A Network Slice Deployment Method Based on Joint Optimization of Resource Scheduling and Adaptation

Abstract

A network slice deployment method based on the joint optimization of resource scheduling and adaptation belongs to the technical field of network slice deployment, and solves the problem of low efficiency of network slice resource scheduling and deployment in the prior art. The method includes: for each network slice in the network slice set, respectively calculating the importance degrees of each physical node in the physical network and each virtual node in the network slice based on the multi-attribute decision-making method, and pre-deploying the network slice based on the importance degrees of each node; jointly optimizing the mapping parameters in the pre-deployment with the maximization of the resource efficiency of the network slice set as the optimization goal to obtain the optimal mapping parameters; for each network slice in the network slice set, respectively recalculating the importance degrees of each physical node in the physical network and each virtual node in the network slice based on the multi-attribute decision-making method according to the optimal mapping parameters, and deploying the network slice based on the importance degrees of each node.

Description

Technical Field

[0001] The present invention relates to the technical field of network slice deployment, and particularly to a network slice deployment method based on joint optimization of resource scheduling and adaptation. Background Art

[0002] The transmission of ultra-high-definition video service data not only has to solve problems such as the diversity of heterogeneous service requirements and the diversity of broadcast / multicast transmission architectures, but also has to face the diversity and dynamics of the underlying basic network, which are manifested as the dynamics and volatility in time, as well as the multi-network and multi-domain characteristics in space. Network slicing is a key technical means to support video transmission and distribution by slicing the infrastructure physical network into dedicated virtual subnets. However, with the increase in the dimension of physical network resources and the number of users / services, how to deploy network slices efficiently and balancedly under the condition of sharing limited physical resources has become the main problem in multi-domain networks.

[0003] In an actual communication network, the video transmission performance will be affected by various resources. Differentiated service requirements result in certain differences in the resource requirements of network slices for different domains. Therefore, end-to-end network slices need to be service-oriented and serve specific services. An efficient network slice deployment strategy is a relatively urgent need at present. Considering the constantly changing business communication requirements and the fragmentation of network resources, the previous methods cannot achieve the overall resource allocation of end-to-end slices, and their feasibility in actual scenarios is limited. In addition, most network slice deployment strategies are relatively independent for node and link deployment. If the link mapping causes slice failure, then the node mapping process will be wasted and consumed, resulting in a reduction in slice efficiency and an increase in delay overhead. There is great room for improvement in the existing network slice resource scheduling and deployment efficiency. Summary of the Invention

[0004] In view of the above analysis, an embodiment of the present invention aims to provide a network slice deployment method based on joint optimization of resource scheduling and adaptation to solve the problem of low efficiency of existing network slice resource scheduling and deployment.

[0005] On the one hand, an embodiment of the present invention provides a network slice deployment method based on joint optimization of resource scheduling and adaptation, including the following steps:

[0006] For each network slice in the network slice set, calculate the importance of each physical node in the physical network and each virtual node in the network slice respectively based on the multi-attribute decision-making method, and pre-deploy the network slice based on the importance of each node;

[0007] Jointly optimize the mapping parameters in the pre-deployment with the goal of maximizing the resource efficiency of the network slice set to obtain the optimal mapping parameters;

[0008] For each network slice in the network slice set, according to the optimal mapping parameters, the importance degrees of each physical node in the physical network and each virtual node in the network slice are recalculated respectively based on the multi-attribute decision-making method, and the network slice is deployed based on the importance degree of each node.

[0009] Based on a further improvement of the above technical solution, the optimization objective function is:

[0010]

[0011]

[0012] Wherein, represents the current remaining resources of the physical node R T represents the total resources of the nodes in the basic physical network, represents the current remaining resources of the physical link B T represents the total link bandwidth resources of the basic physical network, represents the resource requirement of the i-th node in the V-th network slice in the network slice set N represents the resource requirement of the virtual link E V represents the set of virtual nodes in the V-th network slice, V represents the adjacency matrix of the connection relationship between virtual nodes in the V-th network slice, represents the virtual node and in the V-th network slice P represents the set of physical nodes, P represents the adjacency matrix of the connection relationship between physical nodes, (z) represents the set of mapping parameter variables fitted by the z-th search, M represents the number of mapping parameters, PSO (h (z) ) represents the network slice deployment resource efficiency measured by the z-th search fitting, NSR represents the number of network slices in the network slice set; V represents the pre-deployment mapping variable, with a value of 1 if the V-th network slice is successfully deployed, otherwise 0.

[0013] Furthermore, pre-deploying the network slice based on the importance degree of each virtual node includes:

[0014] Taking the virtual node with the highest importance degree in the network slice as the root node, and constructing a search tree for the network slice based on the breadth-first algorithm;

[0015] For each virtual node in the search tree, construct a set of final candidate physical nodes for the current virtual node based on the autonomous domain type and resource requirements of the current virtual node;

[0016] If the current virtual node is the root node of the search tree, select the physical node with the highest importance in the set of final candidate physical nodes of the current virtual node as the mapping node of the current virtual node;

[0017] If the current virtual node is a non-root node of the search tree, use the physical node mapped by the parent virtual node of the current virtual node as the parent physical node, and use the virtual link between the current virtual node and the parent virtual node of the current virtual node as the target virtual link. Select a candidate physical node whose path to the parent physical node meets the resource requirements of the target virtual link from the set of final candidate physical nodes as the mapping node of the current virtual node.

[0018] Furthermore, for each virtual node in the search tree, constructing a set of final candidate physical nodes for the current virtual node based on the autonomous domain type and resource requirements of the current virtual node includes:

[0019] Determine the set of initial candidate physical nodes for the current virtual node based on the autonomous domain type of the current virtual node;

[0020] If the current virtual node is the root node of the search tree, the physical nodes in the set of initial candidate physical nodes that meet the resource requirements of the current virtual node form the set of final candidate physical nodes of the current virtual node;

[0021] If the current virtual node is a non-root node of the search tree, use the physical node mapped by the parent virtual node of the current virtual node as the parent physical node. The physical nodes in the set of initial candidate physical nodes that meet the resource requirements of the current virtual node and are neighbor nodes of the parent physical node form the set of final candidate physical nodes of the current virtual node.

