Cloud Simulation for Large-Scale Agent-Based Traffic Simulations
Translated title:
Cloudbasierte Simulationsausführung von großflächigen und agentenbasierten Simulationen
Author:
Zehe, Daniel
Year:
2019
Document type:
Dissertation
Faculty/School:
Fakultät für Informatik
Advisor:
Knoll, Alois (Prof. Dr. habil.)
Referee:
Knoll, Alois (Prof. Dr. habil.); Cai, Wentong (Prof., Ph.D.)
Language:
en
Subject group:
DAT Datenverarbeitung, Informatik
TUM classification:
DAT 260d; DAT 815d
Abstract:
High-resolution traffic simulators require a large number of hardware resources, which are not accessible to everyone. Cloud computing is a possible solution to the democratization research by enabling access to high-performance resources. Within this thesis, a reference architecture for cloud-based traffic simulation is introduced and evaluated. An agent-based traffic simulation (CityMoS) is used as a use-case. The capabilities of cloud computing are demonstrated, while the challenges in data extraction and search-space exploration are presented and addressed.
«
High-resolution traffic simulators require a large number of hardware resources, which are not accessible to everyone. Cloud computing is a possible solution to the democratization research by enabling access to high-performance resources. Within this thesis, a reference architecture for cloud-based traffic simulation is introduced and evaluated. An agent-based traffic simulation (CityMoS) is used as a use-case. The capabilities of cloud computing are demonstrated, while the challenges in data e...
»
Translated abstract:
Detaillierte Simulatoren benötigen eine große Anzahl an Hardwareressourcen.
Ein Ansatz dieses Problem zu bewältigen ist Cloud-Computing, das den Zugang zu hoch-performanten Ressourcen erleichtert. In dieser Dissertation wird eine cloudbasierte Referenzarchitektur für agenten-basierte Simulationen entworfen, anschließend im Verkehrssimulator CityMoS umgesetzt und evaluiert. Mit der hier vorgestellten Lösung wird somit eine verteilte und zeitsparende Erkundung von großen Parameterräumen möglich.