Estimating rainfall intensity by using vehicles as sensors

Vehicles are key elements in the envisioned Smart Cities, not only providing more efficient mobility, but also becoming mobile network elements able to perform many useful tasks. Environment sensing is a good example where the combination of data coming from vehicles allows achieving insight only comparable to the deployment of hundreds or thousands of sensors in a city. Obtaining rainfall estimations with a high spatial granularity is an example of a task where relying on traditional methods would become too expensive due to the high number of data sources required. Vehicular networking has a great potential to address such challenge by converting every vehicle in a rain sensor. In this paper we carry out a simulation study to estimate the rainfall intensity in a specific area using a vehicular network as data source. To this purpose, we model a rainfall pattern taking real values as reference, and we devise a simulation scenario where the rainfall pattern is deployed. Experimental results using the OMNeT++ simulator show that, even with a low density of vehicles contributing to the proposed monitoring system, rainfall intensity can still be predicted with a high accuracy and granularity, thereby validating the proposed approach.