Task collaboration-oriented distributed node grouping method
By building dynamically updated node packet tables in mobile edge networks and dynamically adjusting the update cycle, the high energy consumption and static problems in traditional architectures are solved, the network service quality and task collaborative execution success rate is improved, and the dynamic network environment is adapted to change.
Patent Information
- Application Number
- CN202411029937.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-07-30
AI Technical Summary
The resource collaboration strategies in existing mobile edge networks face problems such as high energy consumption, staticity and service quality, especially in dynamic network environments, traditional centralized and distributed architectures cannot quickly adapt to changes in network state, resulting in network service quality. Low.
A distributed node grouping method for task collaboration is proposed. By constructing a dynamically updated node grouping table, using periodic updates and task-driven update methods, dynamically adjusting the update cycle to ensure the real-time and effectiveness of the node grouping table. Improve the success rate of task collaborative execution and network service quality.
This method reduces energy consumption, improves network service quality, can quickly adapt to network state changes, improves the success rate of task collaborative execution and overall packet service efficiency, and reduces the energy consumption of equipment.
Smart Images

Figure CN119052237A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of mobile edge networks, specifically to the field of resource collaboration technology, and more specifically to a distributed node grouping method oriented to task collaboration. Background Art
[0002] The resource coordination strategy of mobile edge networks supports nodes to coordinate tasks with other nodes for execution. This can not only alleviate the resource constraints of the device itself, but also improve the efficiency of task processing through task decomposition and parallel execution, such as reference [1]. In the resource coordination process, obtaining the network node grouping information maintained by the current node is the basis of resource coordination. Node grouping information determines which edge nodes can coordinate the current task and is the premise for formulating specific resource coordination plans. The resource usage, location and function of each node or device in the network determine the computing and data processing tasks it can participate in. Therefore, node grouping information is crucial in resource coordination, but obtaining grouping information in actual operations is often challenging.
[0003] Challenge 1: The dynamic network environment means that the status and connectivity of nodes may change frequently, which requires the system to be able to update and maintain node information in real time. However, most current resource coordination solutions assume that node grouping information and node availability information are known, and make resource coordination decisions on this basis. Challenge 2: In actual systems, collecting node information requires computing and communication resources, especially on resource-constrained edge devices. This overhead greatly affects the overall performance and energy efficiency of the device, and ultimately limits the performance of resource coordination strategies. Edge devices typically rely on limited energy supplies, such as batteries. High-frequency data updates and information exchanges may quickly deplete the energy of these devices, leading to an energy crisis. This not only affects the continuous operation time of the node, but may also limit the device's ability to participate in future tasks, ultimately affecting the performance and reliability of the entire network.
[0004] In order to meet the above challenges, a variety of solutions have been proposed in traditional research. Typical node grouping research can be roughly divided into two categories according to its network architecture: centralized node grouping and distributed node grouping.
[0005] In the centralized node grouping, there is a central controller, which is used to integrate the edge node information in the network, and make resource coordination decisions and issue decision instructions for the system based on this. In theory, in order to achieve real-time monitoring of the network status, the central controller updates the global information in a way that will lead to frequent data interaction in the edge network, thereby consuming a large amount of communication resources. At the same time, the central controller needs to be equipped with certain computing resources to ensure the normal data collection and policy formulation.
[0006] Compared with the centralized architecture, some distributed node grouping studies use game theory to maintain grouping information between nodes. By using the analysis and optimization mechanism of game theory, a more efficient node organization strategy is provided in the edge computing environment, achieving balanced resource allocation, taking into account the service quality requirements and optimizing the overall resource utilization efficiency. Under this model, edge nodes only need to make autonomous decisions based on necessary information. This distributed node grouping method reduces the dependence on the central controller and reduces the frequency of data synchronization, thereby showing significant advantages in reducing the operating costs of mobile devices, improving task processing efficiency, and saving communication resources, such as reference [2]. Although using game theory or on-demand routing methods to find suitable resource sharing nodes in edge networks has a certain effect in providing resource coordination information, it is limited by the mechanism of game theory and on-demand routing information interaction. Before making resource coordination decisions, edge nodes need to exchange information with other nodes for at least one round. For some delay-sensitive tasks, the waiting time for the game will affect the service quality. In addition, faced with frequent changes in node resources in edge networks, high mobility, dynamic joining and leaving, etc., multiple rounds of information interaction may be required to make resource coordination decisions, thus affecting the service quality of the edge network and the effective use of resources.
[0007] Based on the above analysis, it can be seen that in traditional network environments, the centralized or distributed architecture in existing technologies to handle network tasks has been widely used. However, when facing the unique challenges of mobile edge networks, the existing methods have obvious limitations and shortcomings, as follows:
[0008] (1) High energy consumption: Centralized processing and data transmission in edge networks may result in excessive energy consumption. Since edge devices are usually powered by batteries, frequent data synchronization and processing consume a lot of energy, which is a severe challenge for energy-constrained edge devices. In addition, single-point processing in a centralized architecture may also lead to low energy efficiency.
[0009] (2) Staticity and lack of flexibility: Traditional centralized and distributed architectures usually rely on static node configurations and preset network models. In edge networks where the network environment changes rapidly, these architectures are often unable to quickly adapt to real-time changes in network status, such as the dynamic joining and exit of nodes, or the immediate fluctuation of network resources.
[0010] (3) Quality of Service (QoS) is difficult to guarantee: In a highly dynamic edge network environment, it becomes more difficult to maintain consistent service quality. Existing architectures are often difficult to predict and adapt to environmental changes, which leads to increased network latency and decreased data processing efficiency, and cannot meet the growing service quality requirements. In particular, in terms of resource allocation and task resource coordination, there is a lack of effective mechanisms to cope with instantaneous network fluctuations and node performance changes.
[0011] In summary, existing methods usually rely on static node configurations and preset network models, and cannot quickly and flexibly adapt to real-time changes in network status, further leading to problems such as low mobile edge network service quality.
[0012] It should be noted that this background technology is only used to introduce the relevant information of the present invention to help understand the technical solution of the present invention, but it does not mean that the relevant information is necessarily the prior art. If there is no evidence that the relevant information has been disclosed before the application date of the present invention, the relevant information shall not be regarded as the prior art.
[0013] References are as follows:
[0014] [1] Gong Yu. Multi-dimensional resource collaboration and intelligent optimization in mobile edge networks[Z]. Beijing University of Posts and Telecommunications, 2023.
[0015] [2] Li T, Qiu Z, Cao L, et al. Privacy-Preserving ParticipantGrouping for Mobile Social Sensing Over Edge Clouds [J]. IEEE Trans. Netw.Sci. Eng., 2021, 8(2): 865-880. Summary of the invention
[0016] Therefore, the purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a distributed node grouping method oriented to task collaboration.
[0017] The objective of the present invention is achieved through the following technical solutions:
[0018] According to a first aspect of the present invention, there is provided a distributed node grouping method for task collaboration, which is used to group multiple nodes for providing edge services in a distributed mobile edge network, each node representing an edge device, and the method is applied to each node, and the method comprises: S1, constructing an initial node grouping table for the current node, the node grouping table comprising the resource status of other discovered nodes; S2, when the current node receives a task, according to the latest node grouping table, utilizing the resources of other nodes to collaboratively execute the task to obtain a task execution result; S3, updating the node grouping table according to the task execution result, and the updating methods comprise: a periodic updating method, configured to update the node grouping table according to an update period, wherein the update period is dynamically adjusted according to the task execution result; and a task-driven updating method, configured to immediately update the node grouping table when the task execution result shows that the task execution has failed.
