Contextaware service design and delivery

1
Context-aware service design and delivery

• Marten van Sinderen
• IS Seminar, 10 July, 2007

2
Outline

• Introduction
• Project overview
• A-MUSE service design
• PhD projects
• AWARENESS service delivery
• PhD projects
• Conclusion

3
Introduction

• Positioning of research
• middleware application level
• specification of interoperability
• Keywords
• service-orientation
• context-awareness

4
Service-orientation

• Next step in evolution of modular middleware
  approaches
• Service as central concept
• Independent of user environment and platform
  technology
• Applies to technical and organizational services
• Requires existence of distributed computing
  platform
• Requires unambiguous definition of service
  concept and theory to support manipulation of
  services

5
Service definitions

• From communication to computing
• Several interpretations
• interaction that creates and captures value
• capability of providing entity
• operation on providing entity
• software system supporting m2m interaction across
  network
• value-added feature on top of basic (telecom)
  facility
• observable behaviour of providing entity

6
Service properties and abstractions

• General aspects
• structure
• behaviour
• information
• goal
• quality
• Abstraction levels
• goal
• choreography
• orchestration

7
Context-awareness

• Application is aware of its own context or the
  context of its users, and it adapts its behaviour
  according to the context situation
• e.g., provide directions to the nearest point of
  interest based on current user location
• e.g., provide bio-feedback on health status
  monitoring dependent on current user activity
• Characteristics of context
• Enables user centric adaptations
• Inherently vague (due to limits in sensors)
• Often needs to be reasoned
• Extremely dynamic
• Is privacy sensitive

8
Quality of Context

• Context sources are never perfect
• Nor are reasoning techniques
• Quality of Context is essential
• application adaptation know when you dont know
  
• infrastructure efficiency e.g., cached context
• privacy enforcement higher quality is more
  privacy sensitive
• Which characteristics
• Freshness
• Temporal resolution
• Probability of correctness
• Precision
• Spatial resolution

9
Context reasoning

• Vertical reasoning - derive higher level
  context information from more primitive context
  information
• Horizontal reasoning - combine context
  information from different sources to improve the
  quality

10
Project overview

• Major national projects
• Freeband BSIK A-MUSE 4.5 M
• Freeband BSIK AWARENESS 10.8 M
• Spin-off European projects
• FP6 IP AMIGO ambient intelligence for the
  networked home environment
• FP6 IP SPICE service platform for innovative
  communication environment
• Other projects
• NWO Casimir reflective middleware support for
  large-scale data-centric systems
• EZ FOVEA food valley eating administrator/adviso
  r

11
A-MUSE

• 4.5 MLN 12 FTE/y
• 5 partners
• University of Twente CTIT
• Telematica Instituut
• Ericsson
• BiZZdesign (since 2006)
• Compuware (since 2006)
• Lucent Bell Labs (till 2005)
• GigSoft (till 2005)

12
A-MUSE goal
To research and propose model-based service
development approaches, specifically for
context-aware mobile applications

• High-level research questions
• What are proper different abstraction levels for
  services?
• How to model services at these abstraction
  levels?
• How to bridge between organizational and
  technical services?
• How to assess consistency of services
  (interoperability, conformance)?
• What are possible composition approaches?
• What are possible transformation approaches?

13
A-MUSE highlights

• Ontological foundation for conceptual modelling
  (SIGMOD nomination) a.o. used for definition of
  context and service interoperability, and applied
  in AWARENESS
• Consistency framework for multi-view
  architectural design idea of using a set of
  common basic concepts has been used in the
  development of transformations
• Guidelines and concepts for model-driven design
  a.o. introduced concept of abstract platform,
  which has been adopted by ISO/IEC SC7/WG19
• Semantic support for service discovery and
  matchmaking ontologies are used to improve
  existing methods, contributed to INTEROP

