Hybrid agent-based simulation for analyzing the National Airspace System

Abstract

Hybrid agent-based simulation is required to provide a mechanism for analyzing large-scale complex systems, such as the National Airspace System (NAS), more accurately and completely. The dynamic behavior of many large-scale complex systems is, in general, hybrid in nature and thus can be best described by a combination of discrete-event and continuous-time models, and their interactions. Correspondingly, hybrid agent-based simulation capable of incorporating different types of models, such as continuous-time and discrete-event models, provides an accurate means of evaluating the reliability and performance of large, complex systems. However, in order to serve as a valuable design and analysis tool, a number of important issues must be addressed. This paper outlines issues in the development of hybrid agent-based simulation architectures capable of providing a scaleable mechanism for simulating the NAS. In particular, an object-oriented approach to hybrid agent-based simulation is described. In addition, methods of improving computational efficiency of updating the simulation are described and compared.