Embedding Intelligence into Coordination: The ACLT Model

   page       attach   

<div>ACLT is a coordination model whose goal is to provide intrinsic support for intelligent coordination of multi-component distributed software systems, such as multi-agent systems. It is based on tuple centres, which are shared communication abstractions a' la Linda enhanced according to the model's goals. In particular, ACLT's tuple centres go beyond Linda tuple spaces for two main reasons:</div>

<div>(1) they promote a twofold interpretation of the logic tuple space as both a communication repository and a theory of the communication. To this end, ACLT provides a family of demo primitives which make it possible to reason about the communication state, thus implicitly providing support for heterogeneous multi-agent architectures. In fact, logic, intelligent agents can here well cope with non-logic agents accomplishing different tasks according to their different perceptions of the communication channels.</div>

<div>(2) they exploit the notion of programmable tuple space as a communication abstraction whose behaviour can be defined according to the system needs. ACLT tuple centres can be programmed by defining reactions to communication events, through the logic tuple-based ReSpecT language (Reaction Specification Tuples). Multi-commponent architectures based on ACLT may be in principle designed around properly defined communication abstractions (intelligent communication channels) embedding the system intelligence.</div>

<div>Some examples of simple multi-agent systems based on the ACLT model will be discussed. A sketch of the novel ACLT-based TuCSoN model for the support of mobile agents will be finally given.</div>