Skip to Main Content
The aim of this work is to demonstrate the suitability of existing toll collection data from the high-ranking road network in Austria for the generation of traffic information. The development of an area-wide, high-quality traffic information service in various implementations as prototypes was target of the project smart mobility analysis of real-time toll-data (GO-SMART). In order to achieve feasible travel time estimations detection and rejection of distorting data is essential. Several methods of outlier detection and filtering have been tested and compared. An algorithm exhibiting good attributes with regard to effectiveness, feasibility and performance has been designed and applied to the data. The developed method is adapted to the specific behaviour and features of the available data that are, among others, caused by system conditions of the tolling system. The proposed method is a combination of moving average, z-score and plausibility checks. For the purpose of displaying fast and up-to-date traffic information a weighted aggregation is suggested. Weights depend on timeliness and related vehicle type of single observations. In order to allow jam detection, a similar behaviour of observed vehicles (chargeable heavy good vehicles) and passenger cars is required in congested situations. This is necessary to assure the same behaviour of the measured part of the traffic (heavy good vehicles) and of the overall traffic. The required assumption could be reproduced for single velocity values as well as for approximation of density-velocity relations. Prototypes of pre-trip and on-trip traveller information services are presented.