CN111800339B

CN111800339B - Route optimization method with path number constraint in hybrid SDN scene

Info

Publication number: CN111800339B
Application number: CN202010633665.1A
Authority: CN
Inventors: 郭迎亚; 郭文忠
Original assignee: Fuzhou University
Current assignee: Fuzhou University
Priority date: 2020-07-02
Filing date: 2020-07-02
Publication date: 2021-06-01
Anticipated expiration: 2040-07-02
Also published as: CN111800339A

Abstract

The invention relates to a route optimization method with path number constraint in a hybrid SDN scene, which comprises the following steps: step S1, determining the deployment position of the SDN node by adopting a greedy algorithm; step S2, finding feasible paths of flow demands among all source destination node pairs according to the deployment positions of the SDN nodes, and step S3, calculating the distribution conditions of the flow on all the feasible paths under the condition of no path constraint; step S4, setting the constraint of the number of paths as h, and selecting the optimal path meeting the constraint of the number of paths from all feasible paths of each flow demand by using random rounding to obtain an optimal path set; and step S5, according to the optimal path set, considering the problem of multi-commodity flow, and calculating the optimal flow distribution of the flow on the path. The invention can effectively reduce the maximum link utilization rate of the network under the condition of the constraint of the path number, and further improve the network performance.

Description

Route optimization method with path number constraint in hybrid SDN scene

Technical Field

The invention belongs to the field of route optimization, and particularly relates to a route optimization method with path number constraint in a hybrid SDN scene.

Background

Software Defined Networking (SDN) is a new network architecture in which decoupling of data and control planes on SDN switches and SDN controllers is implemented. Specifically, the SDN switch is programmable, and forwards network traffic according to a flow entry issued by the controller. An SDN controller is a logically centralized device that obtains network state by collecting network information from an SDN, resulting in a global view. The controller implements fine-grained forwarding control of network traffic by assigning flow entries to SDN switches. Unlike shortest path forwarding in traditional distributed networks, SDN can achieve more flexible, convenient, and intelligent network management and control. Accordingly, research in relevant SDNs has attracted worldwide attention.

Although SDN has many advantages, in practice, it is very difficult to achieve full migration of legacy networks to SDN networks in one step, especially for large-scale legacy networks. The reason is mainly divided into two aspects: economic and technical factors. On the one hand, purchasing and deploying SDN devices to upgrade the infrastructure of an entire legacy network will inevitably bring huge capital expenditure and operational burden to Internet Service Providers (ISPs). Currently, ISPs may be reluctant to implement updates from legacy networks to SDN networks in one step. At the same time, considering the deployment revenue, many studies indicate that it is not necessary to update all old devices in the network to SDN devices. On the other hand, emerging hardware and software related to the SDN are relatively immature, a large-scale appropriate evaluation is lacked, and reliability and stability of the SDN device cannot be guaranteed. This means that updating infrastructure devices of an entire legacy network using dedicated SDN devices may result in potential security and instability risks.

Disclosure of Invention

In view of this, the present invention provides a route optimization method with a path number constraint in a hybrid SDN scenario, which can effectively reduce the maximum link utilization of a network and further improve network performance under the condition of the path number constraint.

In order to achieve the purpose, the invention adopts the following technical scheme:

a route optimization method with path number constraint in a hybrid SDN scene comprises the following steps:

step S1, determining the deployment position of the SDN node by adopting a greedy algorithm;

step S2, finding feasible paths of flow demands among all source destination node pairs according to the deployment positions of the SDN nodes;

step S3, calculating the distribution of the flow on all feasible paths under the condition of no path constraint;

step S4, setting the constraint of the number of paths as h, and selecting the optimal path meeting the constraint of the number of paths from all feasible paths of each flow demand by using random rounding to obtain an optimal path set;

and step S5, according to the optimal path set, considering the problem of multi-commodity flow, and calculating the optimal flow distribution of the flow on the path.

