Fusion of Level of Service Data for Traffic Information Services

Kurzfassung

This contribution describes an approach for fusing level-of-service (LOS) data which takes the statistical properties of the data (including their quality) explicitly into account. In short, the implemented algorithm uses a quality-weighted median to compute fused LOS values. Moreover, a generic quality module assesses the accuracy of the results based on the quality of the input data. This includes a concretization of the term “quality” in context of traffic information from a probabilistic perspective. Finally, this paper presents a simple stochastic algorithm for transforming LOS data between different scales, i.e. scales with 3, 4, 5 or more possible LOS values. This ensures that LOS data from arbitrary sources with arbitrary discretization (in terms of the number of possible LOS values) can be fused efficiently using the proposed median approach. Of course, the effect of such scale transformations on the quality of the related LOS values is taken into account explicitly.