 means the task is executed locally; represents the energy consumption of the nth task executed locally; Represents the transmission energy consumption of the nth task.

[0139] The objective function TA of the total delay is calculated as follows:

[0140]

[0141] in, Indicates the time when the nth task is executed locally; Indicates the transmission time of the nth task; Indicates the time when the nth task was executed non-locally.

[0142] The objective function C of the total cost is calculated as follows:

[0143]

[0144] Among them, r n Represents the transmission cost of the nth task.

[0145] The calculation of the execution energy consumption, transmission energy consumption, execution time, transmission time and transmission cost of the subtask based on the data attributes of each task group includes:

[0146] The energy consumption of executing the subtask locally is calculated based on the computational amount of the subtask, the computing power of the local device, and the coefficient of the device hardware architecture. The energy consumption of executing the subtask locally is calculated as follows:

[0147]

[0148] in, represents the energy consumption of the nth task executed locally; represents the computing power of the nth local device; k represents the coefficient of the device hardware architecture; c n Indicates the computational effort of the nth task;

[0149] The transmission energy consumption of the subtask is calculated according to the transmission power of the subtask when unloading, the data size of the subtask, the transmission rate of the subtask, the interference ratio of the subtask during the transmission period, and the efficiency of the power amplifier of the local device. The transmission energy consumption of the subtask is calculated according to the following formula:

[0150]

[0151] in, represents the transmission energy consumption of the nth task; y n represents the transmission power of the nth task when unloading; s n Indicates the data size of the nth task; Indicates the interference ratio of the nth task during the transmission period, including factors such as signal interference from other users; It represents the efficiency of the power amplifier of the nth local device, which is affected by factors such as device aging and temperature changes; Indicates the transmission rate of the nth task;

[0152] The transmission rate of the nth task Calculate as follows:

[0153]

[0154] The channel bandwidth B is divided into M orthogonal sub-channels, denoted as SC = {1, 2, ..., M}. The mobile device can choose any sub-channel to offload its task to the edge cloud, or choose to execute it locally. When choosing local execution, y n =0 means that no transmission power allocation is required; let x n ={0, 1} for user u n The uninstall decision variable for mobile devices, x n =0 means user u n The mobile device selects a local solution task, x n = 1 indicates that the task should be transmitted through the selected subchannel sc n ∈SC to unload; h n represents the channel gain between the user end and the edge cloud; σ 2 represents the background noise power; represents the sum of interference powers of other users using the same subchannel as the user;

[0155] The local execution time of the subtask is calculated according to the computation amount of the subtask, the computation capability of the local device, and the resource utilization rate of the local device. The local execution time of the subtask is calculated as follows:

[0156]

[0157] in, Indicates the time when the nth task is executed locally; Indicates the resource utilization of the nth local device;

[0158] The non-local execution time of the subtask is calculated according to the computational amount of the subtask and the computing capacity of the edge server. The non-local execution time of the subtask is calculated as follows:

[0159]

[0160] in, Indicates the time when the nth task is executed non-locally; Indicates the computing power of the edge server;

[0161] The transmission time of the subtask is calculated according to the data size of the subtask, the transmission rate of the subtask and the interference ratio of the subtask during the transmission period. The transmission time of the subtask is calculated as follows:

[0162]

[0163] in, Indicates the transmission time of the nth task;

[0164] The transmission cost of the subtask is calculated according to all the costs incurred by the subtask during the transmission process. The transmission cost of the subtask is calculated as follows:

[0165]

[0166] Here, m represents the cost, such as communication cost, computing resource cost, storage and bandwidth cost, service quality cost, maintenance and operation cost, etc.

[0167] The constraints for minimizing the objective function are as follows:

[0168]

[0169]

[0170]

[0171]

[0172]

[0173] Among them, p max Indicates the maximum transmission power; sc n Indicates the subchannel number selected when task n is offloaded; D n represents the delay of task n, that is, d n represents the maximum acceptable delay in completing task n.

[0174] Step 3: Based on the task acceptor model, the multi-target edge server selection algorithm is used to perform task acceptance allocation, and the best task acceptor is selected for the target task, wherein the target task is assigned to the best task acceptor for task acceptance according to the constructed task acceptor model, which lays the foundation for the best task acceptor to combine the security evaluation of local devices and edge servers and select the final task execution processing location for the target task, so as to achieve the dynamic optimization acceptance of the target task while ensuring the security of the target task during execution and processing. Specifically, it includes:

[0175] Based on the task acceptor model, a multi-target edge server selection algorithm with congestion distance is added to select the best task acceptor for the target task according to the task load, transmission power and resource allocation of the task acceptor.

[0176] The multi-target edge server selection algorithm with crowding distance includes the following steps: encoding, initialization, non-dominated sorting, crowding distance calculation, elite retention strategy, selection, crossover and mutation.

[0177] Specifically, Figure 3 As shown, for the multi-objective edge service selection algorithm used in the selection of the task acceptor in the present invention, we use a multi-objective optimization algorithm to select the edge server. The algorithm involves the following key steps: encoding, initializing the population, non-dominated sorting, crowding distance calculation, elite retention strategy, selection, crossover and mutation. In order to improve the convergence speed and computational efficiency, in this algorithm, we combine the objective functions of the total energy consumption, total delay and total cost of the task to ensure that the task acceptor finally selected can optimize the service quality;

[0178] Coding and initializing the population: In the task acceptor, the encoding of the multi-objective edge service selection algorithm adopts a binary coding scheme based on task load, transmission power and resource allocation. Each gene represents the selection of an edge server node, and each allocation scheme of the task acceptor represents an individual in the population; in the initialization stage, the initial population is P0, the initialization generation G = 1, the current population Q0 = P0, and the initial population size N is set according to the problem scale and computing resources. pop and the number of iterations N gen , counter t=0, index i=0, used to select individuals in the population; first check whether the first generation of offspring has been generated, if not, regenerate the population makingNewPop(P0), perform selection, crossover and mutation operations based on the initialized population P0, and generate the first generation offspring population Q0; if the first offspring is generated, record the number of generations G=2.

