Low-delay communication system based on edge computing in distributed network

Abstract

The invention provides a low-delay communication system based on edge computing in a distributed network, which belongs to the field of electrical digital data processing, and comprises a node management module, a computing monitoring module, an intelligent prediction module and a resource allocation module, the calculation monitoring module is used for detecting the calculation state of an edge node, the intelligent prediction module is used for predicting the transmission condition of data, and the resource allocation module allocates communication bandwidth resources based on the prediction condition; according to the system, the transmission starting time of the task result is planned and analyzed, so that the time for the server node to obtain the edge calculation data can be effectively shortened.

Description

Technical Field

[0001] The present invention relates to the field of electrical digital data processing, and particularly to a low-latency communication system based on edge computing in a distributed network. Background Art

[0002] With the rapid development of emerging technologies such as artificial intelligence, distributed network architectures have been widely adopted in various application scenarios, especially in fields such as industrial control, autonomous driving, and cloud gaming where extremely high requirements are placed on the timeliness of data transmission. As a computing paradigm that extends computing power from the data center to the network edge, edge computing can sink some computing tasks to edge nodes close to the data source for processing. However, the processed results need to be transmitted back to the data center, so it is necessary to reduce the transmission latency to meet high-demand scenarios.

[0003] The foregoing discussion of the background art is only intended to facilitate an understanding of the present invention. This discussion does not recognize or admit that any of the materials mentioned are part of common general knowledge.

[0004] Many edge computing systems have now been developed. After a large amount of retrieval and reference, it is found that existing edge computing systems are like the system disclosed in CN115269342B. These systems generally include a cloud monitoring platform, an edge computing gateway, and a local monitoring platform; the cloud monitoring platform is used to obtain data actively uploaded by no less than one edge computing gateway; the edge computing gateway is used to obtain local data collected by the local monitoring platform; the local monitoring platform is used to obtain locally collected monitoring data; distributed computing is performed by executing the calculation closer to the data source. However, this system cannot transmit the data results back to the data center in a timely manner after using edge computing, and has a relatively high latency. Summary of the Invention

[0005] The object of the present invention is to propose a low-latency communication system based on edge computing in a distributed network for the existing deficiencies.

[0006] The present invention adopts the following technical solutions:

[0007] A low-latency communication system based on edge computing in a distributed network includes a node management module, a computing monitoring module, an intelligent prediction module, and a resource allocation module;

[0008] The node management module is used to manage node information, the computing monitoring module is used to detect the computing status of edge nodes, the intelligent prediction module is used to predict the data transmission situation, and the resource allocation module allocates communication bandwidth resources based on the prediction situation;

[0009] The node management module includes a node information configuration unit, a routing information configuration unit, and a node status monitoring unit. The node information configuration unit is used to store basic node information. The routing information configuration unit is used to configure the transmission path between two nodes. The node status monitoring unit is used to monitor the network status of the node;

[0010] The calculation monitoring module includes a task monitoring unit, a progress monitoring unit, and a volume monitoring unit. The task monitoring unit is used to monitor the task information of the edge node. The progress monitoring unit is used to monitor the progress information of each task. The volume monitoring unit is used to monitor the data volume information generated by the task;

[0011] The intelligent prediction module includes a resource management unit, a start prediction unit, and a bandwidth prediction unit. The resource management unit is used to manage the allocation of future bandwidth resources. The start prediction unit is used to predict the transmission start time of the task result. The bandwidth prediction unit is used to predict the bandwidth resource occupancy of the task result transmission;

[0012] The resource allocation module includes an activation monitoring unit, an allocation execution unit, and an information feedback unit. The activation monitoring unit is used to monitor time information and activate result transmission. The allocation execution unit is used to execute the transmission of the data result. The information feedback unit is used to feedback the actual transmission information to the intelligent prediction module.

[0013] Furthermore, the resource management unit includes a bandwidth window record processor, a window time shift processor, and a window update processor. The bandwidth window record processor is used to record the usage of bandwidth resources within a future period of time. The window time shift processor is used to shift the information recorded in the window according to time. The window update processor is used to add the transmission bandwidth occupancy information of new tasks to the window;

[0014] The start prediction unit includes a path preoccupation processor, a bandwidth reception processor, and a start calculation processor. The routing path preoccupation processor sets a transmission line in the corresponding window according to the transmission path information. The bandwidth reception processor is used to receive and manage the predicted bandwidth information. The start calculation processor is used to calculate the start time point of the transmission line.

