Edge computing control method, device, apparatus, medium and product
By segmenting and offloading computing tasks to multiple edge nodes based on the historical and predicted trajectories of mobile devices in edge computing, the problem of service interruption caused by connection failure of mobile devices is solved, and the continuity and stability of computing tasks are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA MOBILE M2M
- Filing Date
- 2026-03-12
- Publication Date
- 2026-06-16
AI Technical Summary
In edge computing scenarios, the decision to offload computing tasks from mobile devices fails due to device movement, leading to service lag, delays, or even failures, affecting the continuity of device services.
By acquiring the historical trajectory of mobile devices, their future trajectory is predicted. Based on the predicted trajectory and the computing resources and channel quality of edge nodes, computing tasks are segmented and offloaded to multiple edge nodes to ensure task continuity.
It achieves continuity of computing tasks while mobile devices are in motion, avoids service interruptions caused by the disconnection of a single node, and improves the service continuity of mobile terminal devices.
Smart Images
Figure CN122227320A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of communication technology, and in particular to a control method, apparatus, device, medium and product for edge computing. Background Technology
[0002] In related technologies, in edge computing scenarios, the computing tasks that the terminal device needs to perform edge computing on are typically offloaded to the edge node with the best current connection quality, based on the connection quality between the current terminal device and the edge node. However, when the terminal device is a mobile device, the offloading decision based on the current connection quality becomes invalid when the device moves. If the device moves out of the signal range of the original edge node halfway through task execution, it must find a new edge node to offload the task. This can cause device service lag, delays, or even failures, severely impacting the continuity of device services. Summary of the Invention
[0003] This application provides a control method, apparatus, device, medium, and product for edge computing, used to improve the continuity of mobile device services.
[0004] In a first aspect, embodiments of this application provide a control method for edge computing, the method comprising: When a computing task is received from a mobile device, the historical trajectory of the mobile device within the historical time window corresponding to the current time point is obtained; wherein, the historical time window is the time period up to the current time point; Based on the historical trajectory, predict the trajectory of the mobile device within the prediction window corresponding to the current time point; wherein, the prediction window is a time period starting from the current time. Based on the predicted trajectory, multiple edge nodes are determined. Based on the predicted trajectory and the available computing resources and channel quality of each edge node at the current time point, the computing task is segmented to determine the sub-task corresponding to each edge node. Based on the subtask corresponding to each edge node, a task unloading instruction is sent to the mobile device, so that the mobile device can unload the computing task to the multiple edge nodes for computing based on the subtask corresponding to each edge node.
[0005] Secondly, embodiments of this application provide a control device for edge computing, the device comprising: The acquisition module is used to acquire the historical trajectory of the mobile device within the historical time window corresponding to the current time point when receiving a computing task from the mobile device; wherein, the historical time window is the time period up to the current time point; The prediction module is used to predict the trajectory of the mobile device within a prediction window corresponding to the current time point based on the historical trajectory; wherein, the prediction window is a time period starting from the current time. The segmentation module is used to determine multiple edge nodes based on the predicted trajectory, and to segment the computing task based on the predicted trajectory and the available computing resources and channel quality of each edge node at the current time point, thereby determining the sub-task corresponding to each edge node. The instruction module is used to send a task unloading instruction to the mobile device based on the subtask corresponding to each edge node, so that the mobile device can unload the computing task to the multiple edge nodes for computing based on the subtask corresponding to each edge node.
[0006] Thirdly, embodiments of this application provide an electronic device, including... A memory on which computer programs are stored; A processor for executing the computer program to implement the steps of the control method for edge computing described above.
[0007] Fourthly, embodiments of this application provide a non-transient readable storage medium storing a computer program thereon, the computer program being executable by a processor to implement the steps of the control method for edge computing described above.
[0008] Fifthly, embodiments of this application provide a computer program product, including a computer program that can be executed by a processor to implement the steps of the control method for edge computing described above.