[0019] In some embodiments of the present invention, in step S3, the dynamic adjustment method includes: counting the number of task execution failures based on the task execution results of all tasks of the current node within a predetermined time period; determining a dynamic adjustment parameter for indicating the service quality of the mobile edge network based on the quantified value of the network status obtained by the current evaluation of the current node and the proportion of unfinished tasks of the node, wherein the proportion is the ratio of the number of task execution failures to the total number of all tasks; dynamically adjusting the update period according to the dynamic adjustment parameter and a preset threshold range, wherein: when the dynamic adjustment parameter is less than the lower limit threshold of the preset threshold range, reducing the update period; when the dynamic adjustment parameter is within the preset threshold range, keeping the update period unchanged; when the dynamic adjustment parameter is greater than the upper limit threshold of the preset threshold range, increasing the update period.
[0020] In some embodiments of the present invention, the dynamic adjustment parameter is determined as follows:
[0021] ,
[0022] in, Indicates dynamic adjustment parameters, represents the scale parameter, Indicates the current node at the current time The previous time period The number of task execution failures in internal statistics With this time period The total number of tasks within The ratio, represents the scale parameter, Indicates the quantitative value of the currently evaluated network status.
[0023] In some embodiments of the present invention, the node grouping table also includes the communication cost between the current node and other nodes. In step S3, the periodic update method includes: receiving updated node grouping tables sent by other nodes directly connected to the current node, and calculating the new communication cost between the current node and other nodes based on the node grouping tables of other nodes and the node grouping table of the current node; when the new communication cost is different from the communication cost between the current node and other nodes before the update, updating the communication cost between the current node and other nodes.
[0024] In some embodiments of the present invention, the resource status includes computing resources and storage resources, and the network status quantization value is calculated as follows:
[0025] ,
[0026] in, Indicates the quantitative value of network status. Indicated in The current node can be satisfied within the hop The probability of computing resource requirements of a task, Indicates the number of the current node. Indicates the maximum number of hops that meets the preset communication cost threshold. Indicates the number of hops. Indicated in The current node can be satisfied within the hop The probability of a task's storage resource requirement, Indicates the current node exist Hop within and except the current node The node connectivity rate of nodes in the network other than , represents the total number of nodes in the network, Indicates the current network except node Node collections other than Indicates that except the current node Networks outside nodes, Indicates that except the current node The number of a node in a network other than Indicates the current node and nodes connectivity.
[0027] In some embodiments of the present invention, in step S3, the task-driven update method includes: when the task execution result of the current node shows that the task execution failed and the task resource collaboration failed, generating a task error report, and updating the node grouping table of the current node according to the task error report; when there are no other nodes that can participate in resource collaboration in the node grouping table of the current node, initiating a new task resource collaboration request in the form of broadcasting, and updating the node grouping table according to the task reply sent by another node indicating that it can participate in resource collaboration.
[0028] In some embodiments of the present invention, in step S2, the method of utilizing the resources of other nodes to collaboratively execute the task includes: selecting a resource collaborative path for the task according to the latest node grouping table, wherein the resource collaborative path includes other nodes that can meet the resource requirements of the task and have the lowest communication cost; the current node unloads the task to other nodes on the corresponding resource collaborative path for collaborative execution to obtain the task execution result.
[0029] According to a second aspect of the present invention, there is provided a distributed node grouping system implemented based on the method according to the first aspect of the present invention, the system being configured in each of a plurality of nodes for providing edge services in a distributed mobile edge network, the system comprising:
[0030] The task collaborative execution module is used to utilize the resources of other nodes in the latest node grouping table to collaboratively execute the task when the current node receives the task, so as to obtain the task execution result; the dynamic adjuster is used to dynamically adjust the update cycle according to the task execution result; the periodic update module is used to build the initial node grouping table for the current node and update the node grouping table according to the update cycle; the task-driven update module is used to instantly update the node grouping table when the task execution result shows that the task execution failed.
[0031] According to a third aspect of the present invention, there is provided an electronic device comprising: one or more processors; and a memory, wherein the memory is used to store executable instructions; the one or more processors are configured to implement the steps of any one of the methods described in the first aspect of the present invention by executing the executable instructions.
[0032] Compared with the prior art, the advantages of the present invention are:
[0033] In the method of the present invention, on the one hand, the node grouping table constructed by the node includes the resource status of other nodes found, and the node grouping table is dynamically updated. When the node receives or generates a task, the latest node grouping table is used to process the task to ensure the collaborative work and resource sharing between nodes, thereby improving the quality of network services. On the other hand, the task-driven update method adopted is: when the task execution result shows that the task execution fails, the node grouping table is updated immediately so that the node grouping table can support the node's task resource collaboration needs in real time. The periodic update method adopted is: the node grouping table is updated according to the update cycle, and the update cycle is dynamically adjusted according to the situation of the task execution result, which can quickly and flexibly adapt to the real-time changes of the network status. The present invention not only ensures the effectiveness and reliability of the node grouping table, but also improves the success rate of task collaborative execution, while reducing energy consumption and further improving the quality of network services. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] The embodiments of the present invention are further described below with reference to the accompanying drawings, in which:
[0035] Figure 1 A schematic diagram of a flow chart of a distributed node grouping method for task collaboration according to an embodiment of the present invention;
[0036] Figure 2 A schematic diagram of the structure of a node grouping table according to an embodiment of the present invention;
[0037] Figure 3 A schematic diagram of an execution flow of a method for grouping edge nodes according to an embodiment of the present invention;
[0038] Figure 4 A schematic diagram of the interaction process of the node grouping method according to an embodiment of the present invention;
[0039] Figure 5 A schematic diagram showing a comparison of task completion rates between the existing method and the method of the present invention under different dynamic node proportions according to an embodiment of the present invention;
[0040] Figure 6 It is a schematic diagram showing the comparison results of task completion rate between the existing method and the method of the present invention when the number of nodes is different and the speed of each dynamic node is 20 m / s according to an embodiment of the present invention;
[0041] Figure 7 It is a schematic diagram showing the comparison results of task completion rate between the existing method and the method of the present invention when the number of nodes is different and the speed of each dynamic node is 40 m / s according to an embodiment of the present invention;
[0042] Figure 8 It is a schematic diagram showing the comparison results of task completion rate between the existing method and the method of the present invention under different task computing requirements according to an embodiment of the present invention;
[0043] Fig. 9 It is a schematic diagram showing the comparison results of task completion rates of the existing method and the method of the present invention at different average moving speeds of 15 nodes according to an embodiment of the present invention;
[0044] Fig.10 It is a schematic diagram showing the comparison results of task completion rates of the existing method and the method of the present invention at different average moving speeds for 30 nodes according to an embodiment of the present invention;
[0045] Fig.11 It is a schematic diagram of energy consumption comparison between the existing method and the method of the present invention when the number of nodes is different and the speed of each dynamic node is 20 m / s according to an embodiment of the present invention;
[0046] Fig.12 It is a schematic diagram of the energy consumption comparison results between the existing method and the method of the present invention when the number of nodes is different and the speed of each dynamic node is 40 m / s according to an embodiment of the present invention. DETAILED DESCRIPTION
[0047] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below through specific embodiments in conjunction with the accompanying drawings. 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.
[0048] As mentioned in the background technology section, existing methods usually rely on static node configurations and preset network models, and are unable to quickly and flexibly adapt to real-time changes in network status, further leading to problems such as low mobile edge network service quality.
[0049] Based on the above problems, the inventor proposes a distributed node grouping method for task collaboration, which is used to group multiple nodes for providing edge services in a distributed mobile edge network, and each node represents an edge device. The method of the present invention is applied to each node, that is, on the one hand, the node grouping table constructed by the node includes the resource status of other nodes found, and the node grouping table is dynamically updated to avoid relying on static node configuration and preset network models. When the node receives a task, the resources of other nodes in the latest node grouping table are used to collaboratively execute the task, ensuring collaborative work and resource sharing between edge nodes, and improving the service quality of the mobile edge network. On the other hand, in response to the problem that the network state will change dynamically, the present invention adopts a periodic update method and a task-driven update method to jointly update the node grouping table, wherein the task-driven update method is: when the task execution result shows that the task execution fails, the node grouping table is updated immediately, so that the node grouping table can support the node's task resource collaboration needs in real time. The periodic update method updates the node grouping table according to the update cycle, and dynamically adjusts the update cycle based on the network task execution status. Updating the node grouping table according to the update cycle can quickly and flexibly adapt to real-time changes in network status. It not only ensures the effectiveness and reliability of the node grouping table, but also improves the success rate of task collaborative execution. The low task failure rate reduces the number of task-driven updates, reduces network energy consumption, improves the service quality of mobile edge networks, and at the same time, improves the overall grouping service efficiency of the node grouping method.