[0022] Furthermore, the following method is used to select a candidate physical node whose path to the parent physical node meets the resource requirements of the target virtual link from the set of final candidate physical nodes as the mapping node of the current virtual node:

[0023] S141. Sort the candidate physical nodes in the set of final candidate physical nodes in ascending order of node load; use the first candidate physical node in the sorted set of final candidate physical nodes as the current candidate physical node;

[0024] S142. Determine whether there is a path that meets the resource requirements of the target virtual link in the physical path from the parent physical node to the current candidate physical node. If there is, map the current virtual node to the current candidate physical node, and map the target virtual link to the shortest path that meets the resource requirements of the target virtual link; otherwise, use the next candidate physical node in the final candidate physical node set as the current candidate physical node; return to step S142;

[0025] S143. If there is no physical path that meets the resource requirements of the target virtual link from the parent physical node to each candidate physical node in the final candidate physical node set, then there is no candidate physical node in the final candidate physical node set whose path to the parent physical node meets the resource requirements of the target virtual link.

[0026] Further, the following method is used to determine whether there is a path that meets the resource requirements of the target virtual link in the physical path from the parent physical node to the current candidate physical node:

[0027] Search for K shortest paths from the parent physical node to the current candidate physical node based on the Dijkstra algorithm;

[0028] Traverse each of the K shortest paths in ascending order of path cost. If the current shortest path meets the SLA constraint, then the current shortest path is the shortest path that meets the resource requirements of the target virtual link, and the traversal ends; otherwise, continue to traverse the next shortest path;

[0029] If none of the K shortest paths meet the SLA constraint, then there is no path that meets the resource requirements of the target virtual link in the physical path from the parent physical node to the current candidate physical node.

[0030] Further, the SLA constraint is:

[0031]

[0032] where, represents the link delay of the physical link of, represents the target virtual link the maximum allowable delay of, represents the physical link the current remaining resources of, represents the virtual link the resource requirements of, represents the current shortest path, E P represents the adjacency matrix of the connection relationship between physical nodes, represents the virtual node in the Vth network slice and the virtual link between N V represents the set of virtual nodes in the V-th network slice, represents the link mapping variable. If the current shortest path includes the physical link then otherwise,

[0033] Furthermore, the following function is used to calculate the path cost of the shortest path

[0034]

[0035] where represents the parent physical node and the current candidate physical node the shortest path between them, is the number of path hops of is the physical link the current workload of is in the basic physical network the set of physical links passed through; represents the mapping parameter, represents the physical link the current remaining resources of represents the physical link the initial total resources of

[0036] Furthermore, the multi-attributes include node resource importance, node degree centrality, betweenness centrality, closeness centrality, and eigenvector centrality; the importance degrees of each physical node in the physical network and each virtual node in the network slice are calculated respectively based on the multi-attribute decision-making method in the following way:

[0037] Based on the node resource importance, node degree centrality, betweenness centrality, closeness centrality, and eigenvector centrality of each physical node in the physical network and each virtual node in the network slice, the decision matrix of the physical network and the decision matrix of the network slice are constructed respectively;

[0038] The decision matrix of the physical network and the decision matrix of the network slice are normalized respectively. Based on the normalized decision matrix of the network slice, the group benefit value and individual regret value of each virtual node are calculated; based on the normalized decision matrix of the physical network, the group benefit value and individual regret value of each physical node are calculated;

[0039] Calculate the compromise decision index value of each virtual node based on the group benefit value and individual regret value of each virtual node; calculate the compromise decision index value of each physical node based on the group benefit value and individual regret value of each physical node; the smaller the compromise decision index value, the greater the importance of the node.

[0040] Further, the following formula is used to calculate the node resource importance of each virtual node

[0041]

[0042] where represents the set of virtual links connected to the virtual node ; represents the resource requirement of the virtual node ; represents the resource requirement of the virtual link ; represents the resource requirement of the virtual link ;

[0043] Compared with the prior art, the present invention can improve the acceptance rate and resource efficiency of slices by dynamically scheduling network resources through an intelligent optimization strategy. The above method proposes a joint optimization strategy for node mapping and link mapping, and at the same time establishes a dynamically schedulable resource model during the node mapping and link mapping processes, and further realizes the efficient scheduling and configuration of resources through intelligent optimization, so as to improve the resource efficiency and support more slice requirements to the greatest extent.

[0044] In the present invention, the above technical solutions can also be combined with each other to achieve more preferred combination schemes. Other features and advantages of the present invention will be described in the subsequent description, and some advantages can be obvious from the description, or can be understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the content specifically pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The drawings are only used for the purpose of showing specific embodiments, and are not considered to be a limitation of the present invention. Throughout the drawings, the same reference signs represent the same components.

[0046] Figure 1 is a flowchart of the network slice deployment method based on resource scheduling and adaptation according to an embodiment of the present invention;

[0047] Figure 2 is a schematic diagram of a search tree in an embodiment of the present invention;

[0048] Figure 3Schematic diagram of the mapping rule for the initial candidate physical node set of the embodiment of the present invention. Detailed implementation manners

[0049] The preferred embodiments of the present invention will be specifically described below in conjunction with the accompanying drawings, where the accompanying drawings form a part of this application and are used together with the embodiments of the present invention to explain the principles of the present invention, rather than to limit the scope of the present invention.

[0050] A specific embodiment of the present invention discloses a network slice deployment method based on joint optimization of resource scheduling and adaptation, as Figure 1 shown, including the following steps:

[0051] S1. For each network slice in the network slice set, calculate the importance of each physical node in the physical network and each virtual node in the network slice respectively based on the multi-attribute decision-making method, and perform pre-deployment of the network slice based on the importance of each node;

[0052] S2. Optimize the mapping parameters in the pre-deployment with the goal of maximizing the resource efficiency of the network slice set to obtain the optimal mapping parameters;

[0053] S3. For each network slice in the network slice set, recalculate the importance of each physical node in the physical network and each virtual node in the network slice respectively based on the optimal mapping parameters according to the multi-attribute decision-making method, and perform deployment of the network slice based on the importance of each node.

[0054] In the existing network slice technology, the node mapping and link mapping from virtual slices to physical resources are run independently. The nodes lack topological logical constraints and cannot build an end-to-end logical connection channel. At the same time, link optimization is not achieved during the node set mapping stage, and there is a large room for optimization in the slice acceptance rate and resource efficiency. The network slice deployment method based on resource scheduling provided in this embodiment can complete slice mapping by dynamically scheduling network resources through an intelligent optimization strategy, which can improve the slice acceptance rate and resource efficiency. The above method proposes a joint optimization strategy for node mapping and link mapping, and at the same time establishes a dynamically schedulable resource model during the node mapping and link mapping processes, and further realizes the efficient scheduling and configuration of resources through intelligent optimization, so as to improve the resource efficiency and support more slice requirements to the greatest extent.

