The study of the models and methods to apply to multi-actor systems is a widely discussed research topic in the related scientific literature. The multi-actor systems are defined as systems that are characterized by the presence of several autonomous elements, of different decision makers and of complex rules that allow the communication, the coordination and the connection of the components of such systems. Commonly, the study of Multi-Actor System, MAS, recalls the well-known issues concerning the multi-agent systems.
The research topic related to the multi-agent system firstly started to appear in scientific literature in 1980s, mainly in relation to the computer science and artificial intelligence. In this dissertation, in particular, the study of the multi-agent systems, and specifically of the multi-actor systems, is taken into account merely in relation to the distinctive features of complexity that characterize such systems and not to the issues concerning the agent-oriented software engineering. Therefore, the research results presented in this thesis are focused on the development and on the realization of innovative models and methodologies to face the management and the decision making mechanisms applied to complex multi-actor systems.
This dissertation especially focuses on two different examples of multi-actor systems in two very diverse perspectives. The former deals with the research problem related to intermodal transportation networks, while the latter with the so called consensus problem in distributed networks of agents.
Concerning the research problem related to the intermodal logistic systems, the research activity addresses the management of their more and more increasing complexity by the applications of the modern Information and Communication Technologies (ICT) tools that are key solutions to achieve the efficiency and to enhance logistics competitiveness. The related scientific literature still seems lacking in addressing with adequate attention the impact of these new techniques on the management of these complex systems and, moreover, there is an apparent lack of a systematic and general methodology to describe in detail the multiplicity of elements that can influence the dynamics and the corresponding information and decision making structure of intermodal transportation systems. The innovative results presented in this dissertation are focused on the development of an Integrated System, IS, devoted to manage intermodal transportation networks at the tactical as well as operational decision level to be used by decision makers both in off-line planning and real time management. To specify the Integrated System, a reference model is developed relying on a top-down metamodeling procedure. These innovative research results are a contribution to bridge the gap and to propose not only a systematic modeling approach devoted to describe a generic multi-actor logistic system, but also a management technique based on a closed loop strategy.
The second example of application is focused on a topic that is widely discussed in scientific literature related to the study of the multi-actor collective behaviors in a distributed network. The interaction protocols that allow the agents to reach the convergence to a common value is called consensus or agreement problem. This research problem is particularly studied in the context of cooperative control of multi-agent systems because the agents are autonomous, independent and have to interact in a distributed network. The presented research results address the investigation of new and fast alignment protocols that enhance the performances of the standard iteration protocols for particular topologies of digraphs on the basis of a triangular splitting of the standard iteration matrix.
The examined examples, the models and the methodologies applied to analyze them, are very different in the two cases and this testifies the large extent of research problems related to the multi-actor systems.