[0050] According to one embodiment of the present invention, a distributed mobile edge network is a mobile edge network system including multiple edge devices, each of which is distributed in different locations, wherein an edge device refers to a computing device located in the edge network or a service node with computing capabilities, which is used to process and manage data and applications in an edge computing environment. In this network system, each edge device is regarded as a node, and the nodes are connected through a wireless network. Data can be transmitted between nodes, and edge network tasks can be collaboratively processed by sharing resources and location information.
[0051] According to an embodiment of the present invention, if the edge network system includes nodes, then use the collection Represents the network system, wherein, are the number of the first node, the number of the second node, and the The nodes are randomly distributed in the network system space. In this network system, some nodes are stationary and some are in motion. For example, drones, vehicles and mobile terminal devices with mobility are nodes in motion. The nodes in motion change the connection status between the nodes in the network. These changes will cause dynamic changes in the network topology, thereby affecting the information exchange and resource sharing between nodes. Therefore, it is necessary not only to delete the old nodes from the network topology and incorporate the new nodes into the network topology, but also to update the connection relationship between the nodes and the node resource information in a timely manner to realize the dynamic grouping of the nodes in the network.
[0052] According to one embodiment of the present invention, see Figure 1 , which is a flow chart of a distributed node grouping method for task collaboration, the method is applied to each node and includes the following steps S1, S2 and S3. In order to better understand the present invention, each step is described in detail below in conjunction with specific embodiments.
[0053] Step S1: construct an initial node grouping table for the current node, where the node grouping table includes resource status of other discovered nodes.
[0054] According to an embodiment of the present invention, the node grouping table (the node grouping table is denoted as ) to build the initial node grouping table. The resource status of other nodes recorded in the node grouping table includes computing resources and storage resources. In addition, the node grouping table also includes the communication cost between the current node and other nodes. Therefore, the node grouping table of each node maintains the grouping information in the network.
[0055] According to an embodiment of the present invention, the process of initializing the node grouping table by periodic updating includes: Fill in to ensure more accurate recording of each node's basic information and the information of its directly connected neighbor nodes. Among them, "neighbor nodes" refer to other directly connected nodes. For other nodes directly connected in the network , the current node and other nodes The physical distance between nodes is set as the communication cost. Among them, a shorter physical distance between nodes usually means low communication cost and fast data transmission speed, thereby ensuring efficient network communication. By linking the communication cost with the distance, it is possible to give priority to paths with lower communication costs for data transmission, thereby improving the communication efficiency and performance of the entire network. For non-directly connected nodes, their communication costs are set to infinity during initialization to indicate that these nodes in the network are unreachable under current conditions. That is, direct communication is not possible without a relay. The node grouping table of the present invention helps to avoid selecting currently inaccessible paths during resource allocation and resource collaborative decision-making, thereby optimizing network performance and improving the efficiency and accuracy of resource management and resource collaborative decision-making.
[0056] According to one embodiment of the present invention, in order to control the resource coordination process in the mobile edge computing scenario and achieve low energy consumption, the transmission power upper limit of the node is set to , and set the maximum communication cost threshold between nodes Among them, since the energy consumption of task transmission is directly related to the distance between nodes, and there are direct connections and indirect connections between nodes, when the distance between two nodes exceeds the threshold When the two parties communicate directly, they cannot communicate directly and need multi-hop relay communication.
[0057] According to one embodiment of the present invention, initially, when the node and nodes Direct connection and distance between Exceeding the threshold When , the communication cost will be set to infinity, as shown in the following formula:
[0058] , (1)
[0059] in, Representation Node and nodes The distance between Indicates the maximum distance between nodes. and nodes The distance between them exceeds the threshold When , the communication cost will be set to infinity, that is, ,in, Representation Node and nodes The communication cost between nodes and nodes The distance between Not exceeding the threshold When , the communication cost between the current node and the directly connected node is: . And the communication cost between a node and itself is 0.
[0060] According to one embodiment of the present invention, at the time of initialization, the communication cost between nodes that are not directly connected to the current node is unknown. When the periodic update method is periodically run, the node grouping table is periodically updated, and the communication cost between the current node and the nodes that are not directly connected is updated to be represented by the multi-hop communication cost. Therefore, for the nodes that are not directly connected, the communication cost is as follows:
[0061] , (2)
[0062] in, Representation Node and nodes The first Hop communication cost, Indicates the maximum number of hops. Needs to be satisfied , Representation Node and nodes The first The hop communication cost needs to meet the maximum communication cost threshold between nodes For example, nodes a and c are indirectly connected through relay node b, that is, the first-hop communication cost between nodes a and c is the distance between nodes a and b, and the second-hop communication cost is the distance between nodes b and c, and the distance between nodes a and b and the distance between nodes b and c are both less than or equal to the maximum communication cost threshold. According to one embodiment of the present invention, in the node grouping table, see Figure 2 , which is a schematic diagram of the structure of the node grouping table. Each row in the figure includes the resource status corresponding to another node and the network connection status between the current node and the other node. The first column is the number of each node, and the middle two columns represent the computing resources and storage resources of different nodes. The last column represents the communication cost, in which the value represents the connection status and communication cost between nodes. When the communication cost exceeds a certain threshold, the connection between the nodes will be disconnected. At this time, the communication cost is set to infinity. Schematically, as shown in Figure 2 As shown in the figure, the computing resources of nodes 1, 2, ..., n are 0, 5, ..., 6, respectively, in units of CU, the storage resources of nodes 1, 2, ..., n are 0, 2, ..., 3, respectively, in units of TB, and the communication cost between nodes 1, 2, ..., n and the current node is 21 meters, 220 meters, ..., infinity. Each item in the node grouping table is dynamic and will change with the dynamic environment of the network and task requirements.
[0063] Step S2: When the current node receives the task, it uses the resources of other nodes to collaboratively execute the task according to the latest node grouping table to obtain the task execution result.
[0064] According to one embodiment of the present invention, the method of using the resources of other nodes to collaboratively execute the task includes: selecting a resource collaborative path for the task according to the latest node grouping table, wherein the resource collaborative path includes other nodes that can meet the resource requirements of the task and have the lowest communication cost; the current node unloads the task to other nodes on the corresponding resource collaborative path for collaborative execution to obtain the task execution result. The technical solution of this embodiment can at least achieve the following beneficial technical effects: the communication cost mainly considers the energy consumption of task transmission, and the node will select the optimal resource collaborative path according to the resource status of other nodes in the node grouping table and the communication cost between the node and other nodes, that is, select the node that can meet the current task computing and storage resource requirements and has the lowest communication cost to collaboratively process the task, thereby improving task processing efficiency and energy utilization.
[0065] Step S3, updating the node grouping table according to the task execution results, the update methods include: a periodic update method, configured to: update the node grouping table according to an update period, wherein the update period is dynamically adjusted according to the task execution results; a task-driven update method, configured to: when the task execution results show that the task execution has failed, immediately update the node grouping table.
[0066] The periodic updating method and the task-driven updating method in step S3 are described below respectively.