Further, the step S1 is specifically:

step S1.1, calculating the network flow distribution condition obtained by the flow according to the shortest path route between the source node and the destination node to obtain the link utilization rate on each link;

and S1.2, according to the sequence from large to small of the link utilization rate, preferentially deploying the nodes with large link utilization rate as SDN nodes.

Further, the step S2 is specifically:

s2.1, aiming at a node a in the hybrid SDN network topology, selecting and using a Dijkstra algorithm to construct a shortest path tree from the node a to each other node, and transposing the found graph to obtain a shortest path tree taking the node a as a destination node;

s2.2, sequentially adding adjacent edges which can be routed by the SDN node flow on the shortest path tree, and checking whether a loop is formed by using a topological sorting algorithm; if an adjacent edge joining the SDN node does not form a loop, joining the edge; otherwise, removing the edge; obtaining a traffic routability graph (PPG) based on the hybrid network topology;

s2.3, repeating the steps S2.1-S2.2, and finding out a routable graph PPG of each node in the network topology;

and S2.4, finding all paths between each traffic demand source destination node pair on the obtained routable graph PPG of all the traffic by adopting a Yen' S algorithm.

Further, the step S3 is specifically:

step S3.1, according to the flow demand in the multi-commodity flow problem model, satisfying the constraints of constraint, link capacity constraint and flow conservation, listing the relevant linear constraints, and aiming at optimizing the maximum link utilization rate, as shown in the following:

where U is the network maximum link utilization, x (p) represents the traffic distribution on path p; c (e) is the link capacity of link e

S3.2, using a linear programming solving tool, CPLEX or Gurobi to complete the solving of the linear programming problem, and adding all paths meeting x (p) > 0 into the set

Further, the step S4 is specifically:

step S4.1, for each flow demand, according to probability

Randomly selecting a path

Join to final set of paths

And will route p from

Removing;

s4.2, adding 1 to the number of the currently selected paths, and repeating the step 3.1 until h paths are selected according to the flow demand;

step S4.3: and repeating the steps S3.1-S3.2, and finding the optimal path set under the path number constraint h for each flow demand.

Compared with the prior art, the invention has the following beneficial effects:

the invention can effectively reduce the maximum link utilization rate of the network under the condition of the constraint of the path number, and further improve the network performance.

Drawings

FIG. 1 is a flow chart of a method in an embodiment of the invention.

Detailed Description

The invention is further explained below with reference to the drawings and the embodiments.

Referring to fig. 1, the present invention provides a route optimization method with path number constraint in a hybrid SDN scenario, including the following steps:

step S1: determining a deployment position of the SDN node by using a greedy algorithm;

step S1.1: firstly, calculating the network flow distribution condition obtained by the flow according to the shortest path route between source and destination nodes, and obtaining the link utilization rate on each link, namely the ratio of the link load to the link bandwidth;

step S1.2: and according to the sequence from large to small of the link utilization rate, preferentially deploying the nodes with the larger link utilization rate as SDN nodes.

Step S2: after the SDN node deployment position is determined, finding feasible paths of flow requirements among all source destination node pairs;

step S2.1: aiming at a node a in a hybrid SDN network topology, selecting and using Dijkstra algorithm to construct a shortest path tree from a to each other node, and transposing a found graph to obtain a shortest path tree taking a as a destination node;

step S2.2: and sequentially adding adjacent edges which can be routed by the SDN node flow on the shortest path tree, and checking whether a loop is formed by using a topological sorting algorithm. Specifically, if an adjacent edge joining an SDN node does not form a loop, the edge is joined; otherwise, the edge is removed. This results in a traffic routability Graph (PPG) based on the hybrid network topology. I.e. a graph containing all paths over which traffic can be rerouted.

Step S2.3: and repeating the steps S1.1-S1.2, traversing each node in the network topology, and finding a routable graph taking the node as a destination node.

Step S2.4: on the obtained routable graph PPG of all the traffic, a Yen's algorithm is used to find all the paths between each traffic demand source destination node pair.

