A method and device for intelligent terminal fusion communication driven by edge computing

Abstract

The present invention relates to an edge computing-driven intelligent terminal converged communication method and device, which aims to solve the problem of converged communication between multiple types of devices in traditional communication methods. The method first collects multi-dimensional data of intelligent terminals, including hardware information, network access methods and network environment information, and normalizes these data to generate detailed terminal feature data. This provides an accurate feature description for each device, helping the system understand its processing power and network status. Subsequently, a classification algorithm is used to classify the terminal feature data, determine the category and feature sub-data of each device, and reasonably allocate computing tasks and network resources. Finally, based on the classification results, the system allocates appropriate resources to each terminal device to achieve efficient task scheduling and resource optimization. This method effectively improves the communication efficiency and coordination capabilities between multiple types of devices, while alleviating the bandwidth limitations and delay problems in traditional communications.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an edge computing-driven intelligent terminal fusion communication method and device. Background Art

[0002] With the rapid development of the Internet of Things, 5G networks, and artificial intelligence technologies, smart terminal devices have been widely used in various application scenarios. Traditional communication methods rely on a centralized cloud computing model, which transmits data from terminal devices to the cloud for processing and then returns the results to the terminal. However, with the surge in data volume and the complexity of network applications, this model faces problems such as bandwidth limitations, high latency, and privacy security. Especially in scenarios involving a large number of smart terminals working together, traditional communication methods have been unable to meet the requirements of real-time and efficiency, resulting in a significant decrease in the response speed and communication quality of terminal devices.

[0003] To address these challenges, the introduction of edge computing provides a new solution. By sinking computing power to network edge nodes, edge computing can process data close to terminal devices, thereby reducing communication delays and bandwidth pressure.

[0004] For example, the patent with application number 202410437818.3 provides an edge computing network host networking interaction system and a real-time data interaction method, which can finely manage the real-time distribution and reception of data, events and commands between edge computing network hosts, and simplify the network programming of host applications. The system includes: edge computing network host: communicates with terminal devices, collects terminal data, and realizes data interaction between hosts through data distribution services; cloud control platform: communicates with edge computing network hosts, and interacts with data through data distribution services; network middleware: embedded in edge computing network hosts and cloud control platforms, interacts with data distribution service systems, and ensures real-time transmission and management of data.

[0005] However, even if the above-mentioned edge computing-based communication method is adopted, the existing technology still has certain limitations, mainly reflected in its insufficient ability to handle the integrated communication of multiple types of intelligent terminal devices. In complex application scenarios, the efficiency of data transmission and collaborative processing between intelligent terminal devices is low, especially when there are many types of devices, different functions and different network access methods, the existing methods are difficult to achieve effective resource allocation and task collaboration. Therefore, for the problem of integrated communication between intelligent terminals, the existing technology still has a lot of room for optimization.

[0006] Therefore, there is still a lot of room for optimization in the existing technology regarding the converged communication problem between smart terminals. Summary of the invention

[0007] The main purpose of the present invention is to provide an edge computing driven intelligent terminal fusion communication method and device to solve the fusion communication problem between intelligent terminals in the prior art.

[0008] To achieve the above-mentioned objectives, the present invention provides an edge computing-driven intelligent terminal converged communication method, comprising the following steps: collecting multi-dimensional data of intelligent terminal devices, and generating terminal feature data of the intelligent terminal devices based on the multi-dimensional data; classifying the intelligent terminal devices based on the terminal feature data, and determining the category and feature sub-data of each of the intelligent terminal devices; and allocating corresponding computing tasks and network resources to the intelligent terminal devices based on the category and feature sub-data of the intelligent terminal devices.

[0009] The present invention also provides an edge computing-driven smart terminal fusion communication device, comprising: a collection unit, used to collect multi-dimensional data of smart terminal devices, and generate terminal feature data of the smart terminal devices based on the multi-dimensional data; a determination unit, used to classify the smart terminal devices based on the terminal feature data, and determine the category and feature sub-data of each of the smart terminal devices; an allocation unit, used to allocate corresponding computing tasks and network resources to the smart terminal devices based on the category and feature sub-data of the smart terminal devices.

[0010] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0011] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned methods are implemented.

[0012] The present invention provides an edge computing-driven intelligent terminal converged communication method and device, which can effectively solve the problem of traditional communication methods in processing converged communication between multiple types of devices by collecting terminal feature data, classifying and processing devices, and allocating resources according to the classification results. First, by comprehensively collecting multi-dimensional data of intelligent terminals, including hardware information (such as CPU model, memory capacity), network access methods (such as Wi-Fi, 5G), and network environment information (such as bandwidth, delay), and normalizing these data, terminal feature data is generated. This detailed data collection provides detailed characteristics of each device, which helps to accurately understand its processing capabilities and network status.

[0013] Secondly, the terminal feature data is classified and processed to determine the category and feature sub-data of each smart terminal. This process uses classification algorithms (such as K-Means) to group terminal devices to ensure that different categories of devices can be properly processed and resource allocated. This classification not only improves the pertinence of task allocation, but also helps the system understand the specific needs of different devices, thereby avoiding the problem of uneven resource allocation caused by a wide variety of devices in traditional methods.

[0014] Finally, based on the classification results of the devices, corresponding computing tasks and network resources are assigned to each terminal. In this way, resources can be reasonably allocated according to the processing capabilities of the devices and the network conditions, and task allocation can be optimized, thereby achieving efficient converged communication among multiple types of devices. This method effectively alleviates the bandwidth limitations and delays in traditional communications and improves overall communication efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of an intelligent terminal converged communication method driven by edge computing in one embodiment of the present invention;

[0016] Figure 2 It is a structural diagram of an intelligent terminal fusion communication device driven by edge computing in one embodiment of the present invention;

[0017] Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0018] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0020] Reference Figure 1 , the embodiment of the present application provides an edge computing driven intelligent terminal fusion communication method, comprising the following steps:

[0021] S1, collecting multi-dimensional data of smart terminal devices, and generating terminal feature data of the smart terminal devices based on the multi-dimensional data.