[0015] Furthermore, the start calculation processor calculates the start progress point Ps of the transmission task result according to the following formula:

[0016]

[0017] where D is the transmission speed corresponding to the predicted bandwidth, Va is the total data volume of the task result, T1 is the time that the task has been running and calculating, and T2 is the remaining time to complete the task.

[0018] Furthermore, the bandwidth prediction unit includes a congestion assessment processor, a bandwidth calculation processor, and a bandwidth feedback processor. The congestion assessment processor is used to evaluate the congestion condition of the window content. The bandwidth calculation processor is used to calculate the bandwidth information occupied by the transmission line. The bandwidth feedback processor is used to feedback the predicted bandwidth information to the start prediction unit;

[0019] The congestion assessment processor calculates the window congestion degree Q according to the following formula:

[0020]

[0021] Where Wa is the total bandwidth of the window, Wu is the bandwidth already in use in the window, and n is the number of transmission lines in the window.

[0022] Furthermore, the bandwidth calculation processor calculates the bandwidth W occupied by the transmission line according to the following formula:

[0023]

[0024] Where m is the congestion level of the window.

[0025] The beneficial effects achieved by the present invention are:

[0026] By monitoring the computing tasks in the edge nodes, this system predicts the transmission status within the prediction window, makes advance planning for the result data transmission of each task, can effectively shorten the time difference between completing the computing task and transmitting the data result to the data center, and at the same time conducts predictive analysis on the bandwidth resources to avoid the occurrence of a large amount of idle resources.

[0027] To enable a further understanding of the features and technical content of the present invention, please refer to the following detailed description of the present invention and the attached drawings. However, the attached drawings are only provided for reference and illustration, and are not used to limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a schematic diagram of the overall structural framework of the present invention;

[0029] Figure 2 It is a schematic diagram of the composition of the node management module of the present invention;

[0030] Figure 3 It is a schematic diagram of the composition of the computing monitoring module of the present invention;

[0031] Figure 4 It is a schematic diagram of the composition of the intelligent prediction module of the present invention;

[0032] Figure 5 It is a schematic diagram of the composition of the resource allocation module of the present invention;

[0033] Figure 6 This is a comparison chart of the latency effects between the present invention and a general system. Detailed implementation manners

[0034] The following are specific embodiments to illustrate the implementation manners of the present invention. Those skilled in the art can understand the advantages and effects of the present invention from the content disclosed in this specification. The present invention can be implemented or applied through other different specific embodiments, and various details in this specification can also be modified and changed based on different viewpoints and applications without departing from the spirit of the present invention. Additionally, the drawings of the present invention are only for simple schematic illustration and are not drawn according to actual sizes, hereby declared. The following implementation manners will further detail the related technical content of the present invention, but the disclosed content is not used to limit the protection scope of the present invention.

[0035] Embodiment 1.

[0036] This embodiment provides a low-latency communication system based on edge computing in a distributed network, combined with Figure 1 , including a node management module, a computing monitoring module, an intelligent prediction module, and a resource allocation module;

[0037] The node management module is used to manage node information, the computing monitoring module is used to detect the computing status of edge nodes, the intelligent prediction module is used to predict the data transmission situation, and the resource allocation module allocates communication bandwidth resources based on the prediction situation;

[0038] The node management module includes a node information configuration unit, a routing information configuration unit, and a node status monitoring unit. The node information configuration unit is used to store basic node information, the routing information configuration unit is used to configure the transmission path between two nodes, and the node status monitoring unit is used to monitor the network status of the node;

[0039] The computing monitoring module includes a task monitoring unit, a progress monitoring unit, and a volume monitoring unit. The task monitoring unit is used to monitor the task information of edge nodes, the progress monitoring unit is used to monitor the progress information of each task, and the volume monitoring unit is used to monitor the data volume information generated by tasks;

[0040] The intelligent prediction module includes a resource management unit, a start prediction unit, and a bandwidth prediction unit. The resource management unit is used to manage the future allocation situation of bandwidth resources, the start prediction unit is used to predict the transmission start time of task results, and the bandwidth prediction unit is used to predict the bandwidth resource occupancy of task result transmissions;

[0041] The resource allocation module includes an activation monitoring unit, an allocation execution unit, and an information feedback unit. The activation monitoring unit is used to monitor time information and activate result transmission. The allocation execution unit is used to execute the transmission of data results. The information feedback unit is used to feedback the actual transmission information to the intelligent prediction module.