[0067] 1. Periodic Update Method
[0068] According to an embodiment of the present invention, in the process of updating the node grouping table according to the update cycle, an adaptive adjustment mechanism is designed to dynamically adjust the update cycle, and the adjustment mechanism can dynamically adjust the execution cycle of the periodic update method according to the network status and task completion status. The method of the present invention can flexibly respond to changes in the network environment and optimize the node grouping process by adjusting the update cycle in real time, thereby improving the energy efficiency and service quality of the overall network.
[0069] According to an embodiment of the present invention, in the adaptive adjustment mechanism, the dynamic adjustment method of the update period includes steps a1, a2 and a3:
[0070] Step a1: according to the task execution results of all tasks of the current node within a predetermined time period, count the number of task execution failures.
[0071] According to an embodiment of the present invention, whether the task is successfully executed is determined based on the task execution result. For example, if the task execution result shows a failure message indicating that resource coordination failed, the task execution is determined to have failed. Since each task has a deadline, if the task is not completed before the deadline, the task execution result shows a timeout failure message, which means that the task execution has failed.
[0072] Step a2: Determine a dynamic adjustment parameter for indicating the service quality of the mobile edge network based on the quantified value of the network status currently evaluated by the current node and the proportion of unfinished tasks of the node, wherein the proportion is the ratio of the number of failed task executions to the total number of all tasks.
[0073] According to one embodiment of the present invention, the network status quantification value is calculated as follows:
[0074] , (3)
[0075] in, Indicates the quantitative value of network status. Indicated in The current node can be satisfied within the hop The probability of computing resource requirements of a task, Indicates the number of the current node. Indicates the maximum number of hops that meets the preset communication cost threshold. Indicates the number of hops. Indicated in The current node can be satisfied within the hop The probability of a task's storage resource requirement, Indicates the current node exist Jump inside and except the current node The node connectivity rate of nodes in the network other than , represents the total number of nodes in the network, Indicates the current network except node Node collections other than Indicates that except the current node Networks outside of nodes, Indicates that except the current node The number of a node in a network other than Indicates the current node and nodes The technical solution of this embodiment can at least achieve the following beneficial technical effects: in the edge network, the connectivity of the task is mainly considered, and the connectivity rate also includes whether the current network topology can meet the task offloading requirements, and whether the current network connection can meet the task's resource requirements. Therefore, the network status of the present invention is reflected by the probability index of meeting the task offloading requirements, thereby reflecting the overall service quality of the network and providing accurate data support for adjusting the execution cycle of the grouping method.
[0076] The calculation principle of formula (3) in the above embodiment is described below:
[0077] First, the connectivity rate can be used to measure the connectivity of the edge network. Therefore, the connectivity rate of the network It is expressed as follows:
[0078] , (4)
[0079] in, Represents the total number of nodes in the network, Representation Node and nodes In actual applications, when the node grouping table is initialized, the node includes its own unique identifier, i.e., MAC address, when broadcasting or replying. The total number of nodes in the edge network can be known by counting the unique identifier. Connectivity indicates whether two nodes are directly connected. The connectivity between nodes can be expressed as follows:
[0080] , (5)
[0081] Secondly, consider the connectivity of the task, that is, whether the current network topology can meet the task offloading requirements. Therefore, it is necessary to consider whether the node grouping table of the current node can meet the resource coordination requirements of the task to determine whether the current network connection status can solve the current task. The current node can be satisfied within the hop The probability of computing resource requirements of a task is expressed as follows:
[0082] , (6)
[0083] in, Indicates the current node A task is transferred to the The set of other nodes directly connected to the node at the time of hopping, Indicates other nodes of computing resources, Indicates the current node The computational resource requirements of a task. The set consists of 4 nodes, and There are 2 nodes in the collection that satisfy the current node The computing resource requirements of a task, then The jump time corresponds to the obtained The probability value of Sum the probability values of each hop in the hop to get .
[0084] exist The current node can be satisfied within the hop The probability of a task's storage resource requirement is expressed as follows:
[0085] , (7)
[0086] in, Indicates other nodes storage resources, Indicates the current node The storage resource requirement of a task, the probability value calculation method is the same as the principle of formula (7).
[0087] Then, according to the above formula (5), the connectivity rate of the current node is calculated as follows:
[0088] , (8)
[0089] According to formula (8), The node connectivity rate of the node where the hop is located can be expressed as follows:
[0090] , (9)
[0091] in, Indicates The number of nodes contained in the network within a hop, Indicates All nodes in the network after removing all hop nodes within the hop. For example, within the 3rd hop, the network contains 3 nodes, namely the 1st hop node 1, the 2nd hop node 2, and the 3rd hop node 3. Therefore, At this time, the value is 3. The value is 7 and includes nodes 1, 2, 3, 4, 5, 6, and 7. The set includes nodes 4, 5, 6, and 7 except for the 3 nodes in the network within the 3rd hop.
[0092] Finally, by combining equations (6), (7), and (8) in the above embodiment, the quantified value of the network status can be expressed as shown in equation (3).
[0093] According to an embodiment of the present invention, the dynamic adjustment parameter is determined as follows:
[0094] , (10)
[0095] in, Indicates dynamic adjustment parameters, represents the scale parameter, Indicates the current node at the current time The previous time period The number of task execution failures in internal statistics With this time period The total number of tasks within The ratio, represents the scale parameter, Indicates the quantitative value of the current network status evaluation. , The technical solution of this embodiment can at least achieve the following beneficial technical effects: the task completion status can reflect the network service quality more intuitively. At the same time, it can determine the dynamic adjustment parameters in combination with the quantified value of the network status, realize the intelligent adjustment update cycle, and can flexibly respond to changes in the network environment and optimize the node grouping process, thereby improving the energy efficiency and service quality of the overall network.
[0096] Step a3, dynamically adjust the update period according to the dynamic adjustment parameter and the preset threshold range, wherein: when the dynamic adjustment parameter is less than the lower limit threshold of the preset threshold range, reduce the update period; when the dynamic adjustment parameter is within the preset threshold range, keep the update period unchanged; when the dynamic adjustment parameter is greater than the upper limit threshold of the preset threshold range, increase the update period.
[0097] According to an embodiment of the present invention, the preset threshold range is set to [ ], the lower threshold is , the upper threshold is Using the dual threshold determination method, the update cycle is dynamically adjusted as follows:
[0098] Dynamic adjustment method 1: When dynamically adjusting parameters Less than the lower threshold When , which means that the current task has a high number of failures or the task completion rate corresponding to the current node grouping table is low, that is, the current node grouping table cannot support the edge network to efficiently process the current task, and the low task completion rate will also lead to an increase in the execution frequency of the task-driven update method. Therefore, it is necessary to reduce the update cycle of the periodic update method to support the edge network to efficiently process the current task and reduce node energy consumption.
[0099] Dynamic adjustment method 2: When the dynamic adjustment parameter is greater than the upper threshold, , indicating that the number of failures of the current task is small or the task network completion rate corresponding to the current node grouping table is high. The current grouping information is sufficient to support the edge network to efficiently process the current task, but the current periodic update method exceeds the energy consumption expectation. Therefore, it is necessary to increase the update cycle and reduce the energy consumption of the node to improve the energy efficiency of the edge network.
[0100] Dynamic adjustment method three: When the dynamic adjustment parameter is within the preset threshold range, , indicating that the current node grouping table can support the execution of edge network tasks, and the current periodic update method meets the energy consumption expectations. Therefore, the update cycle remains unchanged. The present invention not only ensures the efficiency of task execution and improves the service quality, but also improves the energy efficiency of the edge network and reduces the energy consumed by the edge network nodes due to the execution of the task-driven update method.
