Search

Existing organisational centred multi-agent systems (MAS) regulate agents' activities. Nevertheless, population and/or environmental changes may lead to a poor fulfilment of the system's purposes, and therefore, adapting the whole organisation becomes key. This is even more needed in open MAS, where participants are unknown beforehand, they may change over time, and there are no guarantees about their behaviours nor capabilities. Hence, in this paper we focus on endowing an organisation with self-adaptation capabilities instead of expecting agents to increase their behaviour complexity. We regard this organisational adaptation as an assisting service provided by what we call the Assistance Layer. Our abstract Two Level Assisted MAS Architecture (2-LAMA) incorporates such a layer. We empirically evaluate our adaptation mechanism in a P2P scenario by comparing it with the standard BitTorrent protocol. Results provide a performance improvement and show that the cost of introducing an additional layer in charge of system's adaptation is lower than its benefits.

Existing organisational centred multi-agent systems regulate agents’ activities. However, population/environmental changes may lead to a poor fulfilment of system’s goals, and therefore, adapting the whole organisation becomes key. In this paper, we propose to use Case-Based Reasoning learning to adapt norms that regulate agents’ behaviour. Moreover, we empirically evaluate this approach in a P2P scenario., This work is partially funded by IEA (TIN2006-15662-C02-01), EVE (TIN2009-14702-C02-01 / TIN2009-14702-C02-02) and AT (CONSOLIDER CSD2007-0022) projects, EU-FEDER funds, the Catalan Gov. (Grant 2005-SGR-00093), and M. Esteva’s Ramon y Cajal contract.

Organisational centred multi-agent systems (MAS) have proved to be effective to regulate agents' activities. Nevertheless, population and/or environmental changes may lead to a pour fulfilment of the system's purposes, and therefore, adapting the whole organisation becomes key. This paper presents a MAS simulator devoted to test organisations with self-adaptation capabilities in P2P scenarios. More specifically, our simulator implements different sharing P2P methods ---some of them with self-adaptation--- and so, it can be used as a testbed for comparing them.

Similarly to institutions in human societies, an Electronic Institution (EI) provides a structured framework for a Multi-Agent System (MAS) to regulate agents' interactions. However, current EIs cannot regulate a previously existing dynamic social system and deal with its agent population behaviour changes. This paper suggests a solution consisting of two EI extensions to incorporate situatedness and adaptation to the institution. These two properties are usually present at an agent level, but this paper studies how to bring them to an organisational level. While exposing our approach, we use a traffic scenario example to illustrate its concepts.