CN109818865B - SDN enhanced path boxing device and method - Google Patents
SDN enhanced path boxing device and method Download PDFInfo
- Publication number
- CN109818865B CN109818865B CN201910182584.1A CN201910182584A CN109818865B CN 109818865 B CN109818865 B CN 109818865B CN 201910182584 A CN201910182584 A CN 201910182584A CN 109818865 B CN109818865 B CN 109818865B
- Authority
- CN
- China
- Prior art keywords
- network
- service
- particle
- path
- sdn
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
A Software Defined Network (SDN) enhanced path boxing device and a method are used for solving the boxing problem occurring in a path calculation unit based on an SDN controller. The invention comprehensively considers the related constraint conditions of bandwidth resources, disturbance, time delay and the like in a communication network, uses a particle swarm algorithm, is combined with the related technology of an SDN controller network topology planning module, aims to intelligently adjust service path planning under the condition of insufficient bandwidth resources of new services in the network and provides a boxing scheme, and belongs to the field of optimization of intelligent algorithms of telecommunication networks. The main innovation points comprise: 1. the method comprises the following steps of (1) arranging a network boxing service based on an SDN technology; 2. network service scheduling based on SDN is introduced into a boxing decision model; 3. and improving the utilization rate of the network global bandwidth by utilizing a particle swarm algorithm model.
Description
Technical Field
The invention belongs to the field of telecommunication network intelligent algorithm optimization, and particularly relates to a SDN enhanced path boxing device and method.
Background
In the current environment of communication networks, bandwidth resources are the most critical resources in the network. When the network load is unbalanced, that is, the utilization rate of the local network bandwidth in the network is too high, and the utilization rate of the rest bandwidth is too low, unnecessary network congestion and resource waste are caused, and a legal path cannot be calculated for the newly added service under the condition that the current overall network bandwidth resource is sufficient. In order to realize reasonable utilization of network resources, the resource use distribution of the network can be adjusted through some intelligent algorithms, and an optimal path allocation scheme is calculated, namely, a communication link of deployed services in the network is adjusted through global optimization, so that the network resources where new services are deployed are vacated in the whole network, and the disturbance rate of the whole network services is minimum.
In the traditional router solution, the calculation of the service path is mainly realized by a shortest path first algorithm under the condition that the hop count, the delay, the occupied bandwidth resource and the link cost are constraints. The defects of the traditional scheme are as follows: (1) if the traffic from a source node to a destination node exceeds the capacity of the shortest path, the shortest path will become congested, but at the same time a longer path between the two points may not be fully used; (2) in the case where shortest paths from different source nodes overlap on a link, congestion can occur if the total traffic through the link exceeds the capacity of the link. With the increase of network access clients, the scale of the existing network is rapidly increased, and the network congestion situation is increased.
Particle Swarm Optimization (PSO) is a population-based optimization algorithm. The method has the characteristics of quick convergence, simplicity in operation and the like, is widely applied to various fields such as engineering, economic management and the like, and becomes a new hotspot for research in the field of intelligent computing. The particle swarm optimization algorithm is an evolutionary algorithm with the advantages of concise form, rapid convergence, flexible parameter adjusting mechanism and the like, has been successfully applied to a single-target optimization problem, and is considered to be one of the most potential methods for solving the multi-target optimization problem.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides an SDN enhanced path boxing device and method, and solves the technical problem of boxing network load newly added services in an SDN communication network.
In order to achieve the purpose, the invention adopts the following technical scheme:
an SDN enhanced path binning apparatus, comprising: the network service dispatching center is a module for providing basic communication and management functions for the system, and a core component of the network service dispatching center is an SDN controller; the component interface of the network service dispatching center comprises: the interface comprises a northbound interface used for transmitting customized parameters, a data packet interface used for transmitting packet-in or packet-out messages, a flow table interface used for issuing a flow table or forwarding the flow table, and an interface used for acquiring network topology; the functional module of the network service dispatching center comprises: the system comprises a network topology identification module, a service load evaluation module and a network routing calculation module;
the service load evaluation module evaluates index information entering a network, is realized through a big data technology, and constructs a set of service index big database in a network service scheduling center, so that the service index is quickly evaluated according to the service information;
the route calculation module calculates a flow path from the data flow inlet to the forwarding device and then to the data flow outlet based on the topology information of the network and the information of the network security device.
In order to optimize the technical scheme, the specific measures adopted further comprise:
further, the SDN controller employs flodlight, POX, or NOX; the index information comprises the size, time delay and transmission rate of data flow.
Meanwhile, the invention also provides a boxing method of the SDN enhanced path boxing device, which is characterized by comprising the following steps:
the method comprises the following steps: when the data flow of the newly added service in the current working network needs to request network resources in the production environment, if the resources are insufficient, a newly added network service loading request is sent to a network service scheduling center, and after the network service scheduling center receives the service loading request, a network service path boxing mechanism is started;
step two: after a network service path boxing mechanism is started, the OpenFlow switch sends a data flow mirror image to a service load evaluation module in a network service dispatching center and identifies index information;
step three: after receiving a service loading request through a northbound interface, an SDN controller in a network service scheduling center issues a mirror flow table to a flow inlet, a flow outlet and a mounted network switch node in a network;
step four: running a k shortest path algorithm to calculate a plurality of standby paths for each service to be deployed, then taking the standby paths as input, running a particle swarm algorithm again, calculating an optimal scheme for placing a new service into the network, ensuring that the new service can be added into the SDN network, and simultaneously disturbing the deployed service as little as possible;
step five: after a network service dispatching center calculates a dispatching route of flow, an SDN controller issues a flow table to an OpenFolw switch on the route, and data forwarding equipment on the route is connected into a data link;
step six: after receiving the flow table, the OpenFlow switch drains the data flow to the corresponding switch device, the switch adjusts the path of the service flow which does not meet the requirements, modifies the flow table item of the SDN switch through which the service flow passes, and simultaneously issues the newly added service to the flow table of the corresponding switch, so that the switch completes the forwarding of the new service.
In order to optimize the technical scheme, the specific measures adopted further comprise:
further, the calculating of the backup path in the fourth step means that when one or more newly added services cannot be directly deployed, the new services are boxed into the production network through global adjustment based on a particle swarm algorithm.
