Analysis of Aircraft Clusters to Measure Sector-Independent Airspace Congestion

In current air traffic operations, sector controllers separate aircraft manually and airspace congestion is measured in terms of aircraft counts within fixed sector boundaries. Future air traffic management concepts generally include automated separation assurance as a key feature. Since automated separation assurance is independent of airspace geometry, the challenge is to measure local congestion independent of sector boundaries. The first step towards measuring sector-independent airspace congestion is to identify aircraft clusters, i.e., groups of closely spaced aircraft. The objective of this work is to develop a methodology for fully automated identification of aircraft clusters. First, a region-growing clustering technique adapted to the air traffic problem is presented in the paper. Next, an algorithm is designed for determining the best values of a key region-growing parameter to identify aircraft clusters. This algorithm utilizes “natural neighbors” from Delaunay Triangulation, and maximizes a performance metric to determine the best cluster patterns. This technique was implemented in software, and exercised using recorded field data from the Cleveland Center. Preliminary results indicate good performance of the cluster identification methodology presented in this paper.