A terminal side device unloading method and system based on edge orchestration technology

Abstract

The present invention relates to the field of terminal computing technology, and in particular to a terminal device unloading method and system based on edge orchestration technology. The present invention constructs a policy tree and uses the policy tree to perform feature analysis on tasks to determine task types and task characteristics and evaluate the available resource amounts of terminal devices in real time. Based on the determined task types, task characteristics and the available resource amounts of terminal devices evaluated in real time, an execution strategy is generated, and tasks are executed based on the generated execution strategy, thereby helping to achieve efficient and accurate task unloading of terminal devices and improve resource utilization efficiency and performance of embedded systems.

Description

Technical Field

[0001] The present invention relates to the field of edge computing technology, and in particular to an edge device unloading method and system based on edge orchestration technology. Background Art

[0002] As the functions of edge devices (such as smartphones, IoT devices, etc.) continue to increase and application scenarios become more diverse, the demand for computing power is also increasing. However, the resources of edge devices (such as computing power, storage, battery life, etc.) are usually limited, making it difficult to meet complex and large-scale computing tasks. Cloud-based serverless platform computing, as an emerging computing model, provides a new approach to solving this problem. Mobile edge computing provides an effective way to solve this problem by offloading computing tasks to edge servers.

[0003] However, the existing offloading strategies for the limited resources of the terminal devices are not accurate enough in embedded system applications, have poor adaptability to the dynamic changes of the system, and are constrained by various resources. Summary of the invention

[0004] In order to solve the technical problems existing in the above-mentioned background technology, the present invention provides a terminal-side device unloading method and system based on edge orchestration technology, which helps to achieve efficient and accurate task unloading of terminal-side devices and improve the resource utilization efficiency and performance of embedded systems.

[0005] In order to implement the above technical solution, in a first aspect, the present invention provides a terminal device unloading method based on edge orchestration technology, comprising the following steps:

[0006] Step 1: Build a strategy tree;

[0007] Step 2: Perform feature analysis on the task through the constructed strategy tree to determine the task type and task characteristics;

[0008] Step 3: Evaluate the available resources of the end-side device in real time;

[0009] Step 4: Based on the determined task type, task characteristics, and available resource amounts of the end-side device evaluated in real time, an execution strategy is generated, wherein the execution strategy includes a first execution strategy of offloading the task to a cloud serverless platform for execution and a second execution strategy of executing the task on the end-side device;

[0010] Step 5: Execute the task based on the generated execution strategy.

[0011] Furthermore, the task characteristics include computational complexity, data volume, and task delay tolerance;

[0012] The available resources of the end-side device include the computing resources of the end-side device, the CPU idle ratio, the memory idle ratio and the network latency.

[0013] Furthermore, the step 1 comprises:

[0014] Create a dataset that includes task characteristics and processing strategies;

[0015] Calculate the information gain of each task feature in the data set, and use the task feature with the largest information gain as the root node of the strategy tree;

[0016] Recursively split the root node to build a complete strategy tree;

[0017] Furthermore, the step 4 includes:

[0018] Determine whether the computational complexity, data volume, delay tolerance, CPU idle ratio, memory idle ratio, and network delay of each task meet the following preset conditions:

[0019]

[0020] is the computational complexity; is the amount of data; is delay tolerance; is the CPU idle ratio of the end device, is the memory idle ratio of the end device, is the network delay of the end-side device, is the adjustment coefficient, is a positive integer;

[0021] If the conditions are met, the offloading strategy indicator is calculated according to the computational complexity, data volume, delay tolerance and task type of each task; if the conditions are not met, the first execution strategy is generated.

[0022] Furthermore, the step 4 also includes:

[0023] The cost of computing tasks on the end device and in the cloud.

[0024] An execution strategy is generated based on the calculated execution cost of the task on the end device and the cloud and the calculated offloading strategy indicator.

[0025] Furthermore, the step five comprises:

[0026] During the task execution process, the system status and task execution status are continuously monitored to dynamically adjust the weight coefficient and preset indicator threshold.

