A priority-free water-flow cloud-edge collaborative data offloading method

Abstract

This paper proposes a non-priority, water-flow-like, cloud-edge collaborative data offloading method, relating to the technical fields of data offloading and scheduling control. The method mounts nodes in a waiting queue requesting source data to nodes that have already requested the source data, forming multiple data offloading link groups and increasing the number of data being requested / offloaded at the same time. After the preceding node completes data offloading, it offloads data to subsequent nodes to reduce latency. This method can improve data offloading efficiency and user service quality in real-world environments.

Description

Technical Field

[0001] The present invention belongs to the technical field of data unloading and scheduling control, and in particular relates to a priority-free water flow-state cloud-edge collaborative data unloading method. Background Art

[0002] Data offloading through cloud-edge collaboration has been a hot topic in cloud computing and edge computing in recent years. With the increasing proportion of video data on the internet, traditional cloud computing has faced bottlenecks in ensuring user service quality. The emergence of edge computing has addressed this shortcoming. Cloud-edge collaboration will become a core technology for next-generation networks, and data offloading strategies for cloud-edge collaboration are a crucial component. Given limited network bandwidth, adopting data offloading strategies for cloud-edge collaboration can maximize user service quality while ensuring the highest possible user experience. Cloud computing has been widely researched in academia and industry due to its diverse service offerings, powerful resource pools, elastic and flexible applications, dynamic scalability, high performance availability, and managed services.

[0003] Faced with a massive number of internet users, cloud computing's load capacity is no longer sufficient for some real-time-sensitive services. Furthermore, the offload and upload of video streams places significant pressure on the network bandwidth of cloud data centers. Edge computing has emerged to address the data transmission latency and network bandwidth consumption issues faced by cloud computing. Edge computing deploys IaaS resources on the user side, with downlink data representing cloud services and uplink data representing IoT services. Because users are closer to edge computing nodes, receiving services from the edge significantly reduces data processing latency. This also reduces bandwidth consumption in cloud computing. As research on edge computing deepens, it is becoming clear that cloud-edge collaboration, as a new computing model, is emerging as a new research trend.

[0004] In traditional cloud computing service models, data centers primarily use strategies such as fair queuing and first-come, first-served (FCFS) to provide services to users. Fair queuing allocates equal bandwidth to each user based on the amount of data currently being accessed and bandwidth availability. First-come, first-served (FCFS) generates a service stack, providing services based on user time sequence and a first-in, first-out (FIFO) principle.

[0005] The two data offloading models described above are mature and proven in practice, and are widely used in current cloud networks. They have become the foundation of the Internet, but they still have certain limitations. When bandwidth is immutable, an increase in the number of users will eventually lead to congestion, preventing all users from requesting services. Furthermore, when the number of users served is small, service response times are long. Neither of these strategies is suitable for a cloud-edge collaborative network architecture. Currently, these data offloading models remain relatively traditional and inefficient, failing to leverage the advantages of a cloud-edge system architecture. Resource utilization is low, resulting in unnecessary operational overhead and a poor user experience. Summary of the Invention

[0006] To solve the above technical problems, the present invention proposes a non-priority water-flow cloud-edge collaborative data offloading method. In the method, source data located in a cloud computing center is offloaded via the source node of the cloud computing center to each edge node that requests the source data from the cloud computing center.

[0007] The method includes: step S1, determining M edge nodes directly mounted with the source node based on the bandwidth of the source node and the bandwidth of each edge node, and generating M node groups accordingly, wherein the M edge nodes are respectively the initial nodes of the M node groups; step S2, based on the data volume of the source data, the bandwidth of each edge node, the forwarding delay and the maximum delay of receiving the source data, allocating other edge nodes except the M edge nodes to the M node groups, and mounting them in sequence with the preceding nodes; step S3, for each node group, the source node first unloads the requested source data to the initial node, and then the initial node and the idle source node unload the source data to other mounted edge nodes.

[0008] The edge nodes that request the source data from the cloud computing center are queued in a waiting queue in the order of the time when the source data is requested.

[0009] In step S1, the number M of edge nodes directly connected to the source node is determined by calculating the ratio of the bandwidth of the source node to the bandwidth of each edge node and rounding the ratio down.

[0010] Among them, in the step S2: for the other edge nodes to be allocated, first, they are allocated to the M node groups in sequence according to their order in the waiting queue, and mounted in sequence with the preceding nodes; after all the other edge nodes are allocated and mounted, the delay of each of the other edge nodes receiving the source data is calculated, and the delay of each of the other edge nodes receiving the source data does not exceed the maximum delay.

[0011] In step S2, the delay of each of the other edge nodes receiving the source data is calculated in the following manner: the forwarding delay of the predecessor node of each mounted other edge node and the link delay of the link connecting the predecessor nodes are summed; wherein the link delay is the ratio of the data volume of the source data to the bandwidth of the link, and the bandwidth of the link is determined by the forwarding node bandwidth and the receiving node bandwidth.

[0012] In step S2, when the delay of each of the other edge nodes receiving the source data exceeds the maximum delay, the node mounting order is adjusted within the node group set. If the delay still exceeds the maximum delay after adjustment, the node allocation plan is adjusted between each of the node groups.

[0013] In step S3, when the source data starts to be unloaded to the initial node, the initial node immediately forwards the received source data to the next edge node mounted thereto; when the initial node receives all the source data, the source node releases the bandwidth between itself and the initial node, and the idle source node uses the released bandwidth to transmit the source data to other mounted edge nodes.

[0014] Wherein, in the step S3: when the initial node receives all the source data, the initial node releases the bandwidth for receiving the source data, and the initial node in the idle state uses the released bandwidth to transmit the source data to other mounted edge nodes.

[0015] Wherein, in the step S3: when any other mounted edge node receives all the source data, the any other mounted edge node releases the bandwidth for receiving the source data, and the any other mounted edge node in the idle state uses the released bandwidth to transmit the source data to other mounted edge nodes.

[0016] Wherein, in the step S3: for each other mounted edge node, there is no overlapping data between the source data received from the source node, the initial node and / or the other mounted edge nodes.

[0017] In summary, the technical solution provided by the present invention mounts the node requesting source data in the waiting queue to the node that has requested the source data, forming multiple data unloading link groups, increasing the number of requested / unloaded data at the same time; after the previous node finishes unloading data, it unloads data for the subsequent node to reduce latency, thereby improving the data unloading efficiency of efficient operation in a practical environment and improving user service quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 A schematic diagram of node grouping according to an embodiment of the present invention;

[0020] Figure 2 Schematic diagram of data offloading according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0022] The present invention proposes a non-priority water-flow cloud-edge collaborative data offloading method, in which source data located in a cloud computing center is offloaded via a source node in the cloud computing center to each edge node that requests the source data from the cloud computing center.

[0023] The method includes: step S1, determining M edge nodes directly mounted with the source node based on the bandwidth of the source node and the bandwidth of each edge node, and generating M node groups accordingly, wherein the M edge nodes are respectively the initial nodes of the M node groups; step S2, based on the data volume of the source data, the bandwidth of each edge node, the forwarding delay and the maximum delay of receiving the source data, allocating other edge nodes except the M edge nodes to the M node groups, and mounting them in sequence with the preceding nodes; step S3, for each node group, the source node first unloads the requested source data to the initial node, and then the initial node and the idle source node unload the source data to other mounted edge nodes.

[0024] In some embodiments, users are grouped into Figure 1As shown in Figure 1, access nodes are queued and numbered in the order of request time. The bandwidth of the source node is allocated according to the node ID and the actual bandwidth demand until the source node has no remaining bandwidth. The subsequent nodes will be mounted on the previous node within the delay range allowed by the user, forming multiple links. The data forwarding between groups will not affect each other. Then data unloading (as shown in Figure 1) is performed. Figure 2 (As shown in the figure). Data is transmitted along the link, flowing through the nodes within the group like water, until it reaches the last node. When the first node receives all the data, it and the source node jointly offload the data to subsequent nodes until the data offloading task is completed.

[0025] In some embodiments, each edge node that requests the source data from the cloud computing center is queued in a waiting queue in the order of the time when the source data is requested.

[0026] In some embodiments, in step S1, the number M of edge nodes directly mounted to the source node is determined by calculating the ratio of the bandwidth of the source node to the bandwidth of each edge node and rounding down the ratio.

[0027] In some embodiments, in step S2: for other edge nodes to be allocated, they are first allocated to the M node groups in sequence according to their order in the waiting queue, and mounted in sequence with the preceding nodes; after all the other edge nodes are allocated and mounted, the delay for each of the other edge nodes to receive the source data is calculated, and the delay for each of the other edge nodes to receive the source data does not exceed the maximum delay.

