Autonomic computing approach for resource allocation

This paper presents an autonomic solution based in a multi-agent model for resource allocation in a manufacturing environment. This model considers the temporal and synchronism aspects involved in the allocation process. To attend to this purpose a framework was developed with a CIMOSA compliant architecture. To support the framework four agents where defined: the Domain Process Agents (DPA), the Business Process Agents (BPA), the Enterprise Activity Agents (EAA) and the Functional Entity Agents (FEA). Planning networks are used as a strategy to manage the scheduling process complexity. The main contributions of this work are: (i) a user-independent framework allowing to manage the intrinsic scheduling complexity, (ii) a strategy for resource planning allocation based on a multi-agent paradigm including a major complexity element: the real-time disturbances, (iii) the use of multi-agent architecture to support this strategy. The agents' interaction is obtained using the contract net algorithm and market-oriented programming. (C) 2004 Elsevier Ltd. All rights reserved.