| KEYWORD |
Design and development of a framework for large scale distributed Agent- Based simulation systems
Thesis in external company 
Thesis abroad
keywords AGENT BASED SIMULATION, CO-SIMULATION TECHNIQUES, CROWD SIMULATION, DISTRIBUTED SYSTEM, SMART CITIES, TRAFFIC
Reference persons LORENZO BOTTACCIOLI, EDOARDO PATTI
External reference persons Luca Barbierato(luca.barbierato@polito.it), Stefano Bortoli (stefano.bortoli@huawei.com), Alexander Wieder (alexander.wieder@huawei.com)
Research Groups DAUIN - GR-06 - ELECTRONIC DESIGN AUTOMATION - EDA, EDA Group, ELECTRONIC DESIGN AUTOMATION - EDA, Energy Center Lab, GR-06 - ELECTRONIC DESIGN AUTOMATION - EDA, ICT4SS - ICT FOR SMART SOCIETIES
Thesis type EXPERIMENTAL AND MODELING
Description The Lamport Team in Huawei Munich Research Center, led by Dr. Stefano Bortoli, is a multi-disciplinary
research group specialized in large-scale agent-based city simulation systems, starting from automatic
creation of accurate scenarios, until the final post-processing of large datasets of simulation outputs.
In large-scale city simulations, there could be several distinct modeling dynamics involved. As a limited
example, accurate traffic simulations must interact realistically with pedestrian simulations. To tackle
the challenge of dealing with diverse simulation domains, tools, and models, the team is adopting co-
simulation techniques, to couple different simulators under a unified, coherent simulation environment.
This is a well-established approach for studying interaction across multiple continuous simulation
domains (e.g., solid and fluid). Continuous simulation domains allow for mesh-based discretizations and
interpolations serving as interface between simulators. However, agent-based simulators generally
require agent representations (and hence, interactions) going beyond numerical coupling of meshes.
Further, agents moving through the simulation domain during runtime require dynamic approaches for
sharing agent state across simulation domains.
The goal is to design and implement a co-simulation framework for large-scale distributed agent-based
simulations featuring:
• Support for representing agents outside their original domain when required.
• Mechanisms for specifying simulation entities (e.g., other agents) or events that may be of
interest to a given agent (e.g. nearby agents) to accurately compute its state.
• Efficient mechanisms to share agent state as needed.
Deadline 26/05/2025
PROPONI LA TUA CANDIDATURA