From the Ground to the Cloud: Towards an Integrated Transportation Simulation Platform

Abstract

Nowadays, universities and companies all around the world have a huge need for simulation and modelling methodologies. In the particular case of traffic and transportation, making physical modifications in the real traffic networks could be highly expensive, dependent on political decisions and could be highly disruptive to the environment. Therefore, simulation is broadly used in such scenarios. However, while studying a specific domain or problem, analysis through simulation may not be trivial and very often requires several simulation tools, with different resolutions and domain perspectives, hence raising interoperability issues. With the recent evolutions in cloud computing and Software-as-a-Service (SaaS), there is a new paradigm where simulation software is used in the form of services. So, Simulation Software-as-a-Service (SimSaaS) is very beneficial to better exploit the huge amount of platforms and storage that simulation needs per se - and Cloud Computing is able to provide such resources. To address issues arising in this novel perspective the main goal of this dissertation was to present the current state of the art in the field and to propose an agent-directed transportation simulation platform, through the cloud, by means of services. It was used the IEEE standard HLA (High Level Architecture) for simulator interoperability and agents for controlling and coordination. To do so, it was necessary to build, through a systematic literature review, the body of knowledge needed to develop such platform. The reviewed studies were compared and summarised leading to the creation of a taxonomy of the research work, which represent the front research opportunities for the next years. The main scenarios and architecture of the platform were detailed. The proof of concept's implementation was further explained including the used software (OpenStack, Pitch pRTI, SUMO and EBPS) and the chosen simulation scenario. Finally, some experiments were made about the best approach to manage and launch VMs (Virtual Machines). Such analysis is very important to have better performance in simulations under the developed infrastructure.