Task unloading method and device based on edge computing network, computer equipment, storage medium and computer program product

Abstract

The invention relates to a task unloading method and device based on an edge computing network, computer equipment, a storage medium and a computer program product. The method comprises the steps of generating initial task unloading strategy information of other terminals in response to a task unloading request for a to-be-processed terminal; determining a fitness value of the initial task unloading strategy information; screening the initial task unloading strategy information with the minimum fitness value from the initial task unloading strategy information to serve as current task unloading strategy information of other terminals; determining a utility value of the to-be-processed terminal under each piece of preset task unloading strategy information according to the current task unloading strategy information; screening the preset task unloading strategy information with the maximum utility value from the preset task unloading strategy information to serve as target task unloading strategy information of the to-be-processed terminal; and performing corresponding task unloading processing on the to-be-processed terminal according to the target task unloading strategy information. By adopting the method, the task unloading accuracy can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular to a task offloading method and device based on edge computing network, computer equipment, computer readable storage medium and computer program product. BACKGROUND

[0002] With the increasing amount of tasks, how to accurately offload tasks to suitable task processing devices for processing is crucial.

[0003] In the traditional technology, task offloading is generally performed according to the task attributes (such as task data volume) of the tasks; however, this way considers only a single factor, which is prone to errors and results in low accuracy of task offloading. SUMMARY

[0004] Therefore, it is necessary to provide a task offloading method and device based on edge computing network, computer equipment, computer readable storage medium and computer program product, which can improve the accuracy of task offloading.

[0005] In a first aspect, the present application provides a task offloading method based on edge computing network, comprising:

[0006] In response to a task offloading request for a to-be-processed terminal, initial task offloading strategy information of other terminals is generated; the other terminals are used to represent terminals in an edge computing network to which the to-be-processed terminal belongs, except the to-be-processed terminal;

[0007] The fitness value of the initial task offloading strategy information is determined;

[0008] From the initial task offloading strategy information, the initial task offloading strategy information with the smallest fitness value is selected as the current task offloading strategy information of the other terminals;

[0009] According to the current task offloading strategy information, the utility value of the to-be-processed terminal under each preset task offloading strategy information is determined;

[0010] From the preset task offloading strategy information, the preset task offloading strategy information with the largest utility value is selected as the target task offloading strategy information of the to-be-processed terminal;

[0011] According to the target task offloading strategy information, corresponding task offloading processing is performed on the to-be-processed terminal.

[0012] In one embodiment, the determination of the fitness value of the initial task offloading strategy information comprises:

[0013] query a correspondence relationship between the task offloading policy information and historical task latency values, to obtain a plurality of historical task latency values corresponding to the initial task offloading policy information;

[0014] obtain an average value of the plurality of historical task latency values as an average task latency value corresponding to the initial task offloading policy information;

[0015] determine an fitness value of the initial task offloading policy information according to the average task latency value.

[0016] In one of the embodiments, the determining the fitness value of the initial task offloading policy information according to the average task latency value comprises:

[0017] obtain a number of terminals corresponding to the initial task offloading policy information;

[0018] obtain a latency value adjustment value associated with the number of terminals;

[0019] perform summation processing on the average task latency value and the latency value adjustment value to obtain a processed average task latency value as the fitness value of the initial task offloading policy information.

[0020] In one of the embodiments, the determining the utility value of the to-be-processed terminal under each preset task offloading policy information according to the current task offloading policy information comprises:

[0021] determining a current task latency value of the to-be-processed terminal under each preset task offloading policy information according to the current task offloading policy information;

[0022] determining the utility value of the to-be-processed terminal under each preset task offloading policy information according to the current task latency value of the to-be-processed terminal under each preset task offloading policy information.

[0023] In one of the embodiments, the determining the current task latency value of the to-be-processed terminal under each preset task offloading policy information according to the current task offloading policy information comprises:

[0024] determining a first to-be-processed task corresponding to the to-be-processed terminal and a second to-be-processed task corresponding to the other terminal;

[0025] obtaining a first task data volume and a first task computing density corresponding to the first to-be-processed task, and a second task data volume and a second task computing density corresponding to the second to-be-processed task;

[0026] Based on the current task unloading strategy information, the first task data volume, the first task computation density, the second task data volume, and the second task computation density, the current task latency value of the terminal to be processed under each preset task unloading strategy information is determined.

[0027] In one embodiment, the step of performing corresponding task unloading processing on the terminal to be processed according to the target task unloading strategy information includes:

[0028] Extract the first feature vector corresponding to the first task data volume and the second feature vector corresponding to the first task computation density;

[0029] The first feature vector and the second feature vector are fused to obtain the fused feature vector corresponding to the first task to be processed.

[0030] The fused feature vector is input into the trained processing priority prediction model to obtain the processing priority corresponding to each of the first tasks to be processed.

[0031] Based on the processing priority of each of the first tasks to be processed, each of the first tasks to be processed is unloaded to the target device corresponding to the target task unloading strategy information.

[0032] Secondly, this application also provides a task offloading device based on an edge computing network, comprising:

[0033] The strategy information generation module is used to generate initial task unloading strategy information for other terminals in response to a task unloading request for a terminal to be processed; the other terminals are used to refer to terminals other than the terminal to be processed in the edge computing network to which the terminal to be processed belongs.

[0034] The fitness value determination module is used to determine the fitness value of the initial task unloading strategy information;

[0035] The strategy information filtering module is used to filter out the initial task unloading strategy information with the smallest fitness value from each of the initial task unloading strategy information, and use it as the current task unloading strategy information of the other terminals.

[0036] The utility value determination module is used to determine the utility value of the terminal to be processed under each preset task unloading strategy information based on the current task unloading strategy information.

[0037] The strategy information determination module is used to filter out the preset task unloading strategy information with the largest utility value from the preset task unloading strategy information, and use it as the target task unloading strategy information for the terminal to be processed.

[0038] The task unloading processing module is used to perform corresponding task unloading processing on the terminal to be processed according to the target task unloading strategy information.