[0101] According to one embodiment of the present invention, the pseudo code flow of the dynamic adjustment method is as follows:
[0102] Input: Number of mission failures , total number of tasks, node grouping table, adjustment lower limit threshold , upper threshold
[0103] 1: Calculate the network status quantitative value ;
[0104] 2: The proportion of unfinished tasks of the computing node;
[0105] 3: Calculate dynamic adjustment parameters;
[0106] 4: if the dynamic adjustment parameter is less than the lower threshold then
[0107] 5: Output dynamic adjustment parameters And reduce the update cycle;
[0108] 6: else if the dynamic adjustment factor is greater than the upper threshold then
[0109] 7: Output dynamic adjustment parameters And increase the update cycle;
[0110] 8: else
[0111] 9: Output dynamic adjustment parameters And maintain the current update cycle unchanged;
[0112] 10: end if
[0113] 11: Output: Dynamically adjust parameters .
[0114] According to an embodiment of the present invention, the periodic update method includes: 1) initializing the node grouping table using the method of the above embodiment, 2) periodically updating the grouping table, 3) responding to the connection update of the neighbor, and 4) triggering the grouping table update. Among them:
[0115] 2) Periodic update of grouping table: Update the grouping table regularly By sending an updated node grouping table to each neighbor node and checking the response of the neighbor node, the node reachability changes can be discovered and handled in a timely manner. If a neighbor node does not respond within the specified time, the neighbor node will be considered unreachable and the grouping table will be updated accordingly to ensure the accuracy and latest grouping information. This method can cope with dynamic changes in the network, such as node movement, new node joining, or old node leaving.
[0116] 3) Respond to neighbor node connection updates: When the current node receives connection updates from neighbor nodes, receives the updated grouping table of neighbor nodes, and responds to the current node grouping table. Grouping table updated with neighbor nodes Update the current node grouping table , broadcast the updated grouping table to all neighboring nodes. This ensures that every node in the network can obtain the latest grouping table, which is crucial for maintaining the overall communication efficiency and resource allocation efficiency of the network. This mechanism supports the self-maintenance and self-optimization of the network, which is particularly important for the dynamically changing edge network environment.
[0117] 4) Trigger group table update method: in the node group table of the current node When a change occurs, the node grouping table after the change is updated is sent to all neighboring nodes to ensure that each node in the network has the latest grouping table.
[0118] According to one embodiment of the present invention, the current node receives updated node grouping tables sent by other nodes directly connected to the current node, and the current node traverses each item in the updated node grouping tables of other nodes and the node grouping table of the current node to calculate the new communication cost, and updates the node grouping table of the current node according to the new communication cost. Specifically, the new communication cost between the current node and the other nodes is calculated based on the node grouping tables of other nodes and the node grouping table of the current node; when the new communication cost is different from the communication cost between the current node and the other nodes before the update, the communication cost between the current node and the other nodes in the node grouping table of the current node is updated according to the size of the new communication cost. Among them, the process of updating the communication cost between the current node and other nodes in the node grouping table of the current node follows the following principles:
[0119] Principle 1: If the new communication cost is the distance between two directly connected nodes and is greater than the maximum communication cost threshold, it means that the current nodes cannot be directly connected to communicate. Therefore, it is necessary to update the grouping table, that is, delete the relevant information in the grouping table.
[0120] Principle 2: If the new communication cost is smaller than the communication cost between the current node and other nodes, it means that a smaller data transmission path has been found. Therefore, the grouping table needs to be updated, that is, the new communication cost is used as the communication cost between the current node and other nodes.
[0121] Principle 3: If there is no such item in the node grouping table of the node, it means that the two nodes were originally unconnectable but are now connectable. That is, the node grouping table of the current node shows that the current node does not have the communication cost information for connecting with the other node. Therefore, the grouping table needs to be updated, that is, the relevant communication cost information of the current node and the other node needs to be added to the grouping table.
[0122] According to one embodiment of the present invention, the pseudo code flow of the periodic update method is as follows:
[0123] Input: Current node grouping table , the grouping table updated by neighbor nodes
[0124] 1: Initialize the node grouping table ;
[0125] 2: Loop every fixed time interval (according to the update cycle);
[0126] 3: Send the updated node grouping table to each neighbor node ;
[0127] 4: Check the response of neighbor nodes;
[0128] 5: if the neighbor node does not respond do
[0129] 6: Consider the neighbor node to be unreachable and update the grouping table;
[0130] 7: end if
[0131] 8: while receiving connection updates from neighbor nodes do
[0132] 9: for each received neighbor node update group table do
[0133] 10: Calculate the new communication cost according to the above embodiment;
[0134] 11: If the new communication cost is less than the communication cost between the current node and other nodes or there is no such item in the grouping table, then
[0135] 12: Update the grouping table ;
[0136] 13:end if
[0137] 14: if the new communication cost is the distance between two directly connected nodes and the new communication cost ≥ then
[0138] 15: From Delete this item in;
[0139] 16:end if
[0140] 17: end for
[0141] 18: end while
[0142] 19: if node grouping table Changes then
[0143] 20: Send updates to all neighboring nodes;
[0144] 21: end if
[0145] 22: Output updated node grouping table .
[0146] In general, the periodic update method in the method of the present invention can periodically and actively update the network node grouping table, establish a basic network node grouping table, thereby providing a basis for resource coordination for task execution, maintaining basic node grouping information, and having a significant effect on the maintenance of static nodes and node information with stable resource levels. The periodic update method not only improves the efficiency of resource allocation, but also reduces the execution probability of the task-driven update method.
[0147] 2. Task-driven update method
[0148] According to one embodiment of the present invention, the task-driven update method includes: when the task execution result of the current node shows that the task execution has failed and the task resource coordination has failed, a task error report is generated, and the node grouping table of the current node is updated according to the task error report; when there are no other nodes that can participate in resource coordination in the node grouping table of the current node, a new task resource coordination request is initiated in the form of broadcast, and the node grouping table is updated according to the task reply sent by another node indicating that it can participate in resource coordination. The present invention updates the node grouping table on demand according to the execution status of specific tasks based on the task-driven update method. This ensures that resource allocation is closely matched with actual needs, and achieves flexible adaptation to changes in the network environment and optimizes the efficiency of resource coordination.
[0149] According to one embodiment of the present invention, the task-driven update method includes the following five parts:
[0150] 1) Processing resource coordination failure information:
[0151] In actual applications, resource coordination may fail for a variety of reasons, such as disconnection, target node overload, or other network problems. When resource coordination fails, a task error report is immediately generated. , record the cause of the failure and related information in detail, and use Information update group table. For example, task error report If the failure is caused by resource changes, the resource information in the grouping table should be updated; if the communication cost is too high, the node information should be deleted to avoid future attempts to use known failed paths. The message will be broadcast to the affected neighbor nodes so that they can also update their grouping information, ensuring that the nodes in the network can adjust their resource coordination paths in time to avoid repeated failures.
[0152] 2) Initiate a new task request:
[0153] When there is no node in the group table that can perform resource coordination, active measures will be taken to create and broadcast task requests. , to explore new available resource coordination paths to dynamically adapt to network changes. This approach allows the network to find new resource coordination opportunities when the current connection fails or resources are insufficient. If some nodes are already known to be available for resource coordination, it will give priority to these nodes to improve the efficiency of the resource coordination process.
[0154] 3) Respond to task requests:
[0155] When a node receives a task request from another node When a task is requested, the node first evaluates whether it is capable of executing the task request. If it is, the node will create and send a task reply. , directly inform the task request node that the resource collaborative task can be executed here. If the current node cannot execute, the task request Forward it to expand the search scope and increase the possibility of finding resource synergy targets.
[0156] 4) Forwarding task reply:
[0157] Received task reply The node needs to update its node group table to reflect the new connection status and resource coordination path information. If the node is the one that originally initiated the request, the task ends the process of searching for an executable node and resource coordination can begin. It is further forwarded to the original requesting node to ensure the successful resolution of the resource coordination request.
[0158] 5) Node grouping table maintenance:
[0159] Continuously monitor the information in the node grouping table to confirm whether any information in the grouping table is expired. Expired or no longer valid information will be cleared to ensure that the grouping table reflects the latest network status, thereby maintaining the efficiency and accuracy of the network.