[0027] Further, the calculation of the offloading strategy indicator according to the determined computational complexity, data volume, delay tolerance and task type of each task includes:

[0028] The offloading strategy indicator is formulated according to the computational complexity, data volume and delay tolerance of each task. The offloading strategy indicator calculation formula is:

[0029]

[0030] in, To uninstall the strategy indicator; is the computational complexity; is the amount of data; is the delay tolerance, , and is the weight coefficient, and the weight coefficient satisfies the following formula:

[0031]

[0032] in, The value is 3;

[0033] Adjust according to the determined task type. , and Weight coefficient to calculate the uninstall strategy indicator.

[0034] Furthermore, the method also includes monitoring task execution performance and providing feedback.

[0035] In a second aspect, the present invention provides a terminal device unloading system based on edge orchestration technology, comprising:

[0036] Construction module, used to build the strategy tree;

[0037] The task feature analysis module is used to perform feature analysis on the task through the constructed strategy tree to determine the task type, task features and resources required for the task;

[0038] End-side resource evaluation module: used to evaluate the computing resources currently available on the end-side device in real time;

[0039] Offloading strategy formulation module: based on the determined task type, task characteristics and available resource amount of the terminal device evaluated in real time, generating an execution strategy, wherein the execution strategy includes a first execution strategy of offloading the task to the cloud serverless platform for execution and a second execution strategy of executing the task on the terminal device;

[0040] Task execution module: used to execute tasks based on the generated execution strategy.

[0041] In a third aspect, the present invention provides a computer-readable storage medium, which includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the end-side device uninstallation method based on edge orchestration technology described above.

[0042] The beneficial effects of the present invention are:

[0043] (1) The present invention constructs a strategy tree and uses the strategy tree to perform feature analysis on tasks to determine the task type and task characteristics and evaluate the available resources of the terminal device in real time. Based on the determined task type, task characteristics and the available resources of the terminal device evaluated in real time, an execution strategy is generated, and the task is executed based on the generated execution strategy, thereby helping to achieve efficient and accurate task offloading of the terminal device and improve the resource utilization efficiency and performance of the embedded system.

[0044] (2) By constructing a strategy tree, we can determine which tasks are executed on edge devices and which tasks are executed on the cloud serverless platform to achieve more efficient resource allocation. By judging the branches of the strategy tree, we can reasonably allocate limited computing, storage, and network resources to different types of tasks, avoid resource waste, and improve the overall resource utilization efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0046] Figure 1 The present invention is a flow chart of a terminal device unloading method based on edge orchestration technology.

[0047] Figure 2 This is a flow chart of the present invention for constructing a complete strategy tree through recursive splitting of the root node. DETAILED DESCRIPTION

[0048] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0049] It should be noted that the following detailed descriptions are all illustrative and are intended to provide further explanation of the present invention. Unless otherwise specified, each technical and scientific term used in this embodiment has the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0050] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0051] In the present invention, terms such as "upper", "lower", "left", "right", "front", "back", "vertical", "horizontal", "side", "bottom", etc. indicate directions or positional relationships based on the directions or positional relationships shown in the accompanying drawings. They are relational words determined only for the convenience of describing the structural relationships of the various parts or elements of the present invention, and do not specifically refer to any part or element in the present invention and should not be understood as limitations on the present invention.

[0052] In the present invention, terms such as "fixed connection", "connected", "connection", etc. should be understood in a broad sense, indicating that it can be fixedly connected, integrally connected or detachably connected; it can be directly connected or indirectly connected through an intermediate medium. Relevant scientific research or technical personnel in this field can determine the specific meanings of the above terms in the present invention according to specific circumstances, and they should not be understood as limiting the present invention.

[0053] It should be noted that the present invention provides an end-side device unloading method based on edge orchestration technology, which involves an end-side device and a cloud-based serverless platform.

[0054] Among them, cloud-based serverless platform and cloud can be used interchangeably in this article.

[0055] Embodiment 1:

[0056] like Figure 1 As shown, this embodiment provides a terminal side device uninstallation method based on edge orchestration technology, including the following steps:

[0057] S1: Build a strategy tree.

[0058] S11: Create a dataset including task features and processing strategies.

[0059] Specifically, detection software is deployed in the edge devices in the cloud to monitor various tasks connected to the edge devices in the cloud to record various parameters such as the source of the task, device type, computing resources, storage resources, and network resources of the task, so as to obtain rich task characteristics and processing strategies, and create a data set including task characteristics and processing strategies. Among them, task characteristics include computational complexity, data volume, and delay tolerance.

[0060] S12: Calculate the information gain of each task feature in the data set and take the task feature with the largest information gain as the root node of the strategy tree.

[0061] Specifically, the information gain (IG) of each task feature in the data set is calculated, and the task feature with the largest information gain IG is taken as the optimal feature, which is used as the root node, that is, the split point. Each task feature includes computational complexity, data volume, and delay sensitivity.

[0062] The calculation formula of information gain IG is as follows:

[0063]

[0064] in, is a dataset, is the task characteristic, It is a task feature All possible values ​​of It is a task feature The value is A subset of the data, It is a dataset The entropy of .

[0065] Among them, it should be noted that the information gain IG reflects the ability of task features to distinguish classification tasks. By calculating the information gain when different task features are used as root nodes, it can be determined which task feature can reduce the uncertainty of the data to the greatest extent, that is, which task feature is most important for distinguishing different categories (local processing or offloading).

[0066] S13: Recursively split the root node to build a complete strategy tree.

[0067] Specifically, recursive splitting is performed according to the calculated information gain of the feature task to create child nodes, thereby constructing a complete strategy tree.

[0068] For example, the results of information gain for data volume, computational complexity, and delay tolerance are that the information gain for data volume is the largest, the information gain for computational complexity is the second largest, and the information gain for delay tolerance is the smallest.

[0069] like Figure 2 As shown in the figure, the process of recursively splitting the root node to construct a complete strategy tree is:

[0070] Assuming that the data volume feature uses a threshold of 0.7 as the boundary to divide the data volume into large data volume (i.e., if the data volume is not less than 0.7, then the data volume is a large amount) and small data volume (i.e., if the data volume is less than 0.7, then the data volume is a small amount), then according to the threshold of 0.7, all data volumes are divided into two subsets: subset B with a data volume less than 0.7 and subset A with a data volume not less than 0.7.

[0071] A: For subset B, the computational complexity is used as the next split feature. The distribution of computational complexity in this subset is analyzed. Assuming that the split threshold of computational complexity is 0.5, subset B is divided into subset B1 (computational complexity is not less than 0.5) and subset B2 (computational complexity is less than 0.5) with the computational complexity threshold of 0.5.

[0072] For subset B2, the delay tolerance is used as the splitting feature. Assuming that the delay tolerance is divided into "high tolerance (that is, the delay tolerance is not less than 0.6)" and "low tolerance (that is, the delay tolerance is less than 0.6)", subset B2 is divided into subset B21 (delay tolerance is not less than 0.6) and subset B22 (delay tolerance is less than 0.6)

[0073] B: For subset A, continue to split it into subset A1 (computational complexity is not less than 0.5) and subset A2 (computational complexity is less than 0.5) based on the computational complexity;

[0074] For subset A2, it is split into subsets A21 and A22 based on delay tolerance.

[0075] Recursive splitting is performed in this way, and the splitting stops when the number of subsets is less than the preset number of samples.

[0076] Through continuous recursive splitting, a strategy tree with data volume as the root node and multiple levels and branches is finally constructed. Each leaf node represents a set of tasks with a specific combination of data volume, computational complexity, and delay tolerance. :

[0077]

[0078] in, Indicates tasks, each task It has the following task characteristics: computational complexity, data volume, and delay sensitivity.

[0079] Among them, the strategy tree can classify tasks according to their different attributes. The purpose of building a strategy tree is to determine which tasks are executed on edge devices and which tasks are executed on the cloud serverless platform to achieve more efficient resource allocation. Through the branch judgment of the strategy tree, limited computing, storage and network resources can be reasonably allocated to different types of tasks, avoiding resource waste and improving the overall resource utilization efficiency.

[0080] S2: Perform feature analysis on the task through the constructed strategy tree to determine the task type and task characteristics, where the task characteristics include computational complexity, data volume, and delay tolerance.

[0081] S21: Determine the task type of each task.

[0082] Identify the type of task, where the task types include compute-intensive (such as large-scale numerical calculations, complex algorithms in image processing), data-intensive (involving the processing and transmission of large amounts of data, such as big data analysis), delay-sensitive (having strict requirements on response time, such as interactive operations in real-time games, strategy judgment in autonomous driving), etc.

[0083] S22: Determine the computational complexity, data volume, and delay tolerance of each task.

[0084] In order to facilitate explanation and understanding, each task is recorded as .

[0085] S22-1: Analyze each mission The complexity of the computational operations involved is determined by comprehensively evaluating the number of computational steps, the complexity of the algorithm, and the size of the machine learning model to determine the performance of each task. The computational complexity of .

[0086] S22-1: Determine each task The amount of data that needs to be processed .

[0087] Among them, the amount of data It includes the size of the input data, the amount of data generated by the intermediate calculation process, and the amount of data output in the end.

[0088] S22-3: Determine each task Task delay tolerance .

[0089] Among them, the task delay tolerance is obtained by calculating the difference between the absolute deadline of the task and the start time of the task .

[0090] S3: Evaluate the available resources of the end-side device in real time.

[0091] Specifically, calculate the CPU idle ratio of the end-side device (Current CPU unused), memory idle ratio (the total amount of memory not currently occupied), network latency (Current available network bandwidth / network bandwidth limit).

[0092] S4: Determine whether the resource quantity of the terminal device meets the preset conditions. If so, execute step S5; if not, generate a first execution strategy.

[0093] Specifically, judge each task The computational complexity of , Data volume , Delay Tolerance CPU idle ratio of the computing device , Memory Free Ratio , network latency Whether the preset conditions are met, if so, step S5 is executed; if not, a first execution strategy is generated, which is to offload the task to the cloud serverless platform.

[0094] The preset conditions are:

[0095]

[0096] in, is the adjustment coefficient.

[0097] S5: Calculate the offloading strategy indicator according to the determined computational complexity, data volume, delay tolerance and task type of each task.

[0098] S51: According to each task determined The computational complexity of , Data volume and delay tolerance To formulate uninstall strategy indicators (uninstall strategy indicators ), where the uninstallation strategy indicator The calculation formula is:

[0099]

[0100] in, , and is the weight coefficient, and the weight coefficient satisfies the following formula, and the value of m is 3.

[0101]

[0102] More specifically, The value of indicates the resource and data volume required by the task itself. The larger the value, the more difficult it is to process the task on the client side. The larger the value, the higher the delay in end-side processing.

[0103] S52: Adjust according to the determined task type , and Weight coefficient to calculate the uninstall strategy indicator.

[0104] For example, if the task is computationally intensive and the resources of the end device are limited, but the tolerance for latency is high, the weight coefficient and As small as possible, the weight coefficient If the task is sensitive to latency and the resources of the client device can meet the demand, increase the weight coefficient. and As much as possible, Adjust as low as possible, but no matter how you adjust , and The sum must be 1.

[0105] S6: The cost of executing computing tasks on the end device and the cloud serverless platform.

[0106] S61: Determine to execute each task on the terminal device The computational cost of , Network transmission cost and energy costs .

[0107] S62: Determine the execution of each task on the cloud serverless platform The computational cost , network transmission cost , energy cost .

[0108] S63: Calculate the cost of executing the task on the terminal device and on the cloud serverless platform according to the following formula.