[0022] S2: Classify the smart terminal devices based on the terminal feature data to determine the category and feature sub-data of each smart terminal device.

[0023] S3, allocating corresponding computing tasks and network resources to the smart terminal device based on the category and characteristic sub-data of the smart terminal device.

[0024] That is, the system collects multi-dimensional data of smart terminal devices and generates terminal feature data based on this data. This step includes obtaining information such as computing power, storage resources, connected network bandwidth and latency from each smart terminal device. Through this detailed data collection, the system can fully understand the specific capabilities and limitations of each terminal device, thus laying a solid foundation for subsequent processing.

[0025] Next, based on these terminal feature data, the system classifies the intelligent terminal devices. This step divides the terminal devices into different categories according to their characteristics and functions, and assigns characteristic sub-data to each device. This classification not only takes into account the hardware configuration, but also includes network performance, usage scenarios and other aspects, so as to more accurately reflect the capabilities of the device in actual applications. For example, devices with strong computing power and sufficient network bandwidth are classified as high-performance terminals, while devices with weak computing power and limited network bandwidth are classified as low-performance terminals. This classification process helps the system to more accurately match the characteristics and needs of the devices when allocating tasks and scheduling resources.

[0026] Finally, based on the above classification results, the system allocates corresponding computing tasks and network resources to the intelligent terminal devices. In this stage, the edge node uses a multi-objective optimization algorithm to allocate resources based on the classification information and feature sub-data of the device. By comprehensively considering the computing power, network bandwidth and task requirements of the device, the system can achieve the optimal resource configuration.

[0027] In general, this technical solution effectively solves the problem of traditional communication methods in dealing with integrated communication between multiple types of devices by collecting terminal feature data, classifying devices, and allocating resources based on the classification results. Through these steps, the system can achieve more efficient task collaboration and resource management in the edge computing environment, thereby improving the response speed and communication quality of smart terminal devices.

[0028] In one example, multi-dimensional data of a smart terminal device is collected, and terminal feature data of the smart terminal device is generated based on the multi-dimensional data, including: obtaining underlying hardware information of the smart terminal device by accessing a system hardware interface of the smart terminal device, wherein the underlying hardware information is used to characterize the hardware capabilities of the smart terminal device, and includes at least any of the following: CPU model, CPU performance, number of cores, main frequency information, memory capacity, and storage space; controlling the smart terminal device to access a network environment, and obtaining a current network access mode and network environment information of the smart terminal device, wherein the network access mode is used to characterize a network interface protocol supported by the smart terminal device, and the network environment information includes at least network bandwidth, delay, jitter, and packet loss rate; normalizing the underlying hardware information, network access mode, and network environment information of the smart terminal device, and integrating the underlying hardware information, network access mode, and network environment information of the smart terminal device into a unified feature space to obtain the terminal feature data of the smart terminal device.

[0029] That is, by accessing the system hardware interface of the smart terminal device, the system can obtain the underlying hardware information. This information includes the device's CPU model, CPU performance, number of cores, main frequency, memory capacity, and storage space. These hardware parameters provide basic data on the device's computing power and storage capacity, helping to evaluate its ability to handle tasks. For example, a device may have a high-frequency CPU and a large amount of memory, which indicates that it has high performance when processing complex computing tasks.

[0030] Secondly, the system controls the smart terminal device to access the network environment and obtains its network access method and related network environment information. The network access method characterizes the network interface protocol supported by the device, such as Wi-Fi, 4G or 5G, which determines the network connection capability of the device. The network environment information includes data such as network bandwidth, latency, jitter and packet loss rate, which affect the quality and speed of data transmission of the device in the network. For example, a device may access the 5G network, which has high bandwidth and low latency, and is suitable for application scenarios that require high-speed data transmission.

[0031] Finally, the collected underlying hardware information, network access methods, and network environment information are normalized. The purpose of this step is to integrate these different types of data into a unified feature space for unified analysis and comparison. Normalization can convert this information into a relatively consistent format so that different types of data can be compared on the same dimension. For example, hardware performance and network environment information may be standardized into a unified score so that they can be analyzed on a unified scale when comprehensively evaluating device characteristics.

[0032] Through the above process, the system generates terminal feature data of intelligent terminal devices, which integrates hardware performance, network access mode and network environment information. The final terminal feature data will be used in subsequent steps such as device classification, task allocation and resource scheduling to ensure that the device can obtain reasonable resource allocation and task arrangement in actual application.

[0033] In one example, the smart terminal devices are classified based on the terminal feature data to determine the category and feature sub-data of each of the smart terminal devices, including: weighting the terminal feature data of the smart terminal devices to obtain a multidimensional feature matrix with weights; determining the preset number of categories of the smart terminal devices, and constructing a K-Means algorithm model based on the preset number of categories; inputting the multidimensional feature matrix into the K-Means algorithm model, and iteratively calculating the Euclidean distance between the feature vector of each smart terminal device and the center point of the category until convergence to an optimal classification result; based on the optimal classification result, determining the relationship between the feature vector of each smart terminal device and the center point of the category, and determining the category label of the smart terminal device based on the relationship, wherein the category label carries the information content of the smart terminal device in different feature dimensions; using the category label of the smart terminal device as the feature sub-data of the smart terminal device, and outputting the category and feature sub-data of the smart terminal device.

