Distributed collaboration using Multi-Agnet System Artchitectures (DICOMAS)

  • Nowe, Ann (Administrative Promotor)
  • Holvoet, Tom (Coördinator)

Project Details

Description

There is an undeniable trend towards collaborative business processes involving di units. One prototypical example in the domain of logistics is supply chain management. Another example is collaborative health care institutions. Privacy and security issues are typically involved in such distributed collaborative systems. Whereas many initiatives (both in industry and research) aim at enabling communicating business applications, this project targets the development of distributed collaborative applications.

The long-term goal of this project is to establish a knowledge platform on distributed collaboration in Flanders; the concrete objective of the project is to provide a reusable software architecture that supports the development of applications for automated collaboration between business processes.
Multi-agent systems (MAS) is a lively research area that studies modelling and engineering of collaborative systems. A MAS structures a software system as a set of autonomous entities (agents) that are situated in a common (software or physical) environment. Agents have only local access to the environment and interact with one another to achieve the system goals. As such, distributed
collaboration is a core topic of MAS research. Particular qualities associated with MAS-based architectures are adaptability, openness, robustness, and scalability.
Flanders has a lot of expertise in the domain of multi-agent systems, which has not been brought together. This project brings together this expertise and aims to advance our complementary know-how on distributed collaboration, and to provide a valuable platform for valorisation.

The main project objective is to provide a reusable software architecture for supporting the development of automated collaborative systems. To achieve this goal, we acknowledge three related research tracks: (1) collaboration modelling, (2) middleware, and (3) issues of trust, security and privacy.
The targeted applications involve autonomous business units, each with their own (and possibly common) objectives and constraints, that collaborate to realise these objectives. Collaborations are established in which units can play different roles. Collaborations are subject to various kinds of dynamism, strategies of units may change, units may leave or new units may enter, etc. Collaboration modelling requires suitable high-level modelling abstractions. MAS offers abstractions for organisation modelling. However, state-of-the-art approaches offer little or no support for dealing with dynamic organisations. Modelling dynamic organisations needs to take into account changes in the environment and needs to allow entities to enter and leave the organisation dynamically.

Interactions in an organisation are structured by protocols and may be subject to laws that govern what agents are allowed to do. Protocols and laws must be flexible enough to deal with ongoing dynamics and change in the system. Besides investigating suitable abstractions, we study mechanisms to quatifiably assess the added-value of collaboration, both individually as well as collectively.
State-of-the-art middleware platforms mainly aim at enabling communication, yet do not support distributed collaboration. Supporting complex collaboration requires suitable high-level abstractions and corresponding run-time support.
Local applications may also raise particular local requirements and constraints such as trust, privacy and security concerns. Current communication platforms support coarse-grained authentication and authorisation mechanisms. However, local entities may not always want to expose too much information about themselves when authenticating to and/or interacting with other agents and hosts. Such local requirements and constraints clearly a This project provides the opportunity for an in-depth study of these issues, and will integrate the
results into a reusable software architecture. Throughout the research, a continuous validation in a real-world setting will proved suitable feedback to ensure the usability of the software architecture and as such its valorisation potential.
AcronymIWT381
StatusFinished
Effective start/end date1/01/0831/12/11

Keywords

  • informatics

Flemish discipline codes

  • Biological sciences