[0042] The resource management unit includes a bandwidth window recording processor, a window time shift processor, and a window update processor. The bandwidth window recording processor is used to record the usage of bandwidth resources within a future period of time. The window time shift processor is used to shift the information recorded in the window according to time. The window update processor is used to add the transmission bandwidth occupancy information of new tasks into the window.

[0043] The start prediction unit includes a path preemption processor, a bandwidth reception processor, and a start calculation processor. The routing path preemption processor sets a transmission line in the corresponding window according to the transmission path information. The bandwidth reception processor is used to receive and manage the predicted bandwidth information. The start calculation processor is used to calculate the start time point of the transmission line.

[0044] The start calculation processor calculates the start progress point Ps of the transmission task result according to the following formula:

[0045]

[0046] where D is the transmission speed corresponding to the predicted bandwidth, Va is the total data volume of the task result, T1 is the time that the task has been running and calculating, and T2 is the remaining time to complete the task.

[0047] The bandwidth prediction unit includes a congestion assessment processor, a bandwidth calculation processor, and a bandwidth feedback processor. The congestion assessment processor is used to evaluate the congestion situation of the window content. The bandwidth calculation processor is used to calculate the bandwidth information occupied by the transmission line. The bandwidth feedback processor is used to feedback the predicted bandwidth information to the start prediction unit.

[0048] The congestion assessment processor calculates the window congestion degree Q according to the following formula:

[0049]

[0050] where Wa is the total bandwidth of the window, Wu is the bandwidth already in use in the window, and n is the number of transmission lines in the window.

[0051] The bandwidth calculation processor calculates the bandwidth W occupied by the transmission line according to the following formula:

[0052]

[0053] where m is the congestion level of the window.

[0054] Example Two

[0055] This embodiment includes all the content of Embodiment One, and provides a low-latency communication system based on edge computing in a distributed network, including a node management module, a computing monitoring module, an intelligent prediction module, and a resource allocation module;

[0056] The node management module is used to manage node information, the computing monitoring module is used to detect the computing status of edge nodes, the intelligent prediction module is used to predict the data transmission situation, and the resource allocation module allocates communication bandwidth resources based on the prediction situation;

[0057] Combined with Figure 2 , the node management module includes a node information configuration unit, a routing information configuration unit, and a node status monitoring unit. The node information configuration unit is used to store basic node information, the routing information configuration unit is used to configure the transmission path between two nodes, and the node status monitoring unit is used to monitor the network status of the node;

[0058] Combined with Figure 3 , the computing monitoring module includes a task monitoring unit, a progress monitoring unit, and a volume monitoring unit. The task monitoring unit is used to monitor the task information of edge nodes, the progress monitoring unit is used to monitor the progress information of each task, and the volume monitoring unit is used to monitor the data volume information generated by the task;

[0059] Combined with Figure 4 , the intelligent prediction module includes a resource management unit, a start prediction unit, and a bandwidth prediction unit. The resource management unit is used to manage the future allocation of bandwidth resources, the start prediction unit is used to predict the transmission start time of task results, and the bandwidth prediction unit is used to predict the bandwidth resource occupancy of task result transmission;

[0060] Combined with Figure 5 , the resource allocation module includes an activation monitoring unit, an allocation execution unit, and an information feedback unit. The activation monitoring unit is used to monitor time information and activate result transmission, the allocation execution unit is used to execute the transmission of data results, and the information feedback unit is used to feedback the actual transmission information to the intelligent prediction module;

[0061] The node information configuration unit includes a node registrar, a meta-database, and a dynamic updater. The node registrar is responsible for the identity authentication and registration of new nodes, generating a unique node ID. The meta-database is used to store static information such as node hardware parameters and geographical information. The dynamic updater is used to update dynamic attributes such as node online status and load registration in real time;

[0062] The routing information configuration unit includes a topology network register, a routing task receiver, and a path configuration output processor. The topology network register is used to store the communication network relationships between nodes. The routing task receiver is used to receive routing tasks. The path configuration output processor is used to output the transmission path information of each routing task;

[0063] The node status monitoring unit includes a heartbeat detection processor, a bandwidth sampling processor, and a load balancing processor. The heartbeat detection processor is used to periodically send heartbeat packets to detect the survival status of nodes. The bandwidth sampling processor is used to measure the real-time available bandwidth and jitter information between nodes. The load balancing processor is used to count the CPU and memory usage rates of nodes;