[0179] Non-dominated sorting: Using the fast non-dominated sorting algorithm fastNondominatedSort (R t ) Sort the individuals in the population in layers, each layer is denoted as F, where R t The parent population P t and the offspring population Q t The result of the merger; the level of non-dominated solutions of each individual determines its hierarchical position in the population; specifically, the number of solutions dominated by each individual is calculated, and the solution set dominated by the individual is calculated. When adding individuals in the non-dominated level to the new population P t+1 When the new population P is checked t+1 The size of the current layer F i Does the size of the population not exceed the maximum capacity N of the population? pop , until the population R t Completely divided into multiple levels.

[0180] Crowding distance calculation: In the same layer, crowding distance is used to calculate crowdingDistance (F i ) Sort the individuals; the crowding distance indicates the sparseness of the individuals in the target space. The larger the value, the farther the individual is from other individuals. The more unique the individual is in the target space, the less likely it is to be replaced by other individuals, that is, the better; the crowding calculation formula is to be able to sort the individuals in the same layer, and the crowding distance of each individual needs to be calculated. The crowding distance can be obtained by calculating the sum of the distance differences between two adjacent individuals on each sub-objective function. The present invention selects the best task receiver for the target task by utilizing a multi-objective edge service selection algorithm, and adds crowding distance based on the traditional sorting algorithm to further optimize the selection of individuals, maintain the diversity of individuals, and avoid the solution from falling into the local optimum.

[0181] The crowding distance is calculated as follows:

[0182]

[0183] Among them, P[i] distance represents the crowding distance; P[i] E The crowding distance represents the energy consumption, reflecting the power consumption; P[i] TA represents the crowding distance of the delay, including the influence of location distance and transmission power on the delay; P[i] C represents the crowding distance of the cost; E[i+1] and E[i-1] represent the values ​​of the two individuals adjacent to the i-th individual in the energy consumption objective function; TA[i+1] and TA[i-1] represent the values ​​of the two individuals adjacent to the i-th individual in the delay objective function; C[i+1] and C[i-1] represent the values ​​of the two individuals adjacent to the i-th individual in the cost objective function; F represents the number of all individuals in the layer where the current individual is located.

[0184] Elite retention strategy: In order to ensure the retention of the best individuals, we adopt the elite retention strategy; specifically, if the first offspring is generated, the parent population P t and the offspring population Q t Synthesize a new population R t , and then perform Pareto optimal sorting, sorting in order of size Sort(F i ), individuals will be placed in the parent population P from low to high t+1 , until the parent population is full. If the individuals in a certain layer cannot be completely placed in the parent population, the remaining individuals are selected in order from small to large according to the crowding distance;

[0185] Selection, crossover and mutation makingNewPop (P0): In the present invention, binary decision making is used as the selection operation method. The specific steps are: first, randomly select half of the N pop Then, according to the fitness value of each individual, the most suitable individual is selected to enter the next generation population; then crossover and mutation operations are performed to generate a new population Q0. In order to reduce unnecessary calculations, for each new population generated in each generation, individuals with higher fitness are selected to enter the next generation population; the crossover operation generates new individuals through gene crossover to enhance the diversity of the population; the mutation operation randomly changes certain genes of the individual to increase the coverage of the solution space and avoid falling into the local optimum; after each generation of crossover and mutation, individuals that do not meet the constraints (such as individuals with excessive power consumption and delay) are first eliminated, and then the remaining individuals are sorted, crossed and mutated;

[0186] Algorithm selection and optimization objectives: The selection of the best task receiver in the present invention is optimized through a multi-objective edge service selection algorithm. The selection of the best task receiver is carried out under the comprehensive consideration of the total energy consumption, total delay and total cost objective functions, implicitly combining the influence of transmission power, resource allocation and location distance to ensure the balance between service quality and resource utilization. Specifically, population generation and fitness calculation: In the initial stage, the algorithm randomly generates a population of task receivers and calculates the target energy consumption, delay and total cost for each individual. The algorithm will evaluate the performance of each individual on these three objectives; non-dominated sorting and crowding distance: After the individuals are non-dominated sorted, the crowding distance of each individual is calculated to further distinguish the pros and cons of the individuals. Through the crowding distance, the algorithm can avoid premature convergence when selecting the optimal solution; after multiple iterations, the individuals are continuously optimized using binary competition selection, crossover and mutation operations. Until the preset termination conditions are met, such as reaching the maximum number of iterations or the fitness value converges, the optimal task acceptor is finally selected; in order to speed up the convergence speed, the optimization algorithm sets the priority to select edge servers as task acceptors in the initialization stage. This strategy can quickly find the initial solution and effectively reduce the complexity of the search space. Specifically, a random size of N is generated during the initialization process. pop and P0, each individual represents a selection scheme for a task acceptor; in the initialization stage, all individuals select the local server as the task acceptor by default, and randomly generate other possible choices based on their resource conditions; the optimal solution is obtained based on the algorithm results, and the best task acceptor is selected.

[0187] Step 3: Based on the task acceptor model, the multi-target edge server selection algorithm is used to perform task acceptance allocation, and after selecting the best task acceptor for the target task, the method includes:

[0188] When the number of tasks processed by the task acceptor reaches the set maximum task processing number threshold, the task acceptor is reselected for the target task and optimized; otherwise, each time the task acceptor assigns a task, the task counter of the task acceptor is updated.

[0189] Specifically, Figure 4 As shown, set the maximum task processing number threshold N of the task acceptor threshold When the system is initialized, the multi-objective edge service selection algorithm is used to select the best task acceptor and record the currently selected task acceptor and its initial task counter. When the task counter TC of the task acceptor reaches the threshold N threshold When the task acceptor is running, the performance of the current task acceptor is evaluated based on a multi-objective function. The evaluation indicators include but are not limited to total energy consumption, total delay and total cost, and the task acceptor is reselected. This is a process of updating the old processor and optimizing the processor performance. In the process of task acceptor processing tasks, that is, the process of assigning tasks for execution, each time the current task acceptor assigns a task, the task counter TC is updated: TC = TC + 1.