Further, the particle swarm algorithm in the fourth step specifically includes the following steps:
the encoding mode of the particle swarm adopts natural number encoding, the encoding length is only related to the number of services and is not related to the number of the standby path concentration paths of the services, the number of the standby paths of the services is only related to the solution space, and the position x of one particleiThe method is characterized in that a path is randomly selected from a standby path set of each service, and in the position represented by the ith particle of a particle swarm, a service k selects the first particle in the standby path set
A route is then
In this D-dimensional space, there are m particles in total, i.e. the size of the particle population is m; particle i position:
velocity of particle i:
i is more than or equal to 1 and less than or equal to m; historical best position for particle i experienced:
best positions experienced by all particles within the population:
in the particle swarm optimization, an individual Position has two data fields of a p array and a v array which respectively represent Position and speed, the p array represents the Position of the current solution, the solution is n-dimensional, and n represents the existing service quantity; the particle swarm algorithm realizes the boxing problem by globally adjusting a standby path in the existing network, wherein the length of a solution is the same as the number of services of the existing network, the value of the corresponding position of the solution is the selection (0.. once.n.) of the working path of the corresponding service, and the algorithm comprises the following steps:
step 1) initializing a particle swarm, randomly selecting 0-n at each position of an array, wherein n represents the number of current services, and the dimensionality of a solution is determined by the number of the services;
step 2) updating individual pBest according to speed position iteration formulaiAnd then updating gBest according to the global state, wherein pBestiRepresenting the historical optimal solution of the individual particle i, wherein gBest represents the current global optimal solution;
step 3) entering an iterative loop, and jumping out of the loop when the solution is converged;
step 4), outputting an optimal solution gBest;
the final solution is the path serial number in the current network, and the fitness function in the iteration process is determined by the disturbance number.
Meanwhile, the invention also provides a boxing method of the SDN enhanced path boxing device, which is characterized by comprising the following steps:
step 101: a new service enters the system, when a first data packet of a new service enters the SDN, the first data packet reaches the SDN switch, and step 102 is executed;
step 102: the SDN switch sends the data packet or the data packet header to the SDN controller, and step 104 is executed;
step 103: the SDN controller collects all traffic information and resource information in the network it controls, including: entering into step 105, flow table entries in the switch, use of network node resources, bandwidth utilization of the link, topology of the service, and resource requirements of the service;
step 104: the SDN controller starts a PCE, calculates a path which needs to be passed by the service through calculation or policy management, and if the path is successfully calculated, adds path information of the service into a flow entry of an SDN switch to complete the deployment of the service; if the SDN controller cannot find a suitable routing path, execute step 103;
step 105: the SDN controller starts a PCE, firstly runs a genetic algorithm to generate a plurality of standby paths of each service, then uses a particle swarm algorithm again by taking the paths as input, calculates the optimal scheme for placing the services into the network, ensures that the new service can be added into the SDN network, has minimum disturbance on other services in the network, realizes load balancing, and enters step 106;
step 106: the SDN controller transmits the calculated new configuration information to an SDN switch in the network, and the step 107 is entered;
step 107: after receiving the configuration information, the SDN switch adjusts the path of the service flow which does not meet the requirement, modifies the flow table item of the SDN switch through which the service flow passes, and simultaneously issues the newly added service to the flow table of the corresponding switch to enable the switch to complete the forwarding of the new service;
step 108: deployment of the new service is completed and load balancing of the SDN network is completed.
Meanwhile, the invention also provides a boxing method of the SDN enhanced path boxing device, which is characterized by comprising the following steps:
step 201: the newly added service enters the network and requests network resources, and firstly, various constraint conditions and required resource conditions of the newly added service are evaluated;
step 202: load evaluation, before starting network service boxing, firstly judging whether the current residual network resources meet the direct calculation of a newly added service path, if so, entering a step 203;
step 203: calculating a path, namely running a ksp algorithm to calculate the path of the newly added service, and directly calculating the path of the newly added service when all network resources and constraint conditions are met;
step 204: performing boxing operation, namely calling a boxing algorithm when the resource condition of the current network cannot meet the requirement of the newly added service, and adjusting the deployed service to meet the resource requirement of the newly added service;
step 205: and adjusting network service paths, namely after all the old services are adjusted and all available path sets are calculated for the newly added services, formally deploying the old services to be adjusted and the newly added services, and finishing the whole boxing operation.
In order to optimize the technical scheme, the specific measures adopted further comprise:
further, in the step 204, a particle swarm algorithm is adopted to solve the problem of packing, and the specific steps are as follows:
step 301: initializing a population of particles, the population size being m, including random position p and velocity v;
step 302: evaluating the fitness of each microparticle;
step 303: for each particle, comparing the objective function value with the best position pBest passed by the particle, and if the objective function value is better, taking the objective function value as the current best position pBest;
step 304: finding out the particle with the highest objective function value, namely the best position gBest in the current population;
step 305: adjusting the particle velocity and position according to the following formula;
where t denotes the number of iterations, ω denotes the inertial weight, c1、c2Representing an acceleration factor, representing an individual learning factor and a social learning factor, ri、r2Is represented by [0,1 ]]Random numbers, pBest, uniformly distributed over the surfaceiRepresenting the historical optimal solution of the individual particle i, gBest representing the current global optimal solution, vi(i) Representing the velocity, v, of the particle i at the number of iterations ti(t +1) represents the velocity of the particle i at the number of iterations t +1, xi(t) denotes the position of the particle i at the number of iterations t, xi(t +1) represents the position of the particle i at the iteration number t + 1;
step 306: if the end condition is not met, turning to step 302, selecting the iteration end condition as the maximum iteration frequency or the optimal position searched by the particle group so far according to a specific problem to meet a preset minimum adaptation threshold;
in the particle swarm optimization, the particles are composed of position vectors and velocity vectors, and the adaptive value of the current position is calculated according to a target function; in each speed iteration, the particles learn according to the best history record of the particles and the optimal particles in the current population, so that how to adjust and change the flying speed and direction next time is determined, and finally, the position vector of the particles after one iteration is finished is determined;
in the particle swarm packing algorithm, the quality between the particle positions is compared by a target value, pBest of each generation of the particle swarmiAnd gBest is also updated by comparing the objective function values between the particles; the minimum ratio of the disturbed service number to the total service number is required to be solved under the network problem model, so the following objective function is selected:
where θ is the disturbance rate experienced by the traffic; when ω > 1, it means that the link bandwidth is overloaded, i.e. the current network resource cannot carry all the traffic of the current network, so the particle objective function is set to-1, and when one particle objective function value is-1, the particle is an infeasible solution.
The invention has the beneficial effects that: a particle swarm optimization-based enhanced path boxing method in an SDN controller is provided, a high-efficiency path boxing strategy is obtained by combining the flexible deployment capacity of an SDN path computing unit, personalized service deployment service can be provided in a multi-constraint complex network environment, and meanwhile, the routing computing efficiency and the whole bandwidth resource utilization rate of network services are improved by computing an optimal path with lower network cost.
Drawings
Figure 1 is a basic architecture diagram of the SDN network enhanced path binning problem.
Fig. 2 is a flow chart of a binning method.
FIG. 3 is a flow chart of a particle swarm algorithm for solving the bin packing problem.
Detailed Description
The present invention will now be described in further detail with reference to the accompanying drawings.
The invention provides an SDN enhanced path boxing device, which comprises:
the network service dispatching center refers to a module for providing basic communication and management functions for a system, and a core component of the network service dispatching center is an SDN controller, such as Floodlight, POX, NOX and the like. The component interface mainly comprises: 1. a northbound interface for communicating customization parameters; 2. a data packet interface for transmitting packet-in or packet-out messages; 3. a flow table interface for issuing a flow table or forwarding the flow table; 4. an interface for obtaining a network topology. Its main functional module includes: identification of network topology, traffic load assessment, network routing computation, etc.
