Information Technology Reference
In-Depth Information
High-level programming languages are of
prominent use for definition of WSN applica-
tions. There exist a couple of projects trying
to adapt widely used high-level programming
languages like Java (Simon, Cifuentes, Cleal,
Daniels, & White, 2006; Brouwers, Corke, &
Langendoen, 2008) to sensor nodes. Of course,
this enables a rapid development of new WSN
applications by professionals but consequently
requires programming skills in distributed system
design. The macro programming language STOP
(Wada, Boonma, & Suzuki, 2007) is a scripting
language explicitly designed for data collection
in WSNs. It allows to create data queries from a
global viewpoint without considering details of
single nodes. Based on migrating agents, which
collect required data according to a given script,
STOP provides a more comfortable data collec-
tion. Nevertheless, the usage of general languages
requires a complex run-time environment and
Virtual Machines (VMs) on every node. Further,
VMs are usually adapted to certain sensor platform
only and do not support application design across
several platforms.
Finally, these approaches still require to make
use of scripting or programming languages, which
is not feasible for non-scientific deployment. A
straightforward sensor configuration providing a
proper usability of WSNs for non-scientific de-
ployment is still missing. A configuration concept
that aims at ease of use for sensor configuration
must be tailored to the user and self-configure to
defined tasks. Thus, the user only needs to describe
the phenomenon to be sensed without taking care
of WSN properties and deployment conditions.
gets even harder if heterogeneous systems using
miscellaneous sensor nodes need to be configured
to the same task. Hence, sensor nodes may not
provide all sensing capabilities needed for local
detection of the phenomenon to be sensed. In that
case, sensor nodes must collaborate and share
their sensing capabilities to continue with event
detection. For reliable application it is a necessity
to enable sensor nodes to autonomously deal with
different conditions as being expected in pervasive
systems, i.e., heterogeneous distributed sensing
capabilities, missing resources, node mobil-
ity, varying network topology, failed sensors or
sensing units etc. The following discusses some
previously presented mechanisms of collaborative
event detection.
Vu et al. (Vu, Beyah, & Li, 2007) introduced
a composite event detection scheme for sensor
networks composed of different nodes with vary-
ing sensing capabilities. They split complex event
detection among different nodes into sets of so
called atomic events, which are similar to primi-
tive events (threshold values). Atomic events are
merged by special gateway nodes to determine
final results. The gateway nodes however build
SPoFs. This approach provides configurable levels
of fault tolerance by selecting an appropriate
k
for
k
-watching sets of sensors while considering
the energy consumption and the event notifica-
tion time but requires an expensive setup phase.
Phani Kumar et al. (Phani Kumar, Reddy V, &
Janakiram, 2005) present a similar collaboration
scheme. They create event-based trees for complex
events containing all assigned sensor nodes. These
nodes collaborate using a content-based publish/
subscribe communication model, where child
nodes publish readings of interest to parent nodes.
The root node of the event tree obtains all sensor
readings and decides about the monitored event.
Again, this root node is a SPoF and introduces
vulnerability to the system.
Krishnamachari et al. (Krishnamachari, &
Iyengar, 2003; Krishnamachari, & Iyengar,
2004) introduced a self-organizing algorithm that
Autonomous Collaborative
Event Detection
An ease of use for WSN configuration starts with
a straightforward definition process but must also
provide reliable and robust execution of defined
tasks during runtime. Robust application is al-
ready a challenge for homogeneous systems, but
Search WWH ::
Custom Search