Agent-Based Industrial Environments: State of the Art [ READ ]
Hosny Abbas, Samir Shaheen, Mohammed Amin
This paper highlights the state of the art deployments and applications of the agents and multi-agent systems (MAS) technologies within modern industrial environments. It explores and reviews some of the recent multi-agent based industrial applications to identify their advantages and also their common limitations. The key conclusion of this paper is that most of the proposed agent-based industrial applications adopt the traditional basic forms of the agents and multi-agent systems. They only concern the level of individual agents (the system micro level) by starting the development process with the identification of the individual agents’ types and the design of behaviors and interactions of each agent in the system without giving attention to the global system structure and organization (the system macro level) as they assume that the required global system functionality will emerge as a result of the lower level agents interactions and behaviors. This type of development approaches is not adequate for engineering complex, open, and highly distributed modern industrial networks (i.e., modern SCADA). Undesirable system level behaviors can be emerged and can impact system performance and reliability. Present and future complex highly distributed real world applications and especially industrial ones such as those related to critical infrastructures (i.e., power grids, water transportation, etc.) management should be tackled from both of their micro and the macro levels in the same time because they continuously evolve and change in an unpredictable and uncertain manner. This paper highlights the problem and suggests the promising solution.
Hosny Abbas, Samir Shaheen, Mohammed Amin
This paper highlights the state of the art deployments and applications of the agents and multi-agent systems (MAS) technologies within modern industrial environments. It explores and reviews some of the recent multi-agent based industrial applications to identify their advantages and also their common limitations. The key conclusion of this paper is that most of the proposed agent-based industrial applications adopt the traditional basic forms of the agents and multi-agent systems. They only concern the level of individual agents (the system micro level) by starting the development process with the identification of the individual agents’ types and the design of behaviors and interactions of each agent in the system without giving attention to the global system structure and organization (the system macro level) as they assume that the required global system functionality will emerge as a result of the lower level agents interactions and behaviors. This type of development approaches is not adequate for engineering complex, open, and highly distributed modern industrial networks (i.e., modern SCADA). Undesirable system level behaviors can be emerged and can impact system performance and reliability. Present and future complex highly distributed real world applications and especially industrial ones such as those related to critical infrastructures (i.e., power grids, water transportation, etc.) management should be tackled from both of their micro and the macro levels in the same time because they continuously evolve and change in an unpredictable and uncertain manner. This paper highlights the problem and suggests the promising solution.
