Research On Resource Control Technology Based On Business Features

In recent years,the rapid development of the Internet has led to an increase in network capabilities.The substantial increase in data transmission rate and the growth of network resources have provided strong support for the further development of new services.However,some problems have arisen.The demand for network resources of each service is different and difficult to describe.As a resource management party,it's hard to allocate resources for different business needs.In addition,the growth rate of network resources is greatly behind the growth of demand for resources in the business,and the method that increasing network physical resources to meet the needs of all services will undoubtedly encounter bottlenecks.Therefore,resources can only be allocated to all businesses in a relatively fair manner on the basis of existing resources.Under this circumstance,the research on business requirements and network resource control mechanisms is extremely urgent.Based on the above status,this paper proposes a resource control technology solution based on business requirements,which mainly solves two problems.The first is the problem of the description of the business requirements.For the business,the characteristics of the business requirements are reflected in the traffic generated by the business.Therefore,the traffic characteristics can be extracted to obtain the business demand characteristics.Based on the results of convolutional neural networks in computer vision,combined with project requirements,this paper proposes a neural network-based traffic identification scheme,using Moore dataset to train neural networks to achieve traffic identification.Firstly,the Moore data set is processed with some method.According to the actual situation,the data set is divided into a training set and a verification set,and then the data of the training set is input into the neural network,and the neural network model will be trained in multiple rounds.After each round of training,the validation set is used to verify the effectiveness of the training,and finally the trained neural network model is obtained.The second is to solve the problem of network resource allocation.This paper proposes a bandwidth resource allocation scheme based on service type.According to the service type obtained by the traffic identification in the previous solution and the bandwidth resource required for the service acquired in advance,combined with the resource situation in the actual network environment,the network resource is relatively fairly distributed to the service.First,the iterative approximation method is used to obtain the bandwidth required for all types of services.Then,according to the service type obtained by the traffic identification result,all services are divided into two queues,and the total resources are allocated for different queues,and then correspondingly performed inside the queue.The allocation of resources ultimately allocates resources to all businesses in need.In addition to the above two parts,according to the needs of the program,this paper also implements a traffic generator based on the open source framework.The traffic generator can simulate multiple types of traffic and custom some fields in the packet.The traffic generators are used in both traffic identification and resource allocation schemes.At the end of this paper,Mininet and other simulation tools are used to simulate the various functions designed and implemented by the above two schemes.The simulation results are illustrated and analyzed,the simulation results prove the feasibility of the proposed scheme.