The measurement and assessment of traffic conditions on a road network in an accurate and timely manner is fundamental to the efficient operation of any transportation management system. Floating car techniques -- i.e., techniques based on receiving real-time, traffic-related data from probe vehicles travelling through the transportation network -- are unique in that they are intelligent transportation system (ITS) applications that provide very detailed, microscopic traffic data (Floating Car Data, or FCD) in real time. On that basis, FCD show strong evidence of supporting efficient traffic monitoring, incident detection and management, and route guidance applications, to name a few. A considerable body of knowledge and experience on potential applications of FCD has been obtained since 1987 in European programmes through such projects as SOCRATES and EuroScout in Europe, and ADVANCE in the USA. This paper takes a statistical point of view in looking at some of the trade-offs that arise in optimizing sampling and data fusion strategies in the context of floating car data applications in Germany. However, many of the considerations apply with appropriate modifications quite generally.
Abstract