Our Vision: Modeling and Simulation as a Service (MSaaS)

Almost every ecosystem today is tightly coupled with its neighboring economic or social systems and thus these needs to be taken into account when watching the evolution of that ecosystem.

The research interests of the Multi-Agent Research & Simulation (MARS) Group focuses on the development of a generic and scalable framework for massive multi-agent based simulations. This framework shall be applicable for research and educational purposes.

In order to develop a domain-independent platform, the MARS team is involved in interdisciplinary projects across the globe. Together with domain experts and security services are models developed that can brought to life with MARS.

MARS History

The topic of sustainability is especially important in the world of modeling and simulation (M&S) where many research projects are conducted with simulation software crafted specifically and solely for the model at hand. Once the research project is done, the model software usually is abandoned. Often to the effect that the model neither can be executed anymore nor be developed any further.

To democratize modelling and simulation with individual based Multi-Agent Simulation (MAS) as well as to make it more accessible for researchers from across the world, a Modeling and Simulation as a Service (MSaaS) platform is required. Several MAS tools are available which allow for a simplified modeling experience, but usually lack the capability to scale models to real world scenarios while maintaining performance and efficiency. The MARS system thus was developed to provide a scalable, robust and available platform for collaborative modelling and simulation of massive scale scenarios from varying domains. Aside from being used by research projects, this platform should make a great fit for teaching M&S to students.

MARS emerged from the WALK project, a 2D simulation system also developed at the HAW (2011-2013). While WALK was specially geared to movement ecology (e.g. pedestrians or animal hunting), it already featured several architectural core concepts. The original MARS 1.0 system development started in 2013 and focused primarily on the actual simulation system, its distribution and supporting components to simplify model development. The requirements were predominantly driven by socio-ecological models with millions of agents (www.ars-africae.org), but were expanded by applications ranging from medical models to urban simulations.

MARS 2.0 was designed in 2015 to make the transition towards a MSaaS platform, while maintaining compatibility to the already developed models. The development focused on two primary areas: The actual simulation system MARS LIFE 2.0 and the creation of the MARS Cloud as the supporting environment for model creation, simulation configuration and execution as well as result analysis.

As of February 2017 MARS 2.0 reached closed beta status. The system has matured into a scalable and reliable MSaaS platform running inside a research cluster hosted and maintained by the MARS team at Hamburg University of Applied Sciences. The system is being tested together with international partners from Florida and South Africa and aims for a first public release in Summer 2017.