[0190] The step of reselecting and optimizing a task acceptor for a target task includes:

[0191] Selecting a new task acceptor for the target task based on a multi-target edge server selection algorithm, and resetting the task counter of the task acceptor;

[0192] Evaluate the new task acceptor based on a multi-objective function, and dynamically adjust the task processing quantity threshold of the new task acceptor according to the evaluation result;

[0193] The minimization multi-objective function value of the new task acceptor is monitored and optimized in real time.

[0194] Specifically, when the task counter TC reaches the maximum task processing number threshold N threshold When the multi-target edge service optimization selection process is re-executed, a new task acceptor is selected. This method of setting the task number threshold and reselecting the task acceptor after reaching the threshold can dynamically adjust the task acceptor selection frequency while maintaining system stability and improve task processing efficiency. After reselecting the task acceptor, the task counter of the new task acceptor is reset to 0: TC = 0. According to the performance evaluation results of the task acceptor and the changes in the network environment, the task number threshold N is dynamically adjusted. threshold If the performance of the task acceptor is stable and the task processing efficiency is high, the task number threshold N can be appropriately increased. threshold ; If the performance of the task acceptor fluctuates greatly or the task processing efficiency is low, the task number threshold N can be appropriately reduced threshold, according to the evaluation results, dynamically adjust the parameters of the task number threshold to adapt to the current network status and task requirements. Continuously monitor the performance of the task acceptor and the network environment, and periodically optimize the task number threshold N threshold and the task acceptor selection process to ensure that the system minimizes the total energy consumption, total delay and total cost objective functions without affecting the task execution performance.

[0195] Step 4: The optimal task acceptor is combined with safety analysis to allocate the target task to the cloud server, edge server or local device at the edge layer for task execution processing, including:

[0196] Based on the best task acceptor, prioritize the evaluation of whether the comprehensive security of local equipment meets security standards;

[0197] When the comprehensive security level of the local device meets the security standards, the target task is offloaded to the local device for task execution processing;

[0198] When the comprehensive security level of the local device does not meet the security standards, the edge cloud fuzzy reasoning method is used to allocate the target task to the cloud server or edge server based on the judgment factors.

[0199] The present invention selects the best task receiver for the target task by utilizing a multi-objective edge service selection algorithm, dynamically allocates the target task between the cloud server and the edge server nodes, comprehensively considers factors such as real-time resource utilization, transmission power and task delay, uses the edge cloud fuzzy reasoning method to deal with uncertainty, optimizes the task processing location, and analyzes the security of the edge server according to the security judgment method of security analysis, thereby minimizing the energy consumption, time and cost of the system and improving the security of the system.

[0200] Specifically, given that executing tasks on local devices can save delays and costs, priority is given to whether the target task can be executed on the local device. The determining factor for determining whether the target task can be executed on the local device is whether the comprehensive security of the local device meets the security standards, that is, whether the comprehensive security value composed of the computing power, system integrity and data protection capabilities of the local device can meet the security standards. Among them, the computing power of the local device affects the execution delay and energy consumption of the task, and the computing power security value of the local device is evaluated by comparing it with the computational complexity of the task; the system integrity of the local device mainly refers to the health status of the hardware and software of the device, which is similar to the evaluation of the message receiving rate; the data protection capability of the local device mainly refers to the strength of data encryption and data access control. If the comprehensive security value of the local device is within the set local device security threshold, the task is assigned to the local device for execution and processing. Otherwise, the task is assigned to the edge server or cloud server for execution and processing.

[0201] The comprehensive security assessment of the local device includes:

[0202] Total security value TV of local equipment l t , calculated as follows:

[0203]

[0204] Among them, TV comp Indicates the computing power safety value; TV sys Indicates the system integrity security value; TV data Indicates the data protection capability security value.

[0205] The computing capacity safety value TV comp , calculated as follows:

[0206]

[0207] in, Indicates the computing power of the local device; c n represents the computational effort of the task; σ com Indicates the standard deviation of computing power.

[0208] The system integrity safety value TV sys , calculated as follows:

[0209] TV sys =min(S os , S patch , S smalware )

[0210] Among them, S os Indicates the security score of the operating system, reflecting whether the operating system has known vulnerabilities; S patch Indicates the timeliness score of the patch, reflecting whether the patches of the operating system and applications are updated in a timely manner; S smalware Indicates the malware protection score, which reflects whether the device is effectively protected against malware.

[0211] The data protection capability security value TV data , calculated as follows:

[0212]

[0213] Among them, S enc Indicates the encryption strength score, which reflects the encryption protection during data storage and transmission; c enc represents the mean value of encryption strength; σ enc Indicates the standard deviation of encryption strength.

[0214] The edge cloud fuzzy inference method is used to allocate the target task to the cloud server or the edge server according to the determination factors, including:

[0215] The resource utilization, transmission power and task delay of the edge server are used as the input elements of the edge cloud fuzzy reasoning method, and edge cloud fuzzification, edge cloud fuzzy reasoning and edge cloud defuzzification are performed in sequence to obtain the decision of assigning the target task to the cloud server or edge server for execution and processing.

[0216] The step of sequentially performing edge cloud fuzzification, edge cloud fuzzy reasoning, and defuzzification of edge cloud includes:

[0217] The membership function and non-membership function of the input and output variables of the resource utilization, transmission power and task delay of the edge server are established by using trapezoidal membership function and triangular membership function to perform edge cloud fuzzification.

[0218] Using fuzzy inference system to infer the membership and non-membership functions of input variables;

[0219] The center of gravity method of edge cloud defuzzification is used to transform the fuzzy result obtained by the fuzzy reasoning system into a clear decision value, and the edge cloud fuzzy reasoning result of the target task is output.

[0220] Specifically, if all tasks can be processed on the edge server, the latency can be significantly reduced compared to sending them to the cloud server for execution, because the transmission time between the edge and the cloud is saved. Specifically, edge computing reduces the distance of data transmission by processing data closer to the data source, thereby reducing network task latency; however, tasks that arrive at resource-constrained edge nodes are unlikely to be executed immediately, and execution usually requires waiting time, and security issues cannot be guaranteed. Some of the nearest edge servers do not meet the security requirements of the task; therefore, the geographically closest infrastructure is not always a good choice, so it is necessary to consider factors such as real-time edge server resource utilization, transmission power, and computing power to select the optimal task processing location.