The service load evaluation refers to the evaluation of index information such as the size and time delay of data flow entering a network, the function can be realized by adopting the existing technologies such as big data, and a set of service index big database is constructed in a network service scheduling center, so that the service index can be quickly evaluated according to the service information.
And the routing calculation refers to calculating a flow path from a data traffic inlet to a forwarding device and then to a data traffic outlet based on the topology information of the network and the information of the network security device.
The invention provides an SDN enhanced path boxing method, which comprises the following steps:
the method comprises the following steps: when the data traffic of the newly added service in the current working network needs to request the network resource in the production environment, if the resource is insufficient, a newly added network service loading request is sent to the network service scheduling center at this time, and after the network service scheduling center receives the service loading request, a network service path boxing mechanism is started.
Step two: after a network service path boxing mechanism is started, the OpenFlow switch sends a data flow mirror image to a service load evaluation module in a network service scheduling center, and identifies index information such as the size and the transmission rate of data flow.
Step three: and after receiving a service loading request through a northbound interface, an SDN controller in the network service scheduling center issues a mirror flow table to a flow inlet, a flow outlet and a mounted network switch node in the network.
Step four: and operating a k shortest path algorithm to calculate a plurality of standby paths for each service to be deployed, then operating the particle swarm algorithm again by taking the standby paths as input, calculating an optimal scheme for placing a new service into the network, ensuring that the new service can be added into the SDN network, and simultaneously disturbing the deployed service as little as possible. The standby path is calculated by boxing the new service into the production network through global adjustment based on the particle swarm algorithm when one or more new services cannot be directly deployed.
Step five: after the network service dispatching center calculates the dispatching route of the flow, the SDN controller issues the flow table to the OpenFolw switch on the route, and data forwarding devices on the route are connected into a data link.
Step six: after receiving the flow table, the OpenFlow switch drains the data flow to the corresponding switch device, the switch adjusts the path of the service flow which does not meet the requirements, modifies the flow table item of the SDN switch through which the service flow passes, and simultaneously issues the newly added service to the flow table of the corresponding switch, so that the switch completes the forwarding of the new service.
Before the method, the service load optimization is carried out, and the basic steps are as follows:
a. and (3) newly added service data input: new traffic service requirements appear in the network, and the measurement of the network condition comprises the preprocessing of data, such as the bandwidth resource, the node number, the time delay and the like of the network; and on the other hand, the data analysis of the service which is calculated to be loaded into the network by the path, such as the starting node, the required link bandwidth, the number of hops which cannot be exceeded and the like, is specifically described.
b. Establishing, modeling and analyzing an optimization target: a mathematical model is established for actual network characteristics and traffic service characteristics, factors influencing network performance are found out, and network performance, network operation control and network design are improved.
c. Outputting an optimization result: on the basis of abstracting and analyzing the current network and transmission requirements, the deployed service paths are adjusted through control behaviors, and the goal of global load balancing optimization is achieved.
The SDN path enhancement boxing method and device disclosed by the invention have the advantages that a boxing triggering mechanism is issued through newly adding a service request, extracting service constraint and deploying service scheduling; then generating a standby path by a k shortest path algorithm, and calculating a business flow boxing strategy based on a particle swarm algorithm; and finally, the service scheduling center completes the issuing of the scheduling strategy. Compared with a direct boxing method in the traditional method, the method has the advantages that the load disturbance degree of the network before and after the business boxing can be more conveniently understood and compared, and the balanced and low-disturbance boxing can be realized.
The encoding mode of particle swarm currently used encoding technologies include binary encoding, natural number encoding, floating point encoding, character encoding and the like. The packing problem proposed herein is assumed to be natural number coding, where the coding length is only related to the number of services, not to the number of paths in the backup path set of the services, and the number of backup paths of the services is only related to the solution space. Position x of a particleiConsisting of a random selection of a path from the set of alternate paths for each service, e.g. the first in the set of alternate paths selected by service k in the position represented by the ith particle of the group of particles
A route is then
In this D-dimensional space, there are m particles in total, i.e. the size of the particle population is m; particle i position:
velocity of particle i:
i is more than or equal to 1 and less than or equal to m, and the historical best position of the particle i has undergone:
best positions experienced by all particles within a population (or within a domain):
in the particle swarm optimization, an individual Position has two data fields of a p array and a v array which respectively represent Position and speed, the p array represents the Position of the current solution, the solution is n-dimensional in the optimization, and n represents the existing service quantity; the particle swarm algorithm is used for realizing the boxing problem based on the boxing model mentioned above, the standby path in the existing network is also adjusted globally, the length of the solution is the same as the number of the services of the existing network, the numerical value of the corresponding position of the solution is the selection (0.. said., n) of the working path of the corresponding service, and the algorithm comprises the following steps:
step 1) initializing a particle swarm, randomly selecting 0-n at each position of an array, wherein n represents the number of current services, and the dimensionality of a solution is determined by the number of the services.
Step 2) updating individual pBest according to speed position iteration formulaiAnd then updating gBest according to the global state, wherein pBestiRepresenting the historical optimal solution of the individual particle i, and the gBest representing the current global optimal solution.
And 3) entering an iterative loop, and jumping out of the loop when the solution is converged.
And 4) outputting the optimal solution gBest.
The final solution is the path serial number in the current network, and the fitness function in the iteration process is determined by the disturbance number.
In the problem, the disturbance rate is an objective function, so that the pheromone of the current working path of the current network service is set to be 1.5, and other standby paths are set to be 1, so that the probability of selecting the current working path can be increased, and the convergence can be accelerated.
Figure 1 presents a basic architecture diagram of the SDN network enhanced path binning problem. As shown in the figure, a simplified model is adopted to clearly describe the boxing method in the present invention, and the implementation process of the basic framework of the SDN network enhanced path boxing problem is divided into the following steps:
step 101: a new service enters the system; when a first data packet of a new service enters the SDN, it arrives at the SDN switch, and step 102 is executed.
Step 102: the SDN switch sends the data packet or data packet header to the SDN controller, and step 104 is performed.
Step 103: the SDN controller collects all traffic information and resource information in the network it controls: flow table entries in the switch, the use condition of network node resources, the bandwidth utilization rate of the link, the topology of the service, the resource requirements of the service, and the like. Step 105 is entered.
Step 104: the SDN controller starts a PCE, calculates a path which needs to be passed by the service through calculation or policy management, and if the path is successfully calculated, adds path information of the service into a flow entry of an SDN switch to complete the deployment of the service; if the SDN controller cannot find a suitable routing path, step 103 is performed.
Step 105: the SDN controller starts a PCE, first runs a genetic algorithm to generate a plurality of standby paths for each service, and then uses a particle swarm algorithm again with the paths as inputs to calculate an optimal scheme for placing the services into the network, so as to ensure that a new service can be added into the SDN network, and the disturbance to other services in the network is minimal, thereby realizing load balancing, and entering step 106.