[0039] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0040] In response to a task offloading request for a terminal to be processed, initial task offloading strategy information for other terminals is generated; the other terminals are used to refer to terminals other than the terminal to be processed in the edge computing network to which the terminal to be processed belongs.

[0041] Determine the fitness value of the initial task unloading strategy information;

[0042] From the initial task unloading strategy information, the initial task unloading strategy information with the smallest fitness value is selected as the current task unloading strategy information for the other terminals;

[0043] Based on the current task unloading strategy information, the utility value of the terminal to be processed under each preset task unloading strategy information is determined.

[0044] From the preset task unloading strategy information, the preset task unloading strategy information with the largest utility value is selected as the target task unloading strategy information for the terminal to be processed.

[0045] Based on the target task unloading strategy information, the terminal to be processed is subjected to corresponding task unloading processing.

[0046] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0047] In response to a task offloading request for a terminal to be processed, initial task offloading strategy information for other terminals is generated; the other terminals are used to refer to terminals other than the terminal to be processed in the edge computing network to which the terminal to be processed belongs.

[0048] Determine the fitness value of the initial task unloading strategy information;

[0049] From the initial task unloading strategy information, the initial task unloading strategy information with the smallest fitness value is selected as the current task unloading strategy information for the other terminals.

[0050] Based on the current task unloading strategy information, the utility value of the terminal to be processed under each preset task unloading strategy information is determined.

[0051] From the preset task unloading strategy information, the preset task unloading strategy information with the largest utility value is selected as the target task unloading strategy information for the terminal to be processed.

[0052] Based on the target task unloading strategy information, the terminal to be processed is subjected to corresponding task unloading processing.

[0053] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0054] In response to a task offloading request for a terminal to be processed, initial task offloading strategy information for other terminals is generated; the other terminals are used to refer to terminals other than the terminal to be processed in the edge computing network to which the terminal to be processed belongs.

[0055] Determine the fitness value of the initial task unloading strategy information;

[0056] From the initial task unloading strategy information, the initial task unloading strategy information with the smallest fitness value is selected as the current task unloading strategy information for the other terminals.

[0057] Based on the current task unloading strategy information, the utility value of the terminal to be processed under each preset task unloading strategy information is determined.

[0058] From the preset task unloading strategy information, the preset task unloading strategy information with the largest utility value is selected as the target task unloading strategy information for the terminal to be processed.

[0059] Based on the target task unloading strategy information, the terminal to be processed is subjected to corresponding task unloading processing.

[0060] The aforementioned task offloading method, apparatus, computer device, storage medium, and computer program product based on edge computing networks firstly, in response to a task offloading request for a terminal to be processed, treats all terminals in the edge computing network to which the terminal to be processed belongs as other terminals, generates initial task offloading policy information for the other terminals, determines the fitness value of the initial task offloading policy information, and selects the initial task offloading policy information with the smallest fitness value from among the initial task offloading policy information as the current task offloading policy information for the other terminals. Next, based on the current task offloading policy information, the utility value of the terminal to be processed under each preset task offloading policy information is determined. Then, based on each preset task offloading policy information, the preset task offloading policy information with the largest utility value is selected as the target task offloading policy information for the terminal to be processed. Finally, based on the target task offloading policy information, the corresponding task offloading processing is performed on the terminal to be processed. In this way, during the task unloading process, by determining the fitness value of the initial task unloading strategy information of other terminals and selecting the strategy with the smallest fitness value as the initial task unloading strategy information, the most suitable current task unloading strategy information can be determined for other terminals. Based on the above information, the utility value of the terminal to be processed under each preset task unloading strategy information can be determined more accurately. Then, the preset task unloading strategy information with the largest utility value can be selected to determine the most suitable target task unloading strategy information for the terminal to be processed. This further enables more accurate task unloading processing for the terminal to be processed, which helps improve the accuracy of task unloading. Moreover, the whole process comprehensively considers multiple factors such as utility value and fitness value, avoiding the shortcomings of task unloading based solely on task attributes, which considers only one factor and is prone to errors, resulting in low accuracy of task unloading. This further improves the accuracy of task unloading. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0062] Figure 1 This is a flowchart illustrating a task offloading method based on an edge computing network in one embodiment;

[0063] Figure 2 This is a flowchart illustrating a task offloading method based on an edge computing network in another embodiment;

[0064] Figure 3This is a schematic diagram of a network architecture for joint optimization of task offloading and resource allocation by edge computing network terminals in one embodiment.

[0065] Figure 4 This is a schematic diagram illustrating the process of jointly optimizing task offloading and resource allocation for edge computing network terminals in one embodiment;

[0066] Figure 5 This is a schematic diagram of the latency simulation of different algorithms under different task data sizes in one embodiment;

[0067] Figure 6 This is a structural block diagram of a task offloading device based on an edge computing network in one embodiment;

[0068] Figure 7 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0069] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0070] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0071] In one exemplary embodiment, such as Figure 1 As shown, a task offloading method based on edge computing networks is provided. This embodiment illustrates the application of this method to a server; it is understood that this method can also be applied to terminals, and can also be applied to systems including terminals and servers, and is implemented through interaction between the terminal and the server. The terminal can be, but is not limited to, various personal computers, laptops, smartphones, and tablets; the server can be a standalone server or a server cluster composed of multiple servers. In this embodiment, the method includes the following steps:

[0072] Step S101: In response to the task offloading request for the terminal to be processed, generate initial task offloading strategy information for other terminals; other terminals are used to refer to terminals other than the terminal to be processed in the edge computing network to which the terminal to be processed belongs.

[0073] In this context, "task-to-be-processed terminals" refers to terminals that require task offloading. In real-world scenarios, "task-to-be-processed terminals" encompass various types of terminals, including those in healthcare, connected vehicles, smart manufacturing, and smart grids, such as medical terminals, vehicle-mounted terminals, smart terminals, and smart grid terminals.

[0074] Among them, the task unloading request refers to the request information for the terminal to be processed to perform task unloading.

[0075] Among them, other terminals are used to refer to terminals other than the terminal to be processed in the edge computing network to which the terminal to be processed belongs.

[0076] Edge computing network is used to represent a distributed computing network architecture.