[0221] In the task offloading decision, the resource utilization, transmission power and computing power of the edge server are the key factors in determining the location of task processing. The resource utilization of the edge server ensures that the computing resources are not over-occupied, thereby avoiding overload and improving resource utilization efficiency. Appropriate transmission power helps to reduce network latency and energy efficiency loss, ensure the efficient transmission of task data, and has an important impact on bandwidth interference in wireless networks. The computing power directly determines the execution delay of the task. More complex tasks may require powerful computing resources in the cloud, while simple tasks can be processed at the edge, thereby optimizing system load and balance. Taking these three into consideration, it can ensure the reasonable offloading of tasks between the edge server and the cloud, improve processing efficiency, reduce latency, and improve system stability and resource utilization.

[0222] The resource utilization, transmission power and task delay information are used as the input of the edge cloud fuzzy reasoning method for further optimization. The edge cloud fuzzy reasoning method is used to determine whether to choose cloud server processing or edge server processing. The specific steps include edge cloud fuzzification, edge cloud fuzzy reasoning and defuzzification of edge cloud.

[0223] Edge cloud fuzzification: The membership function and non-membership function of the input and output variables are established using trapezoidal membership function and triangular membership function; the membership function and non-membership function of the input and output variables are established using trapezoidal membership function and triangular membership function. For each input variable, an edge cloud fuzzy set is established: in and It is called hesitation degree, which satisfies Represents non-membership, which measures whether element x belongs to fuzzy set A * The value range of non-membership is also in [0, 1]. Compared with membership, if the membership of an element is very high, then its non-membership is very low. represents the degree of membership; represents the hesitation degree, which measures the element x belongs to the fuzzy set A * The degree of uncertainty or hesitation; * represents a fuzzy set; equation Ensure that the sum of membership, non-membership and hesitation does not exceed 1, thus maintaining the consistency of the fuzzy set;

[0224] Edge cloud fuzzy inference: Use the fuzzy inference system Mamdani to infer the membership and non-membership functions of the input variables;

[0225] Deblurring edge clouds: Apply the centroid method for deblurring edge clouds, and its value is:

[0226]

[0227] Among them, w represents the centroid, which is obtained by calculating the weighted average of the input variable x, and the weight is the membership of x μ(x); x represents the input variable, which can be any value that needs to be fuzzy processed; μ(x) represents the membership, which is the value of x belonging to a certain fuzzy set processing; through this method, we convert the result obtained by fuzzy reasoning into a clear decision value;

[0228] Output linear results: Decision-making process is realized through edge cloud fuzzy reasoning:

[0229] The edge cloud fuzzy inference result ECFS is calculated as follows:

[0230] ECFS=(1-π)·FS μ +π·FS v +γ·(FS v -FS μ )

[0231] Among them, π represents the uncertainty of whether an element belongs to or does not belong to a set; γ represents the parameter that adjusts the difference between membership and non-membership, which is a small positive number used to adjust πFS v and πFS μ The difference between them reduces fluctuations; FS μ Membership degree of fuzzy set; FS v The non-membership of the fuzzy set; the purpose of this formula is to find a balance between membership and non-membership so that the distribution of tasks is more reasonable.

[0232] The decision of allocating the target task to the cloud server or the edge server includes:

[0233] If the edge cloud fuzzy inference result of the target task is greater than the threshold r1, the target task is assigned to the cloud server for processing; otherwise, the target task is assigned to the edge server for processing;

[0234] If the edge cloud fuzzy inference result of the target task is greater than the threshold r2, the target task is assigned to the adjacent edge server for processing; otherwise, the target task is assigned to the local edge server for processing.

[0235] Specifically, two thresholds r1 and r2 are set, which are the criteria for making decisions. The decision is formalized by two functions: cloud layer or edge layer decision; cloud layer or edge layer decision f1:

[0236] f1:t i →{Cloud, Edge Server}

[0237]

[0238] If ECFS(ti )>r1, then the task t i Once it is assigned to the cloud server, no other operations are required. The task is directly handed over to the cloud layer for corresponding control arrangements. In the cloud, we do not need to consider its security.

[0239] If ECFS(t i )≤r1, then it is necessary to further determine whether to schedule the task on the local server or on other edge servers.

[0240] Local edge server or neighboring edge server decision f2:

[0241] f2:t i →{local edge server, adjacent edge server}

[0242]

[0243] If ECFS(t i )>r2, then the task t i Deployed to the local edge server, but its security has not been analyzed. It is also necessary to determine whether the current local server meets the security requirements and analyze the security of the current local server. If the security of the local server meets the security requirements, then the task can be deployed to the local server.

[0244] If ECFS(t i )≤r2, it means that the local edge server cannot meet the task requirements, and it is necessary to analyze whether the adjacent edge servers meet the task requirements. If they do, then perform the corresponding security analysis.

[0245] If there are multiple adjacent edge servers to choose from, the delay minimization strategy is adopted. By measuring the network delay between the task and each adjacent edge server, the edge server with the smallest delay is selected as the priority processing node. This strategy can effectively reduce the response time of the task and improve the overall service efficiency. In the specific implementation process, if multiple adjacent servers are detected, the system will dynamically detect the network delay with each adjacent server and prioritize the task to the server with the lowest delay, thereby optimizing resource scheduling and ensuring that the task can be processed in the fastest time.

[0246] The edge cloud fuzzy inference method is used to allocate the target task to the cloud server or the edge server according to the determination factors, including:

[0247] Based on the edge server selected by the target task, a security value evaluation is performed on the message receiving rate, message transmission rate and message tampering rate of the edge server node, and whether the security value of the current edge server node period meets the final security value is determined according to the final security value evaluation result;

[0248] If the security value of the edge server node meets the final security value, the security value evaluation result of the edge server node is used as the historical final security value evaluation result of the next cycle, and the target task is deployed to the edge server node, and the security value of the edge server node is updated at the same time;

[0249] Otherwise, the edge server node is isolated, and a security value evaluation is performed on the message acceptance rate, message transmission rate, and message tampering rate of the adjacent edge server nodes until an edge server node that meets the final security value evaluation result is found.

