Accuracy Assessment for Traffic Information Derived from Floating Phone Data

We describe a layered evaluation approach to assess the accuracy of traffic speed estimation based on floating phone data. We designed a simulation framework to investigate the influence of various parameters on the estimation. Multiple quality metrics and combinations of metrics have been used and intervals of metric values have been mapped to quality ratings to enable a uniform and comprehensible presentation of results. We simulated traffic on several artificial and real maps and identified influencing parameters and their optimal values to achieve higher speed accuracy estimation. The results for artificial maps can be transferred to realistic maps with restrictions.