A Context Framework for Ambient Intelligence

1
A Context Framework for Ambient Intelligence

• A. Dogac, G. B. Laleci, Y. Kabak
• Middle East Technical University

2
Motivation
interoperable
User Context
Machine processable
Security privacy
3
Ambient Intelligence

• Ubiquitous computing
• Ubiquitous communication
• Intelligent User Friendly Interfaces
• Better integration of technology into our
  environment, so that people can freely and
  interactively use it
• Seamless (and secure) delivery of services and
  applications

4
Characteristics of AmI

• Awareness
• ability of the system to locate and recognize
  objects and people, their locations, and their
  needs
• Intelligence
• allows the system to analyze the context, adapt
  to people that live in it, learn from their
  behavior, and eventually to recognize as well as
  show emotion
• Adaptable
• learn about the environment and the people within
  it in order to optimize their own behavior

User Context
5
User Context

• Any information that can be used to characterize
  the user and her situation
• Coming from sensors
• Temporal and spatial location
• Environmental attributes
• Resources nearby
• Physiological measurements
• User preferences and profile
• Schedule, agenda
• Social context

6
An Example Scenario (from ISTAG)

• Maria lands to an airport in a Far Eastern
  Country
• The immigration officer replaced by a device
• Through its sensors, detects identities, performs
  visa and passport control (using Marias P-Com)
• A rented car waiting for her at the exit
• Her Hotel has been reserved by her personal
  software agent

7
Issues to be considered

• Identity information should be understandable by
  any authorized device

• Identity information

• Passport and visa information

• Machine proccesable
• Interoperable

• Context Ontologies should be developed

8
Issues to be considered

• Privacy
• Access should be limited to authorized entities
• The user should be able to state how much
  information to disclose and to whom
• Immigration device
• Location, visa, passport information
• Role or personal identity based privacy mechanisms

9
Issues to be considered

• The context servers needs to recognize the device
  type
• It should be possible to exploit context to
  discover and compose Web Services

10
Proposed System Architecture
11
Context Ontology

• Have formal specification
• Machine processable
• Can be queried through query languages
• Define shared conceptualizations
• Captures consensual knowledge
• Enable knowledge sharing in an open and dynamic
  distributed system

12
Context Ontology

• Define all of the concepts in a taxonomic
  hierarchy

• Define the properties and allowed values (facets)

• Define the relationships among the classes

• Provides a means for intelligent agents to reason
  about the contextual information

13
Web Ontology Language OWL

• OWL is a semantic markup language being developed
  by the World Wide Web Consortium
• for publishing and sharing ontologies
• derived from DAMLOIL

14
Security and Privacy Issues

• Ontologies are stored in knowledge-bases
• Data coming from sensor devices
• Solar System (Minami Kotz)
• Data coming from the context server
• Limited to authorized entities
• Should be possible to state how much information
  will be disclosed to whom

15
Security and Privacy Issues

• View mechanism
• Define views on the user context and grant access
  rights to different users
• CREATE CLASS VIEW ScheduleOfMaria
• SUBCLASSOF Schedule
• SELECT X
• FROM ContextServer
• WHERE X.Name Maria
• GRANT SELECT ON ScheduleOfMaria TO
  PersonalAgentOfMaria
• KAON server (http//kaon.semanticweb.org)

16
Exploiting User Context for Web Service Discovery

• For selecting services in a context-sensitive
  manner
• Services should be discovered based on their
  semantic descriptions
• Agents should
• Query the context of the user
• Maps the preferences of the user with the
  properties of the services advertised

17
Web Service Discovery

• ebXML registry allows metadata to be stored in
  the registry through Classification mechanisms

18
An Example Travel Ontology
TravelService
Entertainment Service
Accommodation Service
Transportation Service
AirTransportation
ReserveAFlight
BuyATicket
Properties of the Generic Service Class
originatingFrom
destinationTo
paymentMethod
19
Relating Services with Ontologies
MyService
20
An example scenario
Understands she doesnt have a valid visa
Checks her Calendar
Checks her Profile
Queries for the visa service of the Country
Finds the service and its WSDL link
Starts arranging her trip
invokes
Retrieves WSDL desc.
21
An example scenario
Prefers air transfer, collects mileage from THY
Queries for the THY reservation service
Finds the service and its WSDL link
Checks her Preferences
Checks her Profile
consults
invokes
Retrieves WSDL desc.
22
Conclusions

• AmI combine
• Ubiquity, context-awareness, intelligence and
  natural interaction
• There is a need for strong mechanisms for storing
  and processing context

• Context Ontologies

23
Conclusions

• To be acceptable AmI should provide
• Security
• Privacy

• Role-based access to context servers

• AmI should exploit user Context for reacting user
  needs automatically

• Web Service discovery and composition based on
  semantics

24
Thank you for your Attention Any Questions?