[0009] The edge computing control method, apparatus, device, medium, and product provided in this application, upon receiving a computing task from a mobile device at the current time point, acquires the historical trajectory of the mobile device within a historical time window corresponding to the current time point. Based on this historical trajectory, it predicts the predicted trajectory of the mobile device within a prediction time window corresponding to the current time point. Then, based on the predicted trajectory, it determines multiple edge nodes. Based on the predicted trajectory and the available computing resources and channel quality of each edge node, it determines the task ratio corresponding to each edge node. Based on the task ratio corresponding to each edge node, it sends a task offloading instruction to the mobile device, enabling the mobile device to offload the computing task to multiple edge nodes for computing based on the task ratio corresponding to each edge node. This can split and offload the computing task of the terminal device to multiple edge nodes that can be reached by communication when the mobile device moves in the future, based on the predicted trajectory of the terminal device, thus solving the service interruption problem caused by terminal mobility and improving the continuity of mobile terminal services. Attached Figure Description
[0010] Figure 1This is a schematic diagram of the implementation environment to which an edge computing control method according to an embodiment of this application is applicable; Figure 2 This is a flowchart illustrating a control method for edge computing according to an embodiment of this application; Figure 3 This is a schematic diagram of a task segmentation process according to an embodiment of this application; Figure 4 This is a schematic diagram of a resource reservation process according to an embodiment of this application; Figure 5 This is a schematic diagram of a control device for edge computing according to an embodiment of this application; Figure 6 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation
[0011] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0012] As will be known to those skilled in the art, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.
[0013] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms are interchangeable where appropriate; this is merely a way of distinguishing objects with the same attributes in the description of embodiments of this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, so that a process, method, system, product, or apparatus that comprises a series of elements is not necessarily limited to those elements but may include other elements not explicitly listed or inherent to those processes, methods, products, or apparatuses.
[0014] As mentioned earlier, in edge computing scenarios, based on the connection quality between the current terminal device and the edge node, the computing tasks that the terminal device needs to perform edge computing are typically offloaded to the edge node with the best current connection quality. If the terminal device is a mobile device, the offloading decision based on the current connection quality will become invalid when the device moves. If the device moves out of the signal range of the original edge node halfway through task execution, the device must find a new edge node to offload the task. In this case, it will cause device service lag, delay, or even failure, seriously affecting the continuity of device service.
[0015] To address this, this application provides a control method for edge computing. In this method, upon receiving a computing task from a mobile device at the current time point, the historical trajectory of the mobile device within a historical time window corresponding to the current time point is obtained. Based on this historical trajectory, the predicted trajectory of the mobile device within a prediction time window corresponding to the current time point is predicted. Then, based on the predicted trajectory, multiple edge nodes are determined. Based on the predicted trajectory and the available computing resources and channel quality of each edge node, the task ratio corresponding to each edge node is determined. Based on the task ratio corresponding to each edge node, a task offloading instruction is sent to the mobile device, causing the mobile device to offload the computing task to multiple edge nodes for computation based on the task ratio corresponding to each edge node. This method can split and offload the computing task of the terminal device to multiple edge nodes that are reachable during future movement of the mobile device, based on the predicted trajectory of the terminal device. This avoids terminal service interruption due to the disconnection of a single node, solves the service interruption problem caused by terminal mobility, and improves the continuity of terminal services.
[0016] To facilitate understanding, let's first combine Figure 1 The implementation environment applicable to the edge computing control method provided in the embodiments of this application will be described. For example... Figure 1 As shown, the implementation environment applicable to the edge computing method provided in this application embodiment can be an environment that requires edge computing services, such as the Internet of Vehicles and the Industrial Internet of Things. This environment may include: mobile devices, edge nodes, and edge computing control devices.
[0017] Mobile devices can be mobile terminal devices that require edge computing, and their capabilities can vary depending on the implementation environment. For example, in a connected vehicle environment, mobile devices can be highly mobile in-vehicle devices.
[0018] Edge nodes can be IoT devices with computing and storage capabilities deployed at the network edge, and their configurations can vary depending on the implementation environment. For example, in a vehicle-to-everything (V2X) environment, edge nodes can be roadside units deployed on the roadside.
[0019] The control unit for edge computing acts as the control center, determining how to break down the computing tasks of mobile devices into multiple tasks and schedule and deploy them to appropriate edge nodes for execution. The control unit can be, for example, a software platform or physical device that centrally manages all edge nodes, such as an edge computing application orchestrator.
[0020] When a mobile terminal needs to perform edge computing, it can send the required computing tasks to the edge computing control device. The control device can predict the mobile terminal's future trajectory based on its historical trajectory, determine multiple edge nodes based on this predicted trajectory, and identify sub-tasks for each edge node based on the computing tasks, available computing resources, and channel quality. Based on these sub-tasks, the control device sends a task offloading command to the mobile device, enabling it to offload the computing tasks to multiple edge nodes. This allows multiple edge nodes to provide highly continuous mobile edge computing services to highly mobile terminal devices, preventing service interruptions due to high mobility.
[0021] The control method for edge computing provided in the embodiments of this application will be described below with reference to the accompanying drawings.