Step 106: the SDN controller transmits the calculated new configuration information to the SDN switch in the network, and then proceeds to step 107.
Step 107: after receiving the configuration information, the SDN switch adjusts the path of the service flow which does not meet the requirements, modifies the flow table item of the SDN switch through which the service flow passes, and simultaneously issues the newly added service to the flow table of the corresponding switch, so that the switch completes the forwarding of the new service.
Step 108: deployment of the new service is completed and load balancing of the SDN network is completed.
Fig. 2 shows a flow chart of the boxing method of the present invention, which includes the following steps:
step 201: the newly added service enters the network and requests network resources, and various constraint conditions and required resource conditions of the newly added service are evaluated at first.
Step 202: and (4) load evaluation, namely, before starting network service boxing, firstly judging whether the current residual network resources meet the direct calculation of a newly added service path, and if the resources meet the requirement, entering a step 203.
Step 203: and (4) path calculation, namely running a ksp algorithm to calculate the path of the newly added service, and directly calculating the path of the newly added service when all network resources and constraint conditions are met.
Step 204: and (4) boxing operation, namely when the resource condition of the current network cannot meet the requirement of the newly added service, calling a boxing algorithm in the text, and adjusting the deployed service to meet the resource requirement of the newly added service.
Step 205: and adjusting network service paths, namely after all the old services are adjusted and all available path sets are calculated for the newly added services, formally deploying the old services to be adjusted and the newly added services, and finishing the whole boxing operation.
Fig. 3 shows a schematic flow chart of solving the boxing problem by the particle swarm algorithm, which specifically includes the following steps:
step 301: a population of particles (population size m) is initialized, including random positions p and velocities v.
Step 302: the fitness of each microparticle was evaluated.
Step 303: for each particle, its objective function value is compared to its past best position pBest, and if better, it is taken as the current best position pBest.
Step 304: and finding out the particle with the highest objective function value, namely the best position gBest in the current population.
Step 305: the particle velocity and position are adjusted according to the formula:
where t denotes the number of iterations, ω denotes the inertial weight, c1、c2Representing an acceleration factor, representing an individual learning factor and a social learning factor, r1、r2Is represented by [0,1 ]]Random numbers, pBest, uniformly distributed over the surfaceiRepresenting the historical optimal solution of the individual particle i, wherein gBest represents the current global optimal solution; v. ofi(t) represents the velocity of the particle i at the number of iterations t, vi(t +1) represents the velocity of the particle i at the number of iterations t + 1; x is the number ofi(t) denotes the position of the particle i at the number of iterations t, xi(t +1) represents the position of the particle i at the number of iterations t + 1.
Step 306: if the end condition is not met, go to step 302.
The iteration termination condition is generally selected according to a specific problem as the maximum number of iterations or the optimal position searched so far by the particle group satisfies a predetermined minimum adaptation threshold.
In solving the optimization problem by using the particle swarm algorithm, the particles are composed of a position vector (the position of the particle in the solution space) and a velocity vector (the direction and the velocity for determining the next flight), and an adaptive value (which can be understood as the distance of the bird from the food) of the current position is calculated according to an objective function. In each speed iteration, the particles can learn according to the best history record of the particles and the optimal particles in the current population, so that how to adjust and change the flying speed and direction next time is determined, and finally, the position vector of the particles after one iteration is completed is determined.
In the particle swarm packing algorithm, the quality between the particle positions is compared by a target value, pBest of each generation of the particle swarmiAnd gBest is also updated by comparing the value of the objective function between the particles. The minimum ratio of the disturbed traffic number to the total traffic number can be solved according to the requirements of the network problem model, so that the following objective function is selected:
where θ is the disturbance rate experienced by the traffic. It should be noted that when ω > 1 indicates that the link bandwidth is overloaded, i.e. the current network resource cannot carry all the traffic of the current network, the particle objective function is set to-1, and when one particle objective function value is-1, the particle is not a feasible solution.
Under the condition, a solution which not only meets the load balance but also meets the newly added service deployment is solved through a particle swarm algorithm. The optimization problem is essentially an NP-hard problem, the solution of the NP-hard problem is extremely difficult, no proper polynomial algorithm is available for solving the problem, and engineering generally tends to comprehensively consider algorithm efficiency and an optimization result to seek a satisfactory solution of the problem. The traditional optimization methods such as branch-and-bound and dynamic planning solve the problem of the combined explosion, and even for the problem with a small scale, the operation time is unacceptable, not to mention the actual problem with a large scale. Therefore, people often employ heuristic algorithms to solve the complete problem in real life. The optimization problem is characterized by large search space, and the particle swarm algorithm capable of rapidly converging is considered for solving.
The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above-mentioned embodiments, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention. It should be noted that modifications and embellishments within the scope of the invention may be made by those skilled in the art without departing from the principle of the invention.
Claims (6)
1. A binning method of an SDN enhanced path binning apparatus, the SDN enhanced path binning apparatus comprising: the network service dispatching center is a module for providing basic communication and management functions for the system, and a core component of the network service dispatching center is an SDN controller; the component interface of the network service dispatching center comprises: the interface comprises a northbound interface used for transmitting customized parameters, a data packet interface used for transmitting packet-in or packet-out messages, a flow table interface used for issuing a flow table or forwarding the flow table, and an interface used for acquiring network topology; the functional module of the network service dispatching center comprises: the system comprises a network topology identification module, a service load evaluation module and a network routing calculation module; the service load evaluation module evaluates index information entering a network, is realized through a big data technology, and constructs a set of service index big database in a network service scheduling center, so that the service index is quickly evaluated according to the service information; the routing calculation module calculates a flow path from a data flow inlet to a forwarding device and then to a data flow outlet based on the topology information of the network and the information of the network security device; the method is characterized by comprising the following steps:
the method comprises the following steps: when the data flow of the newly added service in the current working network needs to request network resources in the production environment, if the resources are insufficient, a newly added network service loading request is sent to a network service scheduling center, and after the network service scheduling center receives the service loading request, a network service path boxing mechanism is started;
step two: after a network service path boxing mechanism is started, the OpenFlow switch sends a data flow mirror image to a service load evaluation module in a network service dispatching center and identifies index information;
step three: after receiving a service loading request through a northbound interface, an SDN controller in a network service scheduling center issues a mirror flow table to a flow inlet, a flow outlet and a mounted network switch node in a network;
step four: the shortest path algorithm is operated to calculate a plurality of standby paths for each service to be deployed, then the standby paths are used as input, the particle swarm algorithm is operated again, the optimal scheme for placing a new service into the network is calculated, the new service can be ensured to be added into the SDN network, and meanwhile, the deployed service is disturbed as little as possible; the method comprises the following specific steps:
step 301, initializing a population of particles, wherein the population size is m and comprises a random position p and a velocity v;
step 302, evaluating the fitness of each particle;
step 303, comparing the objective function value of each particle with the best position pBest passed by the particle, and if the objective function value is better, taking the objective function value as the current best position pBest;
step 304, finding out the particle with the highest objective function value, namely the best position gBest in the current population;
step 305, adjusting the speed and the position of the particles according to the following formula;
where t denotes the number of iterations, ω denotes the inertial weight, c1、c2Representing an acceleration factor, representing an individual learning factor and a social learning factor, r1、r2Is represented by [0,1 ]]Random numbers, pBest, uniformly distributed over the surfaceiRepresenting the historical optimal solution of the individual particle i, gBest representing the current global optimal solution, vi(t) represents the velocity of the particle i at the number of iterations t, vi(t +1) represents the velocity of the particle i at the number of iterations t +1, xi(t) denotes the position of the particle i at the number of iterations t, xi(t +1) represents the position of the particle i at the iteration number t + 1;
step 306: if the end condition is not met, turning to step 302, and selecting the iteration end condition as the maximum iteration frequency or the optimal position searched by the particle group so far to meet the preset minimum adaptation threshold value;
in the particle swarm packing algorithm, the quality between the particle positions is compared by a target value, pBest of each generation of the particle swarmiAnd gBest is also updated by comparing the objective function values between the particles; the minimum ratio of the disturbed service number to the total service number is required to be solved under the network problem model, so the following objective function is selected:
where θ is the disturbance rate experienced by the traffic; when omega is more than 1, the link bandwidth is overloaded, that is, the current network resource can not bear all the services of the current network, so the particle objective function is set to be-1, and when one particle objective function value is-1, the particle is an infeasible solution;
step five: after a network service dispatching center calculates a dispatching route of flow, an SDN controller issues a flow table to an OpenFolw switch on the route, and data forwarding equipment on the route is connected into a data link;
step six: after receiving the flow table, the OpenFlow switch drains the data flow to the corresponding switch device, the switch adjusts the path of the service flow which does not meet the requirements, modifies the flow table item of the SDN switch through which the service flow passes, and simultaneously issues the newly added service to the flow table of the corresponding switch, so that the switch completes the forwarding of the new service.
