Personalizing Pervasive Services on Top of Heterogeneous Networks

1
Personalizing Pervasive Services on Top of
Heterogeneous Networks
Yuping Yang Richard Everson University of
Exeter, UK Rick Dewar M. Howard
Williams Heriot-Watt Universtiy, UK
2
Background (1)

• Daidalos
• - Designing Advanced network Interfaces for the
  Delivery and Administration of Location
  independent, Optimised personal Services
• - The Sixth Framework IST Integrated Project 49
  partners from 16 countries
• - Long-term goal create the support for
  mobility needed by users
• - Focus
• Broadcast and multicast capabilities
• The development of services and applications
  based on pervasive computing, intelligent
  context- awareness and extended personalization.

3
Background (2)

• The aim of Daidalos
• To provide a universal and open service
  platform that can offer pervasive services to
  application developers in such a way that the
  underlying network technology becomes fully
  transparent.

4
Daidalos Pervasive Service Platform (PSP)
5
Overview of Personalization
6
User Preferences in Daidalos

• Non-logic-based preferences
• - Contextual information
• - Attribute/Value Pairs
• Rule-based preferences
• - Containing logic
• - Conforming to the RM specification

Context Mgm
Rule Mgm
7
Three Levels of User Preferences (1)
Generic preferences
Preferences for a certain service type
Preferences for a specific service
8
Three Levels of User Preferences (2)

• Generic rules
• E.g. If I am on a business trip, choose the
  service with the best QoS otherwise, choose the
  one with the cheapest price.
• Rules related to a type of service
• E.g. If it is a small screen, set the number of
  images to be shown simultaneously 1 and display
  frequency 2, and add the image of my family
  otherwise, set the number of images 2 and
  display frequency 4, and remove the image of my
  family.
• Service specific rules
• E.g. If I am in the UK, always choose the
  Vodaphone network for me whenever it is possible.

9
Multiple Identities and Roles
10
Personalization of QoS

• The role of Personalization in determining users
  preferred QoS
• - Before a session starts.
• - At some point in time after a session has
  started, the user context
  changes.
• -The user transfers a session from terminal A
  to terminal B.
• - A session is automatically transferred from
  terminal A to terminal B, based on context and
  rules.
• - The user requests explicitly to
  increase/decrease the current QoS to a certain
  degree.

11
Daidalos QoS Setting
12
Implementation of Dynamic QoS Configuration
13
Personalization Support to Network Layer (1)

• Personalization determines user preferences/rules
  on networks and stores a copy of them in the
  underlying layer via the Intelligent Interface
  Selection (IIS).
• This copy will be updated by Personalization when
  the related user preferences are changed.
• In the case that multiple networks are available,
  the IIS selects one according to the user
  preferences stored locally taking into account
  the price, QoS, provider, etc. of the network.

14
Personalization Support to Network Layer (2)
Bytes in a UDP packet User Network Preference Information
4 bytes(0-gt3) The user preferences in decreasing order (ttechnology, ccosts, pprovider, qquality)
4 bytes(4-gt7) Preferred technologies in decreasing order (WWLAN, DDVB, UUMTS, wwiMax)
1 byte (8) Preferred network Provider
15
Current Status of Daidalos

• The first implementation of the Daidalos PSP was
  completed in Dec. 2004.
• - Two scenarios, diabetes scenario and shared
  wall display scenario, were used to demonstrate
  the prototype.
• A more complete and integrated system with
  advanced functions was evaluated in Nov. 2005.
• - A scenario showing a day in the life of
  business man Bart was used to demonstrate the
  system.
• Future Work In Daidalos Phase II (2006 2008),
  work on Personalization focuses on learning and
  adaptation of composition process.

16
Personalizing Pervasive Services on Top of
Heterogeneous Networks
Thank you. Any questions?