[0034] That is, the terminal feature data of the smart terminal device is weighted to generate a multi-dimensional feature matrix with weights. These weights reflect the importance of different features in device classification. For example, CPU performance and network bandwidth may be given higher weights because they have a greater impact on the overall performance of the device. Next, the system determines the preset number of categories, such as K categories, and builds a K-Means algorithm model based on this preset number. The K-Means algorithm is a commonly used clustering algorithm used to assign data points to K different categories. When building a K-Means model, the system first randomly selects K initial category center points.

[0035] Then, the weighted multidimensional feature matrix is ​​input into the K-Means algorithm model. The algorithm calculates the Euclidean distance from the feature vector of each smart terminal device to the center point of each category, and assigns the device to the category with the closest distance. This process is repeated and iterated until the position of the center point of the category no longer changes significantly, that is, it converges to the optimal classification result. After obtaining the optimal classification result, the system determines the category label of the device based on the relationship between the feature vector of each smart terminal device and the center point of each category. The category label contains the information content of the device in different feature dimensions. For example, a device may be labeled as a "high-performance device", and its label contains detailed information about the device in terms of computing power and network bandwidth.

[0036] Finally, the system uses these category tags as the characteristic sub-data of the device and performs output processing. In this way, the category and characteristic sub-data of each intelligent terminal device are clearly identified and can be used for subsequent resource allocation and task scheduling operations.

[0037] Through this method, the system can effectively classify intelligent terminal devices and assign corresponding feature tags to each device, thereby providing accurate basic data for resource management and task allocation.

[0038] In one example, based on the category and characteristic sub-data of the smart terminal device, corresponding computing tasks and network resources are allocated to the smart terminal device, including: collecting computing resource information and network resource information of the edge node in real time through the internal resource monitoring module of each edge node, and building a resource dynamic model of each edge node based on the computing resource information and network resource information collected in real time; synchronizing and distributing the category and characteristic sub-data of the smart terminal device between each edge node to ensure that each edge node can access the complete characteristic data of each smart terminal device it manages; obtaining the historical task data of each smart terminal device, and predicting the task requirements of the smart terminal device based on the historical task data and the characteristic sub-data of the smart terminal device to obtain the current task requirements of the smart terminal device. The method comprises the following steps: first, a weighted linear weighted method is used to combine the current task requirements and characteristic sub-data of the smart terminal device and the smart terminal devices managed by each edge node to construct a dynamic multi-objective optimization model, wherein the multi-objective optimization model is used to balance the current task requirements of the smart terminal device and the characteristic information of the smart terminal device itself; the multi-objective optimization model is used to dynamically allocate computing tasks among the edge nodes in combination with a load balancing algorithm based on polling, and a task allocation list and a resource allocation plan of each smart terminal device on its edge node are generated based on the computing tasks allocated to each edge node and the dynamic resource model of each edge node; and corresponding computing tasks and network resources are allocated to the smart terminal devices based on the task allocation list and the resource allocation plan of each smart terminal device on its edge node.

[0039] That is, in a typical edge computing task allocation system, the system collects computing resource information and network resource information of edge nodes in real time through the internal resource monitoring module of each edge node, and allocates corresponding computing tasks and network resources to smart terminal devices based on this information. The following explains the relevant nouns and verbs step by step, and refines the description content by adding specific examples and parameter settings.

[0040] First, the so-called "internal resource monitoring module of edge nodes" refers to a software module deployed inside the edge server to monitor the resource status of edge nodes. Edge nodes refer to servers or devices close to end users that are used to share computing tasks. The internal resource monitoring module can collect computing resource information and network resource information of edge nodes. Computing resource information includes CPU usage, memory occupancy, storage space, etc. For example, CPU usage can be measured by the average load (Load Average) collected every second, memory occupancy is represented by regularly obtaining the ratio of available memory to used memory, and storage space is monitored by the remaining available disk space. Network resource information includes parameters such as network bandwidth and network latency. For example, bandwidth can be quantified by monitoring the amount of data sent or received per second (in Mbps), and latency can be represented by the round-trip time measured by the ping command. The internal resource monitoring module can collect this data regularly through SNMP (Simple Network Management Protocol) or other similar protocols.

[0041] Next, the system builds a "resource dynamic model" for each edge node based on the resource information collected in real time. The so-called "resource dynamic model" refers to a mathematical or data model that describes the resource status of edge nodes at different time points. This model can help the system understand the current capabilities and load conditions of edge nodes in order to make reasonable resource allocation decisions. Specifically, the resource dynamic model can be represented by a multidimensional vector, where each dimension corresponds to a different type of resource, such as CPU utilization, memory usage, network bandwidth, and latency. The value of each dimension will be updated over time to form a dynamic resource description. For example, assuming that the resource vector of an edge node is [CPU: 70%, memory: 60%, bandwidth: 50Mbps, latency: 10ms], this information will be used to build a resource dynamic model for subsequent task allocation.

[0042] Then, the system will synchronize and distribute the "categories and characteristic sub-data of smart terminal devices" among the edge nodes. Categories of smart terminal devices include mobile phones, smart watches, smart home devices, etc., and characteristic sub-data refers to specific information that describes the characteristics of these devices, such as the processing power of the device, battery capacity, installed application types, etc. For example, for a smart home device, its characteristic sub-data may include: processor model (such as ARM Cortex-A53), memory capacity (such as 512MB), and current battery power status (such as 80%). The system synchronizes these characteristic data among all edge nodes so that each edge node can obtain complete information about the devices it manages. This step is implemented through a distributed database system or a message queue system to ensure the timeliness and consistency of data synchronization.

