Catálogo de publicaciones - libros

Today, the concept of an environment for multi-agent systems is in its pioneering phase. Consequently, the development of supporting software technologies is still rather primitive and environment technologies reflecting a specific world-of-interest to the agent systems are yet to be developed in full. In contrast, environment technologies that focus on the agent system itself have been in the agenda of MAS research from its very start. Electronic institutions are prominent in this respect for they have been conceived as a type of restricted MAS environment and have had an engineering technology developed around them. In this paper we explore how the restrictions currently imposed by electronic institutions may be overcome when they are seen as a part of a larger environment where agents act. In particular, we focus on electronic institutions by connecting them to a world-of-interest and how this process can facilitate full-fledged environment engineering.

In previous work, we introduced an approach to describe and simulate environments for situated multi-agent systems, based on a language called ELMS. Here, we present extensions to our approach which provide the means to allow normative information to be distributed in environments shared by multiple agents. Organisational structures for multi-agent systems are usually defined independently of any spatial or temporal structure. Therefore, when the multi-agent system is situated in a spatial environment, there is usually a conceptual gap between the definition of the system’s organisational structures and the definition of the environment. Spatially distributing the normative information over the environment is a natural way to simplify the definition of organisational structures and the development of large-scale multi-agent systems. By distributing the normative information in different spatial locations, we allow agents to directly access the relevant information needed in each environmental context. The extensions to our approach for multi-agent environments allow for the definition of spatially distributed normative objects and the means to distribute and handle such objects in a shared environment.

Environment is an essential part of any multi-agent system (MAS), since it provides the surrounding conditions for agents to exist. For some sort of systems, the environment can be viewed as providing a set of services, in which some of them, such as directory facilities, are used explicitly by the agents to perform their tasks, and other such as monitoring, behavioral enforcement and security can be done transparently by the environment. We join the idea that the specification of environments of open multi-agent systems should include laws that define what and when something can happen in an open system. Laws are restrictions imposed by the environment to tame uncertainty and to promote open system dependability. This paper proposes a design approach and application of a middleware based on laws in multi-agent systems. The approach can be viewed as a set of services provided by the environment.

Traffic control can be regarded as a multiagent application in which car-agents and traffic-light-agents need to coordinate with each other to optimize the traffic flow and avoid congestions. Environment abstractions naturally suit this scenario in that agents actions are mainly driven by traffic-related information that are distributed across the environment both at a practical and conceptual level. In this context we present traffic-control mechanisms on the basis of our Co-Fields model and discuss some experimental results we obtained in simulations that validate our proposal.

Self-organisation is being recognised as an effective conceptual framework to deal with the complexity inherent to modern artificial systems. In this article, we explore the applicability of self-organisation principles to the development of multi-agent system (MAS) environments. First, we discuss a methodological approach for the engineering of complex systems, which features emergent properties: this is based on formal modelling and stochastic simulation, used to analyse global system dynamics and tune system parameters at the early stages of design. Then, as a suitable target for this approach, we describe an architecture for self-organising environments featuring artifacts and environmental agents as fundamental entities.

As an example, we analyse a MAS distributed environment made of tuple spaces, where environmental agents are assigned the task of moving tuples across tuples spaces in background and according to local criteria, making complete clustering an emergent property achieved through self-organisation.