View article

Ubiquitous computing environments are highly distributed, heterogeneous, complex and context-aware. Therefore, they require emerging middleware components and paradigms, which significantly extend the capabilities of conventional middleware systems. In this paper we discuss the middleware components required to support contextawareness in ubiquitous environments. Special focus is put on middleware for identifying humans, their location and their activities within in-door environments. A smart room infrastructure comprising a wide range of sensors, perceptual components and other context-aware components is presented as an illustration of the middleware elements constituting an ambient environment. Moreover, we describe the design and implementation of the PolyOmaton system, an ambient system that can coordinate multiple multimedia streams (stemming from multiple cameras) towards selecting the best camera view in the scope of lectures, seminars and meetings. Emphasis is paid into the middleware infrastructure supporting this system. Experimental results demonstrate that this ambient system features extremely high success ratios in selecting the desired camera view.