Decision-making method for computing power balanced distribution between Internet of Things devices

Abstract

The invention relates to the technical field of Internet of Things, in particular to a decision-making method for computing power balanced allocation between Internet of Things devices, which comprises a task allocation module, a resource management module, a dynamic adjustment module and a priority regulation and control module. The task distribution module decomposes tasks and generates a preliminary distribution scheme; the resource management module monitors equipment performance and classifies the equipment performance; the dynamic adjustment module optimizes an allocation strategy by using a particle swarm optimization algorithm; and the priority regulation and control module regulates the task sequence according to the weight and ensures the high-priority task resource demand. The system stability is enhanced through the exception handling unit and the load prediction unit, the computing power distribution efficiency and the resource utilization rate are improved, and efficient and flexible task execution is achieved.

Description

Technical Field

[0001] This invention belongs to the field of Internet of Things and distributed computing technology, specifically a decision-making method for balanced allocation of computing power among Internet of Things devices. Background Technology

[0002] With the rapid development of IoT technology, the balanced allocation of computing power among smart devices has become a crucial aspect of improving overall system performance. Existing computing power allocation methods have certain limitations in resource scheduling, load balancing, and dynamic adaptability, which may lead to problems such as decreased system efficiency, resource waste, and limited response speed.

[0003] A search revealed a patent with publication number CN112769641B, which proposes a blockchain computing power optimization scheduling method for intelligent data processing, published on May 3, 2022. This technology combines massive IoT data processing tasks with the computing power resources of blockchain miners, achieving optimal allocation between external computing tasks and mining computing resources by constructing a Stackelberg game model to solve for Nash equilibrium. However, the application scenarios of this solution are mainly concentrated in the blockchain field, with a relatively narrow scope of application, making it difficult to meet the broader needs for balanced computing power allocation among IoT devices. Furthermore, this method is based on a game model with incomplete information, which may affect the accuracy and dynamic adaptability of computing power allocation in practical applications due to inaccurate or insufficient real-time information acquisition.

[0004] Another patent, CN117424900B, discloses a method, system, device, and storage medium for managing long-connection clusters in the power Internet of Things (IoT), published on July 19, 2024. This technology calculates the load index of each connection unit through a dynamic load balancing allocation algorithm and assigns connection unit addresses to newly connected devices based on resource occupancy information, thereby achieving high system reliability and maximizing resource utilization. However, this solution primarily focuses on long-connection management in the power IoT, with its load balancing strategy paying more attention to resource allocation at the network connection layer, lacking in-depth discussion of fine-grained scheduling for inter-device computing power balancing. Furthermore, this method does not fully consider device heterogeneity and task priority differences, potentially leading to insufficient flexibility in computing power allocation and difficulty in coping with the diverse needs of complex IoT environments.

[0005] The aforementioned problems indicate that existing IoT computing power allocation technologies still need improvement in terms of the universality of application scenarios, the real-time nature of information acquisition, adaptability to device heterogeneity, and dynamic adjustment of task priorities. Therefore, this invention aims to provide a decision-making method for balanced computing power allocation among IoT devices. By comprehensively considering factors such as device performance, task requirements, and network status, it achieves efficient, flexible, and dynamically adaptable balanced computing power allocation, thereby improving the overall performance and resource utilization of the IoT system. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a decision-making method for balanced allocation of computing power among Internet of Things devices.

[0007] The technical solution adopted to solve the above-mentioned technical problems is: a decision-making method for balanced allocation of computing power among IoT devices, including a task allocation module, a resource management module, a dynamic adjustment module, and a priority control module. The task allocation module receives and decomposes computing tasks, dividing them into multiple sub-tasks; the resource management module monitors the performance parameters of IoT devices in real time and records their current load status; the dynamic adjustment module allocates computing power in real time according to task requirements and device performance; and the priority control module adjusts the weight of computing power allocation according to the importance and urgency of the tasks.

[0008] The task allocation module includes a task receiving unit, a task decomposition unit, and a task scheduling unit. The task receiving unit receives externally input task request data packets through a communication interface; the task decomposition unit decomposes the received task into multiple subtasks and generates a list of computing requirements for each subtask; the task scheduling unit matches the subtask requirement lists with the device performance data provided by the resource management module to generate a preliminary computing power allocation scheme.