14
A-MUSE PhD projects
15
Rodrigos PhD project

• Started March 2007
• Dynamic composition of services, considering
• needs and requirements for B2B and B2C
• non-trivial (conversational) behaviours
• semantic interoperability
• Objectives
• identify and define useful model concepts for
  service2service matching and service2need
  matching
• develop algorithm that finds a composition
  matching a need
• apply and validate this in a prototype
• Currently looking at
• existing (dynamic) service composition approaches
• semantic web services challenge

16
AWARENESS

• 10.8 MLN costs 33 FTE/y
• 8 partners
• Alcatel-Lucent Bell Labs
• Telematica Instituut
• University of Twente - CTIT
• Roessingh RD
• Twente Institute for Wireless Mobile
  Communications
• Ericsson
• Yucat
• TMS-I

17
AWARENESS goal
To research and design infrastructure support for
context-aware mobile applications, and validate
this in the m-health domain

• High-level research questions
• How to model context?
• How to reason with context?
• How to do context management (heterogeneity,
  scalability)?
• How to handle the privacy and trust?
• What is added value of context for m-health?

18
AWARENESS infrastructure

• Infrastructure provides generic functionality to
  the applications and users
• Context model reasoning capabilities
• Gathering, management and discovery of context
• User controlled privacy
• Etc.
• Why ?
• Cheaper
• Easier

19
AWARENESS highlights

• Overall Architecture
• Network abstraction layer
• Context and Quality of Context model
• Statistical symbolic reasoning
• Personalized privacy enforcement
• Generic m-health platform including health BAN
• Epilepsy detection application
• Chronic pain application

20
AWARENESS PhD projects
21
Patricias PhD project

• Started August 2003
• Context handling infrastructure
• Research results
• context modeling abstractions
• based on conceptual modeling theory
• intrinsic and relational context types
• situations
• rule-based approach for application behaviour
• Event-Condition-Action pattern
• domain-specific language (ECA-DL)
• ECA2Jess transformation

22
Patricias PhD project

• Research results (contd)
• distribution of Jess rules
• architectural patterns for context management
  infrastructure
• context sources and managers hierarchy
• decoupling of action purposes from action
  implementations
• Ongoing
• case study

23
Ricardos PhD project

• Started Jan 2005
• Trust in context-aware systems
• Research objectives
• Investigate the role of trust in context-aware
  services
• Trust model Identity, context, privacy aspects
• Trust recommendations
• Integration of trust values in management policies

24
Ricardos PhD project

• Research objectives (contd)
• Investigate the role of context in trust/policy
  management
• Use context information to improve available
  trust, privacy and security techniques
• Dynamic trust management using context-aware
  domains
• Security policies using Context and Trust
• Raise users acceptance in the context-aware
  system by providing trust evidences

25
Kamrans PhD project

• Started March 2005
• Enforcing end-user privacy in context-aware
  pervasive environments
• Research questions
• How to collect information about capabilities of
  context management middleware (e.g. context types
  and QoC levels)?
• How to communicate these capabilities to
  services/requesters and collect their context
  requirements?
• How to communicate the capabilities of middleware
  and requirements of context aware services to
  end-users?

26
Kamrans PhD project

• Research questions (contd)
• Privacy policy system for end-users
• Consent is not enough. Users need control over
  disseminated context and QoC.
• Must be easy-to-use and unobtrusive.
• How to control the policy explosion due to
  per-user set of policies?
• How to minimize the performance degradation due
  to enforcement of context-aware privacy policies.
  

27
Conclusion

• presented research is useful complement to
  current IS research projects
• architecture
• requirements
• business process management
• security

28
Additional slides
29
Goal-level service definition
30
choreography-level service definition
31
Orchestration-level service definition
32
Intrinsic context
33
Relational context
34
ECA-DL
Scope (Select (entity.patient., pat, isIncluded
(pat.medConditions, epilepsy)))
Upon EnterTrue(pat.hasSeizure) When
pat.hasSeizure.Accuracy gt 50 Do critical
Foreach (Select (entity.patient.aidPersons, aidP,
aidP.isAvailable
aidP.isNear(pat))) Contact (aidP)always
35
Trust relationships
36
Privacy enforcement in context interaction
37
Actors in privacy agreement