[0022] Please refer to Figure 2 This is a flowchart illustrating a control method for edge computing according to an embodiment of this application. This method can be used in the aforementioned edge computing control device and executed by that device. Figure 2 As shown in the figure, an edge computing control method according to an embodiment of this application includes the following steps: S202, when receiving a computing task from a mobile device, obtain the historical trajectory of the mobile device within the historical time window corresponding to the current time point.
[0023] In this embodiment, the computing task of the mobile device is the edge computing task to be performed by the mobile device. The current time point can be the time point at which the computing task of the mobile device is received.
[0024] The historical time window corresponding to the current time point can be a time period up to the current time point. It is a fixed-length sliding time window with the current time point as the anchor point, and it can slide as the current time point slides. The historical trajectory of movement within the historical time window can include the position information of the mobile device at each time point of the historical time window. This position information reflects the position and motion state of the mobile device in the past period up to the current time point, such as speed, direction, acceleration, etc.
[0025] As an example, a history window can be a time period. Historical trajectories can include the actual location of the mobile device at each point in time within that period; they can be a sequence of locations, i.e. .in, At the current time point, Indicates the time point of the mobile device The true location .
[0026] S204, based on historical trajectories, predicts the trajectory of mobile devices within the prediction window corresponding to the current time point.
[0027] In this embodiment, the prediction window corresponding to the current time point can be a time period starting from the current time point. It is a fixed-length sliding window anchored to the current time point and can slide along with the current time point. The predicted trajectory of the mobile device within the prediction window includes the predicted position information of the mobile device at one or more prediction time points, which can reflect the position and movement state of the mobile device in the future.
[0028] Using the example above, the prediction window can be a time period. The predicted trajectory can include the predicted time points of the mobile device during that time period. Predicted location .
[0029] In this embodiment, the predicted trajectory of a mobile device within a prediction window can be predicted using an appropriate method. In one specific implementation, road topology information of the geographical area where the mobile device is located at the current time point can be obtained; historical trajectories and road topology information are input into a trajectory prediction model, and the predicted trajectory is predicted through the trajectory prediction model.
[0030] The trajectory prediction model is obtained by training a spatiotemporal dual-stream neural network based on a trajectory prediction dataset.
[0031] The trajectory prediction dataset is a dataset for mobile device trajectory prediction tasks. It can include multiple sample data for trajectory prediction. Each sample data can include the historical trajectory of the mobile device within the historical time window corresponding to the historical time point, road topology information, and the actual trajectory of the mobile device within the prediction time window corresponding to the historical time point. The trajectory prediction model can be trained based on the above trajectory prediction dataset using a supervised training method.
[0032] Road topology information is the contextual information of the geographic environment in which the mobile device is located. For example, road topology weights can be obtained based on the regional map information corresponding to the mobile device.
[0033] Using the example above, a spatiotemporal feature matrix of the mobile device at the current time point can be generated based on the historical trajectory of the mobile device within the historical time window corresponding to the current time point. The spatiotemporal feature matrix of the mobile device at the current time point Road topology weight matrix of the geographic environment of the mobile device Input the trajectory prediction model to obtain the movement trajectory of the predicted time window corresponding to the current time window.
[0034] By training a trajectory prediction model based on a spatiotemporal dual-stream neural network, and inputting the historical trajectory of the mobile device and road topology information into the trajectory prediction model trained on the spatiotemporal dual-stream neural network, the historical trajectory of the mobile device can be used as a time stream and the road topology information as a spatial stream, so as to more accurately predict the future trajectory of the mobile device.
[0035] S206. Based on the predicted trajectory, multiple edge nodes are determined. Based on the predicted trajectory and the available computing resources and channel quality of each edge node at the current time point, the computing task is divided to determine the sub-task corresponding to each edge node.
[0036] In this embodiment, the multiple edge nodes can be multiple edge nodes that the mobile device can reach while moving along the predicted trajectory, such as edge nodes whose signals can cover the predicted trajectory. The subtask corresponding to each edge node is the portion of the computation task corresponding to that edge node after the computation task is divided.
[0037] In implementation, the task ratio for each edge node can be determined based on the predicted trajectory, available computing resources, and channel quality. Based on this task ratio, the computing tasks are then segmented to determine the sub-tasks for each edge node. In one specific implementation, this can be achieved through methods such as... Figure 3 The following steps, as shown, determine the subtask corresponding to each edge node: S302, for each edge node, based on the predicted trajectory, determine the predicted distance and the rate of change of the predicted distance between the mobile device and the edge node at the end of the prediction window, and calculate the spatiotemporal coordination factor of the edge node based on the predicted distance, the rate of change of the predicted distance, the available computing resources of the edge node, and the channel quality.