[0250] Specifically, Figure 5 As shown in the figure, the service quality of the edge server node is evaluated mainly from the three dimensions of message transmission rate, message receiving rate and message tampering rate. The security analysis is carried out based on these three dimensions. The three dimensions are combined with the single security value evaluation results and the historical evaluation results to obtain the final security value evaluation results of the three dimensions. At the same time, the security values ​​of the previous cycle and the historical cycle are combined, and the network environment, task urgency, user behavior, network congestion, historical tampering behavior and security event correlation are considered for comprehensive evaluation. According to the final security value evaluation results, it is judged whether the edge server meets the security standards. If any dimension is less than 0.5, the node threshold alarm is triggered, indicating that the edge server node has malicious behavior and does not meet the security requirements. It cannot be used as an edge server for task execution processing, and it is necessary to reselect the edge server for task execution processing.

[0251] The final security value evaluation result of the message receiving rate of the edge server node is calculated as follows:

[0252]

[0253] in, Indicates that edge server node i is in transmission period t n The inside is the final security value evaluation result of the message receiving rate of the edge server node j to be evaluated; ω represents the weight of the security value of the current cycle; β represents the weight of the security value of the historical cycle; TV receive (i,j)(t n ) represents the edge server node i in the transmission period t n The security value evaluation result of the message receiving rate of the edge server node j to be evaluated; θ env Represents the impact factor of network environment changes; θ urgent The influencing factor indicating the urgency of the task; Indicates the edge server node i in the last transmission cycle t n-1The content is the final security value evaluation result of the message receiving rate of the edge server node j to be evaluated;

[0254] The final security value evaluation result of the message transmission rate of the edge server node is calculated as follows:

[0255]

[0256] in, Indicates that edge server node i is in transmission period t n The content is the final security value evaluation result of the message transmission rate of the edge server node j to be evaluated; TV send(i，j) (i,j)(t n ) represents the edge server node i in the transmission period t n The security value evaluation result of the message transmission rate of the edge server node j to be evaluated; φ user Indicates the user's behavior influencing factor; φ congestion Indicator of the impact of network congestion; Indicates the edge server node i in the last transmission cycle t n-1 The content is the final security value evaluation result of the message transmission rate of the edge server node j to be evaluated;

[0257] The final security value evaluation result of the message tampering rate of the edge server node is calculated as follows:

[0258]

[0259] in, Indicates that edge server node i is in transmission period t n The value in the box is the final security value evaluation result of the message tampering rate of the edge server node j to be evaluated; TV tamper(i，j) (i,j)(t n ) represents the edge server node i in the transmission period t n The security value evaluation result of the message tampering rate of the edge server node j to be evaluated; ψ history represents the impact factor of historical tampering behavior; ψ correlation Impact factors that represent the relevance of security events; Indicates the edge server node i in the last transmission cycle t n-1 The content is the final security value evaluation result of the message tampering rate of the edge server node j to be evaluated.

[0260] The safety value evaluation result of the message receiving rate is calculated as follows:

[0261]

[0262] The security value evaluation result of the message transmission rate is calculated as follows:

[0263]

[0264] in,

[0265] The security value evaluation result of the message tampering rate is calculated as follows:

[0266] TV tamper(i，j) (t n )=1-2MTR i，j (t n )

[0267] Among them, TV recrive (i,j)(t n ) represents the edge server node i in the transmission period t n The security value evaluation result of the message receiving rate of the edge server node j to be evaluated; MRR i，j (t n ) indicates that in the transmission period t n The message receiving rate from the inner edge server node i to the edge server node j; ν represents the mean value in the Gaussian function; σ represents the standard deviation in the Gaussian function; TV send(i，j) (i,j)(t n ) represents the edge server node i in the transmission period t n The security value evaluation result of the message transmission rate of the edge server node j to be evaluated; MSR i，j (t n ) indicates that in the transmission period t n The message transmission rate from inner edge server node i to edge server node j; TV tamper(i，j) (i,j)(t n ) represents the edge server node i in the transmission period t n The security value evaluation result of the message tampering rate of the edge server node j to be evaluated; MTR i，j (t n ) indicates that in the transmission period t n The message tampering rate from inner edge server node i to edge server node j.

[0268] The message tampering rate of the edge node to be evaluated is identified based on the comparison of the two data packets received by the edge node. The security assessment of the message tampering rate is as follows: Generally speaking, the difference is within 10%, which is normal. If the difference between the two data packets exceeds 10%, it is considered an edge node that needs to be evaluated. If the data packet is suspected to be tampered, the security value will decrease linearly.

[0269] The message receiving rate is calculated as follows:

[0270]

[0271] Among them, MRR i，j (t n ) indicates that in the transmission period t n R is the message receiving rate from inner edge server node i to edge server node j; i，j (t n ) indicates that in the transmission period t n The number of confirmation messages received from inner edge server node i to edge server node j; R j (t n-1 ) indicates that in the previous transmission cycle t n-1 The number of confirmation messages received by the inner edge server node j;

[0272] The message transmission rate is calculated as follows:

[0273]

[0274] Among them, MSR i，j (t n ) indicates that in the transmission period t n The message transmission rate from inner edge server node i to edge server node j; s i，j (t n ) indicates that in the transmission period t n The number of packets sent from inner edge server node i to edge server node j; s j (t n-1 ) indicates that in the previous transmission cycle t n-1 The number of packets sent by the inner edge server node j;

[0275] The message tampering rate is calculated as follows:

[0276]

[0277] Among them, MTR i，j (t n ) indicates that in the transmission period t n The message tampering rate from inner edge server node i to edge server node j; N j Indicates the number of packets to be evaluated; S jk represents the score of the kth data packet associated with edge node j.

[0278] The scoring mechanism of message tampering rate is based on scoring the abnormality of data packets. The score of each data packet reflects the possibility of its tampering. The final message tampering rate is the weighted average of these scores. The key factors affecting data packet tampering include data packet integrity, abnormal behavior and abnormal content. The scoring range of each dimension is set from 0 to 10 points, where 0 means no abnormality and 10 means high abnormality. Each data packet is evaluated and scored according to the predefined scoring criteria.