2. A boxing method in accordance with claim 1, wherein: and calculating the standby path in the fourth step means that when one or more newly added services cannot be directly deployed, the new services are boxed into the production network through global adjustment based on a particle swarm algorithm.
3. A boxing method in accordance with claim 2, wherein: the particle swarm algorithm in the fourth step specifically comprises the following steps:
the encoding mode of the particle swarm adopts natural number encoding, the encoding length is only related to the number of services and is not related to the number of the standby path concentration paths of the services, the number of the standby paths of the services is only related to the solution space, and the position x of one particleiThe method is characterized in that a path is randomly selected from a standby path set of each service, and in the position represented by the ith particle of a particle swarm, a service k selects the first particle in the standby path set
A route is then
In this D-dimensional space, there are m particles in total, i.e. the size of the particle population is m; particle i position:
velocity of particle i:
i is more than or equal to 1 and less than or equal to m; historical best position for particle i experienced:
best positions experienced by all particles within the population:
in the particle swarm optimization, an individual Position has two data fields of a p array and a v array which respectively represent Position and speed, the p array represents the Position of the current solution, the solution is n-dimensional, and n represents the existing service quantity; the particle swarm algorithm realizes the boxing problem by globally adjusting a standby path in the existing network, wherein the length of a solution is the same as the number of services of the existing network, the value of the corresponding position of the solution is the selection (0.. once.n.) of the working path of the corresponding service, and the algorithm comprises the following steps:
step 1) initializing a particle swarm, randomly selecting 0-n at each position of an array, wherein n represents the number of current services, and the dimensionality of a solution is determined by the number of the services;
step 2) updating individual pBest according to speed position iteration formulaiAnd then updating gBest according to the global state, wherein pBestiRepresenting the historical optimal solution of the individual particle i, wherein gBest represents the current global optimal solution;
step 3) entering an iterative loop, and jumping out of the loop when the solution is converged;
step 4), outputting an optimal solution gBest;
the final solution is the path serial number in the current network, and the fitness function in the iteration process is determined by the disturbance number.
4. A binning method of an SDN enhanced path binning apparatus, the SDN enhanced path binning apparatus comprising: the network service dispatching center is a module for providing basic communication and management functions for the system, and a core component of the network service dispatching center is an SDN controller; the component interface of the network service dispatching center comprises: the interface comprises a northbound interface used for transmitting customized parameters, a data packet interface used for transmitting packet-in or packet-out messages, a flow table interface used for issuing a flow table or forwarding the flow table, and an interface used for acquiring network topology; the functional module of the network service dispatching center comprises: the system comprises a network topology identification module, a service load evaluation module and a network routing calculation module; the service load evaluation module evaluates index information entering a network, is realized through a big data technology, and constructs a set of service index big database in a network service scheduling center, so that the service index is quickly evaluated according to the service information; the routing calculation module calculates a flow path from a data flow inlet to a forwarding device and then to a data flow outlet based on the topology information of the network and the information of the network security device; the method is characterized by comprising the following steps:
step 101: a new service enters the system, when a first data packet of a new service enters the SDN, the first data packet reaches the SDN switch, and step 102 is executed;
step 102: the SDN switch sends the data packet or the data packet header to the SDN controller, and step 104 is executed;
step 103: the SDN controller collects all traffic information and resource information in the network it controls, including: entering into step 105, flow table entries in the switch, use of network node resources, bandwidth utilization of the link, topology of the service, and resource requirements of the service;
step 104: the SDN controller starts a PCE, calculates a path which needs to be passed by the service through calculation or policy management, and if the path is successfully calculated, adds path information of the service into a flow entry of an SDN switch to complete the deployment of the service; if the SDN controller cannot find a suitable routing path, execute step 103;
step 105: the SDN controller starts a PCE, firstly runs a genetic algorithm to generate a plurality of standby paths of each service, then uses a particle swarm algorithm again by taking the paths as input, calculates the optimal scheme for placing the services into the network, ensures that the new service can be added into the SDN network, has minimum disturbance on other services in the network, realizes load balancing, and enters step 106; the method comprises the following specific steps:
step 301, initializing a population of particles, wherein the population size is m and comprises a random position p and a velocity v;
step 302, evaluating the fitness of each particle;
step 303, comparing the objective function value of each particle with the best position pBest passed by the particle, and if the objective function value is better, taking the objective function value as the current best position pBest;
step 304, finding out the particle with the highest objective function value, namely the best position gBest in the current population;
step 305, adjusting the speed and the position of the particles according to the following formula;
where t denotes the number of iterations, ω denotes the inertial weight, c1、c2Representing an acceleration factor, representing an individual learning factor and a social learning factor, r1、r2Is represented by [0,1 ]]Random numbers, pBest, uniformly distributed over the surfaceiRepresenting the historical optimal solution of the individual particle i, gBest representing the current global optimal solution, vi(t) represents the velocity of the particle i at the number of iterations t, vi(t +1) represents the velocity of the particle i at the number of iterations t +1, xi(t) denotes the position of the particle i at the number of iterations t, xi(t +1) represents the position of the particle i at the iteration number t + 1;
step 306: if the end condition is not met, turning to step 302, and selecting the iteration end condition as the maximum iteration frequency or the optimal position searched by the particle group so far to meet the preset minimum adaptation threshold value;
in a particle groupIn the binning algorithm, the goodness of the particle positions is compared by target values, pBest per generation of particle populationiAnd gBest is also updated by comparing the objective function values between the particles; the minimum ratio of the disturbed service number to the total service number is required to be solved under the network problem model, so the following objective function is selected:
where θ is the disturbance rate experienced by the traffic; when omega is more than 1, the link bandwidth is overloaded, that is, the current network resource can not bear all the services of the current network, so the particle objective function is set to be-1, and when one particle objective function value is-1, the particle is an infeasible solution;
step 106: the SDN controller transmits the calculated new configuration information to an SDN switch in the network, and the step 107 is entered;
step 107: after receiving the configuration information, the SDN switch adjusts the path of the service flow which does not meet the requirement, modifies the flow table item of the SDN switch through which the service flow passes, and simultaneously issues the newly added service to the flow table of the corresponding switch to enable the switch to complete the forwarding of the new service;
step 108: deployment of the new service is completed and load balancing of the SDN network is completed.