[0009] The resource management module includes a performance monitoring unit, a load recording unit, and a device grouping unit. The performance monitoring unit collects performance parameters of IoT devices, such as CPU utilization, memory usage, network bandwidth utilization, and remaining battery power, through sensors. The load recording unit stores the collected performance parameters in a local database and updates it periodically. The device grouping unit classifies devices into high, medium, and low categories based on their performance parameters for quick filtering during subsequent computing power allocation.

[0010] The dynamic adjustment module includes a demand analysis unit, an allocation optimization unit, and a feedback correction unit. The demand analysis unit assesses the total computing power required to complete the task based on the computing requirements list generated by the task decomposition unit and the classification results from the device grouping unit. The allocation optimization unit uses the particle swarm optimization algorithm to optimize the computing power allocation scheme, ensuring that the allocation meets the task requirements while minimizing resource waste. The feedback correction unit dynamically adjusts the computing power allocation scheme by monitoring changes in device load during task execution in real time.

[0011] The priority control module includes a weight allocation unit, a priority sorting unit, and a resource preemption unit. The weight allocation unit assigns weight values ​​to tasks based on their importance and urgency; the priority sorting unit sorts all tasks according to their weight values ​​and generates a task execution queue; and the resource preemption unit, during the execution of high-priority tasks, forcibly releases some resources if it detects that low-priority tasks are consuming too many resources to ensure the smooth completion of high-priority tasks.

[0012] Furthermore, the task allocation module also includes a task caching unit for temporarily storing unallocated subtasks. When the system load is too high or device resources are insufficient, the task caching unit stores the unallocated subtasks in a temporary storage area and reallocates them when resources become available. The task caching unit and the task scheduling unit are connected via a bidirectional communication interface to ensure the continuity of task allocation.

[0013] Through the above technical solution, after receiving external computing tasks, the task allocation module decomposes them into multiple sub-tasks through the task decomposition unit and generates a computing requirement list. The resource management module collects the performance parameters of IoT devices in real time and classifies devices into different categories according to the device grouping unit. The dynamic adjustment module generates an initial computing power allocation scheme using the particle swarm optimization algorithm based on the computing requirement list and device performance data, and adjusts the allocation strategy in real time through the feedback correction unit. The priority control module sorts tasks according to task weight values ​​and ensures the resource requirements of high-priority tasks through the resource preemption unit. Specifically, the task receiving unit obtains task request data packets through the communication interface, the task decomposition unit generates a sub-task list, the task scheduling unit reads device performance data from the resource management module, and selects appropriate devices to allocate computing power. At this time, the requirement analysis unit of the dynamic adjustment module evaluates the task requirements, the allocation optimization unit generates the optimal allocation scheme, and the feedback correction unit monitors the task execution status in real time and adjusts the allocation strategy. The weight allocation unit of the priority control module sets weight values ​​according to task attributes, the priority sorting unit generates a task queue, and the resource preemption unit releases resources for low-priority tasks when necessary.

[0014] Furthermore, the resource management module also includes an anomaly handling unit to detect abnormal fluctuations in device performance parameters. The anomaly handling unit compares historical data with currently collected data to determine if the device is experiencing performance degradation or is at risk of failure. When an anomaly is detected, the anomaly handling unit sends a warning signal to the task allocation module and suggests suspending the allocation of new tasks to the device. The anomaly handling unit and the performance monitoring unit are connected via a one-way data transmission interface to ensure timely transmission of anomaly information.

[0015] Through the above technical solution, when IoT devices experience performance anomalies during operation, the anomaly handling unit identifies potential performance degradation or failure risks by comparing historical data with current data. For example, when a device's CPU utilization consistently exceeds a threshold or its network bandwidth utilization suddenly drops, the anomaly handling unit generates a warning signal and notifies the task allocation module to suspend the allocation of new tasks to that device. At this time, the task caching unit of the task allocation module stores unallocated subtasks in a temporary storage area, waiting for reallocation when other device resources become available. The anomaly handling unit also ensures that the load recording unit of the resource management module accurately reflects the actual status of the devices by periodically updating device performance data.

[0016] Furthermore, the dynamic adjustment module also includes a load prediction unit, used to predict the load change trend of the equipment over a future period. Based on historical load data and current task execution status, the load prediction unit uses time series analysis algorithms to predict equipment load changes and generates a prediction report. The load prediction unit is connected to the allocation optimization unit through a data sharing interface to ensure that the prediction results can be used to optimize the computing power allocation scheme.

