Research On Traffic Prediction In WSN

With the rapid development of network communication technology, the Internetbegins to carry more and more application services, it brings forward very highdemands for network quality of service, traffic control and network management.Wireless Sensor Network is featured of on-limits environment, dynamictopology and limited resource, which causes both positive and negative effects to theWSN effectiveness. The Wireless Sensor Networks (WSN) could be flexible andchangeable in an open environment. Also the whole network would worsendramatically due to a single sensor's failure when sensors are scattered randomly andis hard for replace. So when the WSN been used in magnitude scene, build aappropriate traffic prediction model for exceptional traffic detection and the nextefficiency data prediction seemed rather important. This thesis studied light modelsthat apply WSN based on traditional traffic prediction models.First, the thesis comprehensively narrates the current situation about trafficmodel studying, especially models in WSN and features of WSN.Second, the features of current network and the rules of modeling aresummarized, and the paper introduced hot models of traditional networks in detail, thefeatures of WSN and the adaptive models are deeply discussed.Third, modeling scenes of WSN are introduced in detail, especially modeling ininformation collection scene. Analyzed classical models in WSN deeply, in order toestablish the model accord with actual traffic characteristic, a hybrid network flowforecast model that combined with ARMA model and difference technology isproposed, the modeling under information collection scene based on ARIMA isdeeply analyzed, including basic idea, forecasting mothed, reasons for choosingARIMA and the process of modeling.Fourth, the future behavior trend of true traffic data by using the forecastalgorithm of this model is forecasted, and been compared with ARMA modelingresults. The simulation results shows that the ARIMA model can not only fullydescribe and characterize the flow properties, but also forecast network trafficbehavior accurately and effectively.Finally, the potential applications of forecast algorithm in network managementare discussed and analyzed; shortages of this forecast model is briefly analyzed andthe next research target in our future work is pointed out.