[0077] The initial task unloading strategy information is used to represent the task unloading strategy information generated when a task unloading request is received from a terminal to be processed, including selecting to unload the task on the edge computing server, selecting to unload the task on the terminal, and selecting to unload the task on an idle device.

[0078] For example, the server establishes a network path with the terminal to be processed, and receives a task unloading request sent by the terminal to be processed through the network path. Then, the server performs integrity verification processing on the task unloading request and obtains the integrity verification result corresponding to the task unloading request. Then, if the integrity verification result indicates that the task unloading request is complete, the server responds to the task unloading request for the terminal to be processed by treating other terminals in the edge computing network to which the terminal to be processed belongs as other terminals. Then, the server obtains the task types of the historical tasks of other terminals, and queries the correspondence between task types and task unloading policy information to obtain the task unloading policy information corresponding to the task types of the historical tasks of other terminals, which is used as the initial task unloading policy information of other terminals.

[0079] Step S102: Determine the fitness value of the initial task unloading strategy information.

[0080] The fitness value is a quantitative measure of the degree of fit between the initial task unloading strategy information and the edge computing network.

[0081] For example, the server performs feature extraction processing on the initial task unloading strategy information to obtain the feature vector of the initial task unloading strategy information; then, the server inputs the feature vector into multiple trained fitness prediction models to obtain multiple predicted fitness values ​​corresponding to the initial task unloading strategy information; then, the server fuses the multiple predicted fitness values ​​to obtain the fitness value of the initial task unloading strategy information.

[0082] Step S103: Select the initial task unloading strategy information with the smallest fitness value from the initial task unloading strategy information and use it as the current task unloading strategy information for other terminals.

[0083] Among them, the current task unloading strategy information refers to the initial task unloading strategy information with the smallest fitness value.

[0084] For example, the server sorts the initial task unloading policy information according to the fitness value of each initial task unloading policy information to obtain sorted initial task unloading policy information; then, the server selects the initial task unloading policy information with the smallest fitness value from the sorted initial task unloading policy information and uses this initial task unloading policy information as the current task unloading policy information for other terminals.

[0085] Step S104: Determine the utility value of the terminal to be processed under each preset task unloading strategy information based on the current task unloading strategy information.

[0086] The preset task unloading policy information refers to the pre-defined task unloading policy information, including selecting to unload tasks on the edge computing server, selecting to unload tasks on the terminal, or selecting to unload tasks on an idle device. It should be noted that the preset task unloading policy information is subject to change.

[0087] The utility value is used to represent the quantitative value of the overall efficiency of the terminal to be processed under each preset task unloading strategy information.

[0088] For example, the server constructs a utility value prediction model (utility function) associated with the terminal to be processed, and uses this utility model to determine the utility value of the terminal to be processed under each preset task unloading strategy information based on the current task unloading strategy information.

[0089] Step S105: Select the preset task unloading strategy information with the highest utility value from the preset task unloading strategy information and use it as the target task unloading strategy information for the terminal to be processed.

[0090] Among them, the target task unloading strategy information refers to the preset task unloading strategy information with the highest utility value.

[0091] For example, the server sorts the preset task unloading policy information according to the utility value of the terminal to be processed under each preset task unloading policy information to obtain sorted preset task unloading policy information; then, the server selects the preset task unloading policy information with the largest utility value from the sorted preset task unloading policy information and uses the preset task unloading policy information as the target task unloading policy information of the terminal to be processed.

[0092] Step S106: Based on the target task unloading strategy information, perform corresponding task unloading processing on the terminal to be processed.

[0093] For example, the server inputs the target task unloading strategy information into the trained task unloading strategy instruction model to obtain the predicted probability of the terminal to be processed under each preset task unloading strategy instruction; then, the server selects the preset task unloading strategy instruction with the highest predicted probability from each preset task unloading strategy instruction as the target task unloading strategy instruction for the terminal to be processed; then, the server performs the corresponding task unloading processing on the terminal to be processed according to the target task unloading strategy instruction.

[0094] In the aforementioned task offloading method based on edge computing networks, in response to a task offloading request for a terminal to be processed, all terminals in the edge computing network to which the terminal to be processed belongs, excluding the terminal to be processed, are considered as other terminals. Initial task offloading policy information for these other terminals is generated, and the fitness value of the initial task offloading policy information is determined. From these initial task offloading policy information, the initial task offloading policy information with the smallest fitness value is selected as the current task offloading policy information for the other terminals. Next, based on the current task offloading policy information, the utility value of the terminal to be processed under each preset task offloading policy information is determined. Then, from these preset task offloading policy information, the preset task offloading policy information with the largest utility value is selected as the target task offloading policy information for the terminal to be processed. Finally, based on the target task offloading policy information, the corresponding task offloading process is performed on the terminal to be processed. In this way, during the task unloading process, by determining the fitness value of the initial task unloading strategy information of other terminals and selecting the strategy with the smallest fitness value as the initial task unloading strategy information, the most suitable current task unloading strategy information can be determined for other terminals. Based on the above information, the utility value of the terminal to be processed under each preset task unloading strategy information can be determined more accurately. Then, the preset task unloading strategy information with the largest utility value can be selected to determine the most suitable target task unloading strategy information for the terminal to be processed. This further enables more accurate task unloading processing for the terminal to be processed, which helps improve the accuracy of task unloading. Moreover, the whole process comprehensively considers multiple factors such as utility value and fitness value, avoiding the shortcomings of task unloading based solely on task attributes, which considers only one factor and is prone to errors, resulting in low accuracy of task unloading. This further improves the accuracy of task unloading.

[0095] In an exemplary embodiment, step S102, which determines the fitness value of the initial task unloading strategy information, specifically includes the following: querying the correspondence between the task unloading strategy information and historical task latency values ​​to obtain multiple historical task latency values ​​corresponding to the initial task unloading strategy information; obtaining the average value of the multiple historical task latency values ​​as the average task latency value corresponding to the initial task unloading strategy information; and determining the fitness value of the initial task unloading strategy information based on the average task latency value.

