A Framework for Simulating and Analysing Multi-Agent Systems in YODA

Modern software solutions vary in dimension and domain scope and may be composed of a huge and heterogeneous amount of interacting entities. Each of these entities is characterised by a set of respective local or global goals. The entities have to operate and adapt according to a strategy, even cooperating or competing against each other. These interactions may produce many intricacies and unexpected results. Predicting these results should be supported by tools for simulating and analysing. Operating entities are commonly referred to as agents and if we deal with multiple agents, this system is called Multi-Agent System. This is defined as “a loosely coupled network of problem-solving entities (agents) that work together to find answers to problems that are beyond the individual capabilities or knowledge of each entity (agent)”. Agents may be enclosed in an environment and the dynamics between the actors may produce even more complexities, called “emergent behaviours”. A proper definition of environment can help us define the whole system. In this thesis, we introduce YODA, a novel framework for simulating agentbased systems. The framework allows the developer to define how the agents and the environment evolve by providing a set of formal rules and a specification language to model different case scenarios. YODA has been integrated into Sibilla, a Java-based tool for reasoning about Collective Systems, by taking advantage of the modular architecture of the tool. The thesis also includes a set of case studies to understand how YODA works with different scenarios. Additionally, we present two tools for further analysing MultiAgent Systems. In the first case, we introduce Sequit a lightweight visualiser developed using the Unity Game Engine. On the other hand, we show an integration with the Model-Checking framework called GLoTL.