Design And Implementation On Caching Policy Based On Network Traffic Prediction And Timestamp In Smart Collaborative Network

With the development of various Internet applications,network traffic grows rapidly.Thus,the problems of low utilization of network resources and increased user access de-lay have become obstacles to the sustainable development of the network.Therefore,the research of information-centric Smart Collaborative Network is proposed,and the ubiq-uitous caching mechanism of the whole network is deployed.So that the content routers in the network can quickly respond to users' requests,reduce unnecessary access over-head,make full use of network resources,and reduce data acquisition delay to improve the quality of network service.This paper first investigates the current situation of Internet and the factors that re-strict the development of the network to summarize the shortcomings of the traditional caching mechanism in the TCP/IP network architecture.Then,it highlights the necessity of cache research under the information-centric transmission mechanism.Furthermore,this article expounds on the research status of information-centric network architecture and caching mechanisms,focusing on the Smart Collaborative Network architecture and communication mechanism.According to research,this article proposes a caching mech-anism design scheme under Smart Collaborative Network,which aims to solve the cach-ing problem without considering the network topology architecture and that cannot adap-tively adjust according to the dynamic changes of the network in the current research.Secondly,the design of the caching mechanism is mainly for the deployment of four functional modules.The main function of the acquisition module is to classify and count the network traffic and send it to the prediction module.The prediction module uses the prediction algorithm based on RNN(Recurrent Neural Network)model to predict the future network traffic,and calculate the traffic change rate to adapt to the network changes.The resource control module needs to perform initial allocation of the caching space ac-cording to the location characteristics of the content router in the network topology,and adjust the caching size by the network traffic change rate.In addition,the resource control module is able to comprehensively calculate the caching effective time of the data content through the demand and the location information of the node,and identify the timestamp.The caching module initializes and adjusts the size of the caching space according to the caching decision made by the resource control module.Besides,it is responsible for man-aging the content caching by timestamp.Tirdly,the deployment and implementation of each functional module will be carried out in the design scheme.The effectiveness of the prediction algorithm is verified by the network traffic dataset under the real topology.The test results obtained by the prediction model training optimization show that the model can predict accurately according to net-work traffic characteristics.Furthermore,the prediction algorithm is deployed in the pre-diction module to evaluate the performance of the overall caching mechanism.The sim-ulation results show that the caching mechanism based on network traffic prediction and timestamp can fully consider the network topology architecture compared with the exist-ing caching mechanism.It can also dynamically adapt to network changes,thereby providing a higher cache hit rate,reducing unnecessary cache replacement overhead,and the retrieval time of data content in the cache,improving user request response speed,increasing the utilization of network resources,and further improving the network service quality.