The Research of Traffic Analysis and Forecasting Method for Wireless Network
|School||West China University|
|Course||Applied Computer Technology|
|Keywords||Wireless Network Data Analysis Flow projections Grey Theory Chaos Theory Support Vector Machine|
In recent years , the wireless network as a new network structure , more and more applications in the campus area , the military field , the medical field and other public places . With the network in various fields of application , also mentioned the importance of research on wireless network management and security agenda . Network traffic analysis and network traffic prediction is one of the important content of network management and security research in the field . This paper analyzes the characteristics of the wireless network traffic , and further study based on a different prediction methods , thus the traffic prediction method which is suitable for wireless network characteristics . In this paper, research from the following aspects : 1) wireless network traffic data analysis . The part of the two major areas of wireless data network characteristics quantitative and qualitative analysis : on the one hand is the flow of data descriptive analysis statistical analysis of wireless network traffic , on the other hand , from the point of view of the probability statistics , sample flow and overall flow distribution characteristics . This part of the analysis behind the establishment of a forecast model provides the basis and foundation . 2) the prediction model of the wireless network traffic . This paper presents a new network traffic prediction the model - GCSVR model . Flow projections as a new model , we will be chaos theory , gray theory and support vector machine three combined for wireless network traffic prediction . Chaos theory is used to analyze the time series , the reconstruction phase space ; gray theory is used to smooth time sequence , weakening rough random characteristics of the data set , the data has strong regularity to the predictive model to provide a more stable and law strong data ; final fitting and forecasting of traffic using support vector machine . By the new model and the comparison of the original SVR model experiments , we found that the prediction accuracy of the new model and predict the step has significantly improved . The experiments show that GCSVR model for sudden strong regularity poor wireless network traffic is not stationary series , able to do a good smooth processing can effectively improve the prediction accuracy and prediction length .