[0055] During implementation, first convert the physical basic network (physical network) into an undirected weighted graph through the distribution of physical network resources and the network topology information of the physical basic network:

[0056] G P =(N P , E P , R T , B T, D P )

[0057] Among them, N P represents the set of physical nodes. During implementation, physical nodes include access network nodes such as base stations and edge data centers that accommodate virtual mobile edge computing servers transmission network nodes such as optical switches and routers core network nodes such as core data centers that carry out decoupling of the core network E P represents the adjacency matrix of the connection relationship between physical nodes in the physical network; R T represents the resources of physical nodes, such as the channel capacity of base stations on the access network side and the computing power capacity of nodes on the core side, etc.; B T represents the bandwidth information of physical links; D P represents the delay information of physical links.

[0058] During implementation, the network slice set includes uRLLC type slices for video live streaming and eMBB type slices for relay scenarios.

[0059] Convert the virtual network of the Vth network slice requirement into an undirected weighted graph according to the resource information and network topology information of the Vth network slice requirement:

[0060] G V =(N V , E V , R V , B V , D V ), V ≤ N NSR

[0061] Among them, N V represents the set of virtual network function (VNF) nodes of the Vth network slice, that is, the set of virtual nodes, including access network nodes transmission network nodes core network nodes E V represents the adjacency matrix corresponding to the network topology graph of the Vth network slice; R V represents the resource requirements of virtual nodes; B V represents the resource requirements of virtual links; represents the delay constraint of the Vth network slice; N NSR represents the number of network slices in the network slice set.

[0062] During implementation, each network slice in the network slice set is pre-deployed in descending order of importance.

[0063] Specifically, the following formula is used to calculate the importance IN of network slices:

[0064]

[0065] Among them, N V represents the set of virtual nodes of the V-th network slice, represents a virtual node 's resource requirement, represents the virtual link 's resource requirement, represents the virtual link 's maximum allowable delay.

[0066] Specifically, the multiple attributes include node resource importance, node degree centrality, betweenness centrality, closeness centrality, and eigenvector centrality.

[0067] In step S1, the following method is used to calculate the importance of each physical node in the physical network and each virtual node in the network slice based on the multiple attribute decision-making method:

[0068] S111. Based on the node resource importance, node degree centrality, betweenness centrality, closeness centrality, and eigenvector centrality of each physical node in the physical network and each virtual node in the network slice, construct the decision matrix of the physical network and the decision matrix of the network slice respectively;

[0069] The following takes the virtual node of the network slice as an example to illustrate how to calculate the importance of the virtual node. The calculation process of the importance of the physical node is the same as that of it.

[0070] Specifically, the following formula is used to calculate the node resource importance of each virtual node

[0071] Among them, represents the set of virtual links connected to the virtual node , represents the resource requirement of the virtual node , represents 's virtual link in 's resource requirement, represents the virtual link 's maximum allowable delay.

[0072] Specifically, the following formula is used to calculate the node degree centrality of each virtual node

[0073]

[0074] Among them, is the degree of the node , N VDenote the set of virtual nodes of the V-th network slice. N is the current network slice, that is, the number of virtual nodes in the V-th network slice. If node and node are directly connected, δ ij = 1; otherwise, δ ij = 0.

[0075] Specifically, the betweenness centrality of each virtual node is calculated using the following formula

[0076]

[0077] where is the shortest path between node and through node , is the shortest path between node and , and N is the number of virtual nodes in the V-th network slice.

[0078] Specifically, the closeness centrality of each virtual node is calculated using the following formula

[0079]

[0080] where is the shortest path between node and , is the number of hops of the shortest path , and N is the number of virtual nodes in the V-th network slice.

[0081] Specifically, the eigenvector centrality of each virtual node is calculated in the following way

[0082] Calculate the adjacency matrix E V to obtain the maximum eigenvalue λ of E and the eigenvector corresponding to the maximum eigenvalue λ. Each element in the eigenvector is the eigenvector centrality of the corresponding virtual node

[0083] Based on the node resource importance, node degree centrality, betweenness centrality, closeness centrality, and eigenvector centrality of each virtual node in the V-th network slice, construct the decision matrix of the network slice as:

[0084]

[0085] where a ij represents node The value of the j-th attribute, N represents the number of virtual nodes of the V-th network slice, and L represents the number of attributes. For example, L = 5, which are node resource importance, node degree centrality, betweenness centrality, closeness centrality, and eigenvector centrality in sequence.

[0086] S112. Normalize the decision matrix of the physical network and the decision matrix of the network slice respectively, and calculate the group benefit value and individual regret value of each virtual node based on the normalized decision matrix of the network slice; calculate the group benefit value and individual regret value of each physical node based on the normalized decision matrix of the physical network.

[0087] Specifically, the normalized decision matrix is:

[0088]

[0089] Determine the positive ideal solution of B N×L and the negative ideal solution

[0090]

[0091] Calculate the group benefit value S of the virtual node i and the individual regret value R i .

[0092]

[0093] where Ω1 = {ω1,..., ω L} represents the attribute weight coefficient, which is a mapping parameter to be optimized. The value range of the elements is [0, 1], and the sum of all elements is 1.

[0094] S113. Calculate the compromise decision index value of each virtual node based on the group benefit value and individual regret value of each virtual node; calculate the compromise decision index value of each physical node based on the group benefit value and individual regret value of each physical node; the smaller the compromise decision index value, the greater the importance of the node.

[0095] Specifically, use the following formula to calculate the importance compromise decision index Q i V of each virtual node:

[0096]

[0097]

[0098] ​Among them, ψ1 ∈ [0, 1] is a compromise coefficient and also a mapping parameter to be optimized. When ψ1 ≥ 0.5, the current multi-attribute decision-making mechanism is based on maximizing the group utility as the decision basis. When ψ1 < 0.5, the current multi-attribute decision-making mechanism is based on minimizing the individual regret value as the decision mechanism. When ψ1 = 0.5, both the group utility and the individual regret are considered in a compromised way.

