Coordination Models and Languages

Coordination languages are used to express coordination among entities which have to cooperate or to synchronize to perform some set of tasks or achieve some common goal. Such entities are often referred to as agents, hence the resulting systems are named multi-agent systems.

More precisely, according to David Gelertner and Nicholas Carriero, a complete programming model can always be thought as made of two separate, orthogonal components: the computation model and the coordination model. While the first is meant to build a single (sequential) computational activity, the coordination model is the glue that binds separate activities into an ensemble.
A coordination language embodies a coordination model, i.e. it provides operations to create computational activities and to support communication among them.

Ideally, a coordination language should be orthogonal to any computation language, that is, it should be possible to extend any computation language with proper coordination primitives according to such given coordination language.
This feature is particularly relevant when concerned agents are very heterogeneous, and the interaction protocol is the only common denominator among them.

Although a wide range of coordination models have been proposed in literature, a common approach consists of defining some form of shared memory abstraction, which can be used as a message repository. Then, a proper communication protocol is needed which allows agents to cooperate and compete in such framework.

Most well-known coordination models are based on abstractions like blackboards and tuple spaces: some well-known such models are Linda, Shared Prolog, Polis, etc.

More information about Linda may be found at the Linda team at York University, at the Yale Linda group, or at the Linda group at the Parallel Processing Research laboratory, while some references to Shared Prolog and Polis may be the following:

• P. Ciancarini, Coordinating Rule-Based Software Processes with ESP, ACM Trans. on Software Engineering and Methodology 2:3, 1993, 203-227.
• P. Ciancarini, Distributed Programming with Logic Tuple Spaces, New Generation Computing, 12:3, 1994, to appear (TR. CSL-93-7, University of Bologna).

We are working on the ACLT coordination model, which is particularly suited for heterogeneous agent coordination. You may like to know more on ACLT by consulting the ACLT pages, where ACLT-related papers are referenced, too.

• Coordination
• Multi-Agent Systems
• Linda
• Logic Tuple Spaces
• Heterogeneous Agents
• Hybrid Agent Architecture
• Distributed Systems

• Enrico Denti
• Antonio Natali
• Andrea Omicini

• MURST 60%