[0043] Next, the system obtains the "historical task data of smart terminal devices" and predicts the task requirements based on this data. Historical task data refers to the record of tasks that the device has performed in the past, including the type of task, start and end time, the amount of resources consumed, etc. By analyzing these data, the system can predict the tasks that smart terminal devices may need to perform in the future and their resource requirements. For example, if a smart home device performs a video streaming task between 7 and 9 pm every day, the system can predict its network bandwidth requirements during this period based on historical data and allocate sufficient network resources to it in advance. This prediction can use machine learning algorithms, such as LSTM (Long Short-Term Memory Network) based on time series, to model historical data to determine the needs of future tasks.

[0044] In order to achieve the optimal allocation of tasks, the system uses the "linear weighted method of weights" combined with a multi-objective optimization model for calculation. The so-called "linear weighted method of weights" means assigning different weights to different goals, and then weighted summing these goals to obtain a comprehensive optimization goal. For example, when allocating computing resources, the system may consider multiple goals such as computing power, memory usage, and network bandwidth. Each goal is assigned a certain weight (such as computing power weight is 0.4, memory usage weight is 0.3, and network bandwidth weight is 0.3), and then weighted summing each goal to obtain the final optimization result. The multi-objective optimization model is used to balance the current task requirements of smart terminal devices and their own characteristic information to ensure that the device can perform tasks in the most suitable way and achieve the optimal balance of resource utilization.

[0045] In the specific task allocation process, the system combines the "round-robin load balancing algorithm" to dynamically allocate computing tasks among edge nodes. The round-robin algorithm is a simple load balancing method that ensures that tasks are evenly distributed among nodes by assigning tasks to each edge node in turn. For example, if there are three edge nodes, tasks will be assigned to the first, second, and third nodes in turn, and then back to the first node. Combined with the multi-objective optimization model, the round-robin algorithm will take into account the real-time resource status of edge nodes when assigning tasks, thereby achieving more intelligent load balancing.

[0046] Finally, the system generates a "task allocation list and resource allocation plan" for the smart terminal devices on their edge nodes based on the computing tasks and resource dynamic models assigned to each edge node. The task allocation list refers to the specific tasks assigned to each smart terminal device and the time of their execution, while the resource allocation plan describes the computing and network resources required for task execution. For example, the task allocation list may indicate that a smart terminal device will perform a video analysis task during a certain period of time, while the resource allocation plan clearly indicates how much CPU, memory, and network bandwidth need to be allocated.

[0047] Through this comprehensive task and resource allocation mechanism, the system can ensure that each intelligent terminal device obtains appropriate computing power and network resources when performing tasks, thereby improving task execution efficiency, optimizing overall system performance, and achieving dynamic optimization and efficient utilization of resources [1].

[0048] In one example, after allocating corresponding computing tasks and network resources to the smart terminal devices based on the categories and characteristic sub-data of the smart terminal devices, the smart terminal converged communication method further includes: analyzing the task allocation list of the smart terminal devices to determine the dependencies between the tasks, wherein the dependencies include: data dependency, computing dependency and synchronization dependency; using a directed acyclic graph model to describe the dependencies between the tasks to obtain a directed acyclic task graph, wherein each node in the directed acyclic graph model represents a task, and each edge represents the dependency between tasks; based on the directed acyclic task graph, determining the target tasks that need to be coordinated between the smart terminal devices, as well as the synchronization schedule and communication requirement table of the target tasks, wherein the communication requirement table includes the data exchange amount, transmission frequency and data priority required for each task; sending the synchronization schedule and communication requirement table of the target task to the smart terminal device corresponding to the target task, so that the smart terminal device performs the computing task based on the synchronization schedule and the communication requirement table.

[0049] That is, the system analyzes the allocated computing tasks and network resources to identify the dependencies between tasks. Dependencies usually include the following three types: data dependency, computing dependency, and synchronization dependency.

[0050] Data dependency: Data dependency means that the execution of one task depends on the data of another task. For example, task A needs to pre-process the data collected by the sensor before it can be provided to task B for further analysis. In this process, the output of task A becomes the input of task B. In the specific implementation, the output of task A can be a certain data format, such as JSON or CSV file, and the module of task B must have the function of parsing data in this format. The parameters of data transmission can include data size (for example, in bytes), data transmission latency requirements (for example, must respond in milliseconds), and transmission effectiveness (that is, how to deal with data loss during transmission).

[0051] Computational dependency: Computational dependency means that the execution order of tasks is affected by the calculation results of other tasks. For example, in a multi-agent system, the intelligent terminal device may need to perform positioning calculations on its environment, and this positioning information is used by other devices to plan paths. Therefore, the path planning task must wait until the positioning task is completed before it can be executed. The parameters involved here include calculation accuracy (for example, the accuracy of the position is at the centimeter level), calculation time (for example, each positioning takes 10 milliseconds), and the validity period of the calculation results (for example, the position information is valid for the next second).

[0052] Synchronous dependency: Synchronous dependency means that multiple tasks need to be executed synchronously at the same time to ensure the coordination of the system. For example, multiple smart terminal devices need to perform a coordinated task at the same time, such as turning on multiple cameras at the same time to scan the environment to ensure consistent environmental data. In this case, the time synchronization accuracy between the devices must be set, usually through the Network Time Protocol (NTP) or the Precision Time Protocol (PTP) to achieve synchronization. Parameters may include the time error range of synchronization (for example, less than 1 millisecond), the transmission frequency of the synchronization signal (for example, once per second), and the fault recovery mechanism (for example, how to correct when the time deviation exceeds the range).

