EVENT PROCESSING IN SENSOR STREAMS - Managing and Mining Sensor Data - page 78

complex RFID event processing, including event specification languages,
event detection models, event processing methods and their optimiza-
tions. Additionally, we have presented an open discussion on advanced
issues such as processing uncertain and out-of-order sensor events.
Keywords: Sensor streams, RFID streams, event processing
1. Events and Event Processing
We first present an overview of events and event processing in the con-
text of sensor streams, including semantics of events, event processing,
and use cases of sensor event processing.
1.1 Semantics of Events
An “ event ” is a happening of interest [77]. In database applications,
the interest in events comes mostly from the state changes that are
produced by data manipulation operations [54]. Example events in real
world include a financial trade, a web click, a sensor reading and a
social or natural significant happening, and many others. In a monitored
environment deployed with sensors, flows of observation data can be seen
as streams of observable events. When an event takes place, we refer
to its occurrence ; and when an event is recognized by the system, we
refer to its detection .
Events are often interrelated and form complex relationships, such
as temporal, spatial, causal or abstract (or composite) relationships.
A temporal, spatial or causal relationship of events can determine the
partial order between events, and abstract or composite relationship can
be represented as an event that represents or summarizes a collection of
events.
Events can be categorized as atomic events and composite events.
Atomic events or primitive events are the simplest events in a sys-
tem, which are defined to occur at a certain time point or not occur
at all. A composite event or complex event is a high-level derived
event, and it is defined by applying an event operator to constituent
events that are primitive events or other composite events [54].
Events can have their attributes such as type, ID, and time; an event
attribute can have a simple or complex data type. Similar events can be
grouped into an event type, and an event type is denoted by an event
expression. A primitive event (type) name itself is an event expression.
If E 1 ,E 2 , ... ,E n are event expressions, an application of any event
operator over the event expressions is an event expression. An atomic