[0096] The correspondence between task offloading policy information and historical task latency values ​​indicates the association between them. For example, if the task offloading policy is set to offload the task on an edge computing server, the corresponding historical task latency values ​​are 150 milliseconds, 120 milliseconds, and 180 milliseconds; if the policy is set to offload the task on a terminal, the corresponding historical task latency values ​​are 30 milliseconds, 20 milliseconds, and 40 milliseconds; and if the policy is set to offload the task on an idle device, the corresponding historical task latency values ​​are 80 milliseconds, 100 milliseconds, and 60 milliseconds.

[0097] The average task latency value corresponding to the initial task offloading policy information refers to the average of multiple historical task latency values ​​corresponding to the initial task offloading policy information. For example, if the task offloading policy information selects to offload tasks on the edge computing server, the corresponding historical task latency values ​​are 150 milliseconds, 120 milliseconds, and 180 milliseconds, and the average task latency value corresponding to the initial task offloading policy information is 150 milliseconds.

[0098] For example, the server queries the correspondence between the initial task unloading policy information and the historical task latency values ​​based on the initial task unloading policy information to obtain multiple historical task latency values ​​corresponding to the initial task unloading policy information; then, the server obtains the average value of the multiple historical task latency values ​​corresponding to the initial task unloading policy information and uses this average value as the average task latency value corresponding to the initial task unloading policy information; then, the server determines the fitness value of the initial task unloading policy information based on the average task latency value.

[0099] In this embodiment, by querying the corresponding relationship and averaging the latency values ​​of multiple historical tasks corresponding to the initial task unloading strategy information, the average task latency value is obtained, thereby eliminating the influence of individual outliers, making the evaluation results more objective and stable, and thus reflecting the performance of the initial task unloading strategy information in historical task processing, providing a more reliable basis for subsequently determining the fitness value.

[0100] In an exemplary embodiment, the fitness value of the initial task offloading policy information is determined based on the average task latency value. Specifically, this includes: obtaining the number of terminals corresponding to the initial task offloading policy information; obtaining the latency adjustment value associated with the number of terminals; summing the average task latency value and the latency adjustment value to obtain the processed average task latency value, which is used as the fitness value of the initial task offloading policy information.

[0101] The number of terminals is used to represent the number of terminals corresponding to the selection of the initial task unloading strategy information.

[0102] The latency adjustment value is used to represent the numerical value used to adjust the average task latency value.

[0103] For example, the server obtains the number of terminals corresponding to the initial task offloading policy information as the number of terminals corresponding to the initial task offloading policy information; then, the server queries the correspondence between the number of terminals and the latency adjustment value according to the number of terminals corresponding to the initial task offloading policy information to obtain the latency adjustment value associated with the number of terminals; then, the server sums the average task latency value and the latency adjustment value to obtain the processed average task latency value, and uses the processed average task latency value as the fitness value of the initial task offloading policy information.

[0104] In this embodiment, by introducing a latency adjustment value associated with the number of terminals, the effectiveness of the strategy can be measured from a more comprehensive perspective, avoiding the one-sidedness of judging the strategy solely by the average latency. This allows the fitness value to more accurately reflect the performance of the strategy in real-world complex scenarios, which is beneficial for improving the accuracy of determining the fitness value of the initial task unloading strategy information.

[0105] In an exemplary embodiment, step S104, which determines the utility value of the terminal to be processed under each preset task unloading policy information based on the current task unloading policy information, specifically includes the following: determining the current task latency value of the terminal to be processed under each preset task unloading policy information based on the current task unloading policy information; and determining the utility value of the terminal to be processed under each preset task unloading policy information based on the current task latency value of the terminal to be processed under each preset task unloading policy information.

[0106] The current task latency value refers to the task latency value of the terminal to be processed under the preset task unloading policy information, such as 100 milliseconds.

[0107] For example, the server determines the current task latency value of the terminal to be processed under each preset task uninstallation policy information based on the current task uninstallation policy information; then, the server performs reciprocal processing on the current task latency value of the terminal to be processed under each preset task uninstallation policy information to obtain the utility value of the terminal to be processed under each preset task uninstallation policy information.

[0108] In this embodiment, by determining the current task latency value of the terminal to be processed under each preset task offloading strategy information, the impact of different strategies on task processing time can be compared in detail. This allows for a direct presentation of the differences in task execution speed between strategies, helping to assess whether a strategy can respond quickly and complete a task. This is beneficial to improving the accuracy of determining the utility value of the terminal to be processed under each preset task offloading strategy information.

[0109] In an exemplary embodiment, the current task latency value of the terminal to be processed under each preset task offloading policy information is determined according to the current task offloading policy information. Specifically, this includes: determining the first task to be processed corresponding to the terminal to be processed, and the second tasks to be processed corresponding to other terminals; obtaining the first task data volume and the first task computing density corresponding to the first task to be processed, and the second task data volume and the second task computing density corresponding to the second task to be processed; and determining the current task latency value of the terminal to be processed under each preset task offloading policy information based on the current task offloading policy information, the first task data volume, the first task computing density, the second task data volume, and the second task computing density.

[0110] The first task to be processed refers to the task to be processed corresponding to the terminal to be processed.

[0111] The second pending task refers to the pending tasks corresponding to other terminals.

[0112] The first task data volume refers to the size (in bits) of the task data corresponding to the first task to be processed.

[0113] The first task computation density refers to the task computation density corresponding to the first task to be processed, that is, the number of CPU (Central Processing Unit) cycles that can be processed in 1 second.

[0114] The second task data volume refers to the size (in bits) of the task data corresponding to the second task to be processed.

[0115] The second task computation density refers to the task computation density corresponding to the second task to be processed, that is, the number of CPU cycles that can be processed in 1 second.

[0116] For example, the server determines the task to be processed corresponding to the terminal to be processed as the first task to be processed, and determines the tasks to be processed corresponding to other terminals as the second tasks to be processed. Then, the server obtains the first task data volume and the first task computation density corresponding to the first task to be processed, and the second task data volume and the second task computation density corresponding to the second task to be processed. Then, the server determines the third task data volume and the third task computation density of other terminals under each preset task unloading policy information based on the current task unloading policy information, the second task data volume and the second task computation density. Then, the server obtains the computation frequency under each preset task unloading policy information, and determines the current task latency value of the terminal to be processed under each preset task unloading policy information based on the computation frequency, the first task data volume, the first task computation density, the third task data volume and the third task computation density under each preset task unloading policy information.

