| |
72
Chapter 5. General Concepts, Toolkits, and Frameworks
Artifact Layer
In the artifact layer only a single artifact is modeled. Conceptually, we look at
the physical artifact rather than at the sensing or computational devices in this
layer. In some cases the artifact and the computational device are identical (e.g.
a PDA) but in the case of everyday objects the hardware is embedded into the
artifact. For modeling the artifact it is central to identify what the meaningful
context information is. This is related to the type and usage of the artifact. The
following questions help to determine context primitives that need to be supported
by the sensing and perception technology included in the artifact:
What is the artifact and what is its prime use?
Who are the users of such an artifact and in which situation do they use it?
Based on this analysis it is important to identify the technical implications for pro-
viding perception and recognition on the level of the artifact. First and foremost
is the identification of the required sensors. It is also of particular interest whether
the artifact is mobile or embedded in the environment and how energy is supplied.
Communication is another central technical issue, in particular what type of com-
munication is supported and with whom communication is feasible. The technology
embedded in the artifact with regard to processing, communication, and sensing has
to be specified. There are three general tasks that are accomplished in the artifact
layer on any artifact that is modelled.
Sensor data acquisition
This includes reading data from the actual sensors; and potentially also bu er-
ing, time stamping, and controlling the sensor.
Artifact centric perception processing
Perception processing includes low-level signal processing and basic algorithms
to make sensor values more robust. This typically includes the application of
filters to sensor data, the integration of data from di erent redundant sensors,
and the fusion of data from di erent sensors. One aim is to supply more
robust sensor and context information related to the particular artifact. The
other is to provide higher abstraction of the local sensing. Finally, the artifact
structure also allows to represent the real-world structure of the related object.
History and long term bu ers
In various scenarios long term information is of great interest and contains
important context information. Keeping the history and abstractions of the
data on a particular artifact allows the local calculation of temporal contexts.
|  |
|
| |
|
|