[0053] Next, the system uses a directed acyclic graph (DAG) model to describe the dependencies between tasks. DAG is a graph structure in which nodes represent tasks and edges represent dependencies between tasks. Since DAG has no loops, it ensures that the execution order of tasks is linear and there are no circular dependencies. The specific attributes and constraints of each task are represented by the nodes and edges of DAG. For example, when there is a data dependency between task A and task B, an edge in DAG can represent this dependency. This description method makes the priority, dependency, and order of tasks very intuitive.

[0054] For example, suppose we have three tasks: Task A is data collection, Task B is data preprocessing, and Task C is data analysis. In the DAG model, node A refers to data collection, node B refers to data preprocessing, and node C refers to data analysis. The edge from A to B indicates that data collection must be completed before data preprocessing, while the edge from B to C indicates that data analysis can only be performed after data preprocessing. In this way, the system can generate a task execution plan based on the DAG model, clarify the execution order of each task, and avoid execution errors caused by unclear task dependencies.

[0055] After constructing the directed acyclic task graph, the system determines the target tasks that need to be coordinated between smart terminal devices based on the graph model. The system identifies which tasks must be performed in coordination, as well as the synchronization schedule and communication requirements table between these tasks. The synchronization schedule specifies at what point in time each task needs to start or complete to ensure the coordinated progress of the tasks; for example, if task B needs to start immediately after task A is completed, the synchronization schedule will set the time window for tasks A and B. The communication requirements table describes in detail the amount of data exchange between tasks (such as the size of data that needs to be transmitted between tasks, in MB), the transmission frequency (such as the number of transmissions per second), and the data priority (such as critical data needs to be transmitted first). This information is crucial for the efficient execution of tasks because it helps the system optimize task scheduling and data transmission strategies to ensure that all tasks are completed within the scheduled time and meet communication requirements.

[0056] Finally, the system sends the synchronization schedule and communication demand table of the target task to the corresponding intelligent terminal device. Each intelligent terminal device adjusts its own computing task execution method based on this information to ensure that its operation is coordinated with the tasks of other devices. For example, after receiving the communication demand table, intelligent terminal device A will prepare the required data and send the data to device B at the specified time, and device B will start the relevant tasks based on the synchronization schedule to process the data. In this way, the devices can achieve efficient collaborative work and avoid resource waste or processing errors caused by asynchronous task execution.

[0057] Through this method, the system can effectively manage the dependencies of tasks, optimize task synchronization and data exchange, and improve the collaborative processing capabilities of intelligent terminal devices in complex application scenarios. This not only reduces conflicts between tasks, but also ensures the efficiency of data transmission and the accuracy of task execution. For example, in an intelligent transportation system, each intelligent vehicle can collaboratively process traffic data through the above method, which can effectively avoid traffic congestion and improve road traffic efficiency.

[0058] In one example, after determining the target tasks that need to be collaborated between the smart terminal devices, as well as the synchronization schedule and communication requirement table of the target tasks based on the directed acyclic task graph, the smart terminal converged communication method also includes: analyzing the resource dynamic models of the respective edge nodes to determine potential problem information in the network resources of the respective edge nodes, wherein the potential problem information includes at least: network bottleneck conditions and load imbalance conditions; based on the potential problem information in the network resources of the respective edge nodes, as well as the synchronization schedule and communication requirement table of the target tasks, optimizing and adjusting the communication paths of the respective edge nodes, and performing communication transmission of the respective edge nodes based on the optimized communication paths.

[0059] In this example, the further optimization of the intelligent terminal device converged communication method focuses on improving the communication path between edge nodes to achieve efficient data transmission and optimal utilization of system resources. In order to better understand this process, let us explain the nouns and verbs involved in detail and add specific module parameter descriptions.

[0060] First, based on the directed acyclic task graph (DAG), the target tasks that need to be completed collaboratively between smart terminal devices can be determined. This step involves modeling the entire task flow. The DAG graph represents the dependencies between tasks, ensuring that the data processing and transmission process is acyclic and step-by-step, thereby avoiding deadlock problems caused by circular dependencies. Each task node can represent a specific computing or data processing operation, such as encoding and decoding of video data, or complex data analysis. The distributed nodes of the task may include edge servers, smart gateways, user devices, etc., and specific devices may have different computing, storage, and network bandwidth parameters.

[0061] Target tasks, synchronization schedules and communication requirements are the core parts of achieving collaborative work:

[0062] Target tasks: This is a pre-defined list of tasks that need to be completed in the system. Each task contains specific inputs, outputs, and execution logic. For example, a smart gateway may need to process sensor data and transmit it to a central server for further analysis.

[0063] Synchronization schedule: Synchronization schedule refers to the scheduling plan for completing tasks between devices within a specific time. The schedule needs to take into account factors such as network transmission delay of edge nodes and task execution time. For example, a node's schedule may include the need to transmit 1MB of data to a neighboring node within a specific time period, and this operation needs to be completed within a specified millisecond level.

[0064] Communication requirements table: describes the amount of data exchanged between devices, transmission frequency and priority. The amount of data exchange may vary depending on the nature of the task, such as video data flow is large, while text data is relatively small; the transmission frequency determines the time interval for data transmission, such as once per second; the priority determines the execution order of each task, which is used to deal with insufficient network resources.

[0065] On this basis, the system needs to analyze the resource dynamic model of the edge node. The so-called resource dynamic model refers to the real-time monitoring and dynamic modeling of the computing resources and network resources of the edge node, including the following specific contents:

[0066] Computing resources: This refers to the CPU, GPU, and memory usage of the node. The model involves specific parameters such as CPU occupancy, GPU acceleration capability, memory usage, etc. Assuming that an edge node has a quad-core CPU and 8GB of memory, the current CPU usage is 70% and the memory usage is 60%. This information will be dynamically used by the system as the basis for resource analysis.

