A Simple Virtual Sensor Approach for Black Carbon Estimation in Sensor Networks

This paper presents a simple virtual sensor predictive model based on multiple linear regression for the estimation of the equivalent black carbon concentration in an air quality automatic monitoring sensor network. The predictive model uses selected pollutants concentration and meteorological parameters as predictor variables, and equivalent black carbon concentration as target variable. Virtual sensor model is assessed for several different training/test periods. Dataset for training and validation of the model is derived using a vast dataset collected in WeBaSOOP project on Ada Marina, Belgrade supersite.