5. A binning method of an SDN enhanced path binning apparatus, the SDN enhanced path binning apparatus comprising: the network service dispatching center is a module for providing basic communication and management functions for the system, and a core component of the network service dispatching center is an SDN controller; the component interface of the network service dispatching center comprises: the interface comprises a northbound interface used for transmitting customized parameters, a data packet interface used for transmitting packet-in or packet-out messages, a flow table interface used for issuing a flow table or forwarding the flow table, and an interface used for acquiring network topology; the functional module of the network service dispatching center comprises: the system comprises a network topology identification module, a service load evaluation module and a network routing calculation module; the service load evaluation module evaluates index information entering a network, is realized through a big data technology, and constructs a set of service index big database in a network service scheduling center, so that the service index is quickly evaluated according to the service information; the routing calculation module calculates a flow path from a data flow inlet to a forwarding device and then to a data flow outlet based on the topology information of the network and the information of the network security device; the method is characterized by comprising the following steps:
step 201: the newly added service enters the network and requests network resources, and firstly, various constraint conditions and required resource conditions of the newly added service are evaluated;
step 202: load evaluation, before starting network service boxing, firstly judging whether the current residual network resources meet the direct calculation of a newly added service path, if so, entering a step 203;
step 203: calculating a path, namely running a ksp algorithm to calculate the path of the newly added service, and directly calculating the path of the newly added service when all network resources and constraint conditions are met;
step 204: performing boxing operation, namely calling a boxing algorithm when the resource condition of the current network cannot meet the requirement of the newly added service, and adjusting the deployed service to meet the resource requirement of the newly added service; the method comprises the following specific steps:
step 301, initializing a population of particles, wherein the population size is m and comprises a random position p and a velocity v;
step 302, evaluating the fitness of each particle;
step 303, comparing the objective function value of each particle with the best position pBest passed by the particle, and if the objective function value is better, taking the objective function value as the current best position pBest;
step 304, finding out the particle with the highest objective function value, namely the best position gBest in the current population;
step 305, adjusting the speed and the position of the particles according to the following formula;
where t denotes the number of iterations, ω denotes the inertial weight, c1、c2Representing an acceleration factor, representing an individual learning factor and a social learning factor, r1、r2Is represented by [0,1 ]]Random numbers, pBest, uniformly distributed over the surfaceiRepresenting the historical optimal solution of the individual particle i, gBest representing the current global optimal solution, vi(t) represents the velocity of the particle i at the number of iterations t, vi(t +1) represents the velocity of the particle i at the number of iterations t +1, xi(t) denotes the position of the particle i at the number of iterations t, xi(t +1) represents the position of the particle i at the iteration number t + 1;
step 306: if the end condition is not met, turning to step 302, and selecting the iteration end condition as the maximum iteration frequency or the optimal position searched by the particle group so far to meet the preset minimum adaptation threshold value;
in the particle swarm packing algorithm, the quality between the particle positions is compared by a target value, pBest of each generation of the particle swarmiAnd gBest is also updated by comparing the objective function values between the particles; the minimum ratio of the disturbed service number to the total service number is required to be solved under the network problem model, so the following objective function is selected:
where θ is the disturbance rate experienced by the traffic; when omega is more than 1, the link bandwidth is overloaded, that is, the current network resource can not bear all the services of the current network, so the particle objective function is set to be-1, and when one particle objective function value is-1, the particle is an infeasible solution;
step 205: and adjusting network service paths, namely after all the old services are adjusted and all available path sets are calculated for the newly added services, formally deploying the old services to be adjusted and the newly added services, and finishing the whole boxing operation.
6. A boxing method in accordance with claim 1, 4 or 5, wherein: the SDN controller adopts Floodlight, POX or NOX; the index information comprises the size, time delay and transmission rate of data flow.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910182584.1A CN109818865B (en) | 2019-03-11 | 2019-03-11 | SDN enhanced path boxing device and method |
| PCT/CN2019/107872 WO2020181761A1 (en) | 2019-03-11 | 2019-09-25 | Sdn enhanced path allocation device and method employing bin-packing technique |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910182584.1A CN109818865B (en) | 2019-03-11 | 2019-03-11 | SDN enhanced path boxing device and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109818865A CN109818865A (en) | 2019-05-28 |
| CN109818865B true CN109818865B (en) | 2020-09-18 |
Family
ID=66608708
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910182584.1A Active CN109818865B (en) | 2019-03-11 | 2019-03-11 | SDN enhanced path boxing device and method |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN109818865B (en) |
| WO (1) | WO2020181761A1 (en) |
Families Citing this family (57)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109818865B (en) * | 2019-03-11 | 2020-09-18 | 江苏君英天达人工智能研究院有限公司 | SDN enhanced path boxing device and method |
| CN110247845B (en) * | 2019-06-26 | 2021-10-29 | 深圳市中航比特通讯技术股份有限公司 | Method for generating protection route in communication network |
| CN110351145B (en) * | 2019-07-18 | 2020-03-20 | 重庆电子工程职业学院 | Virtualized wireless network function arrangement method based on economic benefits |
