A Characterization of Key Properties of Environment-Mediated Multiagent Systems

The increasing presence of application scenarios which are based on large collections of active components having to adapt continuously to changing environmental requirements has led to several research initiatives with the objective to create new concepts for the design and operation of environment-mediated multiagent systems. In particular, Autonomic Computing (AC) and Organic Computing (OC) have developed the vision of systems possessing life-like properties: They self-organize, adapt to their dynamically changing environments, and establish other so-called self-x properties, like self-healing, self-configuration, self-optimization etc. The impact of these initiatives will depend crucially on our ability to demonstrate the benefits of these systems with respect to some essential properties. Therefore, we need a clear understanding of some key notions like adaptivity, robustness, flexibility, or their degree of autonomy, allowing for self-x properties.

In this paper, a system classification of robust, adaptable, and adaptive systems is presented. Furthermore, a degree of autonomy is characterized to be able to quantify how autonomously a system is working. The degree of autonomy distinguishes and measures external control which is exhibited directly by the user (

no autonomy

) from internal control of a system which might be fully controlled by an observer/controller architecture that is part of the system (

full autonomy

). Finally, learning and of trustworthiness are briefly addressed, since these are further essential aspects of self-organizing, adaptive systems.