[0160] According to one embodiment of the present invention, the pseudo code process based on the task-driven update method is as follows:
[0161] Input: resource coordination failure information, task request , task reply , task error , node grouping table ;
[0162] 1: while there is resource coordination failure information do
[0163] 2: Generate task error report;
[0164] 3: Update the grouping table based on the task error report;
[0165] 4: Broadcast task error report to affected neighbor nodes;
[0166] 5: if there is no node in the node grouping table that can perform resource coordination then
[0167] 6: Create and broadcast task request;
[0168] 7:else
[0169] 8: Use existing nodes that can collaborate on resources;
[0170] 9:end if
[0171] 10: end while
[0172] 11: while receiving task request do
[0173] 12: if the current task request can be executed then
[0174] 13: Create and send task reply;
[0175] 14:else
[0176] 15: forward task request;
[0177] 16:end if
[0178] 17: end while
[0179] 18: while receiving task reply do
[0180] 19: Update the grouping table based on task responses;
[0181] 20: if it is not the original task request node then
[0182] 21: forward the task reply to the original task request node;
[0183] 22:end if
[0184] 23: end while
[0185] 24: Remove expired information from the grouping table;
[0186] 25: Output updated grouping table .
[0187] According to an embodiment of the present invention, the overall process of the node grouping method is now described. Figure 3 , which is a schematic diagram of the execution flow of the edge node grouping method. The execution flow is as follows:
[0188] First, the user generates tasks: including generated tasks 1, 2, and 3, where tasks 1 and 3 are two tasks that can be executed in parallel, and task 2 cannot be executed in parallel with tasks 1 and 3. Task 2 is divided into two subtasks that can be executed in parallel, and task 21 and task 22 are obtained. The node obtains two tasks that can be executed in parallel, such as task 1 and task 3, or task 21 and task 22.
[0189] Secondly, select nodes: adopt resource coordination strategy to coordinate the processing of two tasks, including selecting nodes for each task according to the latest node grouping table, determining resource coordination path, and selecting nodes that can meet the resource requirements of the task with the lowest communication cost.
[0190] Then, attempt to transfer processing: offload the task to the selected node, including attempting to transfer processing the task according to the resource collaborative path to achieve collaborative processing of the task.
[0191] Again, determine whether the task is completed: When the task is completed, the selected node needs to return the task execution result to the source node that requested resource collaboration to determine whether the task is completed. In this process, the source node needs to wait for the response of the selected node to ensure the correct execution of the task. When the source node cannot receive the response of the selected node, the source node needs to reselect the node according to the latest grouping table and unload the task to the new node for processing until the task is fully processed and the processing is ended. However, due to the dynamic changes in the network topology, the selected node may change, which will cause the task execution result to fail, indicating that the task is not completed.
[0192] Finally, when the task is not completed, the task-driven update method is triggered to update the group table, and the source node can update the group table in time according to the dynamic changes of the network topology. At the same time, the dynamic adjustment method in the above embodiment is used to dynamically adjust the update cycle, and the group table is periodically updated according to the adjusted update cycle.
[0193] According to an embodiment of the present invention, based on the above single node grouping method execution process, when multiple nodes respectively execute the grouping method according to the above embodiment, the interaction between multiple nodes is described. Figure 4, which is a schematic diagram of the principle of the node grouping method interaction process. The figure includes nodes A, B and C. Each node selects a node for the task generated or received by it according to its own node grouping table, transmits the task to the selected node for processing, and judges whether the task is completed according to the response of the selected node. If it is not completed, the node needs to be reselected and the node grouping table is updated using the grouping method of the present invention. Among them, since there is an effective connection between node A and node B, an effective connection between node B and node C, and an invalid connection between node A and node C, "effective connection" means that the two nodes can exchange updated grouping table information with each other, complete the information maintenance of the grouping table, and can also transmit resource coordination data. "Effective connection" means that the connectivity between two directly connected nodes is 1, that is, the connected state. "Invalid connection" refers to a connection with too high a communication cost or too few remaining resources of the corresponding node, that is, no resource coordination data is transmitted between the two nodes of the "invalid connection". The present invention supports dynamic adjustment of the update cycle to ensure the update frequency of the information in the node grouping table to maintain the quality of resource coordination.
[0194] According to one embodiment of the present invention, there is provided a distributed node grouping system implemented based on the grouping method described in the above embodiment, the system being used to be configured in each of a plurality of nodes for providing edge services in a distributed mobile edge network, the system comprising: a task collaborative execution module, for utilizing the resources of other nodes in the latest node grouping table to collaboratively execute the task when the current node receives or generates a task, and obtaining a task execution result; a dynamic adjuster, for dynamically adjusting the update cycle according to the task execution result; a periodic update module, for constructing an initial node grouping table for the current node, and updating the node grouping table according to the update cycle; and a task-driven update module, for instantly updating the node grouping table when the task execution result shows that the task execution has failed.
[0195] According to an embodiment of the present invention, the periodic update module is used to execute the periodic update method in step S1 and step S3 in the above embodiment, the task collaborative execution module is used to execute step S2 in the above embodiment, and the task-driven update module is used to execute the task-driven update method in step S3 in the above embodiment. The dynamic adjuster is used to dynamically adjust the update period in the periodic update method in step S3.
[0196] According to one embodiment of the present invention, a main function of the dynamic adjuster is to monitor the number of task failures in real time and calculate the task completion rate of the network. This includes evaluating the completion of various tasks in the current network, thereby providing accurate data support for adjusting the grouping method. By continuously tracking the progress of task completion, the dynamic adjuster can accurately determine the overall service quality of the network, especially when dealing with complex or variable tasks, it can improve the task completion rate. Based on the monitoring of the task network completion rate, another main function of the dynamic adjuster is to dynamically adjust the update cycle of the periodic update method. When the task completion rate is lower than expected, the dynamic adjuster can guide the method to run at a more frequent cycle to quickly respond to changes in network status and increased task requirements, thereby ensuring that the task can be completed on time and improving the service quality of the overall network. On the contrary, when the task completion rate is high, the dynamic adjuster can reduce the execution frequency of the periodic update method, especially in a mobile edge network environment with limited node energy to reduce unnecessary resource consumption, thereby reducing energy consumption and optimizing resource utilization. This is not only conducive to improving energy efficiency, but also helps to extend the service life of the equipment and reduce the operation and maintenance costs of the entire network.
[0197] In order to verify the beneficial effects of the present invention, a simulation experiment is carried out on the present invention below.
[0198] First, an evaluation model is set for the simulation experiment of the method of the present invention, including a resource collaborative transmission energy consumption model and a delay model. The delay model is used to measure the completion time of the task, and then obtain a sign of whether the task has failed. The energy consumption model is used to measure the energy consumption of the collaborative execution of task resources. The two models are only used to evaluate the effect in the simulation experiment. In the actual system, the delay and energy consumption are obtained through actual test records.
[0199] According to an embodiment of the present invention, the energy consumption model not only considers the energy consumption of the node grouping method, but also considers the energy consumption of the task during the offloading transmission and result return process. Therefore, the energy consumption model can be expressed as follows:
[0200] , (11)
[0201] in, Represents the total energy consumption of the entire edge network task during the resource collaborative execution process. Represents the total energy consumption of the entire node grouping method.
[0202] According to one embodiment of the present invention, since the task computing energy consumption has no effect on the performance evaluation of the node grouping method, the computing energy consumption is not taken into consideration, that is, the energy consumption consists of two parts: one is the energy consumption during the unloading process, and the other is the energy consumption during the result return process. At the same time, considering the task unloading and execution for each hop, the model calculates its energy consumption separately, and then accumulates it to obtain the total task energy consumption. The total task energy consumption is expressed as follows:
[0203] , (12)
[0204] in, Indicates that the task is Energy consumption during the jump offloading process (i.e., the task is transferred to the first The transmission energy consumption of the node corresponding to the jump), Indicates that the task is Energy consumption during the jump result transmission process, and Respectively represent the tasks in The transmit power and receive power during the offloading and execution of the jump, Indicates the amount of data for the task, Indicates the amount of result data after task execution. Indicates The transfer rate of the hop. Indicates the maximum number of hops.