[0064] The registered nodes include edge nodes, transmission nodes, and server nodes. The computing monitoring module only monitors edge nodes;

[0065] The task monitoring unit includes a task classification processor, a dependency resolver, and a task queue manager. The task classification processor is used to identify the task type and assign tags. The dependency resolver is used to identify the sequential dependencies between tasks. The task queue manager is used to control the execution order of the task queue;

[0066] The progress monitoring unit includes a computation amount analysis processor, a computation times supervision processor, and a progress calculation processor. The computation amount analysis processor is used to analyze the total computation amount of a task. The computation times supervision processor is used to supervise the real-time computation times during task execution. The progress calculation processor is used to calculate the remaining time to complete the task;

[0067] The progress calculation processor calculates the remaining time T2 of the task according to the following formula:

[0068]

[0069] where T1 is the time that the task has been running for computation, Na is the total computation amount, and Nr is the number of computations that have been executed;

[0070] The volume monitoring unit includes a transferred data register, a progress association recorder, and a volume prediction processor. The transferred data register is used to store the amount of data that needs to be transferred to other nodes. The progress association recorder is used to record the association information between the stored data amount and the corresponding progress. The volume prediction processor predicts the total amount of data that each task needs to transfer based on the association information;

[0071] The volume prediction processor calculates a ratio based on the association information, deletes the ratios with higher discreteness, calculates the mean b of the remaining ratios, and predicts the total data volume Va according to the following formula:

[0072] Va = b · 100;

[0073] The ratio specifically refers to the ratio of the stored data volume to the corresponding progress;

[0074] The resource management unit includes a bandwidth window recording processor, a window time shift processor, and a window update processor. The bandwidth window recording processor is used to record the usage of bandwidth resources within a period of time in the future. The window time shift processor is used to shift the information recorded in the window according to time. The window update processor is used to add the transmission bandwidth occupancy information of new tasks to the window;

[0075] The startup prediction unit includes a path pre - occupation processor, a bandwidth receiving processor, and a startup calculation processor. The routing path pre - occupation processor sets a transmission line in the corresponding window according to the transmission path information. The bandwidth receiving processor is used to receive and manage the predicted bandwidth information. The startup calculation processor is used to calculate the startup time point of the transmission line;

[0076] The startup calculation processor calculates the startup progress point Ps according to the following formula:

[0077]

[0078] where D is the transmission speed corresponding to the predicted bandwidth;

[0079] The calculation result of the task starts to be transmitted when the task progress reaches the startup progress point. Compared with transmitting the result data after the task is completed, it can transmit the data significantly earlier in time and basically complete the transmission of the result when the task calculation is completed, making the computing resources and communication resources in a coupled state;

[0080] The bandwidth prediction unit includes a congestion assessment processor, a bandwidth calculation processor, and a bandwidth feedback processor. The congestion assessment processor is used to evaluate the congestion situation of the window content. The bandwidth calculation processor is used to calculate the bandwidth information occupied by the transmission line. The bandwidth feedback processor is used to feedback the predicted bandwidth information to the startup prediction unit;

[0081] The congestion assessment processor calculates the window congestion degree Q according to the following formula:

[0082]

[0083] where Wa is the total bandwidth of the window, Wu is the bandwidth already in use in the window, and n is the number of transmission lines in the window;

[0084] The congestion assessment processor classifies the window congestion degree into different congestion levels;

[0085] For example, when the window congestion level is negative, the congestion level is 2, and when it is non - negative, the congestion level is 1. In fact, a more detailed classification method can be adopted.

[0086] The bandwidth calculation processor calculates the bandwidth W occupied by the transmission line according to the following formula:

[0087]

[0088] where m is the congestion level of the window;

[0089] Determining the predicted bandwidth based on the window congestion level can effectively prevent the idle of bandwidth resources or the situation of no bandwidth available for use;

[0090] The activation monitoring unit includes a clock synchronization processor, an event - driven trigger, and a transmission window evaluator. The clock synchronization processor is used to keep the time information of all network nodes the same. The event - driven trigger processor is used to trigger the transmission task when the time reaches the starting point. The transmission window evaluator is used to evaluate whether the current window performs the transmission task;

[0091] The allocation and execution unit includes a bandwidth allocation processor, a data source connection processor, and a transmission execution processor. The bandwidth allocation processor is used to allocate the actual transmission bandwidth resources. The data source connection processor is used to connect to the result data source generated by the computing task. The transmission execution processor is used to transmit the result data in the routing path;

[0092] The information feedback unit includes a deviation calculation processor, a deviation feedback processor, and an exception marking processor. The deviation calculation processor is used to calculate the deviation value between the actual transmission information and the predicted information. The deviation feedback processor is used to feedback the deviation information to the intelligent prediction module. The exception marking processor is used to mark the transmission exception situation and feedback it to the intelligent prediction module;

[0093] Part of the code information of this system is as follows:

[0094]

[0095]

[0096]

[0097]

[0098] Now, 10 computing tasks are used for testing, and the latency times in this system and the ordinary system are measured respectively, as Figure 6 shown.

[0099] The content disclosed above is only a preferred and feasible embodiment of the present invention, and does not limit the protection scope of the present invention. Therefore, all equivalent technical changes made by using the content of the specification and drawings of the present invention are included in the protection scope of the present invention. In addition, with the development of technology, the elements therein can be updated.

Claims

1. A low-latency communication system based on edge computing in a distributed network, characterized in that, It includes a node management module, a computing monitoring module, an intelligent prediction module, and a resource allocation module; The node management module is used to manage node information, the computing monitoring module is used to detect the computing status of edge nodes, the intelligent prediction module is used to predict the data transmission situation, and the resource allocation module allocates communication bandwidth resources based on the prediction situation; The node management module includes a node information configuration unit, a routing information configuration unit, and a node status monitoring unit. The node information configuration unit is used to store basic node information, the routing information configuration unit is used to configure the transmission path between two nodes, and the node status monitoring unit is used to monitor the network status of the node; The computing monitoring module includes a task monitoring unit, a progress monitoring unit, and a volume monitoring unit. The task monitoring unit is used to monitor the task information of edge nodes, the progress monitoring unit is used to monitor the progress information of each task, and the volume monitoring unit is used to monitor the data volume information generated by tasks; The intelligent prediction module includes a resource management unit, a start prediction unit, and a bandwidth prediction unit. The resource management unit is used to manage the future bandwidth resource allocation situation, the start prediction unit is used to predict the transmission start time of task results, and the bandwidth prediction unit is used to predict the bandwidth resource occupied by the transmission of task results; The resource allocation module includes an activation monitoring unit, an allocation execution unit, and an information feedback unit. The activation monitoring unit is used to monitor time information and activate result transmission, the allocation execution unit is used to execute the transmission of data results, and the information feedback unit is used to feedback the actual transmission information to the intelligent prediction module.

2. A low-latency communication system based on edge computing in a distributed network according to claim 1, characterized in that, The resource management unit includes a bandwidth window record processor, a window time shift processor, and a window update processor. The bandwidth window record processor is used to record the usage of bandwidth resources in a future period of time, the window time shift processor is used to shift the information recorded in the window according to time, and the window update processor is used to add the transmission bandwidth occupancy information of new tasks to the window; The start prediction unit includes a path preoccupation processor, a bandwidth reception processor, and a start calculation processor. The routing path preoccupation processor sets a transmission line in the corresponding window according to the transmission path information, the bandwidth reception processor is used to receive and manage the predicted bandwidth information, and the start calculation processor is used to calculate the start time point of the transmission line.

3. A low-latency communication system based on edge computing in a distributed network according to claim 2, characterized in that, The start calculation processor calculates the start progress point Ps of the transmission task result according to the following formula: Where D is the transmission speed corresponding to the predicted bandwidth, Va is the total data volume of the task result, T1 is the time that the task has been running and calculating, and T2 is the remaining time to complete the task.

4. A low-latency communication system based on edge computing in a distributed network according to claim 1, wherein, The bandwidth prediction unit includes a congestion assessment processor, a bandwidth calculation processor, and a bandwidth feedback processor. The congestion assessment processor is used to evaluate the congestion situation of the window content, the bandwidth calculation processor is used to calculate the bandwidth information occupied by the transmission line, and the bandwidth feedback processor is used to feedback the predicted bandwidth information to the start prediction unit; The congestion assessment processor calculates the window congestion degree Q according to the following formula: Among them, Wa is the total window bandwidth, Wu is the bandwidth already in use by the window, and n is the number of transmission lines within the window.

5. A low-latency communication system based on edge computing in a distributed network according to claim 4, characterized in that, The bandwidth calculation processor calculates the bandwidth W occupied by the transmission line according to the following formula: Among them, m is the congestion level of the window.

Images