[0279] The data packet p j Rating S j , calculated as follows:

[0280] S j =α·S content +β·S behavior +γ·S integrity

[0281] Among them, S content Indicates whether the content of the data packet meets expectations; S behavior Indicates whether the behavior pattern of the data packet is abnormal; S integrity Indicates whether the integrity of the data packet is abnormal; α, β and γ represent the weight coefficients of each dimension respectively, and α+β+γ=1.

[0282] Embodiment 2:

[0283] The present invention based on the same inventive concept also provides a dynamic task security acceptance system based on cloud-edge collaboration, including:

[0284] The task grouping module is used to determine different task groups based on the data information uploaded by edge layer users;

[0285] A model building module, used for building a task acceptor model based on data attributes in the task group using a multi-objective function;

[0286] An acceptor selection module, used to perform task acceptance allocation based on the task acceptor model through a multi-target edge server selection algorithm, and select the best task acceptor for the target task;

[0287] The task allocation module is used to use the best task acceptor combined with safety analysis to allocate the target task to the cloud server, edge server or local device at the edge layer for task execution processing.

[0288] Specifically, Figure 6As shown, the user's mobile device in the edge layer sends a task request to the edge layer, and the edge layer sensor receives the task request and assigns it to the best task acceptor for acceptance. According to the best task acceptor combined with security analysis, the task request is assigned to the local mobile device of the edge layer, the edge server of the edge layer, or the cloud server of the cloud layer for task execution processing. The cloud layer and the edge layer and the edge layer use wireless local area network WLAN for communication connection, and the task acceptors use metropolitan area network MAN for communication connection.

[0289] The task grouping module is specifically used for:

[0290] According to the data information uploaded by edge layer users to edge layer sensors, different task groups are divided according to data types and processing requirements. Each task group consists of multiple subtasks. Suppose the user mobile device set is U = {u1, u2, ..., u n}, the task group is T = {T1, T2, T3, ..., T N}, subtask is T n =(s n , c n , d n ), where s n Represents subtask T n The size of c n Represents subtask T n The amount of calculation; d n Indicates completion of task T n The maximum acceptable delay.

[0291] The model building module comprises:

[0292] A calculation unit, for calculating the execution energy consumption, transmission energy consumption, execution time, transmission time and transmission cost of the subtasks based on the data attributes of each task group; and calculating the total energy consumption, total delay and total cost objective functions in combination with the task offloading decision based on the execution energy consumption, transmission energy consumption, execution time, transmission time and transmission cost of the subtasks;

[0293] The construction unit is used to construct a task acceptor model according to the minimization objective function and constraint conditions of total energy consumption, total delay and total cost.

[0294] The receiver selection module includes:

[0295] An algorithm unit, for selecting a multi-target edge server based on a task acceptor model by adding a crowding distance;

[0296] A selection unit, used for selecting an optimal task acceptor for a target task according to the task load, transmission power and resource allocation of the task acceptor;

[0297] The multi-target edge server selection algorithm with crowding distance includes the following steps: encoding, initialization, non-dominated sorting, crowding distance calculation, elite retention strategy, selection, crossover and mutation.

[0298] The system further comprises:

[0299] The acceptor optimization module is used to reselect and optimize the task acceptor for the target task when the task processing quantity of the task acceptor reaches the set maximum task processing quantity threshold;

[0300] The timer update module is used to update the task counter of the task acceptor each time the task acceptor assigns a task when the task processing quantity of the task acceptor does not reach the set maximum task processing quantity threshold.

[0301] The receiver optimization module includes:

[0302] A reset unit, configured to select a new task acceptor for the target task based on a multi-target edge server selection algorithm, and reset a task counter of the task acceptor;

[0303] An evaluation unit, used for evaluating the new task acceptor based on a multi-objective function;

[0304] The adjustment unit is used to dynamically adjust the task processing quantity threshold of the new task acceptor according to the evaluation result, and to monitor and optimize the minimized multi-objective function value of the new task acceptor in real time.

[0305] The task allocation module comprises:

[0306] A local assessment unit, used to assess the comprehensive security of local equipment;

[0307] A local judgment unit, used to judge whether the comprehensive security of the local device meets the security standards;

[0308] A local allocation unit, used for offloading the target task to the local device for task execution processing when the comprehensive safety degree of the local device meets the safety standard;

[0309] The non-local allocation unit is used to allocate the target task to the cloud server or edge server according to the judgment factors by using the edge cloud fuzzy reasoning method when the comprehensive security degree of the local device does not meet the security standards.

[0310] The non-local allocation unit comprises:

[0311] The reasoning subunit is used to use the resource utilization, transmission power and task delay of the edge server as input elements of the edge cloud fuzzy reasoning method, and sequentially perform edge cloud fuzzification, edge cloud fuzzy reasoning and defuzzification of the edge cloud;

[0312] The decision subunit is used to generate a decision on assigning the target task to a cloud server or edge server for execution and processing.

[0313] The reasoning subunit is specifically used for:

[0314] The membership function and non-membership function of the input and output variables of the resource utilization, transmission power and task delay of the edge server are established by using trapezoidal membership function and triangular membership function to perform edge cloud fuzzification.

[0315] Using fuzzy inference system to infer the membership and non-membership functions of input variables;

[0316] The center of gravity method of edge cloud defuzzification is used to transform the fuzzy result obtained by the fuzzy reasoning system into a clear decision value, and the edge cloud fuzzy reasoning result of the target task is output.

[0317] The decision subunit is specifically used for:

[0318] If the edge cloud fuzzy inference result of the target task is greater than the threshold r1, the target task is assigned to the cloud server for processing; otherwise, the target task is assigned to the edge server for processing;

[0319] If the edge cloud fuzzy inference result of the target task is greater than the threshold r2, the target task is assigned to the adjacent edge server for processing; otherwise, the target task is assigned to the local edge server for processing.

