Parallel complexity computation based on dynamical systems

Tambet Treimuth (1) , Daniel Delahaye (1) , Sandra Ulrich Ngueveu (2) , Stéphane Puechmorel (3)

1 MAIAA - ENAC - Laboratoire de Mathématiques Appliquées, Informatique et Automatique pour l'Aérien
2 LAAS-ROC - Équipe Recherche Opérationnelle, Optimisation Combinatoire et Contraintes
3 ENAC - Ecole Nationale de l'Aviation Civile

This paper describes a parallel approach for computing an air traffic complexity metric that is based on dynamical systems. It is important in order to quantify the complexity and geometric features of traffic. Lyapunov exponents are used for understanding the geometry of current air traffic in the airspace. It is computed at different points on a spatial grid. Those values can indicate areas where air traffic controllers must pay more attention because the complexity is higher at those points. A parallel implementation of the air traffic complexity metric computation is introduced in this paper. Serial implementation of this problem is too slow for real world problems. This process is parallelizable because Lyapunov exponent values can be computed simultaneously for each grid point in the airspace.

Domaines: Optimisation et contrôle [math.OC]

Tambet Treimuth, Daniel Delahaye, Sandra Ulrich Ngueveu, Stéphane Puechmorel. Parallel complexity computation based on dynamical systems. DASC, 2015 IEEE/AIAA 34th Digital Avionics Systems Conference, Sep 2015, Prague, Czech Republic. pp.1C2-1 - 1C2-8 /978-1-4799-8939-3, ⟨10.1109/DASC.2015.7311340⟩. ⟨hal-01223896⟩