[0038] In this embodiment, the position of the edge node is fixed and remains unchanged. In one specific implementation, the predicted position of the mobile device at the end of the prediction window can be determined based on the predicted trajectory of the mobile device within the prediction window. Based on the predicted position of the mobile device at the end of the prediction window and the position of the edge node, the predicted distance between the mobile device at the end of the prediction window and the edge node is determined. Based on the predicted distance between the mobile terminal at the end of the prediction window and the edge node, and the actual distance between the mobile terminal and the edge node at the current time point, the rate of change of the predicted distance between the mobile terminal and the edge node is calculated. This rate of change of predicted distance is the rate of change of future distance at the current time point.
[0039] Using the example above, the predicted location of the mobile device at the end of the prediction window can be: Assume there are K edge nodes in total. The position can be Prediction window endpoint mobile device and edge node The predicted distance can be Mobile devices and edge nodes at the current time. The actual distance can be The mobile device and edge node at the current time point The predicted distance change rate can be .
[0040] In this embodiment, the spatiotemporal synergy factor of an edge node characterizes the spatiotemporal synergy utility of that edge node. In one specific implementation, the spatiotemporal synergy factor of each edge node can be calculated using the following formula:
[0041] in, The current time point edge nodes Spatiotemporal coordination factors The current time point edge nodes Available computing resources The current time point Mobile devices and edge nodes The predicted rate of change of distance, To predict the absolute value of the rate of change of distance, The spatial attenuation coefficient, For predicting the end of the time window Mobile devices and edge nodes The predicted distance, It is an exponentially decaying function.
[0042] In this embodiment, the channel quality of the edge node at the current time point is the channel quality between the edge node and the mobile device at the current time point, which is determined based on the available bandwidth of the edge node at the current time point and the distance between the edge node and the mobile device at the current time point. In one specific implementation, the channel quality of the edge node at the current time point can be calculated using the following formula:
[0043] in, The current time point edge nodes Available bandwidth, For terminal devices and the first Real-time Euclidean distance between edge nodes This is the wireless channel attenuation factor.
[0044] S304, with the goal of optimizing the weighted sum of the spatiotemporal coordination factors of multiple edge nodes, determine the task proportion corresponding to each edge node in the computing task.
[0045] In practice, a task segmentation ratio decision model can be established to determine the task ratio corresponding to each edge node.
[0046] In one specific implementation, a task segmentation ratio decision model can be established to determine the task ratio corresponding to each edge node by maximizing the function value of the first objective function described below. The task ratio corresponding to each edge node can be determined through this task segmentation ratio decision model.
[0047]
[0048] in, For edge nodes Spatiotemporal coordination factors For edge nodes The corresponding task ratio, This serves as a decision-making smoothing constraint factor (to suppress frequent fluctuations in task allocation). The square of the time rate of change of the task segmentation ratio. .
[0049] In implementation, the set of parameters that maximizes the function value of the first objective function mentioned above. The value is the value of each edge node. The corresponding task ratio.
[0050] The first objective function mentioned above introduces a decision smoothing constraint term. By using decision smoothing constraints It can determine the task allocation ratio with the goal of maximizing the total weighted collaborative utility of all edge nodes, while smoothing the decision constraint term. Suppress frequent fluctuations in task allocation ratios and ensure that task allocation matches spatiotemporal dynamics in real time.
[0051] S306, Based on the task ratio corresponding to each edge node, the computing task is divided to obtain the sub-task corresponding to each edge node.
[0052] Following the example above, edge nodes The corresponding task ratio can be ,in, To calculate the total workload of the task.
[0053] S208, based on the subtask corresponding to each edge node, a task unloading instruction is sent to the mobile device, so that the mobile device can unload the computing task to multiple edge nodes for computing based on the subtask corresponding to each edge node.
[0054] The task unloading command carries subtask information corresponding to each edge node. After receiving the task unloading command, the mobile device can send the task data of the subtask corresponding to each edge node to the corresponding edge node based on the subtask information carried in the task unloading command, so that the edge node can perform edge computing on the corresponding sub-computing task.