| CN111200566B (en) * | 2019-12-17 | 2022-09-30 | 北京邮电大学 | Network service flow information grooming method and electronic equipment |
| CN111598210B (en) * | 2020-04-30 | 2023-06-02 | 浙江工业大学 | Anti-attack defense method for anti-attack based on artificial immune algorithm |
| CN112231705B (en) * | 2020-09-23 | 2023-07-28 | 四川中电启明星信息技术有限公司 | A Reliability Improvement Method of Information System Based on Primary and Secondary Partitioning |
| CN112183893B (en) * | 2020-10-29 | 2024-02-02 | 东北大学 | A two-stage dynamic reactive power optimization method based on PSO-SCA and graph theory assistance |
| CN112448946B (en) * | 2020-11-09 | 2024-03-19 | 北京工业大学 | Log audit method and device based on blockchain |
| CN114650249B (en) * | 2020-12-02 | 2025-01-07 | 南京中兴软件有限责任公司 | Algorithm model and path determination method, electronic device, SDN controller and medium |
| CN112714062B (en) * | 2020-12-07 | 2022-04-29 | 山东省计算中心（国家超级计算济南中心） | A kind of multi-path routing method and device for supercomputing user experience quality |
| CN112637066B (en) * | 2020-12-15 | 2023-01-24 | 南瑞集团有限公司 | Network slicing and path selection optimization method and system for electric power Internet of things |
| CN112615763B (en) * | 2020-12-28 | 2022-09-16 | 广州西麦科技股份有限公司 | Network time delay arrangement system and method based on SPFA algorithm |
| CN112835559B (en) * | 2021-01-28 | 2023-08-11 | 北银金融科技有限责任公司 | Design and implementation method of micro front end in north-silver financial scene construction |
| CN113157431B (en) * | 2021-02-02 | 2022-09-20 | 天津理工大学 | Computing task copy distribution method for edge network application environment |
| CN113065310B (en) * | 2021-03-30 | 2022-06-14 | 福州大学 | Particle swarm algorithm-based high-level comprehensive design method for continuous microfluid biochip |
| CN113315715B (en) * | 2021-04-07 | 2024-01-05 | 北京邮电大学 | Distributed network congestion control method based on QMIX |
| CN113691956B (en) * | 2021-04-13 | 2024-01-09 | 成都天软信息技术有限公司 | Internet of Vehicles Mobility Management Method Based on SDN and MEC |
| CN113191080B (en) * | 2021-04-26 | 2024-02-09 | 辽宁省交通高等专科学校 | Heating furnace billet temperature field prediction model optimization method based on HMPSO algorithm |
| CN113592219A (en) * | 2021-06-11 | 2021-11-02 | 北京理工大学 | Internet of vehicles networking task allocation method based on particle swarm |
| CN113676357B (en) * | 2021-08-30 | 2022-07-29 | 国网内蒙古东部电力有限公司 | Decision method for edge data processing in power internet of things and application thereof |
| CN114024894B (en) * | 2021-11-03 | 2023-04-25 | 苏州全时空信息技术有限公司 | Dynamic calculation method and system in software-defined heaven-earth integrated network |
| CN114116211B (en) * | 2021-11-22 | 2024-04-16 | 中通服咨询设计研究院有限公司 | Business traffic diversion optimization method and system for mobile edge computing |
| CN114205300B (en) * | 2021-12-02 | 2023-09-22 | 南开大学 | A traffic scheduling method that can guarantee the coflow transmission deadline when traffic information is incomplete. |
| CN114244720A (en) * | 2021-12-17 | 2022-03-25 | 湘潭大学 | A multi-controller deployment method based on improved particle swarm optimization in SDN environment |
| CN114466407B (en) * | 2022-01-04 | 2024-05-07 | 中国船舶集团有限公司第七一六研究所 | Network slice arrangement algorithm based on particle swarm inheritance |
| CN114745693B (en) * | 2022-03-03 | 2024-04-26 | 武汉大学 | UAV-assisted vehicle networking resource allocation method based on PSO-GA hybrid algorithm |
| CN114629860B (en) * | 2022-03-17 | 2024-01-30 | 北京恒安嘉新安全技术有限公司 | Data transmission method, device, service line card and storage medium |
| CN114465945B (en) * | 2022-03-25 | 2023-12-08 | 江苏省未来网络创新研究院 | SDN-based identification analysis network construction method |
| CN114785730B (en) * | 2022-04-13 | 2023-12-01 | 东北大学 | Multipath generation method of application layer multipath relay transmission cloud service system |
| CN115002022B (en) * | 2022-04-29 | 2023-10-13 | 中国航空无线电电子研究所 | Route configuration generation method for rapidIO network |
| CN115314355B (en) * | 2022-06-30 | 2023-10-03 | 国网河南省电力公司信息通信公司 | Power communication network architecture system and method based on deterministic network |
| CN115277350A (en) * | 2022-07-26 | 2022-11-01 | 湘潭大学 | A multi-controller deployment method based on improved spectral clustering algorithm in SDN environment |
| CN115412485B (en) * | 2022-08-22 | 2024-09-03 | 中国电信股份有限公司 | Disaster recovery backup method and device, computer storage medium and electronic equipment |
| CN115689412B (en) * | 2022-08-29 | 2025-07-25 | 东北大学 | Multi-period freight pricing and logistics network planning method |
| CN116319270A (en) * | 2022-09-07 | 2023-06-23 | 国网宁夏电力有限公司信息通信公司 | Method and device for deploying backup controller in multi-domain SDN |
| CN115665252B (en) * | 2022-10-20 | 2025-06-27 | 天翼视联科技有限公司 | A cloud storage scheduling system for video surveillance |
| CN115941579B (en) * | 2022-11-10 | 2024-04-26 | 北京工业大学 | A hybrid routing method based on deep reinforcement learning |
| CN115733888B (en) * | 2022-11-23 | 2025-02-14 | 福州大学 | A method for supporting adaptive offloading of multiple IoT applications in edge environments |
| CN116132114B (en) * | 2022-12-26 | 2025-05-16 | 北京航空航天大学 | Security routing method and device based on link and node bearing capacity |
| CN116192655A (en) * | 2022-12-26 | 2023-05-30 | 国网山东省电力公司泰安供电公司 | Security assurance method for power distribution "edge-end" equipment networking based on physical ID |
| CN116016314B (en) * | 2022-12-30 | 2025-06-06 | 东南大学 | A self-healing method for SDN data plane based on bionic mechanism |
| CN116318329B (en) * | 2023-02-02 | 2025-07-01 | 西安电子科技大学 | Joint planning method of satellite network topology and routing based on time-extended graph |
| CN116319600B (en) * | 2023-03-10 | 2026-01-06 | 北京广播电视台 | A set of pathfinding and routing algorithms suitable for resource scheduling and allocation in SDN controllers. |
| CN116032778B (en) * | 2023-03-29 | 2023-06-06 | 南京大学 | Cloud data center intelligent monitoring application task deployment method, device and system |
| CN116720638B (en) * | 2023-04-13 | 2024-03-26 | 广东工业大学 | Method and system for adjusting inverse shortest path weight based on improved evolutionary algorithm |
| CN116938727B (en) * | 2023-06-09 | 2024-08-27 | 中国移动通信集团有限公司研究院 | Scattered protocol processing method and device and readable storage medium |
| CN116709516A (en) * | 2023-07-26 | 2023-09-05 | 沈阳理工大学 | Channel time slot allocation method for wearable wireless sensor network supporting WPT technology |
| CN117221230B (en) * | 2023-09-12 | 2025-06-03 | 腾讯科技（深圳）有限公司 | A data transmission method and related equipment |
| CN117614888A (en) * | 2023-12-06 | 2024-02-27 | 北京理工大学深圳汽车研究院(电动车辆国家工程实验室深圳研究院) | A dynamic scheduling and routing method and system for vehicle TSN network |