[0028] In some embodiments, in step S2, the delay of each of the other edge nodes receiving the source data is calculated in the following manner: summing the forwarding delay of the predecessor node of each mounted other edge node and the link delay of the link connecting the predecessor nodes; wherein the link delay is the ratio between the data volume of the source data and the bandwidth of the link, and the bandwidth of the link is determined by the forwarding node bandwidth and the receiving node bandwidth.

[0029] In some embodiments, in step S2, when the delay for each of the other edge nodes to receive the source data exceeds the maximum delay, the node mounting order is adjusted within the node group set. If the delay still exceeds the maximum delay after adjustment, the node allocation scheme is adjusted between each of the node groups.

[0030] In some embodiments, the primary purpose of node grouping is to increase the number of nodes, allowing more users to receive data simultaneously. The implementation process is as follows: The source node sorts and numbers the access nodes in chronological order. The source node's outbound bandwidth is allocated to the access nodes in numerical order. Nodes not requesting bandwidth are attached to these nodes, forming separate groups of data links. The number of nodes on each link is determined by the tolerance time: the delay in data reaching the last node on each packet link cannot exceed the allowed time.

[0031] like Figure 1 As shown, C is the cloud computing center (source data nodes and data sources are located in C), N i (1≤i≤K, K is the number of edge nodes) represents the edge computing node or user requesting services from the data source. Assuming that the data export bandwidth is 80MB / s, if there are 50 users requesting the source at a certain moment, the source will group {N 1, N2,N3,....,N 50}. The actual bandwidth from each node to the data source is {B1,B2,....,B 50}, which is 20MB / s. The data source will provide services in the order of numbering. According to the above assumptions, only nodes N1, N2, N3 and N4 can directly request data from the data source. N5 and subsequent nodes need to obtain data through the forwarding of the first four nodes. This will generate a link from N1 to N5 to N6, such as Figure 1 This is shown in the first link on the left. The number of nodes on this link depends on the delay time for data to reach N6, that is, the delay for data to reach N6 cannot exceed the data tolerance delay. Assuming that all link delays are the same as the node forwarding delay, the calculation can be simplified to:

[0032] nT r +(n-1)T f =T d

[0033] Where T r T is the delay time for the node to forward data. f is the link delay of data in the network, T d is the latency threshold that users can tolerate. Beyond this threshold, the data becomes worthless to the user. n represents the maximum number of nodes that can be included in a group. This calculation process determines the number of groups and the number of nodes within each group.

[0034] In some embodiments, in step S3: after the source data starts to be unloaded to the initial node, the initial node immediately forwards the received source data to the next edge node mounted thereto; when the initial node receives all the source data, the source node releases the bandwidth between itself and the initial node, and the idle source node uses the released bandwidth to transmit the source data to other mounted edge nodes.

[0035] In some embodiments, in step S3: when the initial node receives all the source data, the initial node releases the bandwidth for receiving the source data, and the initial node in an idle state uses the released bandwidth to transmit the source data to other mounted edge nodes.

[0036] In some embodiments, in step S3: when any other mounted edge node receives all the source data, any other mounted edge node releases the bandwidth for receiving the source data, and any other mounted edge node in an idle state uses the released bandwidth to transmit the source data to other mounted edge nodes.

[0037] In some embodiments, in step S3 : for each other mounted edge node, there is no overlapping data between the source data received from the source node, the initial node and / or the other mounted edge nodes.

[0038] In some embodiments, data offloading can also include the following two steps: First, data is forwarded to the last node in the order of grouping results. Second, multi-node collaborative data offloading is performed. This requires all nodes to be selfless and willing to accept control and scheduling from the data source.

[0039] Assume that the nodes in the group are {N1, N2, N3}. The priority of N1 is greater than N2, and N2 is greater than N3. At the beginning of data unloading, the data source unloads the data to N1, which immediately forwards the received data to N2 and N3. Figure 2 Left. When the data source completes the data offloading task to N1, the data source releases the bandwidth of N1 and transmits the data to N2. N2 has both the data source and N1 that sends data to it. Figure 2 As shown in the middle. Like N2, N3 has more nodes working together to offload data for it, see Figure 2 On the right, when node N1 receives all the data, a new problem arises: how to allocate the amount of data sent by higher-priority nodes. This is calculated using the following formula.

[0040]

[0041] In the above formula, Bi,j represents the bandwidth between node i and node j in the group, B j,C represents the bandwidth between node j and cloud computing center C, S represents the size of data to be unloaded, and D i The table is the size of the data that the i-th node is responsible for unloading during the collaborative data unloading process.

