Distributed identification of the Cell Transmission traffic model: A case study

The problem of the distributed identification of a macroscopic first-order traffic model, viz. the Cell Transmission Model (CTM), is considered in the paper. The parameters to be identified characterize the dynamics of the density in different sections of the freeway (cells). We explore different distributed identification schemes. The purposes of the approach are mainly to obtain good prediction models through the minimization of the one-step ahead prediction error of the densities of the cells, and to reduce the computational time and the effort required to perform the identification. The methodology is validated relying on real-life data measured on a portion of the A12 freeway in The Netherlands. An evaluation of the performance of the identified model used as a set of virtual sensors in different scenarios is presented.