Time- and space-parallel simulation of air traffic networks. (December 2019)

• Record Type: Journal Article
• Authors: Kim, Young Jin; Mavris, Dimitri; Fujimoto, Richard
• Is Part Of: Simulation. Volume 95:Number 12(2019)
• Page Start: 1213
• Page End: 1228
• Publication Date: 2019-12
• Subjects: Time-parallel simulation -- air traffic simulation -- space-parallel simulation; Computer simulation -- Periodicals; 003.3
• DOI: 10.1177/0037549719831358
• Languages: English
• ISSNs: 0037-5497

Abstract

Computer simulations are widely used to design and evaluate air traffic systems. A fast time simulation capability is essential to effectively explore the consequences of decisions in airspace design, air traffic management, and operations. A parallel simulation approach is proposed to accelerate fast time simulation of air traffic networks that exploits both temporal and spatial parallelisms. A time-parallel algorithm is first described that simulates different time intervals concurrently and uses a fix up computation that exploits the scheduled nature of commercial air traffic to address the problem of dependencies between time segments. The time-parallel algorithm is then extended with a space-parallel simulation approach using Time Warp to simulate each time segment in parallel thereby increasing the amount of parallelism that can be exploited. The time and space-parallel algorithms are evaluated using a simulation of the U.S. National Airspace System (NAS). Experimental data is presented demonstrating that this approach can achieve greater acceleration than what can be achieved by exploiting time-parallel or space-parallel simulation techniques alone.