Discovering Probabilistic Models from ADS-B Data Sets using Digital Pheromones

The University of North Dakota is developing airspace within the state where Unmanned Aircraft Systems (UASs) can be flown without an onboard sense and avoid system or Temporary Flight Restrictions (TFRs). With funding from the U.S. Air Force, a mobile ground-based system capable of detecting cooperative aircraft flying in the surrounding airspace, and the software to display information about those aircraft to UAS operators, was developed. An ongoing area of research for this system is its risk mitigation component. The risk mitigation component will be responsible for calculating the total collision risk for an Unmanned Aircraft (UA) flying in a specific Class E airspace configuration. If accurate probabilistic models for pilot behavior are determined, these models could be used in the implementation of the risk mitigation component. In this paper the authors present the results of subpath discovery and data mining of an Automatic Dependent Surveillance – Broadcast (ADS-B) data set from a 3 month period in 2010. Arbitrarily complex subpaths were automatically extracted from the ADS-B data set using a digital pheromone algorithm. Then, probabilistic models of pilot behavior were data mined from this set of discovered subpaths using a variant of the Genetic K-Means Algorithm (GKA).