[0017] Through the above technical solution, the load prediction unit of the dynamic adjustment module predicts the load change trend of devices using time series analysis algorithms and transmits the prediction results to the allocation optimization unit. The allocation optimization unit adjusts the computing power allocation scheme according to the prediction results to avoid a decrease in task execution efficiency due to excessive device load. For example, when the prediction results show that the load of a certain device will increase significantly in the next hour, the allocation optimization unit reduces the number of new tasks assigned to that device and transfers some tasks to other devices with lower loads. The load prediction unit also ensures the accuracy of the prediction results by periodically updating the prediction model.

[0018] Furthermore, the priority control module also includes a task merging unit, used to merge multiple low-priority tasks into a single composite task. The task merging unit integrates multiple low-priority tasks into a composite task based on the task's computational requirements and resource usage, and generates a new list of computational requirements. The task merging unit and the priority sorting unit are connected via a data synchronization interface to ensure that the priorities of the merged tasks are correctly sorted.

[0019] Through the above technical solution, the task merging unit of the priority control module merges multiple low-priority tasks into composite tasks, thereby reducing the resource overhead caused by task switching. For example, when there are multiple low-priority tasks in the system with high resource consumption, the task merging unit integrates these tasks into a composite task and generates a new list of computing requirements. The priority sorting unit reorders the task queue according to the weight value of the composite task, ensuring that high-priority tasks are executed first. The task merging unit also periodically checks the task list to ensure that the merging of low-priority tasks does not affect the overall task execution efficiency.

[0020] The beneficial effects of this invention are as follows: By designing a task allocation module, resource management module, dynamic adjustment module, and priority control module, this invention enables efficient, flexible, and dynamically adaptable balanced allocation of computing power among IoT devices, improving the overall system performance and resource utilization. The task allocation module decomposes computing tasks into multiple sub-tasks and generates a preliminary allocation scheme based on device performance data, ensuring the rationality of task allocation. The resource management module monitors device performance parameters in real time and quickly filters suitable devices through a device grouping unit, improving the efficiency of computing power allocation. The dynamic adjustment module uses a particle swarm optimization algorithm to generate the optimal allocation scheme and adjusts the allocation strategy in real time through a feedback correction unit, ensuring maximum resource utilization during task execution. The priority control module sorts tasks according to their weight values ​​and ensures the resource requirements of high-priority tasks through a resource preemption unit, improving the system's response speed. Furthermore, an anomaly handling unit and a load prediction unit respectively address device performance anomalies and load change trends, further enhancing the system's stability and reliability. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the system architecture of the present invention;

[0022] Figure 2 This is a structural diagram of the resource management module;

[0023] Figure 3 A workflow diagram for dynamically adjusting modules.

[0024] The attached diagram is labeled as follows: 1. Task allocation module; 2. Resource management module; 3. Dynamic adjustment module; 4. Priority control module; 5. Performance monitoring unit; 6. Load recording unit; 7. Equipment grouping unit; 8. Anomaly handling unit; 9. Demand analysis unit; 10. Allocation optimization unit; 11. Feedback correction unit; 12. Load prediction unit. Detailed Implementation

[0025] This invention provides a decision-making method for balanced allocation of computing power among IoT devices. Its core lies in achieving efficient computing power allocation through the collaborative work of a task allocation module 1, a resource management module 2, a dynamic adjustment module 3, and a priority control module 4. The specific embodiments of this invention will be described in detail below with reference to the accompanying drawings.

[0026] like Figure 1 As shown, task allocation module 1 serves as the starting point of the entire system, receiving externally input task request data packets and decomposing them into multiple subtasks. Task allocation module 1 includes a task receiving unit, a task decomposition unit, a task scheduling unit, and a task caching unit. The task receiving unit connects to the external system via a communication interface, responsible for receiving task request data packets and transmitting them to the task decomposition unit. The task decomposition unit generates multiple subtasks based on the content of the task request and creates a computational requirement list for each subtask. These lists contain information such as the CPU utilization, memory usage, and network bandwidth requirements of the subtasks. The task scheduling unit obtains device performance data from resource management module 2 and selects suitable devices for initial computing power allocation based on this data. When the system load is too high or device resources are insufficient, the task caching unit stores unallocated subtasks in a temporary storage area and re-attempts allocation when resources are available. The task caching unit and the task scheduling unit are connected via a bidirectional communication interface to ensure the continuity of task allocation.

