Event-based Middleware for Sensor Networks

1
Event-based Middleware for Sensor Networks

• Bin Wu Roy George
• Department of Computer and Information Science
• Clark Atlanta University
• rkavil_at_cau.edu

2
Overview

• Incorporating Semantics in Sensor Networks (SSN)
  - Driver Applications
• Event-based Multisensor Data Fusion
• Why Use Events?
• Event Definitions
• Event Hierarchical Model
• System Architecture
• Event Ontology
• Previous Work
• Architecture of Event Ontology
• Related Ontologies
• Event Ontology Language
• The Application of SSN Healthcare Application
• Conclusion Future Work

3
Incorporation of Semantics in Sensor Networks

• Enables incorporation of semantics into network
  definitions
• Adaptive with capability to respond to
  environmental changes
• Deals with event streams
• Facilitates automatic processing

4
In the applications of SSN

• Decision support is based on the dynamically
  updated events.
• Events are fused from complex and heterogeneous
  data sources.
• Need for real-time or near real-time data storage
  and retrieval.
• Spatio-temporal data is norm rather than an
  exception.
• Exploration is the predominant mode of
  interaction rather than query.
• Context and state are very important.

5
And The Problems Are

• Explosion of Raw Data from Heterogeneous Data
  Sources
• Need for Real-time Decision Support
• Need for Service Oriented Integration
• Need for High Performance Data Repositories

6
Current Approaches

• Heterogeneous data is located within silos.
• Relationships between events are hard to
  recognize.
• Context information of an event lost.
• Keyword based Queries
• Centralized handling of Events

7
Event-based Systems Applications

• Operating systems
• aDBMS
• Interface design
• Distributed simulation systems

8
Why Event-based Middleware?

• Temporal and Spatial properties are a fundamental
  organizational mechanism for events.
• Provides a natural way of filtering data.
• Real-time decision making.

9
Definition of Event
10
Features of an Event

• Where At or in what place
• When At what time
• What What is the relationship between events?
  How does the event evolve?

11
ER Model of an Event
eID
Event Name
Event
latitude
m
m
1
1
1
1
n
occurs at
Space
has
Location
longitude
name
n
Transcluded Media
has
URI
n
name
Event Topic
has
n
1
Sub-topic
1
occurs at
Time
time
1
1
1
occurs at
Start Time
date
time
1
occurs at
End Time
date
12
Temporal Relationships
Relation Symbol Inverse Symbol Graphic Example
C1 before C2 lt gt
C1 equal C2
C1 meets C2 m mi
C1 overlaps C2 o oi
C1 during C2 d di
C1 starts C2 s si
C1 finishes C2 f fi
Maintaining Knowledge about Temporal
Intervals, James Allen, 1983.
13
Temporal Scenario
1990
1994
1996
2001
2003
14
Spatial Relationships
Spatial co-occurrence C1 (s) C2 (s) All
events in the respective categories overlap in
space.
15
Temporal Spatial Scenario
16
Causality in events

• Causality is the relation between causes and
  effects.
• It is used for describing the evolution of an
  event.

17
Event Hierarchical Model
18
Event-based Edgeware Architecture
19
Graphical User Interface
20
The Event Ontology

• Representation of the semantics of events,
  processes and states
• Basis of sensor-based models of the dynamic
  world.
• Distributed intelligence required to handle the
  transaction at point of its occurrence.
• Common vocabularies to needed to understand and
  share events

21
Event ontology is used for

• Represents the attributes of an event, such as
  time, space, causality, etc.
• Assist the construction of associated context
  where events happen and reasoning the evolution
  of events in enterprise applications.