[0117] In this embodiment, by determining the tasks to be processed of the terminal to be processed and other terminals, and obtaining the corresponding task data volume and computing density, the characteristics of different terminal tasks can be comprehensively and meticulously considered. Combined with the current task unloading strategy information, the current task latency value of the terminal to be processed under each preset task unloading strategy is determined, making the evaluation of the task unloading strategy more accurate.

[0118] In an exemplary embodiment, step S106, which involves performing corresponding task unloading processing on the terminal to be processed according to the target task unloading strategy information, specifically includes the following: extracting a first feature vector corresponding to the first task data volume and a second feature vector corresponding to the first task computation density; fusing the first feature vector and the second feature vector to obtain a fused feature vector corresponding to the first task to be processed; inputting the fused feature vector into the trained processing priority prediction model to obtain the processing priority corresponding to each first task to be processed; and unloading each first task to be processed to the target device corresponding to the target task unloading strategy information according to the processing priority corresponding to each first task to be processed.

[0119] Here, the first feature vector refers to the representation vector corresponding to the amount of data in the first task.

[0120] The second feature vector refers to the representation vector corresponding to the amount of data in the second task.

[0121] Among them, the fused feature vector refers to the feature vector obtained by fusing the first feature vector and the second feature vector.

[0122] Among them, the processing priority prediction model refers to a network model that can use the fused feature vector corresponding to the first task to be processed to obtain the processing priority of the first task to be processed, such as a convolutional neural network model.

[0123] Among them, the processing priority refers to the priority corresponding to the task unloading and processing of the first task to be processed.

[0124] The target devices include edge computing servers, terminals, and idle devices.

[0125] For example, the server inputs the first task data volume and the first task computation density into the feature extraction model, and extracts the first feature vector corresponding to the first task data volume and the second feature vector corresponding to the first task computation density. Then, the server determines the first weight corresponding to the first feature vector and the second weight corresponding to the second feature vector, and sums the first feature vector and the second feature vector according to the first weight and the second weight to obtain the fused feature vector corresponding to the first task to be processed. Then, the server trains the processing priority prediction model to be trained to obtain the trained processing priority prediction model, and inputs the fused feature vector into the trained processing priority prediction model to obtain the processing priority corresponding to each first task to be processed. Then, the server unloads each first task to be processed to the target device corresponding to the target task unloading strategy information according to the processing priority corresponding to each first task to be processed.

[0126] In this embodiment, by extracting the first feature vector corresponding to the first task data volume and the second feature vector corresponding to the first task computation density, and by utilizing the model's learning and prediction capabilities, the processing priorities corresponding to each first task to be processed are more accurate and reliable. This avoids unreasonable priority allocation caused by subjective judgment or simple rules, making task processing more in line with actual needs and resource utilization efficiency.

[0127] In one exemplary embodiment, such as Figure 2 As shown, another task offloading method based on edge computing networks is provided. Taking the application of this method to a server as an example, the specific steps include:

[0128] Step S201: In response to the task offloading request for the terminal to be processed, generate initial task offloading strategy information for other terminals; other terminals are used to refer to terminals other than the terminal to be processed in the edge computing network to which the terminal to be processed belongs.

[0129] Step S202: Query the correspondence between task unloading policy information and historical task latency values ​​to obtain multiple historical task latency values ​​corresponding to the initial task unloading policy information; obtain the average value of multiple historical task latency values ​​as the average task latency value corresponding to the initial task unloading policy information.

[0130] Step S203: Obtain the number of terminals corresponding to the initial task offloading strategy information; obtain the latency adjustment value associated with the number of terminals; sum the average task latency value and the latency adjustment value to obtain the processed average task latency value, which is used as the fitness value of the initial task offloading strategy information.

[0131] Step S204: Select the initial task unloading strategy information with the smallest fitness value from the initial task unloading strategy information and use it as the current task unloading strategy information for other terminals.

[0132] Step S205: Determine the first task to be processed corresponding to the terminal to be processed, and the second task to be processed corresponding to other terminals.

[0133] Step S206: Obtain the first task data volume and first task computation density corresponding to the first task to be processed, and the second task data volume and second task computation density corresponding to the second task to be processed.

[0134] Step S207: Determine the current task latency value of the terminal to be processed under each preset task unloading strategy information based on the current task unloading strategy information, the first task data volume, the first task computation density, the second task data volume, and the second task computation density.

[0135] Step S208: Determine the utility value of the terminal under each preset task unloading strategy information based on the current task latency value of the terminal under each preset task unloading strategy information.

[0136] Step S209: Select the preset task unloading strategy information with the highest utility value from the preset task unloading strategy information and use it as the target task unloading strategy information for the terminal to be processed.

[0137] Step S210: Extract the first feature vector corresponding to the first task data volume and the second feature vector corresponding to the first task computation density; perform fusion processing on the first feature vector and the second feature vector to obtain the fused feature vector corresponding to the first task to be processed.

[0138] Step S211: Input the fused feature vector into the trained processing priority prediction model to obtain the processing priority corresponding to each first task to be processed; according to the processing priority corresponding to each first task to be processed, unload each first task to be processed to the target device corresponding to the target task unloading strategy information.

[0139] In the aforementioned task offloading method based on edge computing networks, during the task offloading process, by determining the fitness value of the initial task offloading strategy information of other terminals and selecting the strategy with the smallest fitness value as the initial task offloading strategy information, the most suitable current task offloading strategy information can be determined for other terminals. Based on the above information, the utility value of the terminal to be processed under each preset task offloading strategy information can be determined more accurately. Furthermore, the preset task offloading strategy information with the largest utility value can be selected to determine the most suitable target task offloading strategy information for the terminal to be processed. This further enables more accurate task offloading processing for the terminal to be processed, which is beneficial to improving the accuracy of task offloading. Moreover, the entire process comprehensively considers multiple factors such as utility value and fitness value, avoiding the shortcomings of task offloading based solely on task attributes, which considers only one factor and is prone to errors, resulting in low accuracy of task offloading. This further improves the accuracy of task offloading.

[0140] In an exemplary embodiment, to more clearly illustrate the task offloading method based on an edge computing network provided in this application, the following specific embodiment will be used to describe the task offloading method based on an edge computing network. In one embodiment, this application also provides a method for jointly optimizing task offloading and resource allocation of edge computing network terminals. During the task offloading process, firstly, in response to a task offloading request for a terminal to be processed, terminals in the edge computing network to which the terminal to be processed belongs, other than the terminal to be processed, are considered as other terminals, and initial task offloading policy information for other terminals is generated. Then, the fitness value of the initial task offloading policy information is determined, and the initial task offloading policy information with the smallest fitness value is selected from each initial task offloading policy information as the current task offloading policy information for other terminals. Next, based on the current task offloading policy information, the utility value of the terminal to be processed under each preset task offloading policy information is determined. Then, based on each preset task offloading policy information, the preset task offloading policy information with the largest utility value is selected as the target task offloading policy information for the terminal to be processed. Finally, based on the target task offloading policy information, the corresponding task offloading process is performed on the terminal to be processed. Specifically, it includes the following:

[0141] Please see Figure 3 , Figure 4 , Figure 5 , Figure 3 This is a network architecture diagram of a method for jointly optimizing task offloading and resource allocation of edge computing network terminals proposed in this embodiment; Figure 4 This is a flowchart of a method for jointly optimizing task offloading and resource allocation of edge computing network terminals proposed in this embodiment; Figure 5This is a time delay simulation diagram for different algorithms under different task data sizes provided in this embodiment.

[0142] Includes the following steps:

[0143] S1, Initialization Phase: Collect basic information on edge computing servers, smart terminal devices, and idle devices;

[0144] S2. Based on the computing resources of idle devices, a mixed-integer nonlinear programming model is established to jointly optimize the task offloading strategy and resource allocation scheme. The optimization objective is to minimize the total task processing latency.

[0145] S3, decompose the optimization problem into a computational resource allocation subproblem and a task unloading strategy subproblem;

[0146] S4. Solve the computational resource allocation subproblem. When the unloading strategy is known, the Lagrange multiplier method is used to obtain the optimal resource allocation scheme.

[0147] S5 solves the task unloading strategy subproblem. First, an improved particle swarm optimization algorithm is used to generate an initial unloading strategy, and then a Nash equilibrium game algorithm is used to obtain the optimal unloading strategy.

[0148] In step S1, the basic information collected includes the task data size, the location information of the edge computing server, terminal and idle device, the number of CPUs required for task computing, and the computing power of the edge computing server, terminal and idle device.

[0149] Step S2 also includes the following steps:

[0150] S21, consider using a regional server A network architecture composed of multiple terminal devices, such as Figure 1 As shown. Terminals are divided into two types: ordinary terminals with computing tasks and idle devices. Ordinary terminals are represented as... Idle devices are represented as The two-dimensional position coordinates of the server can be represented as: ,terminal The position coordinates are Idle devices The position coordinates are A typical terminal has a computing task available. It means that among them It is a terminal The size of the task data (in bits). It is the number of CPU cycles required to compute 1 bit of data. It is a terminal The computing frequency (Hz) is the number of CPU cycles that can be processed per second. It is a server The calculation frequency. Idle devices The computation frequency. Assuming each task is indivisible, ordinary terminal tasks can choose to compute locally, upload to a server for computation, or use an idle device to assist with computation.

[0151] S22, if the terminal If you choose to compute the task locally, the required latency will be the same as the latency required to process the task locally. pass Calculated.

[0152] S23, if the terminal If server-side computation is chosen, the task processing consists of three stages: the terminal uploads the task to the server, the server processes the task, and the server returns the result to the terminal. Since the amount of data in the calculated result is relatively small, the download time is negligible. Therefore, the latency... Due to upload latency and computational delay composition: Server to terminal Channel gain The free-space path loss model can be used for calculation:

[0153] Equation (1)

[0154] in, The channel gain is given when the distance between the server and the terminal is 1m. For the terminal Distance from the server Let be the path loss coefficient. Assume the total bandwidth is... , If each terminal is evenly allocated a frequency, and each terminal occupies a non-overlapping frequency, then the terminal... The transfer rate of the upload task to the server for: .in, For the terminal The transmission power, This represents noise power. Therefore, the upload latency... pass The calculated server computation latency pass The calculation yielded the following result. For server Assigned to the terminal Computing resources.

[0155] S24, Terminal If you select an idle device To assist with computation, the task is first uploaded to an idle device, which then performs the computation. Therefore, the latency consists of upload latency and idle device computation latency, while the latency of sending back the data result is ignored.

[0156] Equation (2)

[0157] in, This refers to time delay. This refers to upload latency. This refers to the computation delay.

[0158] Equation (3)

[0159] Equation (4)

[0160] in, Refers to the terminal Upload task to idle device The transmission rate.

[0161] Equation (5)

[0162] in, For the bandwidth between the terminal and the idle device, Channel gain between the terminal and the idle device , This represents noise power.

[0163] S25, assuming the task is indivisible, a regular terminal task can choose to be computed locally, uploaded to a server for computation, or use an idle device to assist with computation. Define a task offloading variable. for ,in, Refers to the terminal Do you want to compute locally? Refers to the terminal Do you want to upload the data to the server for calculation? Refers to the terminal Choose to use an idle device calculate, , , . Indicates terminal Choose to compute locally. =0 indicates the terminal Local computation was not selected. Indicates terminal Choose to upload to the server for calculation. =0 indicates the terminal The option to upload to the server for calculation was not selected. Indicates terminal Select on idle device calculate, =0 indicates the terminal No choice was made on idle devices Calculation. Because the task is indivisible, a task can only choose one unloading method: .

[0164] Let set To select the terminal for server-side computation The computing resources allocated by the server to these terminals The sum must not exceed the server's total computing resources. : .

[0165] Because idle devices have limited computing resources, they can only process tasks from one terminal device at most within a single cycle. .

[0166] The optimization objective is to minimize the total task processing latency. : .

[0167] In summary, this optimization problem can be expressed as:

[0168] Equation (6)

[0169] In step S3, the optimization problem in step S2 is decomposed into two sub-problems: the computational resource allocation sub-problem and the task unloading strategy sub-problem.

[0170] Step S4 also includes the following steps:

[0171] S41, When the task unloading strategy is fixed, the only optimization variable is the allocation of server computing resources. After selecting fixed latency values ​​for tasks that are uninstalled locally and computed on idle devices, and removing the constant term, the problem persists. This can be transformed into a problem. :

[0172] Equation (7)

[0173] S42, right Find the first derivative: . right Find the second derivative: .so, It is a convex function.

[0174] S43, using the Lagrange multiplier method and the KKT (Karush-Kuhn-Tucker) conditions, can be obtained as follows:

[0175] Equation (8)

[0176] in, These are Lagrange multipliers. Solving for them yields the optimal allocation of computational resources. :

[0177] Equation (9)

[0178] in, This refers to the total computing resources of the server.

[0179] Step S5 also includes the following steps:

[0180] S51, assuming the unloading strategy is known, calculate using the Lagrange multiplier method. This allows us to obtain the optimal computing resources allocated to the terminal by the server. When the task offloading strategy differs, Different, problem The total latency also differs. Therefore, sub-problem two requires optimizing the task unloading strategy. This can be transformed into a problem. :

[0181] Equation (10)

[0182] S52, In the multi-terminal unloading decision problem, the unloading decision of a terminal is related to the unloading strategies of other terminals. Therefore, this embodiment uses game theory to solve the task unloading strategy problem. This task unloading strategy problem can be defined as a game theory problem. ,in It refers to the number of terminals. It is the first Uninstallation strategy for each terminal , It is a terminal utility function In addition to the terminal The set of uninstallation strategies for other terminals, Each terminal will choose an offloading strategy to maximize its own utility function:

[0183] Equation (11)

[0184] S53 employs a Nash equilibrium game theory algorithm to solve the unloading strategy subproblem: First, an initial unloading strategy is generated using a particle swarm optimization algorithm improved with a fitness function, and then an iterative process begins. In each iteration, for each terminal, the utility of each different unloading strategy is independently calculated. If a constraint is not satisfied, the strategy is updated. Then, each terminal selects the unloading strategy that maximizes its utility and updates it. The iteration terminates when the maximum number of iterations is reached or no terminal intends to unilaterally change its unloading strategy to obtain better utility.

[0185] In the above embodiments, during the task unloading process, by determining the fitness value of the initial task unloading strategy information of other terminals and selecting the strategy with the smallest fitness value as the initial task unloading strategy information, the most suitable current task unloading strategy information can be determined for other terminals. Based on this information, the utility value of the terminal to be processed under each preset task unloading strategy information can be determined more accurately. Furthermore, the preset task unloading strategy information with the largest utility value can be selected, determining the most suitable target task unloading strategy information for the terminal to be processed. This further allows for more accurate task unloading processing for the terminal to be processed, improving the accuracy of task unloading. Moreover, the entire process comprehensively considers multiple factors such as utility value and fitness value, avoiding the shortcomings of task unloading based solely on task attributes, which is prone to errors and low accuracy. This further improves the accuracy of task unloading. Simultaneously, this embodiment considers the computing resources of idle devices, avoiding resource waste, and jointly optimizes task unloading strategies and resource allocation schemes to achieve better performance, reduce communication latency, and improve data processing speed, thus meeting users' service quality requirements for low latency and high throughput. Furthermore, this embodiment proposes an improved particle swarm optimization algorithm for solving constrained discrete variable optimization problems, which can be used to generate initial unloading strategies to reduce the number of iterations in game theory algorithms.

[0186] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0187] Based on the same inventive concept, this application also provides an edge computing network-based task offloading device for implementing the above-described task offloading method based on edge computing networks. The solution provided by this device is similar to the solution described in the above-described method. Therefore, the specific limitations of the one or more edge computing network-based task offloading device embodiments provided below can be found in the limitations of the edge computing network-based task offloading method described above, and will not be repeated here.

[0188] In one exemplary embodiment, such as Figure 6 As shown, a task offloading device based on an edge computing network is provided, comprising: a policy information generation module 601, a fitness value determination module 602, a policy information filtering module 603, a utility value determination module 604, a policy information determination module 605, and a task offloading processing module 606, wherein:

[0189] The strategy information generation module 601 is used to generate initial task unloading strategy information for other terminals in response to a task unloading request for the terminal to be processed; other terminals are used to refer to terminals other than the terminal to be processed in the edge computing network to which the terminal to be processed belongs.

[0190] The fitness value determination module 602 is used to determine the fitness value of the initial task unloading strategy information;

[0191] The strategy information filtering module 603 is used to filter out the initial task unloading strategy information with the smallest fitness value from each initial task unloading strategy information, and use it as the current task unloading strategy information for other terminals.

[0192] The utility value determination module 604 is used to determine the utility value of the terminal to be processed under each preset task unloading strategy information based on the current task unloading strategy information.

[0193] The strategy information determination module 605 is used to filter out the preset task unloading strategy information with the highest utility value from each preset task unloading strategy information, and use it as the target task unloading strategy information for the terminal to be processed.

[0194] The task unloading processing module 606 is used to perform corresponding task unloading processing on the terminal to be processed according to the target task unloading strategy information.

[0195] In an exemplary embodiment, the fitness value determination module 602 is further configured to query the correspondence between task unloading strategy information and historical task latency values ​​to obtain multiple historical task latency values ​​corresponding to the initial task unloading strategy information; obtain the average value of multiple historical task latency values ​​as the average task latency value corresponding to the initial task unloading strategy information; and determine the fitness value of the initial task unloading strategy information based on the average task latency value.