[0027] Resource management module 2 is responsible for monitoring the performance parameters of IoT devices in real time and recording their current load status. For example... Figure 2 As shown, the resource management module 2 consists of a performance monitoring unit 5, a load recording unit 6, a device grouping unit 7, and an anomaly handling unit 8. The performance monitoring unit 5 collects performance parameters of IoT devices, such as CPU utilization, memory usage, network bandwidth utilization, and remaining battery power, using sensors, and transmits this data to the load recording unit 6. The load recording unit 6 stores the collected performance parameters in a local database and updates it periodically to reflect the latest status of the devices. The device grouping unit 7 classifies devices into high, medium, and low categories based on their performance parameters to quickly select suitable devices for subsequent computing power allocation. The anomaly handling unit 8 compares historical data with currently collected data to determine if a device has performance degradation or failure risks. For example, when a device's CPU utilization consistently exceeds a threshold or its network bandwidth utilization suddenly drops, the anomaly handling unit 8 sends a warning signal to the task allocation module 1 and suggests suspending the allocation of new tasks to that device. The anomaly handling unit 8 and the performance monitoring unit 5 are connected via a one-way data transmission interface to ensure timely transmission of anomaly information.

[0028] The dynamic adjustment module 3 allocates computing power in real time based on task requirements and equipment performance. For example... Figure 3As shown, the dynamic adjustment module 3 includes a demand analysis unit 9, an allocation optimization unit 10, a feedback correction unit 11, and a load prediction unit 12. The demand analysis unit 9 assesses the total computing power required to complete the task based on the computing demand list generated by the task decomposition unit and the classification results of the device grouping unit 7. The allocation optimization unit 10 uses a particle swarm optimization algorithm to generate an initial computing power allocation scheme, ensuring that the computing power allocation meets the task requirements while minimizing resource waste. The feedback correction unit 11 dynamically adjusts the computing power allocation scheme by monitoring the device load changes during task execution in real time. For example, when the load of a device suddenly increases, the feedback correction unit 11 reduces the number of new tasks allocated to that device and transfers some tasks to other devices with lower loads. The load prediction unit 12 predicts the load change trend of devices based on historical load data and current task execution status using a time series analysis algorithm and transmits the prediction results to the allocation optimization unit 10. The allocation optimization unit 10 adjusts the computing power allocation scheme according to the prediction results to avoid a decrease in task execution efficiency due to excessive device load. The load prediction unit 12 also ensures the accuracy of the prediction results by periodically updating the prediction model.

[0029] Priority control module 4 adjusts the allocation of computing power based on the importance and urgency of tasks. Priority control module 4 includes a weight allocation unit, a priority sorting unit, a resource preemption unit, and a task merging unit. The weight allocation unit assigns weight values ​​to tasks based on their importance and urgency. For example, critical tasks requiring immediate execution receive higher weight values, while non-critical tasks receive lower weight values. The priority sorting unit sorts all tasks according to their weight values, generating a task execution queue. During the execution of high-priority tasks, if the resource preemption unit detects that low-priority tasks are consuming too many resources, it forcibly releases some resources to ensure the smooth completion of high-priority tasks. The task merging unit integrates multiple low-priority tasks into a composite task based on the task's computational requirements and resource usage, generating a new list of computational requirements. The priority sorting unit reorders the task queue according to the composite task's weight values, ensuring that high-priority tasks are executed first. The task merging unit also periodically checks the task list to ensure that merging low-priority tasks does not affect the overall task execution efficiency.

[0030] In a real-world application scenario, suppose a smart factory needs to complete a series of production tasks using IoT devices. After receiving a task request, the task allocation module 1 decomposes the task into multiple sub-tasks, such as controlling a robotic arm or monitoring the production line status. The task scheduling unit reads device performance data from the resource management module 2 and selects suitable devices for computing power allocation. The performance monitoring unit 5 of the resource management module 2 collects parameters such as CPU utilization and memory usage of the devices in real time and transmits this data to the load recording unit 6 for storage. The device grouping unit 7 classifies devices into high, medium, and low categories based on their performance parameters, enabling the task scheduling unit to quickly select suitable devices. When the performance parameters of a device fluctuate abnormally, the exception handling unit 8 sends a warning signal to the task allocation module 1 and suggests suspending the allocation of new tasks to that device. At this time, the task caching unit stores the unallocated sub-tasks in a temporary storage area, waiting for other device resources to become available before reallocating them.

