Research On Network Trattic Trends In Network Resource Management Of Railway Data Network

With the rapid development of railway informatization and the diversification of business,the railway data network becomes the main platform for carrying the communication information such as video conference,video monitoring and data interaction between each management information system(MIS).As more applications are based on the railway data network,its complexity and dependence on the network is increasing.Meanwhile,there is more need for network bandwidth.It become the key problem how to look into the network traffic condition and make the reasonable bandwidth allocation of the network in the research of the network data management of railway data network.The network traffic model which based on stream attributes is the foundation of network performance analysis and network bandwidth allocation.Precise network traffic model is of great significance for traffic flow prediction,network topology design and network performance analysis.Therefore,analyzing the statistics of railway business traffic and the traffic behavior of each subnet,and establishing an efficient network traffic model is an important prerequisite for network bandwidth allocation,and also the primary research topic of the realization of the railway data network's intelligent deployment.However,the current traffic data analysis for the railway data network only remain in the simple and crude monitoring,accurate traffic statistics and traffic prediction technology remains to be studied.Based on the research of network traffic collection and traffic modeling,this paper designs traffic statistics scheme based on service for and utilizes Fractional Differential Autoregressive and Moving Average(FARIMA)model to analyze and forecast the actual network traffic data according to traffic characteristic of railway data network.The main research contents are as follows:(1)Analyze and summarize the business characteristics and IP address allocation rules of railway data network.On this basis,this paper researches and analyzes the advantages and disadvantages of the related traffic statistics methods.The data collection technology,caching technology,aging mechanism and aggregation strategy of NetFlow technology are mainly analyzed.Based on the source IP address prefix match aggregation strategy,the traffic statistics scheme of railway service system is designed,which can provide support for future bandwidth allocation based on traffic flow prediction.(2)According to the characteristics of self-similarity and complexity of railway data network traffic,the paper compares and analyzes the advantages and disadvantages of network traffic modeling method,selects FARIMA model as the model analysis technique.This paper successfully breaks FARIMA model into differential process and ARMA process,which can solve the problem that calculation of FARIMA model is too complex.The FARIMA model's long range correlation is effectively validated by simulation.Meanwhile,the FARIMA model forecasting is applied for the actual traffic data of railway data network and the fitting effect is tested by calculating the mean square error of the autocorrelation function of observational sequence and fitting sequence.The experimental results show that the FARIMA model has a high degree of fit,and can be applied for railway data network traffic to predict and analyze the trend.The prediction model obtained in this paper can dynamically allocate bandwidth for the network.For the railway data network business,it forecasts the traffic trend of each service and can realize the VPN bandwidth allocation of each business subsystem.When the business network is busy,it can predict and expand the bandwidth in advance to avoid packet loss and network delay,ensuring the safe operation of railway data network;when the business network is idle,it can reasonably plan bandwidth and save bandwidth resources.