Step S3: and calculating the distribution of the traffic on all feasible paths under the condition of no path constraint. Specifically, relevant linear constraints are listed according to constraints of flow demand satisfaction, link capacity constraints, and flow conservation in the multi-commodity flow problem model. As follows:

where (1) constraints are satisfied for flow conservation and flow demand. (2) Is a link capacity constraint limit. (3) Is a non-negative limiting constraint. U is the network maximum link utilization. The optimization goal is to minimize the maximum link utilization. x (p) represents the traffic distribution on path p. And c (e) is the link capacity of link e. The solution of the linear programming problem can be accomplished using a linear programming solving tool, CPLEX or Gurobi. Adding all paths satisfying x (p) > 0 to the set

Step S4: the limit on the number of paths is set to h. And according to the constraint of the number of paths, selecting the paths meeting the constraint of the number of paths from all feasible paths of each traffic demand by using random rounding.

Step S4.1, for each flow demand, according to probability

Randomly selecting a path

Join to final set of paths

And will route p from

Is removed.

Step S4.2: and adding 1 to the number of the currently selected paths, and repeating the step 3.1 until h paths are selected for the flow demand.

Step S4.3: repeating the steps 3.1-3.2, and finding the optimal path set under the constraint of the path number h for each flow demand

Step S5, according to the multi-commodity flow problem, calculating the flow on the path

And completing the route optimization by the optimal shunting.

The above description is only a preferred embodiment of the present invention, and all equivalent changes and modifications made in accordance with the claims of the present invention should be covered by the present invention.

Claims

1. A route optimization method with path number constraint in a hybrid SDN scene is characterized by comprising the following steps:

step S1: determining the deployment position of the SDN node by adopting a greedy algorithm;

the step S1 specifically includes:

step S1.1: calculating the network flow distribution condition obtained by the flow according to the shortest path route between the source and destination nodes to obtain the link utilization rate on each link;

step S1.2: according to the sequence from large to small of the link utilization rate, preferentially deploying the nodes with large link utilization rate as SDN nodes;

step S2: finding feasible paths of flow requirements among all source destination node pairs according to the deployment positions of the SDN nodes;

the step S2 specifically includes:

step S2.2: sequentially adding adjacent edges of each SDN node flow routable on the shortest path tree, and checking whether a loop is formed by using a topological sorting algorithm; if an adjacent edge joining the SDN node does not form a loop, joining the edge; otherwise, removing the edge; obtaining a traffic routability graph (PPG) based on the hybrid network topology;

step S2.3: repeating the steps S2.1-S2.2, and finding out a routable graph PPG of each node in the network topology;

step S2.4: on the obtained routable graph PPG of all the flows, all paths between each flow demand source destination node pair are found by adopting a Yen's algorithm;

step S3: calculating the distribution condition of the flow on all feasible paths under the condition of no path constraint;

the step S3 specifically includes:

step S3.1: according to the flow demand satisfying constraint, link capacity constraint and flow conservation constraint in the multi-commodity flow problem model, listing relevant linear constraint, and optimizing the aim to minimize the maximum link utilization, as shown in the following:

minimize U

where U is the network maximum link utilization, x (p) represents the traffic distribution on path p; c (e) is the link capacity of link e;

step S3.2: the solution of the linear programming problem is accomplished using a linear programming solver, CPLEX or Gurobi, adding all paths satisfying x (p) > 0 to the set

；

Step S4: setting the constraint of the number of paths as h, and selecting the optimal path meeting the constraint of the number of paths from all feasible paths of each flow demand by using random rounding to obtain an optimal path set; the step S4 specifically includes:

step S4.1: for each traffic demand, according to probability

Randomly selecting a path

Join to final set of paths

And will route p from

Removing;

step S4.2: adding 1 to the number of the currently selected paths, and repeating the step 4.1 until h paths are selected according to the flow demand;

step S4.3: repeating the steps S4.1-S4.2, and finding an optimal path set under the path number constraint h for each flow demand;

step S5: and according to the optimal path set, considering the problem of multi-commodity flow and calculating the optimal flow distribution of the flow on the path.