22
Related Ontology Work

• Sensor networks ontologies
• Current sensor network ontologies focus on
• Adaptive sensor networks to determine the future
  state of the network (Avancha, 2004)
• General interface between sensor networks and
  Internet services that facilitates bidirectional
  interactions between internet users and sensors,
  as well as interactions between sensor networks
  themselves. (Ota, 2003)
• Describe the major properties of sensor networks
  such as sensor location and sensing mechanism.
  (Jiang, 2003)
• Context ontologies
• The Aspect-Scale-Context (ASC) model describes
  contextual facts and contextual
  interrelationships as well as allow to determine
  service interoperability on the context level.
  (Strang, 2003)
• CONtext ONtology (CONON) an extensible ontology
  for modeling context in pervasive computing
  environments. (Gu, 2004)
• Ontologies in FLAME2008 developed on three
  levels, upper ontology, domain and task
  ontologies, and application ontology, based on
  standards like ISO 19115 (geo metadata) and ISO
  19119 (geo service). (Weißenberg, 2004)
• MIX model is a set of common domain-specific
  vocabularies for the representation of event
  content. (Bornhövd, 2000)
• CORBRA-ONT was developed as a part of the Context
  Broker Architecture (CoBrA) to model places,
  agents, events and their associated properties in
  an intelligent meeting room domain. (Chen, 2003)
• Event-related ontologies
• Video Event Representation Language (VERL)
  ARDA-sponsored Event Taxonomy project provides a
  common representational framework and ontology
  for describing video events. (Nevatia, 2004)
• Discrete-Event Modeling Ontology (DeMO)
  discrete-event modeling (DEM) aiming to assist
  the researchers in simulation area. (Miller,
  2004)
• Versatile Event Logic (VEL) was a semantic
  language to represent temporal relationships and
  events. (Bennett, 2004)

23
Related Ontologies

• Time Ontologies
• DAML-Time
• http//www.cs.rochester.edu/ferguson/daml/
• Entry Sub-ontology of Time
• http//www.isi.edu/pan/OWL-Time.html
• The vocabularies of DAML-Time the Entry
  Sub-ontology of Time are designed for expressing
  temporal concepts and properties common to any
  formalization of time.
• Space Ontologies
• SNAP and SPAN spatial ontologies
• Spatial ontologies de?ne a vocabulary for
  symbolic representation of space.
• The ontology of GIS
• Consists of vocabularies for expressing spatial
  relations for qualitative spatial reasoning.

24
Two-level Model of Event Ontology
25
Event Ontology Language

• Integrates components from related ontologies.
• Based on OWL.
• Supports semantic interoperability to exchange
  and share event knowledge between different
  domains.

26
Global Ontologies

• Time Ontology
• Space Ontology

27
Time Ontologies
Time ontologies are proposed to express time and
temporal relations. They can be used to describe
the temporal properties of different events that
occur in the physical world.

• Adopts the vocabularies of the DAML-Time and the
  Entry Sub-ontology of Time.
• Basic vocabularies are timeTimeInstant and
  timeTimeInterval classes.
• The objects in an event is divided into two
  disjoint classes timeInstantThing and
  timeIntervalThing.

28
Example
lttmeTimeIntervalgt lttmefromgt lttmeTimeInstantgt
lttmeat rdfdatatype"xsddateTime"gt 2004-
02-01T120101 lt/tmeatgt lt/tmeTimeInstantgt
lt/tmefromgt lttmetogt lttmeTimeInstantgt lttme
at rdfdatatype"xsddateTime"gt 2004-02-11T13
4121 lt/tmeatgt lt/tmeTimeInstantgt lt/tmetogt
lt/tmeTimeIntervalgt
29
Temporal Relationship in EOL
EOL de?nes the following properties for
describing the temporal relationships between
events.

• timestartsSoonerThan
• timestartsLaterThan
• timestartsSameTimeAs
• timeendsSoonerThan
• timeendsLaterThan
• timeendsSameTimeAs
• timestartsAfterEndOf
• timeendsBeforeStartOf.

30
Space Ontologies
Space ontologies support reasoning about the
spatial relations between events.

• Adopts the vocabularies of
• SNAP and SPAN spatial ontologies
• OpenGIS
• Two documents
• spatial relationships
• typical geospatial vocabularies
• The objects in an event is described with class
  spaceSpatialThing.

31
Domain Ontologies

• Object Ontology
• Event Ontology

32
Object Ontology
Object Ontology is used for describing objects in
an event by a set of properties. In EOL, it is
included within the objProduct, objPeople,
objRelationship classes.

