FCD

Floating car data is a traffic measurement method that derives road speed estimates from the GPS location and timestamp records of moving vehicles, typically taxis, delivery vehicles, or private cars with onboard connectivity. Rather than installing fixed sensors like inductive loops or radar detectors at specific points, FCD uses the natural flow of traffic itself as the measurement instrument. A vehicle records its position every few seconds; when aggregated across many vehicles on a road segment, these traces reveal how fast traffic is moving.

The method works by partitioning roads into segments, usually 100 to 500 meters long, and calculating average speed from the rate at which instrumented vehicles traverse each segment. Data collection requires a fleet of equipped vehicles or a smartphone app with persistent location sharing. Taxi operators and logistics companies have historically provided the densest datasets because their vehicles operate continuously and systematically cover urban and highway networks. Ride-hailing apps, connected car services, and navigation platforms like Google Maps and HERE now generate massive FCD streams as a byproduct of their core business.

FCD has clear advantages over fixed infrastructure: no installation cost, coverage of every road including minor streets, and responsiveness to temporary congestion. A congestion event shows up in FCD within minutes as travel times spike on affected segments. However, the method depends on sufficient vehicle density; on quiet rural roads with sparse traffic, FCD becomes unreliable or impossible. Data quality also varies with GPS accuracy, which degrades in urban canyons, tunnels, and heavy foliage. Privacy concerns arise because location histories can reveal personal patterns even when anonymized.

The term "floating" refers to the vehicles themselves moving freely through the network rather than being anchored to fixed measurement points. Early research in the 2000s, particularly in Germany and the Netherlands, formalized the method as traffic engineers sought alternatives to expensive sensor networks. The name has stuck despite the rise of connected vehicles and mobile applications that now dominate FCD collection.

FCD integrates into traffic management systems alongside other data sources: loop detectors at highways, toll booth records, incident reports from traffic operations centers, and weather feeds. Modern traffic prediction models blend FCD historical patterns with real-time probe data to estimate congestion and recommend routes. Accuracy typically reaches within 5 to 10 percent of actual average speeds on major roads; uncertainty widens on secondary networks where vehicle sampling is thinner.