[0205] According to an embodiment of the present invention, when the node grouping method maintains and updates the node grouping table information, the amount of data transmitted between nodes is often much smaller than the amount of data in the task. Therefore, the energy consumption of the node grouping method is mainly related to the number of information interactions between nodes when maintaining and updating the node grouping table information. In the method of the present invention, the energy consumption of the node grouping method is mainly affected by the maintenance and update frequency of the node grouping table. Therefore, the energy consumption of the node grouping method can be expressed as follows:
[0206] , (13)
[0207] in, Represents the total energy consumption of the entire node grouping method. represents the energy consumption coefficient of the node grouping method, Indicates the number of node grouping method information exchanges. Number of node grouping method information exchanges Maintenance frequency of node grouping information There is a relationship between: ,in, represents the information interaction coefficient of the node grouping method, represents the execution time of the node grouping method, Indicates the update cycle of the periodic update method. It indicates the number of on-demand detections based on the task execution results in the task-driven update method. Indicates the number of nodes in the edge network. And each node in the edge network can process tasks, so Indicates the number of all nodes in the edge network. Indicates the number of reference nodes, which is a standard value. Since the types of edge devices in the edge network are different and have different functions, such as nodes such as gateways, or security nodes that implement security protocols, data encryption and access control, these nodes with special functions can be called reference nodes. The present invention links the energy consumption of the node grouping method with the maintenance frequency of the node grouping information, and then adjusts the energy consumption of the node grouping method by the maintenance frequency of the node grouping information.
[0208] According to one embodiment of the present invention, in an edge computing environment, the delay model for resource collaboration is as follows:
[0209] , (14)
[0210] in, Represents the total delay of the entire resource coordination process, Indicates the unloading delay, represents the computation delay, Indicates the result return delay. Among them, the transmission delay The time it takes to calculate the task data from the MD or source node to the selected node (i.e., the target node). The transmission delay of each hop depends on the amount of task data. and the transmission rate of that hop Therefore, the total offloading delay is the sum of the offloading delays in all hops. The time required for the target node to process the transmitted task is calculated, and the result is transmitted back. It is the time it takes for the task execution result to be transmitted from the target node to the source node. The result transmission delay depends on the amount of result data. and the transmission rate per hop Therefore, the total result return delay is the sum of the return delays in all hops. The entire delay model of the present invention integrates the transmission, processing and return process of task data in the edge network. Through this model, the key factors affecting resource coordination delay can be analyzed in simulation experiments, and the design and operation of the edge network can be optimized accordingly to reduce task processing time and improve overall efficiency. For example, by improving the processing power of edge nodes Or optimize the performance of computing resources , which can reduce the computational delay Similarly, increasing the transmission rate , or reduce the amount of task data and the resulting data volume , which can effectively reduce the delay of offloading and result return. This delay model has a key impact on the design of efficient edge computing strategies, especially in application scenarios with high real-time requirements.
[0211] Secondly, the simulation parameters in the simulation experiment are set as shown in Table 1:
[0212] Table 1: Simulation parameter settings
[0213] Simulation parameters Numeric unit Number of tasks 1000 indivual Dynamic node ratio 50 % Dynamic node speed 20-40 m / s Total number of nodes 30 indivual
[0214] Finally, simulation verification and comparison are performed:
[0215] In the simulation verification process, the existing method and the method of the present invention are selected to compare the performance when processing mobile edge network resource coordination. The method of the present invention (abbreviated as TADNG) is compared with two existing methods. The first existing method is a scheduled state update (SSU) grouping method, called an SSU aggregation algorithm, and the second existing method is a task-driven resource discovery (TRD) grouping method, called a TRD aggregation algorithm. The simulation process and verification comparison results are described below:
[0216] 1. Task generation and preparation: A set of tasks is randomly generated. These tasks can be split into multiple subtasks that can be executed in parallel. The subtasks are independent of each other. Each task requires a certain amount of calculation and data transmission.
[0217] 2. Develop a resource coordination path based on the node grouping table: When a node receives or generates a task, it selects the optimal resource coordination path based on the information in the grouping table. When the node completes resource coordination, it updates the information in the grouping table to reflect the results of resource coordination. The information in the grouping table is also used to update the topology of the network to ensure the connectivity and stability of the network. When the information in the grouping table changes, the node sends an update to the neighboring nodes directly connected to it to ensure that each node in the network has the latest grouping table. Schematically, after the user device determines the resource coordination decision plan based on the task situation and the grouping table, it can obtain a resource coordination plan. ,in, , , Respectively represent the first task, the second task, and the Each task corresponds to a resource coordination node.
[0218] 3. Execute resource coordination strategy: According to the resource coordination strategy, the task is transferred to the designated resource coordination node for coordinated execution and the node grouping information is updated according to the execution status of the task.
[0219] Verification and comparison results are as follows Figure 5-12 As shown. Among them, Figure 5-10 : This is a task completion rate comparison result diagram. The task completion rate is the ratio of the number of successfully completed tasks to the total number of tasks. Whether the task is successfully completed is determined by whether the delay calculated by the delay model formula (14) provided in the above embodiment exceeds the preset delay threshold. If the calculated delay is within the preset threshold, it means that the task has been successfully completed. Fig.11 and Fig.12 This is an energy consumption comparison result diagram. The energy consumption is calculated using the energy consumption model formula (11) provided in the above embodiment, and the unit of energy consumption is KJ.
[0220] Below Figure 5-10 Each comparison result is described in the figure:
[0221] Figure 5 The figure is a schematic diagram of the task completion rate comparison between the existing method and the method of the present invention under different dynamic node ratios. In the figure, the ordinate is the task completion rate, and the abscissa is the dynamic node ratio. It can be seen that as the number of dynamic nodes increases, the task completion rate of the method of the present invention remains at the highest state, indicating that the present invention can still maintain a relatively high task completion rate level under the condition of a high dynamic node ratio.
[0222] Figure 6 The figure is a schematic diagram of the task completion rate comparison between the existing method and the method of the present invention when the number of nodes is different and the speed of each dynamic node is 20 meters per second. In the figure, the ordinate is the task completion rate, and the abscissa is the number of nodes, in units. It can be seen that as the total number of nodes in the network increases, the task completion rate of the method of the present invention is the highest, indicating that the present invention can still maintain a relatively high level of task completion rate under different numbers of nodes.
[0223] Figure 7 The figure is a schematic diagram of the task completion rate comparison between the existing method and the method of the present invention when the number of nodes is different and the speed of each dynamic node is 40 meters per second. In the figure, the ordinate is the task completion rate, and the abscissa is the number of nodes, in units. It can be seen that when the nodes move at high speed, the task completion rate of the method of the present invention is the highest when the number of nodes is different, indicating that the present invention can always maintain a relatively high level of task completion rate when the nodes move at high speed.
[0224] Figure 8The figure is a schematic diagram of the task completion rate comparison between the existing method and the method of the present invention under different task computing requirements. The ordinate in the figure is the task completion rate, and the abscissa is the task computing requirement. It can be seen that as the task computing requirement in the network increases, the method of the present invention can maintain a relatively high task completion rate level.
[0225] Fig. 9 The figure is a schematic diagram of the task completion rate comparison between the existing method and the method of the present invention at different average moving speeds for 15 nodes. The ordinate in the figure is the task completion rate, and the abscissa is the average moving speed of the node, in m / s. When the number of nodes is 15, it can be seen that as the demand for task computing in the network increases, the method of the present invention can maintain a relatively high level of task completion rate.
