Multilevel modeling and simulation: methodologies and applications

Multilevel modeling and simulation is a methodology that involves the hierarchical decomposition of complex systems into modular, cooperating components, each representing specific aspects of interest. The motivations behind this approach can vary widely, including the ability to leverage existing software, the need for different levels of abstraction and granularity, or the opportunity to better organize model development by separating semantically distinct aspects. This thesis presents a comprehensive study of multilevel modeling and simulation techniques, providing an overview of their use in scientific literature and discussing the design principles and key issues involved. The analysis of the current state of the art reveals that while multilevel modeling is widely used across various scientific fields, little effort was directed toward formal and methodological aspects. As a result, there are no precise standards for the design of such models, and ambiguities exist starting with the terminology. To address this gap and better define potential approaches for building a multilevel framework, the thesis proposes some categories of design patterns that address some critical aspects that may be encountered during the development phase, and a metamodel crafted to enhance multilevel modeling clarity and effectiveness. The proposed design principles were ultimately applied to create multilevel simulations for IoT applications, with a focus on developing a decentralized, efficient sensor data marketplace. This scenario leverages technologies like LoRa and blockchain to support secure and scalable data exchange. Given the complexity of factors involved — including mobility, physical message propagation, data storage and trading — multilevel modeling has proven to be exceptionally effective in managing the complex interactions and dependencies among these components.