Creating predictive models for forecasting the accident rate in mountain roads using VANETs

Resumen

Monitoring the road network status of an entire country in a visual way (as traditionally) is very hard, so different mechanisms to do it in an automatic manner have been investigated. In particular, nomadic pervasive sensing platforms based on VANETs have been recently deployed. However, the level of road damage is a relative variable, and it is necessary to predict the particular impact of the same in each case, in order to prioritize the conditioning works. Therefore, in this paper a predictive model for forecasting the accident rate in mountain roads, considering the measures previously obtained through a nomadic sensing environment (and through the weather office) is defined. The model considers the type of road under study as well as different analysis scales to perform the calculations. The model is based on Taylor's series and multivariate functions. Real data related to Valais (Switzerland) road network is employed to construct and validate the proposed model.