Research On Cloud-side Data Reliable Transmission In Industrial Software-defined Networks

In recent years,with the widespread application of Internet technology and big data technology,industrial Internet technology has flourished.The Industrial Internet promotes the self-transformation and reorganization of global industries,becoming more diverse in terms of products,systems,factories and regions.From the perspective of the industrial economy,the Industrial Internet can be regarded as a new source of productivity and competitiveness.It can help enterprises achieve digital transformation,from traditional manufacturing to intelligent manufacturing,and realize the transformation from a simple product supplier to an intelligent product and service provider.This transformation can provide enterprises with more value-added services,help enterprises better understand market and customer needs,and further improve their competitiveness.In addition,the Industrial Internet can also promote collaborative innovation and development of the ecosystem.By establishing a digital industrial chain and value chain,enterprises can better work together,jointly develop new products and services,and share data,resources and technologies,thereby improving the efficiency and competitiveness of the entire ecosystem.Real-time data transmission in the Industrial Internet has strict requirements on network service quality.Achieving safe and reliable real-time data transmission in complex industrial Internet networks,especially backbone networks,has always been one of the difficulties restricting the development of the Industrial Internet.One way to increase the reliability of transmission paths is to provide dedicated backup paths.In the event of a failure,transmission tasks on failed links can be rerouted on predetermined paths in the backup network.Firstly,cloud-side data transmission is divided into real-time data transmission and non-real-time data transmission.The former requires an active method to quickly switch to a backup path when a link failure occurs;the latter requires a passive method to switch to a new path when a link failure occurs.This thesis focuses on real-time data transmission,so it is necessary to solve the problem of network traffic classification.It is known that Deep Convolutional Neural Network has good image classification performance,while network traffic can be normalized to remove the dimensional relationship and then converted into a grey release graph.Based on this,the network traffic classification problem can be transformed into a grey release graph classification problem,and then a network traffic classification model can be built through the VGGNet model.The model is designed and implemented based on the pytorch machine learning framework to improve the accuracy of network traffic classification.Secondly,the bandwidth information of real-time data transmission is a nonlinear time series sequence.Based on the excellent time series prediction performance of Long Short-Term Memory Recurrent Neural Network,at the same time,transfer learning is used to transfer knowledge from the source domain to the target domain,which can be used in the case of insufficient data.Build a network traffic prediction model.The model is designed and implemented based on the pytorch machine learning framework,which improves the prediction accuracy and reduces the time cost compared with the direct training model.The model can not only be used in the field of network traffic prediction,but also meet the task requirements of other similar application scenarios.Finally,under the condition that the delay is known and the bandwidth is predictable,the alternate path design problem is formulated as an integer linear programming problem of the main network with general link capacity and independent and co-distributed link failure probability,so as to improve the real-time data transmission.Reliability and ensure that data cannot be transmitted due to link failure.It is very difficult to actually build a network environment,requiring enough switches and servers,and paying a great price.Therefore,this thesis uses the topology generation framework to generate the topology,which includes information such as the number of nodes,the number of links,link bandwidth,and link delay.According to the above conditions,this thesis proposes a backup path construction scheme based on swarm intelligence algorithm.The constraints of the backup path are that the link bandwidth and delay are satisfied,and the failure healing ability of the specified main path is maximized.After the algorithm is constructed,a comparative experiment is carried out with the baseline algorithm.From the perspective of robustness,the expected goal is achieved,the healing ability of link failures is improved,and the link cost is reduced.