[0055] In this embodiment, upon receiving a computing task from a mobile device at the current time, the historical trajectory of the mobile device within the historical time window corresponding to the current time is obtained. Based on this historical trajectory, the predicted trajectory of the mobile device within the prediction time window corresponding to the current time is predicted. Then, based on the predicted trajectory, multiple edge nodes are determined. Based on the predicted trajectory and the available computing resources and available bandwidth of each edge node, the task ratio corresponding to each edge node is determined. Based on the task ratio corresponding to each edge node, a task offloading instruction is sent to the mobile device, enabling the mobile device to offload the computing task to multiple edge nodes for computing based on the task ratio corresponding to each edge node. This allows the computing tasks of the terminal device to be distributed and offloaded to multiple edge nodes that can be reached by communication when the mobile device moves in the future, based on the predicted trajectory of the terminal device. This avoids terminal service interruption due to the disconnection of a single node, solves the service interruption problem caused by terminal mobility, and improves the continuity of terminal services.
[0056] To ensure that edge nodes have sufficient computing and spectrum resources to perform edge computing on their corresponding subtasks, improve the stability of edge computing, and guarantee the continuity of terminal services, in one implementation, before sending the task offload instruction to the mobile device, the following can be performed before step S208 above: Figure 4 The following steps are shown to instruct each edge node to reserve computing and spectrum resources for its corresponding subtask.
[0057] S402, based on the sub-tasks corresponding to each edge node and the spectral efficiency and available computing resources of each edge node at the current time point, with the goal of optimizing the spatiotemporal utility of multiple edge nodes, determines the reserved computing resources and reserved spectrum resources corresponding to each edge node.
[0058] In implementation, a joint optimization model of computing resources and spectrum resources can be established. Through this model, the reserved computing resources and reserved spectrum resources corresponding to each edge node can be determined.
[0059] In one specific implementation, a joint optimization model can be established to determine the reserved computing resources and reserved spectrum resources corresponding to each edge node, with the objective of maximizing the function value of the second objective function described below. The reserved computing resources and reserved spectrum resources corresponding to each edge node are determined through this model.
[0060]
[0061] in, For edge nodes The corresponding task ratio, To calculate the total workload of the task, For edge nodes The workload of the corresponding subtasks The current time point edge nodes Spectral efficiency factor For edge nodes The corresponding reserved computing resources, For edge nodes The corresponding reserved spectrum resources, The current time point edge nodes Available computing resources This is the variance penalty coefficient.
[0062] In implementation, the set of methods that maximizes the function value of the aforementioned second objective function. The value is the value of each edge node. The corresponding reserved computing resources and reserved spectrum resources.
[0063] In the second objective function mentioned above, The total spatiotemporal utility of multiple edge nodes, This is a variance penalty term. By introducing a variance penalty term... Determine the reserved computing resources and reserved spectrum resources corresponding to each edge node, which can be achieved through variance penalty coefficient. Adjusting the balance of computing resource allocation among edge nodes avoids excessive resource concentration. In practical implementation, to ensure the feasibility of resource allocation, constraints can be applied. and Next, determine a set of parameters that maximizes the second objective function. .in, The total computing resources required for the computation task. The total spectrum resources required for the computation task.
[0064] In one specific implementation, the set of steps that maximizes the second objective function can be determined by performing steps one through five below. .
[0065] Step 1: Initialize parameters.
[0066] Specifically, set the initial resource allocation vector. The constraints are satisfied, where... For the first The initial computing resource allocation for each edge node. For the first Initial spectrum resource allocation for each edge node; number of initial iterations. Maximum number of iterations Convergence threshold .
[0067] Step 2: Perform dynamic gradient calculation and direction search.
[0068] Specifically, calculate the gradient of the objective function: .
[0069] In the formula, , , For the second objective function gradient vector, To calculate the average amount of resource allocation, It is the sum of the reciprocals of the squares of the available computing resources for all edge nodes.
[0070] Acceleration is achieved through gradient approximation. Specifically, this can be achieved using the spatiotemporal feature matrix. Historical data for predicting gradient direction: ; In the formula, This is an approximate gradient vector. These are the resource allocation vectors for the current time step and the previous time step, respectively.
[0071] Step 3: Perform constrained projection optimization.
[0072] Specifically, solving for the linear search direction : ; In the formula, For the constraint domain, This is the vector transpose operator; Where the constraint domain ; The high-dimensional constrained projection is simplified using the Lagrange dual decomposition method as follows: ; In the formula, This refers to the amount of computing resources allocated after projection. Due to the overall system computing resource constraints, This refers to the amount of spectrum resources allocated after projection. Constraints on total system spectrum resources; Step 4: Perform adaptive step size adjustment.
[0073] Specifically, calculate the dynamic step size: ; In the formula, It is the Lipschitz constant; Step 5: Perform iterative updates and determine the termination.
[0074] Specifically, update the resource allocation scheme: ; When satisfied or When the optimal solution is output. Otherwise Return to step two.
[0075] In the above scheme, a joint optimization model for computing resources and spectrum resources is established based on the second objective function, and dynamic factors are used to optimize the model. The time-varying characteristics of spectral efficiency with device movement are quantified, and an improved optimization algorithm is used to solve for the optimal resource allocation scheme, achieving dual optimization of computational task execution latency and resource utilization balance.
[0076] S404: For each edge node, generate a resource reservation instruction based on the reserved computing resources and reserved spectrum resources corresponding to the edge node, and send the resource reservation instruction to the edge node so that the edge node can reserve resources for its corresponding subtask.
[0077] The resource reservation instruction sent to the edge node carries information about the reserved computing resources and spectrum resources corresponding to that edge node. After receiving the resource reservation instruction, the edge node can reserve the corresponding computing resources and spectrum resources for its subtasks based on the reserved computing resources and spectrum resources information carried in the resource reservation instruction.
[0078] The above describes the edge computing control method according to an embodiment of this application. In implementation, the mobile device may not move according to the predicted trajectory. To cope with sudden changes in the movement trajectory and improve robustness in complex movement scenarios, the actual movement trajectory of the mobile device can be detected in real time. When the deviation between the actual movement trajectory and the predicted value exceeds a threshold... In such cases, an emergency recalculation is triggered to re-unload the task, for example, to re-execute it. Figure 2 The steps are shown.
[0079] Based on the same inventive concept as the foregoing embodiments of this application, this application also provides a control device for edge computing. Figure 5 This is a schematic diagram of a control device for edge computing according to an embodiment of this application. Figure 5 As shown, an edge computing control device 500 according to an embodiment of this application may include: The acquisition module 510 is used to acquire the historical trajectory of the mobile device within the historical time window corresponding to the current time point when receiving the computing task of the mobile device; wherein, the historical time window is the time period up to the current time point.
[0080] The prediction module 520 is used to predict the predicted trajectory of the mobile device within the prediction window corresponding to the current time point based on the historical trajectory; wherein the prediction window is a time period starting from the current time.
[0081] The segmentation module 530 is used to determine multiple edge nodes based on the predicted trajectory, and to segment the computing task based on the predicted trajectory and the available computing resources and channel quality of each edge node at the current time point, thereby determining the sub-task corresponding to each edge node.
[0082] The instruction module 540 is used to send a task unloading instruction to the mobile device based on the subtask corresponding to each edge node, so that the mobile device can unload the computing task to the multiple edge nodes for computing based on the subtask corresponding to each edge node.
[0083] In one embodiment, the segmentation module 530 is specifically used for: For each edge node, based on the predicted trajectory, the predicted distance and the rate of change of the predicted distance between the mobile device and the edge node at the end of the prediction window are determined. Based on the predicted distance, the rate of change of the predicted distance, and the available computing resources and channel quality of the edge node, the spatiotemporal coordination factor of the edge node is calculated. With the goal of optimizing the weighted sum of the spatiotemporal coordination factors of the multiple edge nodes, the task proportion corresponding to each edge node in the computing task is determined; Based on the task ratio corresponding to each edge node, the computing task is divided to obtain the sub-task corresponding to each edge node.
[0084] In one embodiment, the segmentation module 530 is specifically used for: To maximize the value of the first objective function, the task proportion corresponding to each edge node is determined as follows:
[0085] in, For edge nodes Spatiotemporal coordination factors The proportion of tasks corresponding to edge nodes .
[0086] In one embodiment, the control device 500 further includes: The reserved resource determination module is used to determine the reserved computing resources and reserved spectrum resources corresponding to each edge node based on the sub-task corresponding to each edge node and the spectral efficiency and available computing resources of each edge node at the current time point, with the goal of optimizing the spatiotemporal utility of the multiple edge nodes. The instruction module 540 is further configured to generate a resource reservation instruction for each edge node based on the reserved computing resources and reserved spectrum resources corresponding to the edge node, and send the resource reservation instruction to the edge node so that the edge node reserves resources for its corresponding subtask.
[0087] In one implementation, the reserved resource determination module is specifically used for: To maximize the value of the second objective function, the reserved computing resources and reserved spectrum resources corresponding to each edge node are determined as follows:
[0088] in, For edge nodes The corresponding task ratio, To calculate the total workload of the task, For edge nodes The workload of the corresponding subtasks The current time point edge nodes Spectral efficiency factor For edge nodes The corresponding reserved computing resources, For edge nodes The corresponding reserved spectrum resources, The current time point edge nodes Available computing resources This is the variance penalty coefficient.
[0089] In one embodiment, the prediction module 520 is specifically used for: Obtain road topology information for the geographic area where the mobile device is located at the current time. The historical trajectory and the road topology information are input into the trajectory prediction model, and the predicted trajectory is predicted by the trajectory prediction model; wherein, the trajectory prediction model is obtained by training a spatiotemporal dual-stream neural network based on the trajectory prediction dataset.
[0090] The edge computing control device in this application embodiment can serve as the execution subject of the above-described edge computing control method and can realize the function of the above-described edge computing control method. Since the principle is the same, it will not be described again here.
[0091] Based on the same inventive concept as the foregoing embodiments of this application, embodiments of this application also provide electronic devices. For example... Figure 6 As shown, the electronic device 600 includes a memory 610 and a processor 620.
[0092] In some embodiments, memory 610 may be configured to store various other data to support operation on the electronic device. Examples of such data include instructions for any application or method used to operate on the electronic device. Memory 610 may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. Processor 620, coupled to memory 610, is used to execute the computer program stored in memory 610 to implement the steps of the control method for edge computing described above. To avoid repetition, further details are omitted here.
[0093] Furthermore, based on the same inventive concept as the foregoing embodiments of this application, this application also provides a non-transient computer-readable storage medium storing a computer program, which, when executed by a computer, enables the implementation of the steps of the aforementioned edge computing control method. To avoid repetition, these details will not be elaborated upon here.
[0094] Accordingly, this application also provides a computer program product that stores instructions, which, when executed by a computer, cause the computer to implement the steps of the edge computing control method described in the above embodiments. To avoid repetition, these will not be repeated here.
[0095] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.
[0096] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.
[0097] It should be understood that the training and prediction processes of the AI models involved in the various embodiments of this application all adhere to the principles of legal data sources, compliant data content, and compliant data governance, and comply with the requirements of Article 5 of the Patent Law. Specifically: Data source legitimacy: All datasets used for AI model training were obtained through legal means, covering three categories: publicly authorized data, data authorized by partners, and self-collected compliant data. Publicly authorized data comes from compliant data sources following open-source licenses such as Apache 2.0, with complete copyright attribution and authorization scope clearly marked, and no unauthorized open-source code or data reuse. Data authorized by partners has been subject to formal data usage agreements, clearly defining the scope, duration, and confidentiality obligations, and possessing a complete authorization chain. For self-collected data involving personal information, strict informed consent procedures have been followed, and anonymization processes (including but not limited to field masking, feature anonymization, and differential privacy technology applications) have been implemented to remove personally identifiable information, fully complying with the requirements of relevant laws and regulations such as the "Interim Measures for the Administration of Generative Artificial Intelligence Services" and the "Personal Information Protection Law."
[0098] Data content compliance: The AI model's dataset undergoes multiple screenings and cleaning processes to remove all content that may violate social morality or harm public interests. It contains no obscene, pornographic, violent, discriminatory, or information that endangers national or public safety, nor does it involve the illegal acquisition or use of genetic resources. For data in sensitive fields (such as healthcare and finance), an additional privacy-preserving computation module (including federated learning and secure multi-party computation technologies) ensures that the data is "usable but not visible," avoiding compliance risks during the original data transmission process and ensuring that the data application scenarios and uses comply with public order and good morals and industry regulatory requirements.
[0099] Data governance norms: A complete data traceability system is established during the AI model training process to automatically record the source, collection time, annotation process, cleaning rules, and permission allocation of training data, generating traceable compliance reports to ensure that the data is verifiable throughout its entire lifecycle. The dataset annotation process for AI models is completed by a professional human R&D team, clearly defining the proportion of human creative contributions and avoiding reliance on AI-generated data that has not undergone substantial human modification, thus meeting the examination requirements for "human main contributions" in AI patent applications.
[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.
Claims
1. A control method for edge computing, characterized in that, The method includes: When a computing task is received from a mobile device, the historical trajectory of the mobile device within the historical time window corresponding to the current time point is obtained; wherein, the historical time window is the time period up to the current time point; Based on the historical trajectory, predict the trajectory of the mobile device within the prediction window corresponding to the current time point; wherein, the prediction window is a time period starting from the current time. Based on the predicted trajectory, multiple edge nodes are determined. Based on the predicted trajectory and the available computing resources and channel quality of each edge node at the current time point, the computing task is segmented to determine the sub-task corresponding to each edge node. Based on the subtask corresponding to each edge node, a task unloading instruction is sent to the mobile device, so that the mobile device can unload the computing task to the multiple edge nodes for computing based on the subtask corresponding to each edge node.