[0109]

[0110] in, is a strategy variable, when When , the calculated Z value is the execution cost of executing the task on the client device. When , the Z value is the execution cost when the task is offloaded to the cloud serverless platform for execution.

[0111] S7: Generate an execution strategy based on the calculated execution cost of the task on the end-side device and the cloud serverless platform and the calculated offloading strategy indicator, wherein the execution strategy includes a first execution strategy and a second execution strategy.

[0112] Specifically, it is determined whether the execution cost of executing the task on the end-side device is less than the execution cost of executing the task on the cloud serverless platform, and the following determination result is generated:

[0113] A: If yes, determine whether the calculated uninstallation strategy index is less than the preset index. If yes, generate the second execution strategy; if no, generate the first execution strategy;

[0114] B: If not, directly generate the first execution strategy.

[0115] Among them, the second execution strategy is to execute the task on the cloud serverless platform.

[0116] S8: Execute the task based on the generated first execution strategy or the second generation strategy.

[0117] For example, if the second execution strategy is generated, the computing resources of the end-side device are used to complete the computing task; if the first execution strategy is generated, the task is offloaded to the cloud-based serverless platform, and the task-related data and execution requirements are sent to the cloud-based serverless platform through the network, and the computing results returned by the cloud-based serverless platform are received.

[0118] S9: During the task execution process, the system status and task execution status are continuously monitored to dynamically adjust the weight coefficient and the preset indicator threshold.

[0119] Specifically, if during the task execution process, it is found that the actual effect does not meet the expectations, the weight coefficient and threshold can be dynamically adjusted according to the system performance indicators over a period of time to optimize the subsequent unloading strategy.

[0120] Note: For example, in a network transmission task, although the data transmission speed is high, it exceeds the hardware computing power of the CPU, causing computing jams. In this case, it is necessary to dynamically adjust the weight coefficients of the remaining memory and delay tolerance.

[0121] A) Dynamically adjusting the weight coefficient includes the following steps:

[0122] For example, if the client device often experiences high latency when executing tasks, it may be necessary to increase the task latency tolerance, i.e. the weight of the task tolerance. , assuming that the average delay of the client-side task in the preset time period is , the target average delay is , then the weight adjustment formula is:

[0123] ;

[0124] in is the adjustment factor.

[0125] It should be noted that in order to keep the total weight equal to 1, other weight coefficients also need to be adjusted accordingly. The adjustment calculation method is as follows:

[0126] ;

[0127] in, ; is the number of weight coefficients. In this embodiment, , and There are 3 weight coefficients in total. After that, the corresponding , Adjustments also need to be made.

[0128] Based on the adjusted weight coefficient, the index value of the uninstallation policy is recalculated according to the calculation formula of the index value of the uninstallation policy (mentioned above and not repeated here).

[0129] B) Dynamically adjusting the preset indicator threshold includes the following steps:

[0130] Among them, the preset indicator threshold is an empirical value. For example, taking the data volume as the main task feature, the indicator value E of the unloading strategy is obtained through calculation, and E is compared with the preset indicator threshold. According to the comparison result, the execution strategy is generated.

[0131] Assume that the initial preset indicator threshold is , the system's task processing success rate within the preset time period is The target success rate is , is the threshold adjustment factor.

[0132] The adjusted preset indicator threshold is obtained by the following preset indicator threshold adjustment formula :

[0133]

[0134] Specifically, if the task processing success rate Below target success rate , indicating that the current preset indicator threshold may be too high and needs to be lowered so that more tasks can be processed on the end device or offloaded to the cloud serverless platform, thereby improving the success rate.

[0135] S6: Performance monitoring and feedback.

[0136] During the task execution process, the execution performance of the task is continuously monitored, such as execution time, resource consumption, result accuracy, etc., and based on the monitoring results, the offloading strategy algorithm is fed back and optimized to improve the accuracy and efficiency of the strategy.

[0137] Embodiment 2:

[0138] This embodiment provides an end-side device unloading system based on edge orchestration technology, including: an end-side device and a cloud-based serverless platform, wherein the end-side device and the cloud-based serverless platform are used to execute tasks.

[0139] Construction module, used to build the strategy tree;

[0140] The task feature analysis module is used to perform feature analysis on the task through the constructed strategy tree to determine the task type, task features and resources required for the task;

[0141] End-side resource evaluation module: used to evaluate the computing resources currently available on the end-side device in real time;

[0142] An offloading strategy formulation module generates an execution strategy based on the determined task type, task characteristics, and available resource amounts of the end-side device evaluated in real time, wherein the execution strategy includes a first execution strategy of offloading the task to a cloud serverless platform for execution and a second execution strategy of executing the task on the end-side device;

[0143] Task execution module: used to execute tasks based on the generated execution strategy.

[0144] Performance monitoring and feedback module: used to monitor task execution performance and provide feedback.

[0145] Embodiment 3:

[0146] This embodiment provides a computer-readable storage medium, which includes a stored program. When the program is running, the device where the computer-readable storage medium is located is controlled to execute the terminal-side device uninstallation method based on edge orchestration technology described in Embodiment 1.

[0147] In this specification, the same or similar parts between the various embodiments can be referred to each other. In particular, for the terminal embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the description in the method embodiment.

[0148] In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of systems or units, which can be electrical, mechanical or other forms.

[0149] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0150] In addition, it should be noted that the flowchart in the accompanying drawings shows the method of the embodiment of the present disclosure. In the corresponding description in the flowchart or block diagram in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in an order different from that disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be performed substantially in parallel, and they can sometimes be performed in the opposite order, which can depend on the functions involved. Each block in the block diagram and / or flow chart, and the combination of blocks in the block diagram and / or flow chart, can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0151] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A terminal device uninstallation method based on edge orchestration technology, characterized in that: The following steps are involved: Step 1: Build a strategy tree; Step 2: Perform feature analysis on the task through the constructed strategy tree to determine the task type and task characteristics; Step 3: Evaluate the available resources of the end-side device in real time; Step 4: Based on the determined task type, task characteristics, and available resource amounts of the end-side device evaluated in real time, an execution strategy is generated, wherein the execution strategy includes a first execution strategy of offloading the task to a cloud serverless platform for execution and a second execution strategy of executing the task on the end-side device; Step 5: Execute the task based on the generated execution strategy; The step one comprises: Create a dataset that includes task characteristics and processing strategies; Calculate the information gain of each task feature in the data set, and use the task feature with the largest information gain as the root node of the strategy tree; Recursively split the root node to build a complete strategy tree; The recursive splitting of the root node to construct a complete strategy tree includes: recursively splitting the root node according to the calculated information gain of the feature task to create child nodes, thereby constructing a complete strategy tree; Among them, task characteristics include computational complexity, data volume, and delay tolerance; The fourth step comprises: Determine whether the resource quantity of the terminal device meets the preset conditions. If not, generate a first execution strategy. If satisfied, perform the following steps: Calculate the offloading strategy indicator based on the computational complexity, data volume, delay tolerance and task type of each task; The cost of computing tasks on the client device and the serverless cloud platform. Generate an execution strategy based on the calculated execution cost of the task on the end-side device and the cloud serverless platform and the calculated offloading strategy indicator, wherein the execution strategy includes a first execution strategy and a second execution strategy; The generating of the execution strategy based on the calculated execution cost of the task on the terminal device and the cloud serverless platform and the calculated offloading strategy indicator includes: Determine whether the execution cost of executing the task on the client device is less than the execution cost of executing the task on the cloud serverless platform, and generate the following judgment results: A: If yes, determine whether the calculated uninstallation strategy index is less than the preset index. If yes, generate the second execution strategy; if no, generate the first execution strategy; B: If not, directly generate the first execution strategy.

2. The terminal device uninstallation method based on edge orchestration technology according to claim 1 is characterized in that: The available resources of the end-side device include the computing resources of the end-side device, the CPU idle ratio, the memory idle ratio and the network latency.

3. The terminal device uninstallation method based on edge orchestration technology according to claim 2 is characterized in that: The fourth step comprises: Determine whether the computational complexity, data volume, delay tolerance, CPU idle ratio, memory idle ratio, and network delay of each task meet the following preset conditions: is the computational complexity; is the amount of data; is delay tolerance; is the CPU idle ratio of the end device, is the memory idle ratio of the end device, is the network delay of the end-side device, is the adjustment coefficient, is a positive integer; If the conditions are met, the offloading strategy indicator is calculated according to the computational complexity, data volume, delay tolerance and task type of each task; if the conditions are not met, the first execution strategy is generated.

4. The terminal device uninstallation method based on edge orchestration technology according to claim 1 is characterized in that: The step five comprises: During the task execution process, the system status and task execution status are continuously monitored to dynamically adjust the weight coefficient and preset indicator threshold.

5. The terminal device uninstallation method based on edge orchestration technology according to claim 3 is characterized in that: The calculation of the offloading strategy index according to the determined computational complexity, data volume, delay tolerance and task type of each task includes: The offloading strategy indicator is formulated according to the computational complexity, data volume and delay tolerance of each task. The offloading strategy indicator calculation formula is: in, To uninstall the strategy indicator; is the computational complexity; is the amount of data; is the delay tolerance, , and is the weight coefficient, and the weight coefficient satisfies the following formula: in, The value is 3; According to the determined task type, adjust , and Weight coefficient to calculate the uninstall strategy indicator.

6. The terminal device uninstallation method based on edge orchestration technology according to claim 1 is characterized in that: The method also includes monitoring task execution performance and providing feedback.

7. A terminal device unloading system based on edge orchestration technology, characterized in that: include: Construction module, used to build the strategy tree; The task feature analysis module is used to perform feature analysis on the task through the constructed strategy tree to determine the task type, task features and resources required for the task; End-side resource evaluation module: used to evaluate the computing resources currently available on the end-side device in real time; Offloading strategy formulation module: based on the determined task type, task characteristics and available resource amount of the terminal device evaluated in real time, generating an execution strategy, wherein the execution strategy includes a first execution strategy of offloading the task to the cloud serverless platform for execution and a second execution strategy of executing the task on the terminal device; Task execution module: used to execute tasks based on the generated execution strategy; The strategy tree construction includes: Create a dataset that includes task characteristics and processing strategies; Calculate the information gain of each task feature in the data set, and use the task feature with the largest information gain as the root node of the strategy tree; Recursively split the root node to build a complete strategy tree; The recursive splitting of the root node to construct a complete strategy tree includes: recursively splitting the root node according to the calculated information gain of the feature task to create child nodes, thereby constructing a complete strategy tree; Among them, task characteristics include computational complexity, data volume, and delay tolerance; The generating of the execution strategy based on the determined task type, task characteristics and the available resource amount of the terminal device evaluated in real time includes: Determine whether the resource quantity of the terminal device meets the preset conditions. If not, generate a first execution strategy. If satisfied, perform the following steps: Calculate the offloading strategy indicator based on the computational complexity, data volume, delay tolerance and task type of each task; The cost of computing tasks on the client device and the serverless cloud platform. Generate an execution strategy based on the calculated execution cost of the task on the end-side device and the cloud serverless platform and the calculated offloading strategy indicator, wherein the execution strategy includes a first execution strategy and a second execution strategy; The generating of the execution strategy based on the calculated execution cost of the task on the terminal device and the cloud serverless platform and the calculated offloading strategy indicator includes: Determine whether the execution cost of executing the task on the client device is less than the execution cost of executing the task on the cloud serverless platform, and generate the following judgment results: A: If yes, determine whether the calculated uninstallation strategy index is less than the preset index. If yes, generate the second execution strategy; if no, generate the first execution strategy; B: If not, directly generate the first execution strategy.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the terminal-side device uninstallation method based on edge orchestration technology according to any one of claims 1 to 6.

Images