[0099] Using the same method as in steps S111 - S113, calculate the compromise decision index value of each physical node in the physical network. Calculate the compromise decision index value of the physical node. During the process, the attribute weight coefficient is expressed as Ω2 = {ω'1,..., ω' L}, and the compromise coefficient is expressed as ψ2.

[0100] During pre-deployment, these mapping parameters Ω1, Ω2, ψ1, ψ2 are preset values. It should be noted that the smaller the compromise decision index value of the node, the greater the importance of the node, and the more important the node is.

[0101] After calculating the importance of each virtual node in the V-th network slice, pre-deploy the network slice based on the importance of each virtual node, which specifically includes:

[0102] S12. Taking the virtual node with the greatest importance in this network slice as the root node, construct a search tree for this network slice based on the breadth-first algorithm;

[0103] As Figure 2 shown, the virtual node with the greatest importance (i.e., the smallest Q value) in this network slice is node Taking this node as the root node of the search tree, based on the topological structure of the virtual nodes, construct a BFS search tree as shown in the tree structure in Figure 2 .

[0104] It should be noted that in the search tree, the nodes in the same layer are arranged in descending order of importance. For example, nodes and node are in the same layer, The importance of is greater than that of Therefore, node is ranked before

[0105] After constructing the search tree, starting from the root node of the search tree, traverse each virtual node in order from top to bottom and from left to right for mapping.

[0106] S131. Determine the initial candidate physical node set of the current virtual node based on the autonomous domain type of the current virtual node;

[0107] Specifically, as Figure 3 shown, virtual nodes are divided into access network nodes transport network nodes core network nodes and managed in independent autonomous domains. Node resource mapping is performed in the autonomous domains of access network nodes P in physical resource G transport network nodes core network nodes respectively, and the initial candidate physical node set is determined through the autonomous domain type of the virtual node. That is, the access virtual nodes in the network slice are mapped to the access side of the physical basic network, the transport virtual nodes are mapped to the transport network, and the core virtual nodes are mapped to the core network. It should be noted that since the uRLLC type slice has high requirements for latency, the user plane function (UPF) and the control plane in the core network are decoupled and preferentially mapped to the access network side to reduce the user-side latency.

[0108] After obtaining the initial candidate physical node set of the current virtual node, further determine the final candidate physical node set of the current virtual node according to the position of the current virtual node in the search tree.

[0109] If the current virtual node is the root node of the search tree, the physical nodes in the initial candidate physical node set that meet the resource requirements of the current virtual node constitute the final candidate physical node set of the current virtual node.

[0110] If the current virtual node is a non-root node of the search tree, the physical node mapped by the parent virtual node of the current virtual node is used as the parent physical node, and the physical nodes in the initial candidate physical node set that meet the resource requirements of the current virtual node and are neighbor nodes of the parent physical node constitute the final candidate physical node set of the current virtual node.

[0111] Specifically, the current virtual node is a non-root node of the BFS search tree, and its parent node (parent virtual node) in the search tree is parent virtual node mapped to the physical node as the parent physical node of the current virtual node. The candidate physical nodes in the initial candidate physical node set that meet the resource requirements of the current virtual node and are neighbor nodes of constitute the final candidate physical node set of the current physical node. During implementation, the neighbor nodes of are k-hop neighbor nodes. For example, if k = 2, then the nodes two hops away from the parent physical node Physical nodes with a path hop count less than or equal to 2 to the parent physical node are 2-hop neighbor nodes of the parent physical node.

[0112] It should be noted that if the set of candidate physical nodes of the current virtual node is empty, the deployment of the current virtual node fails, and the deployment of the current network slice fails, that is, the current network slice requirement is rejected, and the deployment of the next network slice continues. Otherwise, proceed to the next step.

[0113] S14. If the current virtual node is the root node of the search tree, select the physical node with the highest importance in the final candidate physical node set of the current virtual node as the mapping node of the current virtual node;

[0114] If the current virtual node is a non-root node of the search tree, use the physical node mapped by the parent virtual node of the current virtual node as the parent physical node, use the virtual link between the current virtual node and the parent virtual node of the current virtual node as the target virtual link, and select from the final candidate physical node set a candidate physical node whose path to the parent physical node meets the resource requirements of the target virtual link as the mapping node of the current virtual node.

[0115] For the root node, directly select the physical node with the highest importance from the final candidate physical node set as its mapping node. For non-root nodes, use the physical node mapped by the parent virtual node of the current virtual node as the parent physical node, and use the virtual link between the current virtual and the parent virtual node of the current virtual node as the target virtual link. For example, the current virtual node whose parent virtual node is parent virtual node the physical node mapped to is the virtual link between the current virtual node and the parent virtual node is then the parent physical node is the target virtual link is

[0116] Specifically, the following method is used to select from the final candidate physical node set a candidate physical node whose path to the parent physical node meets the resource requirements of the target virtual link as the mapping node of the current virtual node:

[0117] S141. Sort the candidate physical nodes in the final candidate physical node set in ascending order of node load; use the first candidate physical node in the sorted final candidate physical node set as the current candidate physical node.

[0118] Specifically, the following formula is used to calculate the load of the candidate physical node:

[0119] where is the physical node The current remaining resources, Is a physical node The total initial resources.

[0120] S142. Determine whether there is a path in the physical path from the parent physical node to the current candidate physical node that meets the resource requirements of the target virtual link. If so, map the current virtual node to the current candidate physical node, and map the target virtual link to the shortest path that meets the resource requirements of the target virtual link; otherwise, take the next candidate physical node in the final candidate physical node set as the current candidate physical node, and return to step S142.

[0121] S143. If there is no physical path from the parent physical node to each candidate physical node in the final candidate physical node set that meets the resource requirements of the target virtual link, then there is no candidate physical node in the final candidate physical node set whose path with the parent physical node meets the resource requirements of the target virtual link.

[0122] During implementation, the first candidate physical node in the final candidate physical node set is used as the current candidate physical node. There is a virtual link that satisfies the target to the current candidate physical node resource demand path, the current virtual node is mapped to the current candidate physical node, and the target virtual link Mapping to slave physical nodes To the current candidate physical node that satisfies the virtual link On the shortest path of resource requirements, the current virtual node mapping is completed, and the next virtual node mapping is performed; otherwise, the second candidate physical node is used as the current candidate physical node, and the physical node is judged. Whether there is a virtual link that satisfies the target to the current candidate physical node The path of resource requirements, and so on. There is no virtual link that satisfies the target to each candidate physical node in the final candidate physical node set. If the resource requirement path is not met, the current virtual node mapping fails, the current network slice deployment fails, and the deployment of the next network slice continues. By combining node mapping and link mapping, the efficiency of network slice deployment is improved.

