Map-matching of GPS traces on high-resolution navigation networks using the Multiple Hypothesis Technique (MHT)

In recent years, data from GPS-based surveys has become increasingly important since transport modellers benefit from more accurate and reliable information about times, geographic locations, and routes. At the same time, participants’ burden is reduced substantially if the GPS data collection does not involve time-consuming questions to derive additional information. However, without respondent provided information, extensive data processing is required to derive results that can be used for analysis and model estimation. Since the first GPS survey (Wagner, 1997), one of the key postprocessing steps is the map-matching, i.e. the association of the GPS points to the links of a network, in order to establish the routes travelled by the survey participants. The algorithm proposed in this paper is an adaptation of the algorithm proposed by Marchal et al. (2005). Like the algorithm by Marchal et al. (2005), it is designed to match large-scale GPS data sets on a high-resolution navigation network in an acceptable computation time. This paper describes the implementation of the algorithm after a short overview of the existing literature. Afterwards, the performance of the algorithm is evaluated both in terms of accuracy and computational efficiency for about 36,000 car trips from 2,434 persons living in and around Zurich, Switzerland, which are mapped on the Swiss Navteq network, a high resolution navigation network covering all regions of Switzerland and containing 408,636 nodes and 882,120 unidirectional links.