[0067] Network resources: These refer to parameters such as network bandwidth, latency, and packet loss rate between each node and other nodes. For example, the bandwidth between node A and node B is 10Mbps, the latency is 50ms, and the current packet loss rate is 2%. These parameters are used to evaluate the feasibility and effectiveness of the communication path.

[0068] Next, the system will pay special attention to potential problem information in network resources. Potential problem information mainly includes network bottlenecks and load imbalance:

[0069] Network bottleneck: This refers to a part of the network where the transmission rate drops significantly due to excessive data traffic. Bottleneck conditions can be identified by monitoring the network bandwidth usage between nodes in real time and comparing it with the maximum bandwidth. When the actual usage of a path approaches or exceeds its maximum bandwidth (for example, more than 90%), it is determined to be a bottleneck path.

[0070] Load imbalance: This means that the resource utilization rate of some nodes or links in the network is significantly higher than that of other parts, which may lead to a decrease in the overall efficiency of the system. Load imbalance is usually evaluated by the distribution of tasks on each node. For example, if the CPU load of a node reaches more than 90%, while the load of other nodes is only 40%, the system will consider that there is load imbalance.

[0071] In order to optimize the communication path, the system first considers these potential problem information with the synchronization schedule and communication demand table of the previously determined target task. The purpose of this is to ensure that the execution of the task is not affected by network problems. For example, suppose the target task requires node A to transmit data to node B within milliseconds, and there is a network bottleneck in the path between nodes A and B. The system will adjust the path according to the actual situation, such as rerouting the data to node C, and then passing it from node C to B. If the network status from C to B is good, this can avoid the bottleneck effect.

[0072] The technical means involved in the optimization process are:

[0073] Rerouting: Data that was originally transmitted directly through the congested path is transmitted through other paths with lower loads. Taking graph theory as an example, the system will calculate a path with the least load and send the data packet to the final destination.

[0074] Load balancing: Distribute certain tasks or data flows to other nodes with lower loads to reduce the pressure on the current high-load nodes. For example, when the computing resources of node A reach a bottleneck, part of the computing tasks can be offloaded to the adjacent node D to achieve the purpose of load balancing.

[0075] Finally, the system performs communication transmission based on the optimized communication path. In actual communication, the system will transmit data according to the newly planned path to ensure that the data can be quickly and reliably transmitted to the target node. For example, if the data of node A is transferred to node B via node C, the system will start data transmission and monitor the delay and packet loss rate in real time during the transmission process to ensure that it meets the requirements.

[0076] Through such an optimization and adjustment mechanism, the system can maximize the efficiency of data transmission, avoid performance problems caused by network bottlenecks and load imbalance, and make the collaboration between smart terminal devices more efficient, thereby significantly improving the performance and stability of the overall system. This optimization method is particularly important in edge computing scenarios, especially for tasks that require high real-time performance and high reliability, such as traffic control systems in smart cities or data sharing between autonomous vehicles.

[0077] In one example, based on the potential problem information in the network resources of the edge nodes, as well as the synchronization schedule and communication requirement table of the target task, the communication paths of the edge nodes are optimized and adjusted, including: based on the potential problem information in the network resources of the edge nodes, as well as the synchronization schedule and communication requirement table of the target task, a particle swarm-based communication path optimization model is constructed; and the communication paths of the edge nodes are optimized and adjusted using the particle swarm-based communication path optimization model.

[0078] In this example, the process of optimizing the communication path first needs to comprehensively combine the potential problem information in the network resources of the edge nodes, as well as the synchronization schedule and communication demand table of the target task, and then build a communication path optimization model based on the particle swarm optimization algorithm (PSO).

[0079] First, it is necessary to conduct a detailed analysis of potential problems in the network resources of edge nodes, including factors such as network bottlenecks, link congestion, and node load imbalance. This network resource information is obtained by monitoring the specific performance parameters of each edge node, usually involving the latency, packet loss rate, bandwidth utilization, CPU occupancy, and storage resource availability of the network link. For example, in a specific network link, latency can be measured by sending ping packets at regular intervals, while packet loss rate can be calculated by counting the ratio of the number of packets lost to the total number of packets sent within a certain period of time. For bandwidth utilization, it is obtained by monitoring the data traffic transmitted on the link in real time and comparing it with the total bandwidth. By measuring these parameters, the system can identify potential bottlenecks and areas that may affect network performance.

[0080] Secondly, data collection and modeling are also required based on the synchronization schedule and communication demand table of the target task. The synchronization schedule of the target task involves the precise arrangement of the execution time of different tasks, which is usually the responsibility of the scheduler module. The main parameters include the start time, end time and dependencies between tasks. For example, if the result of task A is the input of task B, then the end time of task A must be earlier than the start time of task B. The precise configuration of the synchronization schedule ensures that tasks between multiple edge nodes can work together to achieve the best performance of the overall system.

[0081] The communication requirement table is managed by the Task Requirement Collection Module, which includes specific parameters such as the data exchange volume, transmission frequency, and data priority required for each task. For example, some tasks may involve high-frequency data acquisition, such as video streaming, which requires high bandwidth and low latency, while some low-priority data updates can tolerate higher latency. These parameters are used to describe the specific requirements for communication between different tasks and provide guidance for path optimization.

[0082] Based on the data collected above, the system built a communication path optimization model based on the particle swarm optimization algorithm. The particle swarm optimization algorithm is a meta-heuristic algorithm that simulates the foraging behavior of bird flocks in nature. It searches for the optimal solution through the movement of particles in the search space. In this model, each possible communication path is regarded as a "particle", and the state of each particle is described by multiple parameters, such as the link bandwidth, delay, packet loss rate, node load, etc. of the path. The performance parameters of the path are measured by these state parameters, which in turn affect the efficiency of the entire network communication.