[0042] As can be seen, the above method mounts users in the service queue who have not yet requested service to nodes that have already requested service, forming multiple data offload link groups, greatly increasing the number of user requests at a time. As the requests of the preceding nodes complete, they jointly offload data to the subsequent nodes, thereby reducing latency. This non-priority data offloading method can effectively operate in real-world environments, improving data center data offloading efficiency and improving user service quality.

[0043] Please note that the various technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification. The above-mentioned embodiments only express several implementation methods of the present application. The description is relatively specific and detailed, but it cannot be understood as a limitation on the scope of the invention patent. It should be pointed out that for ordinary technicians in this field, without departing from the concept of this application, several variations and improvements can be made, which all fall within the scope of protection of this application. Therefore, the scope of protection of the patent in this application shall be based on the attached claims.

Claims

1. A non-priority water flow cloud-edge collaborative data offloading method, characterized in that: In the method: The source nodes of the cloud computing center unload the source data located in the cloud computing center to the edge nodes that request the source data from the cloud computing center; wherein the edge nodes that request the source data from the cloud computing center are queued in a waiting queue in the order of the time when the source data was requested; The method comprises: Step S1: Determine M edge nodes directly connected to the source node based on the bandwidth of the source node and the bandwidth of each edge node, and generate M node groups accordingly, where the M edge nodes are initial nodes of the M node groups respectively; Step S2: Based on the data volume of the source data, the bandwidth of each edge node, the forwarding delay, and the maximum delay of receiving the source data, allocating the edge nodes other than the M edge nodes to the M node groups and mounting them in sequence with the preceding nodes; Wherein, in said step S2: For other edge nodes to be allocated, first allocate them to the M node groups in order according to their order in the waiting queue, and mount them in sequence with the preceding nodes; After all the other edge nodes have completed allocation and mounting, calculating the delay of each of the other edge nodes receiving the source data, and the delay of each of the other edge nodes receiving the source data does not exceed the maximum delay; Step S3: For each node group, the source node first unloads the requested source data to the initial node, and then the initial node and the idle source nodes unload the source data to other mounted edge nodes.

2. The method for non-priority water flow cloud-edge collaborative data offloading according to claim 1 is characterized in that: In step S1, the number M of edge nodes directly connected to the source node is determined by calculating the ratio of the bandwidth of the source node to the bandwidth of each edge node and rounding the ratio down.

3. The non-priority water flow cloud-edge collaborative data offloading method according to claim 2 is characterized in that: In step S2, the time delay for each of the other edge nodes to receive the source data is calculated in the following manner: Summing the forwarding delay of the predecessor node of each mounted other edge node and the link delay of the link connecting the predecessor nodes; The link delay is the ratio between the data volume of the source data and the bandwidth of the link, and the bandwidth of the link is determined by the forwarding node bandwidth and the receiving node bandwidth.

4. The method for cloud-edge collaborative data offloading without priority in a water flow state according to claim 3 is characterized in that: In step S2, when the delay of each of the other edge nodes receiving the source data exceeds the maximum delay, the node mounting order is adjusted within the node group set. If the delay still exceeds the maximum delay after adjustment, the node allocation plan is adjusted between each of the node groups.

5. The method for cloud-edge collaborative data offloading without priority in a water flow state according to claim 4 is characterized in that: In step S3: When the source data starts to be unloaded to the initial node, the initial node immediately forwards the received source data to the next edge node mounted thereto; When the initial node receives all the source data, the source node releases the bandwidth between itself and the initial node, and the idle source node uses the released bandwidth to transmit the source data to other mounted edge nodes.

6. A non-priority water flow cloud-edge collaborative data offloading method according to claim 5, characterized in that: In step S3: when the initial node receives all the source data, the initial node releases the bandwidth for receiving the source data, and the initial node in an idle state uses the released bandwidth to transmit the source data to other mounted edge nodes.

7. The non-priority water flow cloud-edge collaborative data offloading method according to claim 6 is characterized in that: In step S3: when any other mounted edge node receives all the source data, the any other mounted edge node releases the bandwidth for receiving the source data, and the any other mounted edge node in an idle state uses the released bandwidth to transmit the source data to other mounted edge nodes.

8. The non-priority water flow cloud-edge collaborative data offloading method according to claim 6 is characterized in that: In step S3: for each other mounted edge node, there is no overlapping data between the source data received from the source node, the initial node and / or the other mounted edge nodes.

Images