Knowledge Networks

For future network scenarios to exhibit autonomic behaviour, both networks and application components and services need to be aware of their computational and environmental context, and must tune their activities accordingly. In this position paper, we propose an abstract architecture for knowledge networks that addresses the key issues of how both physical contextual knowledge and social knowledge from the users of communication networks can be used to form a knowledge space in support of autonomic agents dealing with network elements and applications. We discuss that the availability of raw contextual data is not enough to achieve meaningful autonomie behaviours. Rather, contextual information should be properly organised into 'networks of knowledge', to be exploited by both network and application components as the basic 'nervous system' in which situational stimuli reify into digital knowledge, and by means of which components can properly orchestrate their activities in a globally meaningful way. Here we firstly discuss the fundamental role of knowledge networks, and try to sketch what actual form and position such knowledge networks could assume. Then, we analyse some simple scenarios of use, showing how it is possible for the components of an autonomic communication system to build such knowledge networks autonomously; and, at the same time, to exploit them for orchestrating their activities in a type of stigmergy-based knowledge-rich system. Eventually, we sketch a rough research agenda and discuss the relations with other research areas. © IFIP International Federation for Information Processing 2006.