Information Technology Reference
In-Depth Information
electronics for monitoring and switching individual cells), and intelligent power
electronic devices (thermal management, i.e. sensing, cooling, the use of thermally
stable materials like SiC and/or GaN, packaging technology, passive components,
semiconductor devices.
R
oadmap 9.
Energy scavenging and management.
Short term
(2015)
Mid term (2020)
Long term (2025)
Power
Energy scavenging
systems
Advanced battery
management system
Intelligent power
electronic devices
Advanced energy
scavenging systems
Endless power to mobile sensors, actuators
and processors (wireless/battery)
5.2. Reasoning
5.2.1.
Reasoning for AAL
A core function of AAL systems is the conclusion of knowledge about the activities of
the user and the current situation in this environment from low-level sensor data and to
plan the appropriate short-term and long-term reaction. Not only daily activities and
situations have to be considered, but also emergency situations have to be detected
which need an immediate reaction as well as analysing the user' s mid-term and long-
term behaviour to assess the development of the physical, mental, and psycho-social
status. The characteristics needed for this task are to reliably and quickly detect
situations based on input information provided by sensors which must be considered as
inherently imprecise and unreliable. Multiple heterogeneous information streams must
be fused. The reasoning system must be adaptable to varying environments and users.
But most importantly, the reasoning in complex human activities or behaviour requires
highly expressive models.
5.2.2.
Models
The main requirement for valuable reasoning is that any information characterizing the
situation of an entity can be used. A.K. Dey (Anind 2001, S.4-7) uses the term
“context” to define this information. An entity includes a person, a device, a location or
a computing application. Entities are characterized by attributes, e.g. the location could
be characterized not only by its topology, but also by its noise level, light intensity,
temperature and humidity.
Recent approaches have described this information using models, because they
improve the separation of concerns of application logic from structural characteristics.
Constructing platform-independent models and using mapping and transformation
techniques to transform and refine models improves interoperability, extensibility and
reuse in software applications. Issues of model generation, evolution and complexity
have still to be resolved. Model generation is currently a demanding, time consuming
and error-prone task requiring highly trained specialists. In order to make model-based
approaches economically viable, automatisms as an alternative to manual model
encoding are needed.
To refine reasoning and to integrate context as broad as possible, the models used
will be extended and additional background and common sense knowledge will be
