New Concepts for Constructing Agent Applications
Alexander Pokahr, Lars Braubach · 2010
Agent-based approaches offer promising solutions for recent demands in enterprise software like adaptability cooperation and coordination (Jennings 2001). Industrial strength agent platforms, such as the JADE platform or Whitesteins LS/TS tool suite, serve as an established technological basis for agent-based programming and software development (Unland et al. 2005). While these systems consider agents as first-class entities, they currently offer only limited conceptual and technological support for 1) dealing with the agent application as a whole and 2) coping with other non-agent components of an application (i.e. the agent environment). As a result, today‟s agent-based software systems often employ ad-hoc solutions for these issues, suffering from a lack of tool support as well as poor maintainability. The concepts and tools presented in this paper allow for defining an agent application as a conceptual entity, which is composed of agents as well as additional non-agent components. A pluggable architecture is proposed that allows defining common types of environment models, thereby facilitating seamless integration of the corresponding environment components into an agent application. The architecture is implemented as part of the Jadex agent framework.1 The paper is organized as follows. In Section 2 the conceptual framework for application description as well as the pluggable environment architecture are presented. Furthermore, as an example, the environment support space is presented. In Section 3 an example application is presented that illustrates the advantages of the proposed framework. Section 4 discusses related work and Section 5 concludes the paper with a summary and an outlook.