[0083] The first step in the optimization process is to initialize the state of the particles. In this process, an initial communication path is randomly selected for each particle, and an initial velocity and position are assigned to each particle. During the initialization process, potential problems in the network are also converted into constraints in the particle swarm optimization model. For example, insufficient bandwidth on a path or overloaded nodes are considered infeasible solutions. These constraints can effectively limit the search space of the particle swarm, allowing the algorithm to converge to the optimal solution more quickly.

[0084] The working principle of the particle swarm optimization model is to find the optimal communication path by continuously iterating and simulating the movement of each particle in the search space. During each iteration, the particle will adjust according to its current position and the best historical position, and the formula is as follows:

[0085] Among them, vi(t+1) is the speed of particle i in the next generation, w is the inertia weight, which is used to control the degree of inheritance of the particle's historical speed, c1 and c2 are learning factors, r1 and r2 are random numbers, which are used to increase the randomness of the search. is the best historical position of particle i, is the global best position, is the current position of particle i. Through this speed update mechanism, particles can gradually move to their historical best position and global best position, thereby continuously improving the value of the fitness function.

[0086] The fitness function is the core of the entire model, which is used to evaluate the quality of each communication path. The setting of the fitness function needs to comprehensively consider multiple factors such as delay, bandwidth, load balance, etc., which can be specifically expressed as:

[0087] ;

[0088] Among them, α, β, and γ are weight factors used to adjust the impact of various indicators on fitness, delay is the delay of the path, bandwidth is the available bandwidth of the path, and load is the load of the node. In this way, the fitness function can accurately reflect the performance of each path, thereby helping particles find the optimal communication path.

[0089] After multiple iterations, the algorithm will eventually select the communication path configuration with the best performance, which is then applied to the actual communication process of the edge nodes. Through such optimization, the system can reduce the delay of data transmission, improve bandwidth utilization efficiency, and solve problems such as network bottlenecks and load imbalance. For example, if a path has increased delay due to excessive load, the system will automatically select other feasible paths for data transmission, thereby ensuring efficient data transmission and collaborative execution of tasks.

[0090] The particle swarm optimization algorithm is used to optimize and adjust the communication paths of edge nodes, thereby improving the overall performance of the edge computing system. By flexibly adjusting the communication paths, the system can dynamically adapt to the changing network environment and task requirements, ultimately achieving more efficient and reliable communication.

[0091] like Figure 2 As shown, the embodiment of the present invention also provides an edge computing driven intelligent terminal fusion communication device, including:

[0092] A collecting unit 1 is used to collect multi-dimensional data of smart terminal devices and generate terminal feature data of the smart terminal devices based on the multi-dimensional data;

[0093] Determining unit 2, used for classifying the intelligent terminal devices based on the terminal characteristic data, and determining the category and characteristic sub-data of each of the intelligent terminal devices;

[0094] The allocation unit 3 is used to allocate corresponding computing tasks and network resources to the intelligent terminal device based on the category and characteristic sub-data of the intelligent terminal device.

[0095] In this embodiment, for the specific implementation of each unit in the above device embodiment, please refer to the above method embodiment, which will not be repeated here.

[0096] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0097] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0098] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0099] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0100] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0101] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. An edge computing driven intelligent terminal fusion communication method, characterized in that: The following steps are involved: Collecting multi-dimensional data of the smart terminal device, and generating terminal feature data of the smart terminal device based on the multi-dimensional data; Classify the smart terminal devices based on the terminal feature data to determine the category and feature sub-data of each smart terminal device; The computing resource information and network resource information of the edge nodes are collected in real time through the internal resource monitoring module of each edge node, and the resource dynamic model of each edge node is constructed based on the computing resource information and network resource information collected in real time; the category and feature sub-data of the smart terminal device are synchronized and distributed among the edge nodes to ensure that each edge node can access the complete feature data of each smart terminal device it manages; Acquire historical task data of each of the smart terminal devices, and perform prediction processing on the task requirements of the smart terminal devices based on the historical task data of the smart terminal devices and the characteristic sub-data to obtain current task requirements of the smart terminal devices; Based on the weighted linear weighting method, a dynamic multi-objective optimization model is constructed in combination with the current task requirements and characteristic sub-data of the smart terminal devices, as well as the smart terminal devices managed by the edge nodes, wherein the multi-objective optimization model is used to balance the current task requirements of the smart terminal devices and the characteristic information of the smart terminal devices themselves; the multi-objective optimization model is used in combination with the polling-based load balancing algorithm to dynamically allocate computing tasks among the edge nodes, and based on the computing tasks allocated to the edge nodes and the resource dynamic model of the edge nodes, a task allocation list and a resource allocation plan for each of the smart terminal devices on its edge node are generated; based on the task allocation list and the resource allocation plan for each of the smart terminal devices on its edge node, corresponding computing tasks and network resources are allocated to the smart terminal devices; The method also includes: analyzing the task allocation list of the smart terminal device to determine the dependency relationship between each task, wherein the dependency relationship includes: data dependency, computing dependency and synchronization dependency; using a directed acyclic graph model to describe the dependency relationship between the tasks to obtain a directed acyclic task graph, wherein each node in the directed acyclic graph model represents a task, and each edge represents the dependency relationship between tasks; based on the directed acyclic task graph, determining the target tasks that need to be coordinated between the smart terminal devices, as well as the synchronization schedule and communication requirement table of the target tasks, wherein the communication requirement table includes the data exchange volume, transmission frequency and data priority required for each task; sending the synchronization schedule and communication requirement table of the target task to the smart terminal device corresponding to the target task, so that the smart terminal device performs the computing task based on the synchronization schedule and the communication requirement table.

