Sensor Networks’ Integration - Application and Multidisciplinary Aspects of Wireless Sensor Networks - page 96

enabler may have to interact with multiple sensor network systems to derive the
required context information components. Based on an incoming context informa-
tion request of an IMS application context service enabler performs a decomposi-
tion of high-level context information requests into low-level service task graphs to
be executed by one or more wireless sensor networks. It then requests the low-level
services from the identified sensor network gateway and composes the context
information response from the required context information components of the
service responses. Communication between sensor networks and gateways is
enabled by publish/subscribe mechanisms based on the SIP event framework.
The e-SENSE framework offers several desired features such as discovery of
available sensor networks and their service capabilities. It enables a decoupling of
application from the underlying sensor networks - that is, applications can enquire
the contextual information without requiring knowledge of the underlying sensor
networks. Application queries can encompass information that may be jointly pro-
vided by multiple sensor networks and processing and composition can be achieved
either in the sensor networks/gateways or in the service enabler. While making use
of scalable well-understood signaling mechanisms, it raises scalability concerns as
communication requires always the involvement of the centralized service enabler.
Although multiple physical instances of a context service enabler can be deployed,
the architecture does not address how coordination between different such instances
is achieved. Optimization of data flows across different queries are currently not
addressed by the proposed frameworks, as are not adoptions of existing queries to
changing conditions in the system. Furthermore, the architecture of the e-SENSE
framework falls short in considering security, privacy, trust, and accounting issues.

5.2.2.2 Peer-to-Peer

Global Sensor Networks

Global Sensor Networks [ 18 ] is an approach of providing a distributed middleware
platform for integrating heterogeneous sensors into a “sensor web” providing internal
stream processing capabilities on the exchanged sensor information. The architec-
ture of the GSN framework is based on distributed peer entities called GSN con-
tainers. GSN containers are typically deployed at normal Internet hosts or servers
and communicate with each other via point-to-point connections. Core element in
GSN is the so-called virtual sensor abstraction. Virtual sensors abstract implemen-
tation details to access information from physical sensors and allow a unified way
of treating sensor services or composed sensor service by the middleware. The
specification of a virtual sensor includes metadata for identification and discovery,
the structure of input and output streams, SQL-like internal streaming processing
and properties related to life-cycle management and physical deployment. Virtual
sensors have one or multiple input streams and produce exactly one output stream.
Input streams can come from physical sensors interfaced via implementation-specific
wrappers or other virtual sensors. Virtual sensors can manipulate and combine