[0320] The non-local allocation unit further comprises:

[0321] An edge evaluation subunit, configured to perform security value evaluation on a message receiving rate, a message transmission rate, and a message tampering rate of an edge server node based on the edge server selected by the target task;

[0322] The edge judgment subunit is used to determine whether the security value of the current edge server node cycle meets the final security value based on the final security value evaluation result; if the security value of the edge server node meets the final security value, the security value evaluation result of the edge server node is used as the historical final security value evaluation result of the next cycle, and the target task is deployed to the edge server node, and the security value of the edge server node is updated at the same time; otherwise, the edge server node is isolated, and the message acceptance rate, message transmission rate and message tampering rate of the adjacent edge server nodes are evaluated for security values ​​until an edge server node that meets the final security value evaluation result is found.

[0323] Embodiment 3:

[0324] like Figure 7As shown, the present invention also provides an electronic device, which may be a computer device, a single-chip device, an intelligent mobile device, etc. The electronic device in this embodiment may include a processor, a memory, a transceiver component, etc. The memory, the processor, and the transceiver component are connected via a bus; the memory may be used to store an execution program, and an exemplary execution program may include instructions; the processor is used to execute the instructions stored in the memory. The memory may also be used to store data, which may be called and / or modified when the instructions are executed.

[0325] The processor may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in a storage medium to implement the corresponding method flow or corresponding functions, so as to implement the steps of a dynamic task security acceptance method based on cloud-edge collaboration in the above embodiment.

[0326] Embodiment 4:

[0327] Based on the same inventive concept, the present invention also provides a readable storage medium, specifically an electronic device readable storage medium (Memory), which is a memory device in the electronic device for storing programs and data. It can be understood that the storage medium here can include both built-in storage media in the electronic device and, of course, extended storage media supported by the electronic device. The storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more execution programs (including program codes). It should be noted that the storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk storage. The processor loads and executes one or more instructions stored in the storage medium, which can implement the steps of a dynamic task security acceptance method based on cloud-edge collaboration in the above embodiment.

[0328] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0329] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0330] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0331] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0332] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention are included in the scope of the claims of the present invention to be approved.

Claims

1. A dynamic task security acceptance method based on cloud-edge collaboration, characterized in that: include: Determine different task groups based on the data information uploaded by edge layer users; Based on the data attributes in the task group, a task acceptor model is constructed using a multi-objective function; Based on the task acceptor model, task acceptance allocation is performed through a multi-target edge server selection algorithm to select the best task acceptor for the target task; Utilize the best task acceptor combined with safety analysis to distribute the target tasks to cloud servers, edge servers or local devices at the edge layer for task execution and processing.

2. The method according to claim 1, characterized in that Determining different task groups according to the data information uploaded by edge layer users includes: According to the data information uploaded by edge layer users to edge layer sensors, different task groups are divided according to data types and processing requirements. Each task group consists of multiple subtasks. Suppose the user mobile device set is U = {u1, u2, ..., u n }, the task group is T = {T1, T2, T3, ..., T N }, subtask is T n =(s n , c n , d n ), where s n Represents subtask T n The size of c n Represents subtask T n The amount of calculation; d n Indicates completion of task T n The maximum acceptable delay.

3. The method according to claim 2, characterized in that The method of constructing a task acceptor model based on data attributes in the task group using a multi-objective function includes: Based on the data attributes of each task group, the execution energy consumption, transmission energy consumption, execution time, transmission time and transmission cost of the subtask are calculated; Based on the execution energy consumption, transmission energy consumption, execution time, transmission time and transmission cost of the subtask, the total energy consumption, total delay and total cost objective functions are calculated in combination with the task offloading decision; The task acceptor model is constructed according to the objective function and constraints of minimizing total energy consumption, total delay and total cost.

4. The method according to claim 3, characterized in that The calculation of the execution energy consumption, transmission energy consumption, execution time, transmission time and transmission cost of the subtask based on the data attributes of each task group includes: Calculate the energy consumption of the local execution of the subtask based on the computational load of the subtask, the computing power of the local device, and the coefficient of the device hardware architecture; The transmission energy consumption of the subtask is calculated according to the transmission power of the subtask when unloading, the data size of the subtask, the transmission rate of the subtask, the interference ratio suffered by the subtask during the transmission period, and the efficiency of the power amplifier of the local device; Calculate the local execution time of the subtask according to the computation amount of the subtask, the computing power of the local device, and the resource utilization rate of the local device; The time for the subtask to be executed non-locally is calculated based on the computational effort of the subtask and the computing power of the edge server; The transmission time of the subtask is calculated according to the data size of the subtask, the transmission rate of the subtask and the interference ratio of the subtask during the transmission period; The transmission cost of the subtask is calculated based on all the costs incurred by the subtask during the transmission process.

5. The method according to claim 1, characterized in that: The method of performing task acceptance allocation based on the task acceptor model by using a multi-target edge server selection algorithm and selecting the best task acceptor for the target task includes: Based on the task acceptor model, a multi-target edge server selection algorithm with congestion distance is added to select the best task acceptor for the target task according to the task load, transmission power and resource allocation of the task acceptor. The multi-target edge server selection algorithm with crowding distance includes the following steps: encoding, initialization, non-dominated sorting, crowding distance calculation, elite retention strategy, selection, crossover and mutation.

6. The method according to claim 5, characterized in that The crowding distance is calculated as follows: Among them, P[i] distance represents the crowding distance; P[i] E represents the crowding distance of energy consumption; P[i] TC represents the crowding distance of delay; P[i] c represents the crowding distance of the cost; E[i+1] and E[i-1] represent the values ​​of the two individuals adjacent to the i-th individual in the energy consumption objective function; TA[i+1] and TA[i-1] represent the values ​​of the two individuals adjacent to the i-th individual in the delay objective function; C[i+1] and C[i-1] represent the values ​​of the two individuals adjacent to the i-th individual in the cost objective function; F represents the number of all individuals in the layer where the current individual is located.

7. The method according to claim 3, characterized in that The method comprises: performing task acceptance allocation based on the task acceptor model by using a multi-target edge server selection algorithm, and selecting the best task acceptor for the target task, comprising: When the number of tasks processed by the task acceptor reaches the set maximum task processing number threshold, the task acceptor is reselected for the target task and optimized; otherwise, each time the task acceptor assigns a task, the task counter of the task acceptor is updated.

