Ambient Assisted Living 5. AAL-Kongress 2012 Berlin, Germany, Ja
In this book, leading authors in the field discuss development of Ambient Assisted Living. The contributions have been chosen and invited at the 5th AAL congress, Berlin. It presents new technological developments which support the autonomy and independen
- PDF / 321,363 Bytes
- 11 Pages / 429.442 x 659.895 pts Page_size
- 105 Downloads / 205 Views
Fraunhofer Institute for Computer Graphics Research, 64283 Darmstadt, Germany {gee.fung.sit,chenbin.shen,cristian.hofmann}@igd.fraunhofer.de 2 Fraunhofer Institute for Experimental Software Engineering, 67663 Kaiserslautern, Germany [email protected]
Abstract. A glance at the today’s research and industry community shows that AAL installations are normally offered as “complete solutions”, often including overlapping of almost equal or homogeneous sensors. Thus, redundant sensors are integrated in one single space when purchasing different AAL solutions, leading to an increase of acquisition costs and higher data volume. In order to counteract this problem, we present a method for application-oriented fusion and aggregation of sensor data. Here, the main contribution is a reference model and a semiautomatic approach for the determination of applicability of sensors to predefined AAL applications.
1
Introduction
In specific areas such as home automation, activity recognition, or smart metering, a multitude of sensor and actuator technologies are employed which have different properties, e.g., wireless vs. wired communication, microsystems and terminals, etc. Here, methods for sensor fusion and aggregation have a supporting effect: the combination or junction of sensor data leads to better quality of gathered information in comparison with the consideration of individual devices. In this scope, improvement of information refers to a more stable behavior with perturbations and an increased clearness of statement by means of enhancement of the measurement range and the resolution of measured data [1]. Furthermore, additional information, like characteristic activities of daily living (ADL) for long-term behavior monitoring, can be gathered from the combination of sensor data, which cannot be captured from the single data streams [2][4]. So, main directions of sensor fusion focus on activity recognition, context recognition, or personal identification. A consideration of the research community shows that, from a methodological point of view, especially the area of sensor technology for activity and context recognition is not sufficiently understood and supported. Current publications on AAL applications indicate that sensor data is often integrated in an application or system in a proprietary way (tailored to specific requirements) or, if basic
4
G.F. Sit et al.
interfaces are defined, no detailed specification of sensor types and data formats are provided. Common users and deployers of respective systems, for instance programmers, system integrators or engineers still rely on ad hoc defined interfaces and methods of data processing. Thus, AAL solutions or installations are normally offered as complete packages consisting of a collection of sensors (and actors) along with software supplying complex behavior. As these are tailored for specific needs it stands to reason that multiple systems may be required within an AAL space. In essence, each particular need, i.e. each application, requires the acquisition of an e
Data Loading...