2. The intelligent terminal integrated communication method according to claim 1, characterized in that: Collecting multi-dimensional data of the smart terminal device and generating terminal feature data of the smart terminal device based on the multi-dimensional data includes: Obtaining underlying hardware information of the smart terminal device by accessing the system hardware interface of the smart terminal device, wherein the underlying hardware information is used to characterize the hardware capabilities of the smart terminal device, and includes at least any one of the following: CPU model, CPU performance, number of cores, main frequency information, memory capacity, and storage space; Controlling the smart terminal device to access the network environment, and obtaining the current network access mode and network environment information of the smart terminal device, wherein the network access mode is used to characterize the network interface protocol supported by the smart terminal device, and the network environment information at least includes: network bandwidth, delay, jitter, and packet loss rate; The underlying hardware information, network access method and network environment information of the smart terminal device are normalized, and the underlying hardware information, network access method and network environment information of the smart terminal device are integrated into a unified feature space to obtain terminal feature data of the smart terminal device.

3. The intelligent terminal integrated communication method according to claim 1, characterized in that: Classifying the smart terminal devices based on the terminal feature data to determine the category and feature sub-data of each smart terminal device includes: Performing weighted processing on the terminal feature data of the intelligent terminal device to obtain a multi-dimensional feature matrix with weights; Determine the number of preset categories of the smart terminal device, and construct a K-Means algorithm model based on the number of preset categories; Input the multidimensional feature matrix into the K-Means algorithm model, and iteratively calculate the Euclidean distance between the feature vector of each smart terminal device and the category center point until converging to the optimal classification result; Based on the optimal classification result, determining the relationship between the feature vector of each of the smart terminal devices and the center point of the category, and determining the category label of the smart terminal device based on the relationship, wherein the category label carries the information content of the smart terminal device in different feature dimensions; The category label of the smart terminal device is used as the characteristic sub-data of the smart terminal device, and the category and the characteristic sub-data of the smart terminal device are output.

4. The intelligent terminal integrated communication method according to claim 1, characterized in that: After determining the target tasks that need to be coordinated between the intelligent terminal devices, and the synchronization schedule and communication requirement table of the target tasks based on the directed acyclic task graph, the intelligent terminal converged communication method further includes: Analyze the resource dynamic model of each edge node to determine potential problem information in the network resources of each edge node, wherein the potential problem information at least includes: network bottleneck situation and load imbalance situation; Based on the potential problem information in the network resources of each edge node, as well as the synchronization schedule and communication requirement table of the target task, the communication path of each edge node is optimized and adjusted, and communication transmission of each edge node is performed based on the optimized communication path.

5. The intelligent terminal integrated communication method according to claim 4, characterized in that: Based on the potential problem information in the network resources of each edge node, as well as the synchronization schedule and communication requirement table of the target task, the communication path of each edge node is optimized and adjusted, including: Based on the potential problem information in the network resources of each edge node, as well as the synchronization schedule and communication demand table of the target task, a communication path optimization model based on a particle swarm is constructed; The communication path optimization model based on particle swarm is adopted to optimize and adjust the communication path of each edge node.

6. An edge computing-driven intelligent terminal fusion communication device, characterized in that: include: A collecting unit, used to collect multi-dimensional data of the intelligent terminal device, and generate terminal feature data of the intelligent terminal device based on the multi-dimensional data; A determination unit, configured to classify the intelligent terminal devices based on the terminal characteristic data, and determine the category and characteristic sub-data of each of the intelligent terminal devices; The allocation unit is used to collect computing resource information and network resource information of each edge node in real time through the internal resource monitoring module of each edge node, and build a resource dynamic model of each edge node based on the computing resource information and network resource information collected in real time; synchronize and distribute the category and feature sub-data of the smart terminal device among each edge node to ensure that each edge node can access the complete feature data of each smart terminal device it manages; Acquire historical task data of each of the smart terminal devices, and perform prediction processing on the task requirements of the smart terminal devices based on the historical task data of the smart terminal devices and the characteristic sub-data to obtain current task requirements of the smart terminal devices; Based on the weighted linear weighting method, a dynamic multi-objective optimization model is constructed in combination with the current task requirements and characteristic sub-data of the smart terminal devices, as well as the smart terminal devices managed by the edge nodes, wherein the multi-objective optimization model is used to balance the current task requirements of the smart terminal devices and the characteristic information of the smart terminal devices themselves; the multi-objective optimization model is used in combination with the polling-based load balancing algorithm to dynamically allocate computing tasks among the edge nodes, and based on the computing tasks allocated to the edge nodes and the resource dynamic model of the edge nodes, a task allocation list and a resource allocation plan for each of the smart terminal devices on its edge node are generated; based on the task allocation list and the resource allocation plan for each of the smart terminal devices on its edge node, corresponding computing tasks and network resources are allocated to the smart terminal devices; The device also includes: analyzing the task allocation list of the smart terminal device to determine the dependency relationship between each task, wherein the dependency relationship includes: data dependency, computing dependency and synchronization dependency; using a directed acyclic graph model to describe the dependency relationship between the tasks to obtain a directed acyclic task graph, wherein each node in the directed acyclic graph model represents a task, and each edge represents the dependency relationship between tasks; based on the directed acyclic task graph, determining the target tasks that need to be coordinated between the smart terminal devices, as well as the synchronization schedule and communication requirement table of the target tasks, wherein the communication requirement table includes the data exchange volume, transmission frequency and data priority required for each task; sending the synchronization schedule and communication requirement table of the target task to the smart terminal device corresponding to the target task, so that the smart terminal device performs the computing task based on the synchronization schedule and the communication requirement table.

7. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images