8. The method according to claim 7, characterized in that The step of reselecting and optimizing a task acceptor for a target task includes: Selecting a new task acceptor for the target task based on a multi-target edge server selection algorithm, and resetting the task counter of the task acceptor; Evaluate the new task acceptor based on a multi-objective function, and dynamically adjust the task processing quantity threshold of the new task acceptor according to the evaluation result; The minimization multi-objective function value of the new task acceptor is monitored and optimized in real time.

9. The method according to claim 1, characterized in that: The method of using the best task acceptor in combination with safety analysis to allocate the target task to the cloud server, edge server or local device at the edge layer for task execution processing includes: Based on the best task acceptor, prioritize the evaluation of whether the comprehensive security of local equipment meets security standards; When the comprehensive security level of the local device meets the security standards, the target task is offloaded to the local device for task execution processing; When the comprehensive security level of the local device does not meet the security standards, the edge cloud fuzzy reasoning method is used to allocate the target task to the cloud server or edge server based on the judgment factors.

10. The method according to claim 9, characterized in that The comprehensive security assessment of the local device includes: Total security value of local equipment Calculate as follows: Among them, TV comp Indicates the computing power safety value; TV sys Indicates the system integrity security value; TV data Indicates the data protection capability security value.

11. The method according to claim 10, characterized in that The computing capacity safety value TV comp , calculated as follows: in, Indicates the computing power of the local device; c n represents the computational effort of the task; σ com Indicates the standard deviation of computing power.

12. The method according to claim 10, characterized in that The system integrity safety value TV sys , calculated as follows: TV sys =min(S os ,S patch ,S smalware ) Among them, S os Indicates the security score of the operating system; S patch Indicates the timeliness score of the patch; S smalware Indicates the malware protection score.

13. The method according to claim 10, characterized in that The data protection capability security value TV data , calculated as follows: Among them, S enc Indicates the encryption strength score; c emc represents the mean value of encryption strength; σ enc Indicates the standard deviation of encryption strength.

14. The method according to claim 9, characterized in that The edge cloud fuzzy inference method is used to allocate the target task to the cloud server or the edge server according to the determination factors, including: The resource utilization, transmission power and task delay of the edge server are used as the input elements of the edge cloud fuzzy reasoning method, and edge cloud fuzzification, edge cloud fuzzy reasoning and edge cloud defuzzification are performed in sequence to obtain the decision of assigning the target task to the cloud server or edge server for execution and processing.

15. The method according to claim 14, characterized in that The step of sequentially performing edge cloud fuzzification, edge cloud fuzzy reasoning, and defuzzification of edge cloud includes: The membership function and non-membership function of the input and output variables of the resource utilization, transmission power and task delay of the edge server are established by using trapezoidal membership function and triangular membership function to perform edge cloud fuzzification. Using fuzzy inference system to infer the membership and non-membership functions of input variables; The center of gravity method of edge cloud defuzzification is used to transform the fuzzy result obtained by the fuzzy reasoning system into a clear decision value, and the edge cloud fuzzy reasoning result of the target task is output.

16. The method according to claim 15, characterized in that The decision of allocating the target task to the cloud server or the edge server includes: If the edge cloud fuzzy inference result of the target task is greater than the threshold r1, the target task is assigned to the cloud server for processing; otherwise, the target task is assigned to the edge server for processing; If the edge cloud fuzzy inference result of the target task is greater than the threshold r2, the target task is assigned to the adjacent edge server for processing; otherwise, the target task is assigned to the local edge server for processing.

17. The method according to claim 14, characterized in that The edge cloud fuzzy inference method is used to allocate the target task to the cloud server or the edge server according to the determination factor, including: Based on the edge server selected by the target task, a security value evaluation is performed on the message receiving rate, message transmission rate and message tampering rate of the edge server node, and whether the security value of the current edge server node period meets the final security value is determined according to the final security value evaluation result; If the security value of the edge server node meets the final security value, the security value evaluation result of the edge server node is used as the historical final security value evaluation result of the next cycle, and the target task is deployed to the edge server node, and the security value of the edge server node is updated at the same time; Otherwise, the edge server node is isolated, and a security value evaluation is performed on the message acceptance rate, message transmission rate, and message tampering rate of the adjacent edge server nodes until an edge server node that meets the final security value evaluation result is found.

18. The method according to claim 17, characterized in that The final safety value evaluation result of the message reception rate of the edge server node is calculated based on the safety value evaluation result of the message reception rate of the edge server node in the current cycle, the final safety value evaluation result of the message reception rate in the previous cycle, the influencing factor of the network environment change, the influencing factor of the task urgency, and the weight of each evaluation result to obtain the final safety value evaluation result of the message reception rate of the edge server node; The final security value evaluation result of the message transmission rate of the edge server node is calculated based on the security value evaluation result of the message transmission rate of the edge server node in the current cycle, the final security value evaluation result of the message transmission rate in the previous cycle, the user's behavior influencing factor, the influencing factor of the network congestion situation, and the weight of each evaluation result to obtain the final security value evaluation result of the message transmission rate of the edge server node; The final security value assessment result of the message tampering rate of the edge server node is obtained based on the security value assessment result of the message tampering rate of the edge server node in the current cycle, the final security value assessment result of the message tampering rate in the previous cycle, the influencing factor of historical tampering behavior, the influencing factor of the security event correlation and the weight of each assessment result.

19. A dynamic task security acceptance system based on cloud-edge collaboration, characterized in that: include: The task grouping module is used to determine different task groups based on the data information uploaded by edge layer users; A model building module, used for building a task acceptor model based on data attributes in the task group using a multi-objective function; An acceptor selection module, used to perform task acceptance allocation based on the task acceptor model through a multi-target edge server selection algorithm, and select the best task acceptor for the target task; The task allocation module is used to use the best task acceptor combined with safety analysis to allocate the target task to the cloud server, edge server or local device at the edge layer for task execution processing.

20. An electronic device, characterized in that: include: at least one processor and memory; The memory and the processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, a dynamic task security acceptance method based on cloud-edge collaboration as described in any one of claims 1 to 18 is implemented.

21. A readable storage medium, characterized in that: An execution program is stored thereon, and when the execution program is executed, a dynamic task security acceptance method based on cloud-edge collaboration as described in any one of claims 1 to 18 is implemented.

Images