[0031] The demand analysis unit 9 of the dynamic adjustment module 3 assesses the total computing power required to complete the tasks based on the computational requirements list of the subtasks and the classification results of the device grouping unit 7. The allocation optimization unit 10 generates an initial computing power allocation scheme using the particle swarm optimization algorithm and adjusts the allocation strategy in real time through the feedback correction unit 11. For example, when the load on a device suddenly increases, the feedback correction unit 11 reduces the number of new tasks allocated to that device and transfers some tasks to other devices with lower loads. The load prediction unit 12 predicts the load change trend of the devices using a time series analysis algorithm and transmits the prediction results to the allocation optimization unit 10. The allocation optimization unit 10 adjusts the computing power allocation scheme according to the prediction results to avoid a decrease in task execution efficiency due to excessive device load.

[0032] The weight allocation unit of priority control module 4 assigns weight values ​​to tasks based on their importance and urgency. For example, critical tasks that need immediate execution are assigned higher weight values, while non-critical tasks are assigned lower weight values. The priority sorting unit sorts all tasks according to their weight values, generating a task execution queue. During the execution of high-priority tasks, if the resource preemption unit detects that low-priority tasks are consuming too many resources, it forcibly releases some resources to ensure the smooth completion of high-priority tasks. The task merging unit combines multiple low-priority tasks into a composite task based on their computational requirements and resource consumption, generating a new list of computational requirements. The priority sorting unit reorders the task queue according to the weight values ​​of the composite task, ensuring that high-priority tasks are executed first.

[0033] As can be seen from the above embodiments, the present invention, through the collaborative work of the task allocation module 1, resource management module 2, dynamic adjustment module 3, and priority control module 4, can achieve efficient, flexible, and dynamically adaptive balanced allocation of computing power among IoT devices. The task allocation module 1 decomposes computing tasks into multiple sub-tasks and generates a preliminary allocation scheme based on device performance data, ensuring the rationality of task allocation. The resource management module 2 monitors device performance parameters in real time and quickly filters suitable devices through the device grouping unit 7, improving the efficiency of computing power allocation. The dynamic adjustment module 3 uses the particle swarm optimization algorithm to generate the optimal allocation scheme and adjusts the allocation strategy in real time through the feedback correction unit 11, ensuring maximum resource utilization during task execution. The priority control module 4 sorts tasks according to their weight values ​​and ensures the resource requirements of high-priority tasks through the resource preemption unit, improving the system's response speed. Furthermore, the anomaly handling unit 8 and the load prediction unit 12 respectively address device performance anomalies and load change trends, further enhancing the system's stability and reliability.

[0034] To enable those skilled in the art to fully understand and implement this invention, the specific implementation principles of this invention are further supplemented below with a specific application scenario.

[0035] In a smart factory production environment, suppose a series of complex production tasks need to be completed through IoT devices, such as controlling robotic arms to perform precise operations, monitoring production line status, and adjusting processing parameters in real time. These tasks have different computing requirements and priorities, and require balanced allocation of computing power among multiple heterogeneous devices.

[0036] First, the task receiving unit of task allocation module 1 receives the task request data packet sent by the external system through the communication interface. This data packet contains overall description information of the task, such as the total amount of computing resources required, the task type, and its urgency. Then, the task decomposition unit breaks down the task into multiple subtasks, each corresponding to a specific list of computing requirements, including CPU utilization, memory usage, and network bandwidth requirements. For example, controlling a robotic arm may require high real-time performance and low latency, while monitoring the production line status may prioritize the stability and accuracy of data acquisition. The task scheduling unit reads the current equipment performance data from resource management module 2 and selects suitable equipment for initial computing power allocation. If a device cannot meet the allocation requirements due to excessive load or insufficient resources, the task caching unit stores the unallocated subtasks in a temporary storage area, awaiting subsequent reallocation.

[0037] The performance monitoring unit 5 in resource management module 2 continuously collects performance parameters of IoT devices through sensors, such as CPU utilization, memory usage, network bandwidth utilization, and remaining battery power. This data is transmitted to the load recording unit 6 and stored in a local database for real-time querying and analysis. The device grouping unit 7 categorizes devices into high, medium, and low performance classes based on the collected performance parameters, enabling the task scheduling unit to quickly select suitable devices for specific subtasks. For example, for subtasks requiring high computing power, the device grouping unit 7 prioritizes high-performance devices; while for tasks with lower computing demands, medium- to low-performance devices can be selected to optimize resource utilization. Furthermore, the anomaly handling unit 8 compares historical data with currently collected data to determine if a device is experiencing performance degradation or malfunction risk. For example, if the CPU utilization of a device consistently exceeds a set threshold, the anomaly handling unit 8 sends a warning signal to the task allocation module 1 and suggests suspending the allocation of new tasks to that device. At this time, the task caching unit temporarily stores unallocated subtasks, retrying allocation when other device resources become available.

[0038] The dynamic adjustment module 3's requirement analysis unit 9, based on the computational requirement list generated by the task decomposition unit and combined with the classification results of the device grouping unit 7, assesses the total computing power required to complete the task. The allocation optimization unit 10 uses the particle swarm optimization algorithm to generate an initial computing power allocation scheme, ensuring that the computing power allocation meets the task requirements while minimizing resource waste. For example, the particle swarm optimization algorithm finds the optimal computing power allocation scheme by simulating the movement of particles in the solution space, thereby preventing some devices from experiencing a decrease in task execution efficiency due to excessive load. The feedback correction unit 11 dynamically adjusts the computing power allocation scheme by monitoring changes in device load during task execution in real time. For example, when the load of a device suddenly increases, the feedback correction unit 11 reduces the number of new tasks allocated to that device and transfers some tasks to other devices with lower loads. The load prediction unit 12 predicts the load change trend of devices based on historical load data and current task execution status using time series analysis algorithms. For example, when the prediction results show that the load of a device will increase significantly in the future, the allocation optimization unit 10 adjusts the computing power allocation scheme in advance to avoid a decrease in task execution efficiency due to excessive device load.

[0039] The weight allocation unit of priority control module 4 assigns weight values ​​to tasks based on their importance and urgency. For example, critical tasks requiring immediate execution, such as controlling a robotic arm to perform high-precision operations, are assigned higher weight values; while non-critical tasks, such as monitoring production line status, are assigned lower weight values. The priority sorting unit sorts all tasks according to their weight values, generating a task execution queue. During the execution of high-priority tasks, if the resource preemption unit detects that low-priority tasks are consuming too many resources, it forcibly releases some resources to ensure the smooth completion of the high-priority tasks. For example, if a device is executing a low-priority task and receives a request for a high-priority task, the resource preemption unit will forcibly interrupt the low-priority task and allocate resources to the high-priority task. The task merging unit integrates multiple low-priority tasks into a composite task based on their computational requirements and resource consumption, generating a new list of computational requirements. For example, when multiple low-priority tasks exist in the system and are consuming a lot of resources, the task merging unit integrates these tasks into a composite task and generates a new list of computational requirements. The priority sorting unit reorders the task queue according to the weight values ​​of the composite task, ensuring that high-priority tasks are executed first.

[0040] As can be seen from the above steps, this invention, through the collaborative work of the task allocation module 1, resource management module 2, dynamic adjustment module 3, and priority control module 4, can achieve efficient, flexible, and dynamically adaptive balanced allocation of computing power among IoT devices. The task allocation module 1 decomposes computing tasks into multiple sub-tasks and generates a preliminary allocation scheme based on device performance data, ensuring the rationality of task allocation. The resource management module 2 monitors device performance parameters in real time and quickly filters suitable devices through the device grouping unit 7, improving the efficiency of computing power allocation. The dynamic adjustment module 3 uses the particle swarm optimization algorithm to generate the optimal allocation scheme and adjusts the allocation strategy in real time through the feedback correction unit 11, ensuring maximum resource utilization during task execution. The priority control module 4 sorts tasks according to their weight values ​​and ensures the resource requirements of high-priority tasks through the resource preemption unit, improving the system's response speed. Furthermore, the anomaly handling unit 8 and load prediction unit 12 respectively address device performance anomalies and load change trends, further enhancing the system's stability and reliability.