[0123] Specifically, in step S142, the following method is used to determine whether there is a path that meets the resource requirements of the target virtual link in the physical path from the parent physical node to the current candidate physical node:

[0124] S1421. Search and obtain K shortest paths from the parent physical node to the current candidate physical node based on the Dijkstra algorithm;

[0125] It should be noted that the search for the shortest path is carried out on the adjacency matrix C E constructed according to the link resource situation of the current physical network. The adjacency matrix C E is obtained in the following way: If the link between nodes and nodes in the physical network meets the resource requirements of the target virtual link , then the value at the corresponding position of nodes E and nodes in the adjacency matrix C is the load value of the link between the two nodes, otherwise it is infinity. The load of the physical link is calculated by the following formula:

[0126]

[0127] where represents the current remaining resource of the physical link , represents the initial total resource of the physical link . Searching for K shortest paths from the parent physical node to the current candidate physical node using the Dijkstra algorithm is an existing algorithm, which will not be elaborated in this application.

[0128] S1422. Traverse each of the K shortest paths in ascending order of path cost. If the current shortest path meets the SLA constraint, then the current shortest path is the shortest path that meets the resource requirements of the target virtual link, and the traversal ends; otherwise, continue to traverse the next shortest path.

[0129] The following function is used to calculate the path cost of the shortest path

[0130]

[0131]

[0132] where represents the shortest path between the parent physical node and the current candidate physical node , is the number of hops of the path, is the current workload of the physical link , is the set of physical links passed through in the basic physical network ; represents the link resource allocation coefficient, Indicates the current remaining resources of the physical link , Indicates the initial total resources of the physical link . The link resource allocation coefficient is also a mapping parameter that needs to be optimized. During pre-deployment, the parameter is a preset value.

[0133] SLA constraint, that is, Service Level Agreement (SLA) constraint. During implementation, the SLA constraint includes link delay constraint and link resource constraint, and the link resource can be link bandwidth. Specifically, the SLA constraint is as follows:

[0134]

[0135] Among them, indicates the link delay of the physical link , indicates the maximum allowable delay of the target virtual link , indicates the current remaining resources of the physical link , indicates the resource requirement of the virtual link , indicates the current shortest path, E V represents the adjacency matrix of the connection relationship of virtual nodes in the Vth network slice, indicates the virtual node in the Vth network slice and the virtual link between V represents the set of virtual nodes in the Vth network slice, indicates the link mapping variable. If the virtual link is mapped to the physical link then otherwise That is, the sum of the physical path delays of the selected shortest path should be less than the maximum allowable delay of the virtual link in the network slice, and the bandwidth should be greater than the virtual link requirement.

[0136] S1423. If none of the K shortest paths satisfy the SLA constraint, then there is no path in the physical path from the parent physical node to the current candidate physical node that satisfies the resource requirement of the target virtual link.

[0137] After pre-deploying with each network slice in the network slice set, optimize the mapping parameters according to the pre-deployment situation to improve the slice acceptance rate and resource efficiency.

[0138] Specifically, the optimization objective function in step S2 is as follows:

[0139]

[0140]

[0141] Among them, represents the current remaining resources of the physical node R T represents the total resources of the nodes in the basic physical network, represents the physical link B T represents the total link bandwidth resources of the basic physical network, represents the resource requirement of the i-th node in the V-th network slice in the network slice set represents the virtual link N V represents the set of virtual nodes in the V-th network slice, E V represents the adjacency matrix of the connection relationship between virtual nodes in the V-th network slice, represents the virtual node in the V-th network slice and the virtual link between P represents the set of physical nodes, E P represents the adjacency matrix of the connection relationship between physical nodes, h (z) represents the set of mapping parameter variables fitted by the z-th search, M represents the number of mapping parameters, f PSO (h (z) ) represents the network slice deployment resource efficiency measured by the z-th search fitting, N NSR represents the number of network slices in the network slice set; m V represents the pre-deployed mapping variable, with a value of 1 if the V-th network slice is successfully deployed, otherwise 0.

[0142] During implementation, the particle swarm optimization algorithm can be used to optimize the mapping parameters. The x-th particle in the particle swarm is represented as h x = [h x,1 , h x,2 ,.., h x,M , and after the algorithm ends, the parameter value corresponding to the optimal particle is the optimal mapping parameter

[0143] Through the joint optimization of the mapping parameters of node mapping and link mapping, the efficient scheduling and configuration of resources are improved, the resource efficiency is enhanced, and more network slice requirements are supported to the maximum extent.

[0144] ​According to the optimal mapping parameters, for each network slice in the network slice set, the importance degrees of each physical node in the physical network and each virtual node in the network slice are recalculated respectively based on the multi-attribute decision-making method, and the network slice is deployed based on the importance degree of each virtual node.

[0145] Specifically, in step S3, the importance degree of each physical node is recalculated according to the optimal mapping parameters by the same process as steps S111 - S113. For each network slice in the network slice set, the importance degree of each virtual node in it is recalculated according to the optimal mapping parameters by the same process as steps S111 - S113. Each network slice is redeployed according to the process of steps S12 - S14. Since the mapping parameters at this time are the optimal mapping parameters obtained with the maximization of the resource efficiency of the network slice set as the optimization goal, the resource efficiency can be maximally improved and the acceptance rate of the slice can be increased.

[0146] Based on the optimal mapping parameters, each network slice in the network slice set is deployed in descending order of importance degree. If the current network slice deployment fails, according to the foregoing process, there must be a virtual node in the current network slice whose deployment fails. Based on this, an attempt can be made to reconstruct the successfully deployed network slices and dredge the congested physical nodes, thereby increasing the acceptance rate of the network slice. Specifically, the following method is used for reconstruction:

[0147] S41. If the current virtual node of the current network slice fails to be deployed, determine the congested physical node that causes the deployment failure, sort the congested physical nodes in descending order of congestion degree, and use the first congested physical node after sorting as the current congested physical node; sort the successfully deployed network slices in ascending order of importance degree, and use the first successfully deployed network slice after sorting as the current successfully deployed network slice;

[0148] S42. For the current congested physical node, if the current successfully deployed network slice can be reconstructed, reconstruct the current successfully deployed network slice and redeploy the current virtual node of the current network slice; otherwise, use the next successfully deployed network slice as the current successfully deployed network slice and return to step S42; if there is no next successfully deployed network slice, use the next congested physical node as the current congested physical node and return to step S42;

[0149] S43. If there is no successfully deployed network slice that can be reconstructed for each congested physical node, reject the deployment of the current network slice.