[0196] In an exemplary embodiment, the fitness value determination module 602 is further configured to obtain the number of terminals corresponding to the initial task unloading strategy information; obtain the latency adjustment value associated with the number of terminals; and sum the average task latency value and the latency adjustment value to obtain the processed average task latency value, which is used as the fitness value of the initial task unloading strategy information.

[0197] In an exemplary embodiment, the utility value determination module 604 is further configured to determine the current task latency value of the terminal to be processed under each preset task uninstallation policy information based on the current task uninstallation policy information; and to determine the utility value of the terminal to be processed under each preset task uninstallation policy information based on the current task latency value of the terminal to be processed under each preset task uninstallation policy information.

[0198] In an exemplary embodiment, the utility value determination module 604 is further configured to determine the first task to be processed corresponding to the terminal to be processed, and the second tasks to be processed corresponding to other terminals; obtain the first task data volume and the first task computing density corresponding to the first task to be processed, and the second task data volume and the second task computing density corresponding to the second task to be processed; and determine the current task latency value of the terminal to be processed under each preset task unloading strategy information based on the current task unloading strategy information, the first task data volume, the first task computing density, the second task data volume and the second task computing density.

[0199] In an exemplary embodiment, the task unloading processing module 606 is further configured to extract a first feature vector corresponding to the first task data volume and a second feature vector corresponding to the first task computation density; perform fusion processing on the first feature vector and the second feature vector to obtain a fused feature vector corresponding to the first task to be processed; input the fused feature vector into the trained processing priority prediction model to obtain the processing priority corresponding to each first task to be processed; and unload each first task to be processed to the target device corresponding to the target task unloading strategy information according to the processing priority corresponding to each first task to be processed.

[0200] Each module in the aforementioned task offloading device based on edge computing networks can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can invoke and execute the operations corresponding to each module.

[0201] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 7As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores data such as fitness values ​​and utility values. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a task offloading method based on an edge computing network.

[0202] Those skilled in the art will understand that Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0203] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0204] In one exemplary embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above-described method embodiments.

[0205] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0206] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0207] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0208] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A task offloading method based on edge computing networks, characterized in that, The method includes: In response to a task offloading request for a terminal to be processed, initial task offloading strategy information for other terminals is generated; the other terminals are used to refer to terminals other than the terminal to be processed in the edge computing network to which the terminal to be processed belongs. Determine the fitness value of the initial task unloading strategy information; From the initial task unloading strategy information, the initial task unloading strategy information with the smallest fitness value is selected as the current task unloading strategy information for the other terminals. Based on the current task unloading strategy information, the utility value of the terminal to be processed under each preset task unloading strategy information is determined. From the preset task unloading strategy information, the preset task unloading strategy information with the largest utility value is selected as the target task unloading strategy information for the terminal to be processed. Based on the target task unloading strategy information, the terminal to be processed is subjected to corresponding task unloading processing.

2. The method according to claim 1, characterized in that, The determination of the fitness value of the initial task unloading strategy information includes: Query the correspondence between task unloading policy information and historical task latency values ​​to obtain multiple historical task latency values ​​corresponding to the initial task unloading policy information; The average value of the multiple historical task latency values ​​is obtained and used as the average task latency value corresponding to the initial task unloading strategy information. The fitness value of the initial task unloading strategy information is determined based on the average task latency value.

3. The method according to claim 2, characterized in that, The step of determining the fitness value of the initial task unloading strategy information based on the average task latency value includes: Obtain the number of terminals corresponding to the initial task unloading strategy information; Obtain the latency adjustment value associated with the number of terminals; The average task latency value and the latency adjustment value are summed to obtain the processed average task latency value, which is used as the fitness value of the initial task unloading strategy information.

4. The method according to claim 1, characterized in that, The step of determining the utility value of the terminal to be processed under each preset task unloading strategy based on the current task unloading strategy information includes: Based on the current task unloading strategy information, the current task latency value of the terminal to be processed under each preset task unloading strategy information is determined. Based on the current task latency value of the terminal to be processed under each preset task unloading policy information, the utility value of the terminal to be processed under each preset task unloading policy information is determined.

5. The method according to claim 4, characterized in that, The step of determining the current task latency value of the terminal to be processed under each preset task unloading policy information based on the current task unloading policy information includes: The first task to be processed corresponding to the terminal to be processed and the second task to be processed corresponding to the other terminals are determined. Obtain the first task data volume and first task computation density corresponding to the first task to be processed, and the second task data volume and second task computation density corresponding to the second task to be processed; Based on the current task unloading strategy information, the first task data volume, the first task computation density, the second task data volume, and the second task computation density, the current task latency value of the terminal to be processed under each preset task unloading strategy information is determined.

6. The method according to claim 5, characterized in that, The step of performing corresponding task unloading processing on the terminal to be processed according to the target task unloading strategy information includes: Extract the first feature vector corresponding to the first task data volume and the second feature vector corresponding to the first task computation density; The first feature vector and the second feature vector are fused to obtain the fused feature vector corresponding to the first task to be processed. The fused feature vector is input into the trained processing priority prediction model to obtain the processing priority corresponding to each of the first tasks to be processed. Based on the processing priority of each of the first tasks to be processed, each of the first tasks to be processed is unloaded to the target device corresponding to the target task unloading strategy information.

7. A task offloading device based on an edge computing network, characterized in that, The device includes: The strategy information generation module is used to generate initial task unloading strategy information for other terminals in response to a task unloading request for a terminal to be processed; the other terminals are used to refer to terminals other than the terminal to be processed in the edge computing network to which the terminal to be processed belongs. The fitness value determination module is used to determine the fitness value of the initial task unloading strategy information; The strategy information filtering module is used to filter out the initial task unloading strategy information with the smallest fitness value from each of the initial task unloading strategy information, and use it as the current task unloading strategy information of the other terminals. The utility value determination module is used to determine the utility value of the terminal to be processed under each preset task unloading strategy information based on the current task unloading strategy information. The strategy information determination module is used to filter out the preset task unloading strategy information with the largest utility value from the preset task unloading strategy information, and use it as the target task unloading strategy information for the terminal to be processed. The task unloading processing module is used to perform corresponding task unloading processing on the terminal to be processed according to the target task unloading strategy information.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images