[0041] All content not described in detail in this specification is prior art known to those skilled in the art, and the model parameters of each electrical appliance are not specifically limited; conventional equipment can be used. Electrical control components not mentioned in this technical solution are not shown in the figures because they are prior art, and will not be described further here.

[0042] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A decision-making method for balanced allocation of computing power among Internet of Things (IoT) devices, characterized in that... It includes a task allocation module (1), a resource management module (2), a dynamic adjustment module (3), and a priority control module (4). The task allocation module (1) is used to receive computing tasks and decompose them into multiple sub-tasks, and generate a computing requirement list for each sub-task. The resource management module (2) is used to monitor the performance parameters of IoT devices in real time and record the current load status. The dynamic adjustment module (3) is used to allocate computing power in real time according to task requirements and device performance. The priority control module (4) is used to adjust the weight of computing power allocation according to the importance and urgency of the task.

2. The decision-making method for balanced allocation of computing power among IoT devices according to claim 1, characterized in that... The task allocation module (1) includes a task receiving unit, a task decomposition unit, a task scheduling unit, and a task caching unit. The task receiving unit receives externally input task request data packets through a communication interface. The task decomposition unit decomposes the received task into multiple sub-tasks and generates a computing requirement list. The task scheduling unit generates a preliminary computing power allocation scheme by matching the computing requirement list with the device performance data provided by the resource management module (2). The task caching unit is used to temporarily store unallocated sub-tasks and reallocate them when resources are idle.

3. The decision-making method for balanced allocation of computing power among IoT devices according to claim 1, characterized in that... The resource management module (2) includes a performance monitoring unit (5), a load recording unit (6), a device grouping unit (7), and an anomaly handling unit (8). The performance monitoring unit (5) collects the CPU utilization rate, memory usage rate, network bandwidth utilization rate, and remaining power performance parameters of IoT devices through sensors. The load recording unit (6) stores the collected performance parameters in a local database and updates them periodically. The device grouping unit (7) divides the devices into high, medium, and low categories according to their performance parameters. The anomaly handling unit (8) compares historical data with the currently collected data to determine whether the device has a performance degradation or failure risk and sends a warning signal to the task allocation module (1).

4. The decision-making method for balanced allocation of computing power among IoT devices according to claim 1, characterized in that... The dynamic adjustment module (3) includes a demand analysis unit (9), an allocation optimization unit (10), a feedback correction unit (11), and a load prediction unit (12). The demand analysis unit (9) evaluates the total computing power required to complete the task based on the computing demand list generated by the task decomposition unit and the classification results of the equipment grouping unit (7). The allocation optimization unit (10) generates an initial computing power allocation scheme using the particle swarm optimization algorithm. The feedback correction unit (11) dynamically adjusts the computing power allocation scheme by monitoring the changes in equipment load during the task execution process in real time. The load prediction unit (12) predicts the load change trend of the equipment based on historical load data and the current task execution status using a time series analysis algorithm and transmits the prediction results to the allocation optimization unit (10).

5. The decision-making method for balanced allocation of computing power among IoT devices according to claim 1, characterized in that... The priority control module (4) includes a weight allocation unit, a priority sorting unit, a resource preemption unit, and a task merging unit. The weight allocation unit assigns weight values ​​to tasks based on their importance and urgency. The priority sorting unit sorts all tasks according to their weight values ​​to generate a task execution queue. If the resource preemption unit detects that a low-priority task is occupying too many resources during the execution of a high-priority task, it forcibly releases some resources to ensure the smooth completion of the high-priority task. The task merging unit integrates multiple low-priority tasks into a composite task and generates a new list of computational requirements based on the task's computational needs and resource usage.

6. The decision-making method for balanced allocation of computing power among IoT devices according to claim 3, characterized in that... The anomaly handling unit (8) ensures that the load recording unit (6) can accurately reflect the actual status of the equipment by periodically updating the equipment performance data, and connects to the performance monitoring unit (5) through a one-way data transmission interface to transmit anomaly information.

7. The decision-making method for balanced allocation of computing power among IoT devices according to claim 4, characterized in that... The load prediction unit (12) ensures the accuracy of the prediction results by periodically updating the prediction model, and connects to the allocation optimization unit (10) through the data sharing interface to transmit the prediction results.

Images