[0150] Specifically, in step S41, the following method is used to determine the congested physical node that causes the deployment failure:

[0151] S411. If the current virtual node is the root node of the search tree corresponding to the current network slice, the congested physical node is a physical node that conforms to the mapping rule of the current virtual node but does not meet the resource requirements of the current virtual node.

[0152] S412. If the current virtual node is a non-root node of the search tree corresponding to the current network slice, the physical node mapped by the parent virtual node of the current virtual node is used as the parent physical node, and the congested physical node is a neighbor node of the parent physical node that conforms to the mapping rule of the current virtual node but does not meet the resource requirements of the current virtual node.

[0153] Among them, the search tree corresponding to the current network slice is the search tree constructed in step S3 according to the recalculated node importance in the same process as in step S12.

[0154] Specifically, the mapping rule of the virtual node is the mapping rule in step S131, that is: the access virtual node in the network slice is mapped to the access side of the physical basic network, the transmission virtual node is mapped to the transmission network, the core virtual node is mapped to the core network, and the user plane function (UPF) and the control plane in the core network of the uRLLC type network slice are decoupled and mapped to the edge data center node on the access network side.

[0155] If the current virtual node is the root node in the search tree, the congested physical node is a physical node that conforms to the mapping rule of the current virtual node but does not meet the resource requirements of the current virtual node. If the current virtual node is a non-root node in the search tree, its parent node (parent virtual node) in the search tree is parent virtual node mapped to the physical node as the parent physical node of the current virtual node. The congested physical node is among the neighbor nodes of that conform to the mapping rule of the current virtual node but do not meet the resource requirements of the current virtual node. During implementation,

[0156] After obtaining the congested physical nodes, sort the congested physical nodes in descending order of congestion degree, and use the first congested physical node after sorting as the current congested physical node. Considering the overload situation of the congested physical nodes and the load situation of the physical links connected to them, calculate the congestion degree of the congested nodes. Specifically, use the following formula to calculate the congestion degree C of the congested physical node degree :

[0157]

[0158] Among them, Denote the i-th node in the V-th network slice 's resource requirement. The V-th network slice is the currently deployed failed network slice, and the node is the currently failed virtual node in the deployment, Denote the i'-th node in the V'-th network slice 's resource requirement, Denote the physical node 's initial total resources, N V Denote the set of virtual nodes in the V-th network slice, N V' Denote the set of virtual nodes in the V'-th network slice, Denote the physical link 's current remaining resources, Denote the physical link 's initial total resources, Denote the set of physical links connected to the current congested physical node . If the virtual node in the V'-th virtual slice is successfully deployed to the physical node then Otherwise

[0159] For the successfully deployed network slices, sort them in ascending order of importance, and take the first successfully deployed network slice in the sorted order as the current successfully deployed network slice. Specifically, the importance of the network slice can refer to the calculation process of the importance IN of the aforementioned network slice.

[0160] Specifically, in step S42, the following method is used to determine whether the current successfully deployed network slice can be reconfigured for the current congested physical node:

[0161] S421. Construct a set of virtual nodes to be migrated τ with the virtual nodes deployed on the current congested physical node in the current successfully deployed network slice as the virtual nodes to be migrated, and construct a set of candidate physical nodes to be migrated τ with the neighbor nodes of the current congested physical node as the candidate physical nodes to be migrated;

[0162] Specifically, assume that the current congested physical node is and the current successfully deployed network slice is the V″-th network slice. Then, the virtual nodes deployed on (i.e., mapped to) the current congested physical node in the V″-th network slice are the virtual nodes to be migrated, and construct a set of virtual nodes to be migrated τ v .

[0163] The neighbor nodes of the current congested physical node are used to construct a set of candidate physical nodes to be migrated τ p. During implementation, the neighbor nodes are k-hop neighbor nodes, for example, k = 2.

[0164] If the set of virtual nodes to be migrated is τ v , and the virtual nodes in the currently successfully deployed network slice do not affect the currently congested physical node, for the currently congested physical node, the currently successfully deployed network slice cannot be reconfigured.

[0165] S422. Determine whether the resource slack generated by the virtual nodes to be migrated in the set of virtual nodes to be migrated is greater than or equal to the resource deficit generated after the currently congested physical node joins the current virtual node. If not, the currently successfully deployed network slice cannot be reconfigured; otherwise, execute step S423.

[0166] Specifically, determine whether the slack is greater than or equal to the resource deficit according to the following formula.

[0167]

[0168] The left side of the formula is the resource deficit generated after the currently congested physical node joins the current virtual node, and the right side of the formula is the resource slack generated by migrating the virtual nodes to be migrated in the set of virtual nodes to be migrated.

[0169] Among them, represents the resource requirement of the i-th node in the V-th network slice. The V-th network slice is the currently failed deployed network slice, and the node is the currently failed virtual node. represents the resource requirement of the i'-th node in the V'-th network slice. represents the initial total resources of the physical node . N V represents the set of virtual nodes in the V-th network slice, and N V' represents the set of virtual nodes in the V'-th network slice. represents the currently congested physical node 's initial total resources. represents the resource requirement of the j-th node in the V''-th network slice. The V''-th network slice is the currently successfully deployed network slice.

[0170] If the virtual node in the V'-th virtual slice is successfully deployed to the physical node , then Otherwise

[0171] If the slack is less than the resource deficit, that is, the current physical node still cannot meet the resource requirements of the current virtual node after being unblocked, then the currently successfully deployed network slice cannot be reconfigured. Otherwise, further judgment is made according to step S423.

[0172] S423. Determine whether each virtual node to be migrated in the set of virtual nodes to be migrated can be migrated to a physical node in the set of candidate migrated physical nodes. If the resource slack generated by all virtual nodes to be migrated that can be migrated to the physical nodes in the set of candidate migrated physical nodes is less than the resource deficit, then the currently successfully deployed network slice cannot be reconfigured. Otherwise, the currently successfully deployed network slice can be reconfigured.

