Abstract

The shared data space coordination paradigm, introduced by Linda’s generative communication in 1985, is a popular alternative to the conventional point-to-point communication approaches. Linda, with its basic entities of tuples and tuplespaces, is renowned for its temporal and spatial separation properties, as well as independent of any computation language and machine architectures — very attractive traits for coordination in open systems. Reaching its maturity, the research in tuplespace-based coordination is now in the phase of providing general-purpose data spaces to create efficient large-scale implementations of open distributed multi-component systems, one of which is the autonomous multi-agents systems. This paper presents a large variety of Linda-like models that have been experimented on agents, with different focus and approach to meet different requirements in solving the problem identified. Towards the end, we outline some research questions pertaining to the possibility of coordinating multiple agents using Scope, the latest variant of Linda. 1

Citations