2. The method according to claim 1, characterized in that, The computational task is segmented based on the predicted trajectory and the available computing resources and channel quality of each edge node at the current time point, determining the sub-task corresponding to each edge node, including: For each edge node, based on the predicted trajectory, the predicted distance and the rate of change of the predicted distance between the mobile device and the edge node at the end of the prediction window are determined. Based on the predicted distance, the rate of change of the predicted distance, and the available computing resources and channel quality of the edge node, the spatiotemporal coordination factor of the edge node is calculated. With the goal of optimizing the weighted sum of the spatiotemporal coordination factors of the multiple edge nodes, the task proportion corresponding to each edge node in the computing task is determined; Based on the task ratio corresponding to each edge node, the computing task is divided to obtain the sub-task corresponding to each edge node.
3. The method according to claim 2, characterized in that, The step of determining the task proportion corresponding to each edge node in the computational task, with the objective of optimizing the weighted sum of the spatiotemporal coordination factors of the multiple edge nodes, includes: To maximize the value of the first objective function, the task proportion corresponding to each edge node is determined as follows: in, For edge nodes Spatiotemporal coordination factors The proportion of tasks corresponding to edge nodes As a decision smoothing constraint factor, .
4. The method according to claim 1, characterized in that, Before sending the task unload instruction to the mobile device, the method further includes: Based on the subtasks corresponding to each edge node and the spectral efficiency and available computing resources of each edge node at the current time point, with the goal of optimizing the spatiotemporal utility of the multiple edge nodes, the reserved computing resources and reserved spectrum resources corresponding to each edge node are determined. For each edge node, a resource reservation instruction is generated based on the reserved computing resources and reserved spectrum resources corresponding to the edge node, and the resource reservation instruction is sent to the edge node so that the edge node can reserve resources for its corresponding subtask.
5. The method according to claim 4, characterized in that, Based on the subtask corresponding to each edge node and the spectral efficiency and available computing resources of each edge node at the current time point, with the goal of optimizing the spatiotemporal utility of the multiple edge nodes, the reserved computing resources and reserved spectral resources corresponding to each edge node are determined, including: To maximize the value of the second objective function, the reserved computing resources and reserved spectrum resources corresponding to each edge node are determined as follows: in, For edge nodes The corresponding task ratio, To calculate the total workload of the task, For edge nodes The workload of the corresponding subtasks The current time point edge nodes Spectral efficiency factor For edge nodes The corresponding reserved computing resources, For edge nodes The corresponding reserved spectrum resources, The current time point edge nodes Available computing resources This is the variance penalty coefficient.
6. The method according to claim 1, characterized in that, The step of predicting the trajectory of the mobile device within the prediction window corresponding to the current time point based on the historical trajectory includes: Obtain road topology information for the geographic area where the mobile device is located at the current time. The historical trajectory and the road topology information are input into the trajectory prediction model, and the predicted trajectory is predicted by the trajectory prediction model; wherein, the trajectory prediction model is obtained by training a spatiotemporal dual-stream neural network based on the trajectory prediction dataset.
7. A control device for edge computing, characterized in that, The device includes: The acquisition module is used to acquire the historical trajectory of the mobile device within the historical time window corresponding to the current time point when receiving a computing task from the mobile device; wherein, the historical time window is the time period up to the current time point; The prediction module is used to predict the trajectory of the mobile device within a prediction window corresponding to the current time point based on the historical trajectory; wherein, the prediction window is a time period starting from the current time. The segmentation module is used to determine multiple edge nodes based on the predicted trajectory, and to segment the computing task based on the predicted trajectory and the available computing resources and channel quality of each edge node at the current time point, thereby determining the sub-task corresponding to each edge node. The instruction module is used to send a task unloading instruction to the mobile device based on the subtask corresponding to each edge node, so that the mobile device can unload the computing task to the multiple edge nodes for computing based on the subtask corresponding to each edge node.
8. An electronic device, characterized in that, include A memory on which computer programs are stored; A processor for executing the computer program to implement the steps of the edge computing control method according to any one of claims 1 to 6.
9. A non-transient readable storage medium, characterized in that, It stores a computer program that can be executed by a processor to implement the steps of the edge computing control method according to any one of claims 1 to 6.
10. A computer program product, characterized in that, It includes a computer program that can be executed by a processor to implement the steps of the edge computing control method according to any one of claims 1 to 6.