High Level Definition of Event-Based Applications for Pervasive Systems - Advanced Design Approaches to Emerging Software Systems - page 144

the application context. Detailed descriptions and
configurations of these elements are presented in
later Sections introducing publish/subscribe and
leasing time.
Of course, proper adjustment of both param-
eters requires skills in distributed computing or
experience in WSN configuration. Hence, the user
assistant shall be able to customize these param-
eters automatically based on experience data or
available patterns. In addition, the assistant should
provide an expert mode, which allows users to
manually customize each parameter.

execution constraints for each phenomenon. This
element defines the event evaluation frequency
as a time interval. Time intervals can be quanti-
fied by acceptable periods or exact time slots that
must be adhered to.
The <CONSEQUENCE> element is the last
mandatory component of an event specification. It
links procedures to the event specification, which
have to be executed upon phenomenon detection.
These procedures are called event handlers. Event
handlers are listed as <TRIGGERHANDLER>
elements, containing the name of the event
handler. Specifying several event handlers in a
single event specification is allowed and all of
them are executed in the sequence listed, when
the respective phenomenon occurs. Since event
handlers trigger available functions or processes
at the sensor nodes, those must also be adapted
by the language interpreter to the target platform
and the respective Operating System (OS). For
example, a general event handler such as sendalert
could be mapped to a respective interrupt routine
of the OS. This is automatically done by the event
configuration system.

Addition of Execution Constraints
and Associated Handlers

Next, the temporal expansion of the phenomenon
is to be configured. Real-world phenomena are
usually subject to different temporal expansion,
which must be considered for event specifica-
tion as well. For example, the acoustic wave of
an explosion can only be detected within a few
milliseconds and hence, requires a short sensing
interval. The frequency of event evaluation and
coupled collaboration processes consequently
affects the energy consumption of the sensor
nodes. Energy consumption is an essential and
very critical issue when designing WSN-based
applications. Sensor nodes provide different
modes of operation that result in significantly
different amounts of energy consumption. Active
modes like data processing or data transmission
are draining the energy resources much more
than passive modes such as sleeping. Thus, ac-
tive periods must be kept as short as possible to
reduce energy consumption to a very minimum.
On the other hand, extensive passive periods may
reduce the accuracy and reliability of event detec-
tion. When a node may switch to a power saving
mode highly depends on the application running.
Therefore the ESL provides means that help to
adjust the update rates of sensor readings. Thus,
an event specification contains an <TIMEINTER-
VAL> element to configure application-oriented

Determining the Region of Event

Besides the temporal resolution of the event
detection, configuration of the expected spatial
expansion of the phenomenon is necessary.
Wireless sensor nodes can communicate up to
approximately 300 meters but many phenomena
usually appear only locally. For example in an
environmental surveillance scenario, temperature
changes usually appear widely, whereas the size
of an emerging fire is relatively small but has to
be detected as well. As we learned, a suitable col-
laboration region closely relates to the expected
spatial expansion of the phenomenon. Hence, the
ESL allows to describe this collaboration region.
If sensor nodes may jointly share their resources
for collaborative event detection, these nodes must
know whether they share a certain collaboration
region. The ESL configures this valid region within