A joint optimization method for multi-modal networks

Abstract

The application discloses a kind of joint optimization methods for multi-modal network, belong to computer network technical field.The application includes: by joint optimization the routing selection of traffic in multi-modal network and the allocation strategy of multi-modal network element to controller, realize control plane load balancing and the promotion of system throughput;On the basis of considering controller resource limit, link bandwidth constraint and multi-modal network element characteristics, adopt the optimization strategy of "two steps walk ": first, based on historical traffic data, determine the matching relationship between controller and multi-modal network element, optimize the load distribution of controller;Second, combined with the result of first step, design the online traffic routing algorithm of original-dual form, realize system throughput maximization under resource constraint.The method of the application can significantly reduce controller load and improve overall system throughput performance under the condition of maintaining resource constraint.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer networks, in particular to a control plane load balancing and traffic routing joint optimization method suitable for multi-modal networks, and specifically relates to a multi-modal network-oriented joint optimization method. BACKGROUND

[0002] With the rapid development of emerging businesses such as industrial internet, Internet of Vehicles, and remote medical treatment, the network has higher requirements for large bandwidth, low latency, and high reliability. New network modalities such as identity networks, naming center networks, and Mobility First are difficult to meet the needs of multi-modal network coexistence and dynamic evolution, although they have been applied in some scenarios. Therefore, the architecture idea of "multi-modal network environment" is proposed to realize the parallel operation and flexible loading of different network modalities on a unified physical network platform.

[0003] In a multi-modal network, the control plane is composed of multiple controllers, which are responsible for perceiving network states and scheduling resources, and the data plane is composed of multi-modal network element nodes, which perform packet analysis and forwarding. The current control plane mostly adopts the SDN architecture idea, that is, the controller receives the first packet of a new flow and issues a forwarding path. However, with the increase in the number of network elements and traffic, a single controller is prone to processing bottlenecks, leading to a decrease in system throughput. Although a distributed control plane can alleviate this problem, it still faces the problem of uneven controller load caused by dynamic changes in traffic in actual operation.

[0004] Existing research mainly focuses on the architecture, compilation, and system design of multi-modal networks, and pays less attention to the load balancing problem of distributed control planes. Some solutions introduce a controller migration mechanism to dynamically adjust the load, but have problems such as high migration cost and poor real-time performance, making it difficult to be applied to actual multi-modal network scenarios. Therefore, there is an urgent need for an efficient optimization method that can combine traffic routing and controller allocation, taking into account control plane load balancing and system throughput. SUMMARY

[0005] In view of the problem of uneven load distribution of the distributed control plane and the decrease in system throughput caused by dynamic changes in traffic, the purpose of the present application is to provide a multi-modal network-oriented joint optimization method.

[0006] The purpose of the present application is achieved by the following technical solution: a multi-modal network-oriented joint optimization method, comprising the following steps:

[0007] According to the multi-modal network environment, a system model is constructed; wherein the multi-modal network comprises a control plane and a data plane, the control plane comprises logically centralized controllers responsible for resource management and traffic scheduling, and the data plane is composed of multi-modal network elements and links responsible for forwarding data packets of multiple network modalities according to the instructions of the control plane;

[0008] The joint optimization of traffic routing of multiple network modalities and multi-modal network element-controller allocation takes the load balancing of the control plane and the maximization of system throughput as the optimization objectives; the matching relationship between the multi-modal network elements and the controllers is determined according to historical traffic, so as to balance the loads of the controllers; and the multi-modal network elements are sorted according to the amount of control resources consumed by them, and the multi-modal network elements with large resource requirements are preferentially allocated to the controllers with low current loads;

[0009] An online traffic routing algorithm based on the primal-dual theory is designed to find the revenue of the path corresponding to each traffic, and the path with the maximum revenue greater than 0 is adopted to output the routing scheme of the traffic corresponding to the path, so as to prevent the controllers from being overloaded due to the dynamics of the traffic and further improve the system throughput under the condition of load balancing of the controllers.

[0010] Further, the joint optimization of traffic routing of multiple network modalities and multi-modal network element-controller allocation comprises the following constraints:

[0011] The allocation constraint allocates a controller to each multi-modal network element for management;

[0012] The routing constraint selects a routing path for each traffic for forwarding;

[0013] The controller resource constraint is that when the controller issues a flow table to the multi-modal network, the resource capacity of the controller is not exceeded;

[0014] The link resource constraint is that the total traffic size passing through each link does not exceed the bandwidth capacity of the link.

[0015] Further, the joint optimization of traffic routing of multiple network modalities and multi-modal network element-controller allocation comprises:

[0016] Under the premise of given controller quantity, multi-modal topology, terminal distribution and traffic set, the optimization objective is decomposed into two sub-problems: which multi-modal network elements are managed by each controller; and which routing path is taken by each flow to forward data packets to the corresponding destination terminal.

[0017] Further, the load balancing of the control plane and maximizing system throughput are taken as optimization objectives, including introducing a weight coefficient to adjust the relative influence of the load balancing of the control plane and maximizing system throughput in optimization.

[0018] Further, the matching relationship between the multi-modal network elements and the controllers is determined according to historical traffic, and the load balancing of each controller includes:

[0019] A multi-modal network element-controller allocation algorithm is designed based on allocation constraints and controller load balancing constraints, wherein the allocation constraint is that each multi-modal network element must be allocated a controller for management, and the controller load balancing constraint is that minimizing the maximum load of the controller is taken as an optimization objective.

[0020] Based on the multi-modal network element-controller allocation algorithm, it is determined which multi-modal network elements each controller needs to manage, so as to determine the matching relationship between the multi-modal network elements and the controllers, and realize the load balancing of each controller.

[0021] Further, the multi-modal network elements are sorted according to the amount of control resources consumed by the multi-modal network elements, including:

[0022] The control resource demand of each multi-modal network element is calculated, and all multi-modal network elements are sorted according to the control resource demand;

[0023] The load of each controller is initialized to 0;

[0024] The multi-modal network elements sorted are allocated controllers with the minimum load one by one, and the load of each controller is ensured not to exceed the processing capacity during the allocation.

[0025] Further, the online traffic routing algorithm based on the primal-dual theory includes:

[0026] Constants and dual variables are initialized, wherein the constants represent the maximum usage of each resource in all routing schemes;

[0027] A path is selected from a candidate path set for each traffic for routing, the revenue of all paths is calculated, and it is calculated whether the maximum revenue is greater than 0, if less than 0, the traffic is discarded, and the dual variable is reset, if greater than 0, the path is taken as the best path, and the routing state and the dual variable are updated;

[0028] After all the traffics are processed, the routing scheme of the traffic of the best path is output.

[0029] The application also provides a joint optimization system for a multi-modal network, including a control plane load balancing module and a data plane throughput optimization module.

[0030] The application further provides an electronic device, comprising a memory and a processor, wherein the memory is coupled to the processor; the memory is used for storing program data, and the processor is used for executing the program data to realize the method.

[0031] The application further provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to realize the method.

[0032] The application has the following beneficial effects:

[0033] (1) The application constructs a joint optimization model considering the multi-modal network element-controller distribution relationship and the traffic routing path, and the target is to realize the control plane load balancing and the system throughput maximization. The model is a non-convex integer programming problem, and considers multiple key constraints such as controller processing capacity, link bandwidth, and topology connection relationship.

[0034] (2) The application proposes a 'two-step' solving method, which decomposes the complex problem into two stages:

[0035] First, based on historical traffic information, the static matching relationship between the multi-modal network element and the controller is determined, and the preliminary load balancing of the control plane is realized.

[0036] Second, an online traffic routing algorithm based on the primal-dual theory is designed to further improve the system throughput under the condition of meeting the resource constraint condition.

[0037] (3) The application realizes the load transfer between controllers through the dynamic adjustment of the traffic path, avoids local overload, and improves the robustness and scalability of the control plane without relying on the high-cost controller migration mechanism. BRIEF DESCRIPTION OF DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0039] Figure 1 It is a multi-modal network environment diagram of the method of the application;

[0040] Figure 2 It is an application environment diagram of the control channel network architecture of the application;

[0041] Figure 3 It is a multi-modal network element-controller distribution algorithm flowchart based on maximum load priority of the application;

[0042] Figure 4A flow chart of a real-time traffic routing algorithm based on the original dual method of the present application;

[0043] Figure 5 A flow chart of a joint algorithm of the present application. DETAILED DESCRIPTION

[0044] The present application will be described in detail below with reference to the accompanying drawings. The features in the following examples and embodiments can be combined with each other without conflict.

[0045] The present application, in a multi-modal network operating environment, in view of the actual situation that the controller needs to cope with the dynamic changes of traffic for a long time, jointly optimizes the multi-modal network element-controller allocation strategy and the traffic routing path design, and proposes an online optimization method with resource perception capability, so as to dynamically balance the controller load and improve the throughput performance and control efficiency of the overall system under the premise of guaranteeing resource constraints.

[0046] The embodiment of the present application provides a joint optimization method for a multi-modal network, which is suitable for the scenario that the traffic of the multi-modal network continuously and dynamically changes in actual operation, such as Figure 1 As shown in the figure. The method comprises the following steps:

[0047] S1, constructing a system model according to a multi-modal network environment.

[0048] S2, jointly optimizing the traffic routing of multiple network modalities and the multi-modal network element-controller allocation, taking the load balancing of the control plane and maximizing the system throughput as the optimization goal.

[0049] S3, determining the matching relationship between the multi-modal network elements and the controllers according to the historical traffic, so as to balance the load of each controller.

[0050] S4, sorting the multi-modal network elements according to the amount of control resources consumed by the multi-modal network elements, and preferentially allocating the multi-modal network elements with large resource demand to the controllers with low current load.

[0051] S5, designing an online traffic routing algorithm based on the original-dual theory, finding the revenue of the path corresponding to each traffic, adopting the path with the maximum revenue and greater than 0, and outputting the routing scheme of the traffic corresponding to the path, so as to prevent the controller from being overloaded due to the dynamic nature of the traffic, and further improve the system throughput under the condition of balancing the load of the controller.

[0052] In one of the embodiments, step S1 is specifically: according to Figure 1The system modeling of the multimodal network environment shown is performed. The multimodal network consists of a control plane and a data plane. The control plane contains a logically centralized controller responsible for resource management, traffic scheduling, and other tasks. The data plane consists of multimodal network elements and links, responsible for forwarding data packets of various network modes according to the instructions of the control plane. The network architecture of the control plane and data plane in the application environment is as follows: Figure 2 As shown.

[0053] Among them, this article uses Indicates controller, using This represents the set of controllers in the control plane, and the data plane is modeled as a directed acyclic graph. ,in Indicates a link. Represents a set of links. Represents multimodal network elements. This represents a set of multimodal network elements. The link is responsible for connecting the terminal and the multimodal network elements. The bandwidth capacity is denoted as Controller Processing power usage Let the set of network modes in a multimodal network be denoted as . Multimodal networks contain traffic from multiple network modes. The network modes are defined as follows: The set of traffic is represented as The set of traffic for all network modes in the network is represented as Every stream of traffic Can be formed by triples It means that, among them Represents a stream The source terminal, Represents a stream The destination terminal Represents a stream The network mode to which it belongs. Let the flow... The flow rate is and using binary constants Indicates network mode as Traffic The size of the control flow table information is issued. Table 1 below summarizes the parameter symbols used in this paper.

[0054] Table 1: Explanation of Parameter Symbols

[0055] In one embodiment, step S2 specifically involves: defining the problem based on the multimodal network environment described in step S1 and the problem to be solved by the present invention; optimizing the control plane load distribution in the multimodal network from the perspective of traffic routing; and performing joint optimization of traffic routing and multimodal network element-controller allocation (JRECA) for multiple network modes in the network. The joint algorithm flow is as follows: Figure 5 As shown. Specifically, given the number of controllers, multimodal topology and terminal distribution, and traffic set, JRECA needs to solve the following two sub-problems: a. Which multimodal network elements each controller is responsible for managing; b. Which routing path each flow takes to forward data packets to the corresponding destination terminal.

[0056] Furthermore, using binary variables Indicates controller Are you responsible for multimodal network elements? Traffic processing. If multimodal network elements Controller Management, then ;otherwise, Each stream There is a set of candidate paths This set can be obtained using the K-shortest path algorithm. Represents a stream The flow rate is determined using a binary variable. Represents a stream Use path Perform routing. If the flow... The routing path is ,but ;otherwise, .

[0057] Preferably, when jointly optimizing traffic routing and multimodal network element-controller allocation, the following constraints are considered:

[0058] Assignment constraints: For any multimodal network element, a controller must be assigned to manage it, that is:

[0059] ;

[0060] Routing constraints require that for any given traffic, a routing path must be selected for forwarding, i.e.:

[0061] ;

[0062] Controller resource constraints: When the controller distributes flow tables to the multimodal network, the resource capacity of the controller should not be exceeded.

[0063] ;

[0064] Link resource constraints stipulate that the total traffic passing through each link should not exceed the bandwidth capacity of that link, i.e.:

[0065] .

[0066] In one embodiment, the load balancing of the control plane and the maximization of system throughput described in step S2 are used as optimization objectives and calculated using the following formula, with weighting coefficients used to adjust the relative influence of these two components in the joint objective, namely:

[0067] .

[0068] In one embodiment, step S3 specifically involves designing a multimodal network element-controller allocation algorithm. This algorithm needs to determine which multimodal network elements each controller needs to manage, i.e., determine the matching relationship between multimodal network elements and controllers, so that the loads of each controller are similar. Define the controller load factor. , representing the proportion of controller resource consumption. The multimodal network element-controller matching problem is formalized as follows, where the first constraint indicates that each multimodal network element must be assigned a controller for management, the second constraint is the controller load balancing constraint, and the optimization objective is to minimize the maximum load on the controller, i.e. :

[0069] .

[0070] In one embodiment, step S4 specifically involves: based on the multimodal network element-controller allocation algorithm and considering the controller's processing capacity limitations, sorting the multimodal network elements according to their control resource consumption, and prioritizing the allocation of multimodal network elements with higher resource requirements to controllers with lower current loads, thereby achieving load balancing. A detailed algorithm flowchart is shown below. Figure 3 As shown, it includes the following steps:

[0071] S401, First calculate each multimodal network element. Controlling resource demand and according to Sort all multimodal network elements.

[0072] S402, Load each controller Initialize to 0.

[0073] S403. Assign the controller with the smallest load to each of the sorted multimodal network elements, ensuring that the load of each controller does not exceed its processing capacity during the assignment.

[0074] In one embodiment, step S5 specifically involves: In the multimodal network element-controller allocation algorithm, the present invention determines the matching relationship between multimodal network elements and controllers based on historical traffic information. However, due to the dynamic nature of traffic, controllers may still experience overload. Therefore, the controller's load must still be considered when routing traffic. The real-time routing problem for multiple network modal traffic is formalized as follows:

[0075] (1);

[0076] In formula (1), the first constraint is the routing constraint, which means that each traffic flow can select at most one path for routing; the second constraint is the controller processing capacity constraint; the third constraint is the data plane link constraint; and the optimization objective is to maximize the system throughput.

[0077] To address the aforementioned real-time routing problem, this invention designs a real-time traffic routing algorithm based on the primal-dual method. First, a dual problem is constructed for the linear relaxation problem of the aforementioned real-time routing problem formula. Let... , and Let be the dual variables of the first, second, and third inequalities, respectively, and all dual variables are non-negative. The dual problem can be formulated as:

[0078] (2);

[0079] The solution to the primal problem defined by formula (1) is constructed based on the dual problem. The first step of the algorithm is to initialize all dual variables and constants. , and Among them, constants This represents the maximum usage of each resource across all routing schemes, calculated using the following formula:

[0080] (3);

[0081] Where, constant The number of multimodal networks representing the number of the second and third inequalities in formula (2) is, i.e. ,constant This indicates the degree of reward or punishment for resource consumption.

[0082] For each traffic from the candidate path set Select a path The algorithm performs routing. It calculates each candidate path. Benefits The calculation method is shown in the following formula:

[0083] (4);

[0084] Among them, use This indicates the path with the highest profit among all paths. express The profit value. If This would violate the duality feasibility in formula (2), at which point the flow rate would be... It will not be processed. If The algorithm is based on the path Traffic Perform routing and update according to formula (4). For the controller and path The update rules for its dual variable are as follows:

[0085] ;

[0086] ;

[0087] in, and They represent flow rates respectively. After occupying resources and The value of . The specific process of the real-time traffic routing algorithm is as follows: Figure 4 As shown.

[0088] This invention also provides a joint optimization system for multimodal networks, including a control plane load balancing module and a data plane throughput optimization module. Since it basically corresponds to the method embodiment, relevant parts can be found in the description of the method embodiment.

[0089] This invention also provides an electronic device, including a memory and a processor, wherein the memory is coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to implement the method described in any of the above embodiments.

[0090] This invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the methods described in any of the above embodiments.

[0091] The computer-readable storage medium can be an internal storage unit of any data processing device described in any of the foregoing embodiments, such as a hard disk or memory. The computer-readable storage medium can also be any data processing device, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc., equipped on the device. Furthermore, the computer-readable storage medium can include both internal storage units of any data processing device and external storage devices. The computer-readable storage medium is used to store the computer program and other programs and data required by the data processing device, and can also be used to temporarily store data that has been output or will be output.

[0092] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0093] The above embodiments are only used to illustrate the design concept and features of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The protection scope of the present invention is not limited to the above embodiments. Therefore, all equivalent changes or modifications made based on the principles and design ideas disclosed in the present invention are within the protection scope of the present invention.

Claims

1. A joint optimization method for multimodal networks, characterized in that, The steps include the following: A system model is constructed based on a multimodal network environment; wherein, the multimodal network includes a control plane and a data plane, the control plane includes a logically centralized controller responsible for resource management and traffic scheduling, and the data plane consists of multimodal network elements and links, responsible for forwarding data packets of various network modes according to the instructions of the control plane; The system performs joint optimization of traffic routing and multimodal network element-controller allocation for multiple network modes, with the optimization objectives being to achieve load balancing of the control plane and maximize system throughput. The matching relationship between multimodal network elements and controllers is determined based on historical traffic to achieve load balancing among controllers. Multimodal network elements are sorted according to the amount of control resources they consume, and multimodal network elements with high resource requirements are preferentially allocated to controllers with lower current loads. We design an online traffic routing algorithm based on primal-dual theory to find the benefit of each path for each traffic flow. We select the path with the maximum benefit (greater than 0) and output the routing scheme for the traffic corresponding to that path. This prevents the controller from being overloaded due to the dynamic nature of traffic and further improves the system throughput under the condition of controller load balancing.

2. The method according to claim 1, characterized in that, The joint optimization of traffic routing and multimodal network element-controller allocation for multiple network modes includes the following constraints: Assignment constraints: For any multimodal network element, assign a controller to manage it; Routing constraints: For any given traffic, select a routing path for forwarding. Controller resource constraints: when the controller issues flow tables to the multimodal network, the resource capacity of the controller shall not be exceeded. Link resource constraints: the total traffic passing through each link cannot exceed the bandwidth capacity of that link.

3. The method according to claim 1, characterized in that, The joint optimization of traffic routing and multimodal network element-controller allocation for multiple network modes includes: Given the number of controllers, multimodal topology, terminal distribution, and traffic set, the optimization objective is decomposed into two sub-problems: which multimodal network elements each controller is responsible for managing; and which routing path each flow takes to forward data packets to the corresponding destination terminal.

4. The method according to claim 1, characterized in that, The optimization objective of achieving load balancing of the control plane and maximizing system throughput includes introducing weighting coefficients to adjust the relative influence of load balancing of the control plane and maximizing system throughput in the optimization.

5. The method according to claim 1, characterized in that, The step of determining the matching relationship between multimodal network elements and controllers based on historical traffic to achieve load balancing among controllers includes: Based on allocation constraints and controller load balancing constraints, a multimodal network element-controller allocation algorithm is designed. The allocation constraint requires that each multimodal network element be assigned a controller for management, and the controller load balancing constraint is to minimize the maximum load of the controller as the optimization objective. Based on the multimodal network element-controller allocation algorithm, it is determined which multimodal network elements each controller needs to manage, so as to determine the matching relationship between multimodal network elements and controllers and realize that the load of each controller is similar.

6. The method according to claim 1, characterized in that, The sorting based on the amount of control resources consumed by multimodal network elements includes: Calculate the control resource requirements for each multimodal network element, and sort all multimodal network elements according to the control resource requirements; Initialize the load of each controller to 0; Assign the controller with the least load to each of the sorted multimodal network elements, while ensuring that the load of each controller does not exceed its processing capacity.

7. The method according to claim 1, characterized in that, The online traffic routing algorithm based on primal-dual theory includes: Initialize constants and dual variables, where the constants represent the maximum usage of each resource in all routing schemes; For each traffic, select a path from the candidate path set for routing, calculate the revenue of all paths and check if the maximum revenue is greater than 0. If it is less than 0, discard the traffic and reset the dual variable. If it is greater than 0, take the path as the best path and update the routing state and dual variable. After processing all traffic, output the routing scheme for the traffic with the best path.

8. A joint optimization system for multimodal networks, used to implement the method according to any one of claims 1-7, characterized in that, It includes a control plane load balancing module and a data plane throughput optimization module.

9. An electronic device comprising a memory and a processor, characterized in that, The memory is coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to implement the method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-7.

Images