[0173] Specifically, step S423 determines whether each virtual node to be migrated in the set of virtual nodes to be migrated can be migrated to a physical node in the set of candidate migrated physical nodes in the following way:

[0174] S4231. Sort the virtual nodes to be migrated in the set of virtual nodes to be migrated in ascending order of migration cost;

[0175] Specifically, the following formula is used to calculate the migration cost of each virtual node to be migrated

[0176] where represents the migration cost of the node type corresponding to the virtual node to be migrated, which is related to the sharing degree of the node and the position of the node in the network topology. represents the resource requirement of the virtual node to be migrated represents the virtual node to be migrated 's resource requirement. represents the set of virtual links connected to the virtual node to be migrated represents the set of virtual links connected to the virtual node to be migrated represents the virtual link 's resource requirement.

[0177] S4232. For each virtual node to be migrated, use the physical node mapped by the parent virtual node of the current virtual node to be migrated as the parent physical node, and use the virtual link between the current virtual node to be migrated and the parent virtual node of the current virtual node to be migrated as the virtual link to be migrated. The physical nodes in the set of candidate migrated physical nodes that meet the resource requirements of the current virtual node to be migrated form the set of target candidate migrated physical nodes;

[0178] S4233. If there is a physical node in the set of target candidate migrated physical nodes whose path to the parent physical node meets the resource requirements of the virtual link to be migrated, then the current virtual node to be migrated can be migrated to a physical node in the set of candidate migrated physical nodes. Otherwise, the current virtual node to be migrated cannot be migrated to a physical node in the set of candidate migrated physical nodes.

[0179] Specifically, in step S4233, the following method is used to determine whether there is a physical node in the target candidate migration physical node set whose path to the parent physical node meets the resource requirements of the virtual link to be migrated:

[0180] S42331. Sort the physical nodes in the target candidate migration physical node set in ascending order of node load; take the first physical node in the sorted target candidate migration physical node set as the current target physical node. The calculation of node load can refer to step S141.

[0181] S42332. Determine whether there is a path that meets the resource requirements of the virtual link to be migrated in the physical path from the parent physical node to the current target physical node. If there is, there is a physical node in the target candidate migration physical node set whose path to the parent physical node meets the resource requirements of the virtual link to be migrated; otherwise, take the next physical node in the target candidate migration physical node set as the current target physical node, and return to step S42332.

[0182] S42333. If there is no physical path that meets the resource requirements of the virtual link to be migrated from the parent physical node to each target physical node in the target candidate migration physical node set, then there is no physical node in the target candidate migration physical node set whose path to the parent physical node meets the resource requirements of the virtual link to be migrated.

[0183] Specifically, in step S42332, to determine whether there is a path that meets the resource requirements of the virtual link to be migrated in the physical path from the parent physical node to the current target physical node, the same process as in steps S1421 - S1423 can be used for judgment, which will not be elaborated here.

[0184] For the current congested physical node, if the currently deployed successful network slice is reconfigurable, then reconfigure the currently deployed successful network slice. According to the process in step S423, find the target physical node corresponding to each node to be migrated, map it to the corresponding physical node, and map the virtual link to be migrated to the physical link that meets its resource requirements. At this time, the resource slack generated by all the nodes to be migrated that can be migrated to the physical nodes in the candidate migration physical node set is greater than or equal to the resource deficit. Therefore, the current virtual node of the current network slice, that is, the virtual node with deployment failure, can be deployed to the current congested physical node. At this time, redeploy the current virtual node. After the current virtual node is deployed, continue to deploy the next virtual node of the current network slice. If the deployment of the next virtual node fails, then return to step S41 for reconfiguration.

[0185] If the currently successfully deployed network slice is not reconfigurable, then use the next successfully deployed network slice as the currently successfully deployed network slice and return to step S42; if there is no next successfully deployed network slice in the network slice set, then use the next congested physical node as the current congested physical node and return to step S42. For all congested physical nodes, if all successfully deployed network slices are not reconfigurable, then the current network slice deployment fails completely and the deployment of the current network slice is rejected.

[0186] Through the dynamic deployment process of network slice mapping and reconstruction, when a network slice is rejected, the slice reconstruction mechanism is triggered. The slice reconstruction process is based on the slice deployment process and uses the slice deployment process to find the congested physical nodes that cause the network slice to be rejected, so as to clear the congested physical nodes, reconstruct the successfully deployed network slices, and enable the currently failed network slice to be redeployed, thereby improving the acceptance rate of the network slice. In addition, to ensure the stability within the slice and the isolation between slices, virtual nodes on the congested physical nodes are migrated in units of network slices, and virtual links related to the virtual nodes are migrated for slice reconfiguration.

[0187] Those skilled in the art can understand that all or part of the processes of implementing the methods of the above embodiments can be completed by instructing relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. Among them, the computer-readable storage medium is a magnetic disk, an optical disc, a read-only memory, or a random access memory, etc.

[0188] As mentioned above, only the specific preferred embodiments of the present invention are described, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered by the protection scope of the present invention.