| CN117880218B (en) * | 2024-03-11 | 2024-05-10 | 中联云港数据科技股份有限公司 | High-power network resource scheduling system |
| CN117938864B (en) * | 2024-03-22 | 2024-05-28 | 无锡九方科技有限公司 | Node load matching method and system based on machine learning |
| CN119012282A (en) * | 2024-08-07 | 2024-11-22 | 广东邮电职业技术学院 | Time-sensitive network-oriented side collaborative task scheduling strategy method |
| CN119449610B (en) * | 2024-11-14 | 2026-01-06 | 上海交通大学 | Hierarchical control network architecture and multi-domain resource collaborative management method based on industrial SDN |
| CN119629178B (en) * | 2024-12-04 | 2025-06-13 | 北京邮电大学 | An edge data transmission optimization method based on energy consumption minimization and load balancing |
| CN119316362B (en) * | 2024-12-16 | 2025-03-18 | 清创网御(合肥)科技有限公司 | Controller load balancing method based on hierarchical load sensing and adaptive inheritance |
| CN120179414B (en) * | 2025-05-19 | 2025-07-22 | 深圳软银思创科技有限公司 | Resource scheduling method, device and equipment |
| CN120512373B (en) * | 2025-07-21 | 2025-09-30 | 成都秦川物联网科技股份有限公司 | Network resource scheduling method, system, device and medium based on industrial Internet of Things |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105119830A (en) * | 2015-09-14 | 2015-12-02 | 珠海赞同科技有限公司 | Load balancing software defined networking routing aggregation method based on packing optimization |
| CN108400890A (en) * | 2018-01-30 | 2018-08-14 | 天津城建大学 | A kind of software defined network multi-controller dispositions method |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101980470B (en) * | 2010-10-03 | 2013-12-04 | 鲁东大学 | Chaotic particle swarm optimization-based OFDM system resource allocation algorithm |
| CN105337853A (en) * | 2014-06-11 | 2016-02-17 | 杭州华三通信技术有限公司 | Instance establishing method and apparatus in software defined network (SDN) |
| US20160294699A1 (en) * | 2015-03-30 | 2016-10-06 | Alcatel-Lucent Usa, Inc. | Traffic oblivious offline optimization for traffic engineering with segment routing |
| CN105071947B (en) * | 2015-07-06 | 2019-03-08 | 山东超越数控电子有限公司 | A kind of link optimizing method of SDN network |
| CN105429877B (en) * | 2015-11-10 | 2018-08-10 | 中山大学 | A kind of optimum path search method based on particle group optimizing |
| CN106100990B (en) * | 2016-06-06 | 2019-04-05 | 中国电子科技集团公司第三十研究所 | A kind of communication network various dimensions resources regulation method and system based on SDN framework |
| CN107517155A (en) * | 2016-06-15 | 2017-12-26 | 中兴通讯股份有限公司 | A kind of software defined network, node, path calculation method and device |
| CN106789707B (en) * | 2016-11-28 | 2019-10-29 | 北京邮电大学 | A kind of software definition Packet Transport Network overall situation service dynamic optimization method and device |
| CN107040605B (en) * | 2017-05-10 | 2020-05-01 | 安徽大学 | Cloud platform resource scheduling and management system based on SDN and application method thereof |
| CN108521375B (en) * | 2018-04-17 | 2020-12-15 | 中国矿业大学 | A SDN-based network multi-service traffic QoS transmission and scheduling method |
| CN108809857B (en) * | 2018-05-23 | 2020-04-28 | 中国石油大学(华东) | A method for SDN-based traffic monitoring and business service quality assurance strategy |
| CN109818865B (en) * | 2019-03-11 | 2020-09-18 | 江苏君英天达人工智能研究院有限公司 | SDN enhanced path boxing device and method |
-
2019
- 2019-03-11 CN CN201910182584.1A patent/CN109818865B/en active Active
- 2019-09-25 WO PCT/CN2019/107872 patent/WO2020181761A1/en not_active Ceased
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105119830A (en) * | 2015-09-14 | 2015-12-02 | 珠海赞同科技有限公司 | Load balancing software defined networking routing aggregation method based on packing optimization |
| CN108400890A (en) * | 2018-01-30 | 2018-08-14 | 天津城建大学 | A kind of software defined network multi-controller dispositions method |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2020181761A1 (en) | 2020-09-17 |
| CN109818865A (en) | 2019-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109818865B (en) | SDN enhanced path boxing device and method | |
| CN116390164B (en) | Low orbit satellite network trusted load balancing routing method, system, equipment and medium | |
| CN112020103B (en) | A content cache deployment method in mobile edge cloud | |
| CN116208567A (en) | Method and system for flow scheduling of SDN network resources of cross-domain data center | |
| Liu et al. | Scalable deep reinforcement learning-based online routing for multi-type service requirements | |
| CN115665258A (en) | Deep reinforcement learning-based priority perception deployment method for multi-target service function chain | |
| CN113472420A (en) | Satellite network cache placement method based on regional user interest perception | |
| Chen et al. | Traffic engineering based on deep reinforcement learning in hybrid IP/SR network | |
| CN118113484B (en) | Resource scheduling method, system, storage medium and vehicle | |
| Nguyen et al. | Efficient virtual network embedding with node ranking and intelligent link mapping | |
| Isravel et al. | Centrality based congestion detection using reinforcement learning approach for traffic engineering in hybrid SDN | |
| Meng et al. | Intelligent routing orchestration for ultra-low latency transport networks | |
| CN119341967B (en) | A blockchain-enabled intelligent trusted routing method for drone networks | |
| Belkout et al. | A load balancing and routing strategy in fog computing using deep reinforcement learning | |
| CN112333102B (en) | Software defined network routing method and system based on knowledge graph | |
| CN113094857A (en) | Energy-saving controller layout method for software defined vehicle network | |
| Ma et al. | Dynamic routing via reinforcement learning for network traffic optimization | |
| Balasubramanian et al. | An IoT-Driven Reinforcement Learning Framework for Optimized Flow Management in Autonomous Systems | |
| CN116192746B (en) | A method, device and storage medium for selecting a routing path based on SDN | |
| Zhang et al. | Flowlet-level routing optimization with GNN-based multi-agent deep reinforcement learning | |
| García et al. | Multicast routing and virtual network function placement in NFV-SDN networks: A genetic algorithms approach | |
| Cai et al. | Multipath routing for traffic engineering with hypergraph attention enhanced multi-agent reinforcement learning | |
| Zheng et al. | Research on intelligent routing technology based on improved dqn | |
| CN119211889B (en) | Time-sensitive network routing planning method and system for intelligent connected vehicles | |
| Wang et al. | DeepVNP: Virtual network placing with deep reinforcement learning in industrial IoT |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |