Decentralized Coordination in Self-Organizing Systems based on Peer-to-Peer Coordination Spaces

Abstract

Coordination is an important aspect to realize Self-organizing Systems usually implemented as part of the functional properties of the system. This paper promotes a separation of concerns via a declarative approach to realize decentralized coordination in Self-organizing Multi-Agent Systems (MAS). In previous work the concept of Coordination Spaces was developed which provides explicit support for the task of coordination in MAS. Coordination Spaces are part of the agent environment and handle a declarative description of the coordination process. Thereby, this approach allows developers rather focusing on what to coordinate than on how to coordinate. Also by releasing the developer from programming coordination manually, the approach offers benefits like reusability and interoperability of coordination processes. This paper extends the approach by a distribution concept for coordination spaces. By using different techniques like remote service calls, group communication and publish/subscribe models the distribution of information among multiple platforms in a peer-to-peer like approach is achieved.