• ObjSense Describes the sensing element.
• ObjPeople Describes the actors in an event.
• ObjRelationship Describes the relationships

33
Example
ltobjSensegt ltobjname rdfdatatype"xsdstring"gt
Temperature Sensor lt/objnamegt ltobjmanu
rdfdatatype"xsdstring"gtOregon
lt/objmangt ltobjmodel rdfdatatype"xsdstring"gt
THC268 lt/objmodelgt ltobj manfcDate
rdfdatatype"xsddate"gt2004-09-12lt/obj
manfcDategt ltobjprice rdfdatatype"xsdstring"gt
23.97lt/objpricegt ltobjspec rdfresource"http/
/www.amazon.com/exec"/gt ltobjpicture
rdfresource"http//www.amazon.com/exec1"/gt lt/obj
Sensegt
34
Event Ontology

• The event ontology can be used to describe the
  occurrence of different activities, schedules,
  and sensing events.
• In the event ontology document, the eveEvent
  class represents a set of all events in the
  domain.
• eveSpatialTemporalEvent class is de?ned to
  speci?cally describe events that have both
  temporal and spatial extensions.

35
An example
ltowlClass rdfID"DetectedBluetoothDev"gt ltrdfss
ubClassOf rdfresource"eveTemporalSpatialEvent"
/gt lt/owlClassgt ltowlObjectProperty
rdfID"foundDevice"gt ltrdfsdomain
rdfresource"DetectedBluetoothDev"/gt lt/owlObjec
tPropertygt ltDetectedBluetoothDevgt ltspacehasCoord
inatesgt ltgeoLocationCoordinatesgt ltgeolongitud
e rdfdatatype...gt -76.7113 lt/geolongitudegt
ltgeomlatitude rdfdatatype...gt 39.2524 lt/g
eomlatitudegt lt/geoLocationCoordinatesgt lt/sapce
hasCoordinatesgt ltfoundDevice rdfresource"url-x
-some-device"/gt lttimeatgt lttimeTimeInstantgt ltt
imeat rdfdatatype"xsddateTime"gt 2004-02-01T
120101 lt/timeatgt lt/timeTimeInstantgt lt/time
atgt ltDetectedBluetoothDevgt
36
Application
37
Emergency Care Sensor Network
38
(No Transcript)
39
2200 02/11/2005
919 02/12/2005
delivery
rest (in recovery room)
go to operating room
episode 1
episode 0
episode 2
Domain event level
2200
1230
1240
605
745
919
thread 1
thread 2
thread 3
thread 4
thread 5
Element event level
2200
1230
1240
745
919
2215
2230



state 3
state i
state 1
state 2
state j
state j1
state k1
state m
state n
Data event level
300
305
600
605
610
620
level
medical care domain
domain
Timeline (Unit hour)
40
(No Transcript)
41
Conclusions

• Events are used as the container to encapsulate
  the data stream.
• EOL is presented a formal and extensible event
  model based on OWL to represent, manipulate and
  access event streams and their properties in
  intelligent environment.
• Numerous applications of Event-driven Semantic
  Sensor Networks.

42
Future Work

• Semantic query for event streams will be
  developed.
• Event-based Reasoning
• The domain ontology of event ontology will be
  extend to the C4ISR application domain.

43
Related Publications

• B. Wu, Z.J. Liu, and R. George, Event-based
  Edgeware Managing Data from RFID Networks,
  International Conference on Sensor Networks,
  Montreal, Canada, 2005.
• B. Wu, Z. Liu, R. George, and K. Shujaee.
  eWellness Building a Smart Hospital by
  Leveraging RFID Networks, Sep. 2005. IEEE EMBS
  2005 Conference in Shanghai, China.
• B. Wu, R. George, Event-based Edgeware in
  Hospital Networks, submitted to Journal of UCS,
  2005
• Bin Wu, Rahul Singh, Punit Gupta, Ramesh Jain.
  eVitae An Event-Based Electronic Chronicle.
  Demo paper, 9th International Conference on
  Extending Database Technology, EDBT04.
  Heraklion, Crete, Greece, March 2004.