The process begins with mobility data preparation. Raw data is first indexed using geohashes to support efficient spatial processing and then re-indexed by anonymized vehicle ID to extract individual vehicle tracks. Each track passes through a sequence of transformations, including position validation and timestamp assignment, to ensure that the observations are spatially and temporally consistent.

Through a separate process, the road network map is prepared as a connected road graph. OpenStreetMap data is transformed into a directed graph composed of interconnected road edges and corresponding from/to nodes, with all relevant parameters assigned to each road element.

The information from all available hardware-based sources (detectors, cameras, etc.) as well as from virtual detectors is mapped to the road network. Ticon’s proprietary algorithm identifies potential road-network matches, after which the resulting candidates are reviewed for suitability. A special procedure is applied to confirm that the selected locations and distances meet the codified mapping requirements.

The extracted vehicle tracks are then processed and normalized. This stage includes filtering spatial clusters and removing observations that create unrealistic movement angles. Additional filtering is applied to identify tracks that were not properly separated based on distance.

Each remaining track is then mapped to the road graph. The algorithm identifies the most appropriate points on the network and constructs graph-based references linking the observed track to the corresponding road segments. A track is accepted only when its calculated probability exceeds a predefined threshold.

Vehicle speed is calculated using the distance travelled, elapsed time, and applicable road-network constraints. Pedestrian tracks are separated into a distinct dataset.

The mapped detectors are also filtered to remove locations with an estimated Annual Average Daily Traffic volume below 300 vehicles, to prevent insufficiently supported observations from affecting the final calculations.

The resulting network values are then rebalanced with the use of a least-squares approximation for the available set of counts, minimizing the assigned objective function and improving consistency across connected road segments.

Traffic volumes are subsequently calculated for each edge using a weighted-volume formula that accounts for both observed counts and mobile data penetration rates. Through an inverse spanning mechanism, each detector is extended across the relevant road network, and the resulting values are applied to the associated set of road edges.

In addition to AADT estimation, Ticon generates detailed traffic profiles, including hourly, weekly, and monthly traffic distributions, as well as origin-destination patterns. These mobility insights, combined with road network characteristics and map data, are used in Ticon’s proprietary truck percentage model to estimate the share of heavy vehicles traffic across the network.

Finally, a simplification algorithm consolidates individual road edges into practical blocks of continuous road segments. These simplified segments form the final structure used to display traffic information on Ticon’s traffic map.