Claims

1. A network slice deployment method based on the joint optimization of resource scheduling and adaptation, characterized in that It includes the following steps: For each network slice in the network slice set, calculate the importance degree of each physical node in the physical network and each virtual node in the network slice respectively based on the multi-attribute decision-making method, and perform pre-deployment on the network slice based on the importance degree of each node; Jointly optimize the mapping parameters in the pre-deployment with the goal of maximizing the resource efficiency of the network slice set to obtain the optimal mapping parameters; For each network slice in the network slice set, recalculate the importance degree of each physical node in the physical network and each virtual node in the network slice respectively based on the optimal mapping parameters and the multi-attribute decision-making method, and perform deployment on the network slice based on the importance degree of each node; The optimization objective function is: Among them, represents the current remaining resources of the physical node R T represents the total resources of the nodes in the basic physical network, represents the physical link B T represents the total link bandwidth resources of the basic physical network, represents the resource requirement of the i-th node in the V-th network slice in the network slice set ; represents the virtual link N V represents the set of virtual nodes in the V-th network slice, E V represents the adjacency matrix of the connection relationship between virtual nodes in the V-th network slice, represents the virtual node in the V-th network slice and the virtual link between them, N P represents the set of physical nodes, E P represents the adjacency matrix of the connection relationship between physical nodes, h (z) represents the set of mapping parameter variables fitted by the z-th search, M represents the number of mapping parameters, f PSO (h (z) ) represents the network slice deployment resource efficiency measured by the z-th search fitting, N NSR represents the number of network slices in the network slice set; m V represents the pre-deployed mapping variable, with a value of 1 if the V-th network slice is successfully deployed, otherwise 0; Performing pre-deployment on the network slice based on the importance degree of each node and performing deployment on the network slice based on the importance degree of each node includes: Taking the virtual node with the highest importance degree in the network slice as the root node, and constructing a search tree for the network slice based on the breadth-first algorithm; For each virtual node in the search tree, construct a final candidate physical node set for the current virtual node based on the autonomous domain type and resource requirements of the current virtual node; If the current virtual node is the root node of the search tree, select the physical node with the highest importance degree in the final candidate physical node set of the current virtual node as the mapping node of the current virtual node; If the current virtual node is a non-root node of the search tree, take the physical node mapped by the parent virtual node of the current virtual node as the parent physical node, take the virtual link between the current virtual node and the parent virtual node of the current virtual node as the target virtual link, and select from the final candidate physical node set a candidate physical node whose path to the parent physical node meets the resource requirements of the target virtual link as the mapping node of the current virtual node.

2. The network slice deployment method based on joint optimization of resource scheduling and adaptation according to claim 1, characterized in that Constructing a final candidate physical node set for the current virtual node based on the autonomous domain type and resource requirements of the current virtual node includes: Determine the initial candidate physical node set of the current virtual node based on the autonomous domain type of the current virtual node; If the current virtual node is the root node of the search tree, the physical nodes in the initial candidate physical node set that meet the resource requirements of the current virtual node constitute the final candidate physical node set of the current virtual node; If the current virtual node is a non-root node of the search tree, take the physical node mapped by the parent virtual node of the current virtual node as the parent physical node, and the physical nodes in the initial candidate physical node set that meet the resource requirements of the current virtual node and are neighbor nodes of the parent physical node constitute the final candidate physical node set of the current virtual node.

3. The network slice deployment method based on joint optimization of resource scheduling and adaptation according to claim 1, characterized in that Select a candidate physical node whose path to the parent physical node meets the resource requirements of the target virtual link from the final candidate physical node set as the mapping node of the current virtual node in the following way: S141. Sort the candidate physical nodes in the final candidate physical node set in ascending order of node load; take the first candidate physical node in the sorted final candidate physical node set as the current candidate physical node; S142. Determine whether there is a path that meets the resource requirements of the target virtual link in the physical path from the parent physical node to the current candidate physical node. If so, map the current virtual node to the current candidate physical node and map the target virtual link to the shortest path that meets the resource requirements of the target virtual link; otherwise, take the next candidate physical node in the final candidate physical node set as the current candidate physical node, and return to step S142; S143. If there is no physical path that meets the resource requirements of the target virtual link from the parent physical node to each candidate physical node in the final candidate physical node set, then there is no candidate physical node in the final candidate physical node set whose path to the parent physical node meets the resource requirements of the target virtual link.

4. The network slice deployment method based on joint optimization of resource scheduling and adaptation according to claim 3, characterized in that, The following method is used to determine whether there is a path that meets the resource requirements of the target virtual link in the physical path from the parent physical node to the current candidate physical node: Based on the Dijkstra algorithm, search for K shortest paths from the parent physical node to the current candidate physical node; Traverse each of the K shortest paths in ascending order of path cost. If the current shortest path meets the SLA constraint, then the current shortest path is the shortest path that meets the resource requirements of the target virtual link, and the traversal ends; otherwise, continue to traverse the next shortest path; If none of the K shortest paths meet the SLA constraint, then there is no path that meets the resource requirements of the target virtual link in the physical path from the parent physical node to the current candidate physical node.

5. The network slice deployment method based on joint optimization of resource scheduling and adaptation according to claim 4, wherein The SLA constraint is: Among them, represents the link delay of the physical link . represents the maximum allowable delay of the target virtual link . represents the current remaining resources of the physical link . represents the resource requirement of the virtual link . represents the current shortest path, E P represents the adjacency matrix of the connection relationship between physical nodes represents the virtual node in the Vth network slice and the virtual link between them, N V represents the set of virtual nodes in the Vth network slice represents the link mapping variable. If the current shortest path includes the physical link then otherwise, 6. The network slice deployment method based on joint optimization of resource scheduling and adaptation according to claim 4, wherein The path cost of the shortest path is calculated using the following function Among them, represents the shortest path between the parent physical node and the current candidate physical node, is the number of path hops, is the current workload of the physical link, is the set of physical links passed through in the underlying physical network; represents the mapping parameter, represents the current remaining resources of the physical link, represents the initial total resources of the physical link.

7. The network slice deployment method based on joint optimization of resource scheduling and adaptation according to claim 1, characterized in that, The multiple attributes include node resource importance, node degree centrality, betweenness centrality, closeness centrality, and eigenvector centrality; The following method is used to calculate the importance of each physical node in the physical network and each virtual node in the network slice based on the multi-attribute decision-making method: Based on the node resource importance, node degree centrality, betweenness centrality, closeness centrality, and eigenvector centrality of each physical node in the physical network and each virtual node in the network slice, construct the decision matrix of the physical network and the decision matrix of the network slice respectively; Normalize the decision matrix of the physical network and the decision matrix of the network slice respectively. Calculate the group benefit value and individual regret value of each virtual node based on the normalized decision matrix of the network slice; calculate the group benefit value and individual regret value of each physical node based on the normalized decision matrix of the physical network; Calculate the compromise decision index value of each virtual node based on the group benefit value and individual regret value of each virtual node; Calculate the compromise decision index value of each physical node based on the group benefit value and individual regret value of each physical node; the smaller the compromise decision index value, the greater the importance of the node.

8. The method for network slice deployment based on joint optimization of resource scheduling and adaptation according to claim 7, wherein The node resource importance of each virtual node is calculated using the following formula Among them, represents the set of virtual links connected to the virtual node, represents the resource requirement of the virtual node , represents the resource requirement of the virtual link , represents the resource requirement of the virtual link , and represents the maximum allowable delay of the virtual link.

Images