[0226] Fig.10 The figure is a schematic diagram of the task completion rate comparison between the existing method and the method of the present invention at different average moving speeds for 30 nodes. In the figure, the ordinate is the task completion rate, and the abscissa is the average moving speed of the nodes, in m / s. It can be seen that as the average moving speed of the nodes in the network increases, the task completion rate of the existing method decreases significantly, while the method of the present invention can maintain a relatively high level of task completion rate.
[0227] comprehensive Figure 5-10 From the comparison results, it can be seen that compared with the traditional grouping method, the percentage of task completion rate improvement of the method of the present invention in scenarios such as mobility and increased task demand is 12.7%~36.7%, showing strong adaptability and robustness.
[0228] Below Fig.11 and Fig.12 Explain them separately:
[0229] Fig.11 The figure is a schematic diagram of the energy consumption comparison between the existing method and the method of the present invention when the number of nodes is different and the speed of each dynamic node is 20 m / s. In the figure, the ordinate is the task completion rate, and the abscissa is the number of nodes, in units. It can be seen that when the node moves at 20 m / s, the energy consumption of the method of the present invention is lower than that of the SSU aggregation algorithm, and when the number of nodes is large, the energy consumption is significantly lower than the two existing methods.
[0230] Fig.12 The figure is a schematic diagram of the energy consumption comparison between the existing method and the method of the present invention when the number of nodes is different and the speed of each dynamic node is 40 m / s. In the figure, the ordinate is the task completion rate, and the abscissa is the number of nodes, in units. It can be seen that when the nodes move at a higher speed of 40 m / s, the energy consumption of the method of the present invention is lower than that of the SSU aggregation algorithm. As the number of nodes increases, the energy consumption of the method of the present invention is significantly lower than that of the two existing methods.
[0231] comprehensive Figure 11-12 From the comparison results, it can be seen that compared with the traditional grouping method, the energy consumption of the method of the present invention is reduced by 10.15%~33.38%. In high-speed mobile scenarios and when there are a large number of nodes, the energy consumption of resource coordination can be significantly reduced, making the method of the present invention more competitive in performance.
[0232] It should be noted that although the above describes the various steps in a specific order, it does not mean that the various steps must be executed in the above specific order. In fact, some of these steps can be executed concurrently or even in a different order as long as the required functions can be achieved.
[0233] The present invention may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present invention.
[0234] A computer-readable storage medium may be a tangible device that holds and stores instructions used by an instruction execution device. Computer-readable storage media may include, for example, but are not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination thereof. More specific examples (non-exhaustive list) of computer-readable storage media include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a protruding structure in a groove on which instructions are stored, and any suitable combination thereof.
[0235] The embodiments of the present invention have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.
Claims
1. A distributed node grouping method for task collaboration, for grouping a plurality of nodes for providing edge services in a distributed mobile edge network, each node representing an edge device, the method being applied to each node, the method comprising: S1. Build an initial node grouping table for the current node, which includes the resource status of other nodes discovered; S2. When the current node receives a task, it uses the resources of other nodes to collaboratively execute the task according to the latest node grouping table to obtain the task execution result; S3. Update the node grouping table according to the task execution result. The updating method includes: The periodic update mode is configured to: update the node grouping table according to an update period, wherein the update period is dynamically adjusted according to the task execution result; The task-driven update method is configured to update the node grouping table immediately when the task execution result shows that the task execution fails.
2. The method according to claim 1, characterized in that In step S3, the dynamic adjustment method includes: According to the task execution results of all tasks of the current node within the scheduled time period, the number of task execution failures is counted; Determine a dynamic adjustment parameter for indicating the quality of service of the mobile edge network according to a quantified value of the network status obtained by the current node in the current evaluation and a proportion of unfinished tasks of the node, wherein the proportion is a ratio of the number of failed executions of the task to the total number of all tasks; The update cycle is dynamically adjusted according to the dynamic adjustment parameters and the preset threshold range, where: When the dynamic adjustment parameter is less than the lower limit threshold of the preset threshold range, reducing the update period; When the dynamic adjustment parameter is within the preset threshold range, the update cycle remains unchanged; When the dynamic adjustment parameter is greater than an upper threshold of a preset threshold range, the update period is increased.
3. The method according to claim 2, characterized in that The dynamic adjustment parameter is determined as follows: , in, Indicates dynamic adjustment parameters, represents the scale parameter, Indicates the current node at the current time The previous time period The number of task execution failures in internal statistics With this time period The total number of tasks within The ratio, represents the scale parameter, Indicates the quantitative value of the currently evaluated network status.
4. The method according to claim 2, characterized in that: The node grouping table also includes the communication cost between the current node and other nodes. In step S3, the periodic update method includes: Receive updated node grouping tables sent by other nodes directly connected to the current node, and calculate new communication costs between the current node and other nodes based on the node grouping tables of other nodes and the node grouping table of the current node; When the new communication cost is different from the communication cost between the current node and the other nodes before the update, the communication cost between the current node and the other nodes is updated.
5. The method according to claim 4, characterized in that The resource status includes computing resources and storage resources, and the network status quantization value is calculated as follows: , in, Indicates the quantitative value of network status. Indicated in The current node can be satisfied within the hop The probability of computing resource requirements of a task, Indicates the number of the current node. Indicates the maximum number of hops that meets the preset communication cost threshold. Indicates the number of hops. Indicated in The current node can be satisfied within the hop The probability of a task's storage resource requirement, Indicates the current node exist Hop within and except the current node The node connectivity rate of nodes in the network other than , represents the total number of nodes in the network, Indicates the current network except node Node collections other than Indicates that except the current node Networks outside nodes, Indicates that except the current node The number of a node in a network other than Indicates the current node and nodes connectivity.
6. The method according to claim 1, characterized in that In step S3, the task-driven updating method includes: When the task execution result of the current node shows that the task execution failed and the task resource coordination failed, a task error report is generated, and the node grouping table of the current node is updated according to the task error report; When there are no other nodes that can participate in resource collaboration in the node grouping table of the current node, a new task resource collaboration request is initiated by broadcasting, and the node grouping table is updated according to the task reply received from another node indicating that it can participate in resource collaboration.
7. The method according to claim 4, characterized in that In step S2, the method of utilizing resources of other nodes to collaboratively execute the task includes: Select a resource coordination path for the task according to the latest node grouping table, where the resource coordination path includes other nodes that can meet the resource requirements of the task and have the lowest communication cost; The current node offloads the task to other nodes on the corresponding resource collaborative path for collaborative execution to obtain the task execution result.
8. A distributed node grouping system implemented based on the method according to any one of claims 1 to 7, the system being configured in each of a plurality of nodes for providing edge services in a distributed mobile edge network, the system comprising: The task collaborative execution module is used to use the resources of other nodes in the latest node grouping table to collaboratively execute the task when the current node receives the task, and obtain the task execution result; A dynamic adjuster is used to dynamically adjust the update cycle according to the task execution results; The periodic update module is used to construct an initial node grouping table for the current node and update the node grouping table according to the update cycle; The task-driven update module is used to instantly update the node grouping table when the task execution result shows that the task execution has failed.
9. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and the computer program can be executed by a processor to implement the steps of the method according to any one of claims 1 to 7.
10. An electronic device, characterized in that: include: one or more processors; as well as A memory, wherein the memory is used to store executable instructions; The one or more processors are configured to implement the steps of the method of any one of claims 1 to 7 by executing the executable instructions.
Citation Information
Patent Citations
Coprocessing method for space-based network heterogeneous calculation power resources and storage medium
CN113271137A
Dynamic task scheduling method based on computing node resource condition
CN115048203A
Cloud and multi-edge network node collaborative micro-service deployment method
CN118337640A
Packet full life cycle-oriented decentralized security guarantee method and device
WO2021212851A1
