Introduction to Agent and MAS

Abdul Samad Shibghatullah 40 30052008 selected tool should be able to fulfil these requirements in order to achieve the aim of the proposed system. Current approaches such as heuristics, mathematical and others have some limitations as discussed in the critiques of current approaches. Heuristics approaches are not easily adaptable, and they were not suitable for general optimisation Wren and Rouseau, 1995; Wren, 1998. Mathematical approaches are usually slow to produce results because they are computationally intensive when it comes to complex situations Kwan et al., 1999. Other approaches genetic algorithm, tabu search, ant system, and constraint programming were not reported to have capabilities in operating in uncertain environments. These limitations as mentioned above prevented us from adopting such approaches for the proposed system. Alternatively, a MAS can fulfil the requirements mentioned above and be capable of reacting in uncertain environments and providing quick solutions in real time Jennings et al., 1998; Ferber, 1999; Wooldridge, 2002. A MAS has been used for other scheduling fields meeting scheduling, manufacturing scheduling, events scheduling and etc. where similar problem are faced i.e. uncertain environment. The following subsections explain details of a MAS theoretical description and the use of a MAS in scheduling.

2.5.2 Introduction to Agent and MAS

MAS is a relatively new field of research. These systems have only been studied since about 1980, and the field has only gained widespread recognition since about the 1990s Oliveira et al., 1998; Wooldridge, 2002. MAS have become more important in many aspects of computer science by introducing the issues of distributed intelligence and interaction. MAS seem to be a natural metaphor for understanding and building a range of what were called artificial social systems. They represent a new way of analysing, designing, and implementing complex software systems Jennings et al., 1998. What is an agent? Ranges of definitions from different disciplines have been proposed for the term agent. There is no commonly accepted definition of the term, and there is much continuing debate on this matter Jennings et al., 1998; Wooldridge, 2002. Abdul Samad Shibghatullah 41 30052008 According to Maes 1995, agents are computational systems that inhabit some complex dynamic environment, sense and act autonomously in this environment, and by doing so realise a set of goals or tasks for which they are designed. Wooldridge and Jennings 1995 define an agent as a computer system that is situated in some environment and that is capable of flexible and autonomous action in this environment in order to meet its design objectives. By flexible, it means that the system must be responsive, proactive, and social. Ferber 1999 describes an agent as a physical or virtual entity which is capable of acting and perceiving in an environment, can communicate directly with other agents, possesses resources, skills and can offer services. Its behaviour tends towards satisfying its objectives, taking account of the resources and depending on its perception, and the communication its receives. What is MAS? Ferber 1999 defined MAS as a system composed of a population of autonomous agents, which interact with each other to reach common objectives, while simultaneously each agent pursues individual objectives. Oliveira et al. 1998 defined MAS as a collection of possibly heterogeneous, computational entities, having their own problem-solving capabilities able to interact in order to reach an overall goal. According to Jennings et al. 1998 a MAS’s main characteristics are that each agent has incomplete information, or capabilities for solving the problem, each agent has a limited viewpoint, there is no global system control, data is decentralised, and computation is asynchronous. Two main MAS architectures have been addressed in the literature: blackboard and autonomous architectures Jennings et al., 1998; Ferber, 1999. Early MAS were based on the blackboard model proposed by Hayes-Roth 1985. The blackboard architecture is based on the idea that problem solving could result from the opportunistic activation of specialists, the “knowledge sources”. The activity of the “knowledge sources” consists of putting down, modifying, and withdrawing solution elements within a common working area, called a blackboard. A centralised control mechanism is used to activate the “knowledge sources”. According to Ferber 1999, blackboard architectures cannot be considered as MAS as they do not respond to the characteristics of MAS. In autonomous architectures, the agents are not controlled or managed by any other agents, rather they communicate and interact directly with any other agent in the system to Abdul Samad Shibghatullah 42 30052008 achieve the global objective Jennings et al., 1998; Ferber, 1999. Knowledge and control are distributed, in the sense that each agent embodies its own knowledge and control.

2.5.3 MAS interactions