An Infrastructure for Contextawareness Based on First Order Logic Anand Ranganathan

1
An Infrastructure for Context-awareness Based on
First Order LogicAnand Ranganathan

• Jan. 4, 2005
• Choi, Chang Ik

2
Overview

• Introduction
• Key Features of First Order Context Model
• The Context Model
• Protocols for Obtaining Context
• The Context Infrastructure
• Context Synthesizer
• Configuration File
• Optimizing Context Expression Evaluation
• Conclusion

3
Introduction

• Context simplifies and enriches human-computer
  interaction
• It is necessary to develop a common, reusable
  model for context used by ubiquitous environment
• First order calculus
• Allows complex rules
• Enables automated inductive and deductive
  reasoning
• Express description of context
• Part of GAIA project (UIUC)

4
Key Features of First Order Context Model

• Supports gathering of context information from
  different sensors and delivery of appropriate
  context information to applications
• Allows complex reasoning
• Allows applications to specify different
  behaviors in different contexts

5
The Context Model (1/4)

• The basic structure-the context predicate
• Example
• Location (chris, entering, room 3231)
• Temperature (room 3231, , 98 F)
• StockQuote (msft, gt, 60)
• Arguments are constrained by the type of context
• Structure of context is specified in ontology
• Check validity of context predicate

6
The Context Model (2/4)

• Operations on context
• Using Boolean operation conjunction,
  disjunction, negation
• Example
• Location (Manuel, Entering, Room 3211) ? Social
  Activity (Room 3211, Meeting) refers to the
  context that Manuel is entering Rom 3211 and that
  there is a meeting going on in that room
• EnvironmentLighting (Room 3234, Off) ?
  EnvironnmentLighting (Room 3234, Dim) refers to
  the context tat the lighting in Room 3234 is
  either off or dim

7
The Context Model (3/4)

• Existential quantifier (There exists)
• Context is true for at least one value of the
  variable w/i the indicated scope of the variable
• Example
• ?Location y Location (Chris, In, y)
• Chris is in some location
• Universal quantifier (For all)
• Context is true for all values of the variable
  that lie in the scope of the variable
• Example
• ?people x Location (x, In , Room 3231)
• All people in Room 3231

8
The Context Model (4/4)

• Additional Features
• Arguments in the context predicate can be
  functions that return some value
• ?Person s Location (s, Entering, currentRoom())
• currentRoom() returns the room in which the Room
  Controller application is running gt can be used
  for all rooms
• Derive new context from other sensed contexts
• Sound (Room 3234, gt, 40dB)?Lighting (Room 3234,
  Stroboscopic)?People (Room 3234, gt, 6)gt
  Social Activity (Room 3234, Party)

9
Protocols for Obtaining Context (1/2)

• Query-answer protocol
• Application sends a query to the Context Provider
• One or more fields of the context predicate is
  replaced by a variable
• Example
• Who is in Room 3231?
• Location (X, In, Room 3231)

10
Protocols for Obtaining Context (2/2)

• Subscribe-notify protocol
• To subscribe for certain contexts and gets
  notified whenever that context becomes true
• Simple callback from the Context Provider
• Event-based mechanism
• Bulletin board kind of system
• In our model, CORBA event mechanism
• Using event channel
• Any interested Context Consumers can listen on
  these channels and get notified whenever the
  context changes

11
The Context Infrastructure

• Context Providers
• Obtain context from sensors or other data source
• Context Consumers
• Context-sensitive applications
• Context Synthesizers
• Derive higher level contexts
• Context Provider Lookup Service
• Context Providers advertise the context they
  provide
• Allows applications to find appropriate Context
  Providers
• Context History
• All past contexts are maintained in a database

12
Context Synthesizer

• Rule-based synthesizer
• More than one rule can be true at the same time
• If we want the Context Synthesizer to return just
  a single value for the context, we use
  priority-based mechanism for resolving conflicts
• Context Synthesizer re-evaluates the rules
  whenever it gets an event from one of the Context
  Providers that supply the basic context
• May not deduce the right context simply because
  the real world may not follow the rules

13
Example Scenario
SocialActivity (Room 2401, Lecture)
Activity Context Synthesizer
Location (Manuel, In, Room 2401) Location
(Chris, In, Room 2401) Location (Roy, Entering,
Room 2401)
Temperature (Room 2401, , 68) Lighting (Room
2401, Is, Dim)
Application (PowerPoint, Is, Running) Application
(MP3 Player, Is, Running)
Location Context Provider
RoomEnvironmentContext Provider
Applications Context Provider
14
Configuration File

• Gives a set of rules that specify the behavior of
  the application in different context
• Action with higher priority wins, if same, then
  action is randomly executed

15
Optimizing Context Expression Evaluation (1/2)

• Whenever a context changes the Context
  Synthesizer has to re-evaluate the inferred
  context
• Re-evaluation takes O(n) time and can be very
  inefficient if the number of rules is very large
• Use Poset Structure (Partially Ordered Set)
• If context expression of a node in the poset
  evaluates to false, then the context expression
  of all its descendents will also be false

16
Optimizing Context Expression Evaluation (2/2)
Top-down search
Hierarchy of context rules for Jukebox application
17
Conclusion

• We can enhance the context model by associating
  probabilities with context expressions to handle
  the uncertainty in deducing context
• An efficient algorithm to generate new inferred
  context when changes occur in real-world
• A well-defined grammar to represent context is
  important for generation, aggregation, inference,
  and querying of contexts