COCOA : ConversationBased Service Composition for Pervasive Computing Environments Sonia Ben Mokhtar

1
COCOA Conversation-Based Service Composition
for Pervasive Computing EnvironmentsSonia Ben
Mokhtar, Nikolaos Georgantas and Valérie
IssarnyICPS'06 IEEE International Conference
on Pervasive Services 200626-29 June 2006, Lyon,
France
2
Outline

• Background
• COCOA COnversation based service COmposition
  for pervAsive computing environments
• Prototype Implementation and Evaluation
• Conclusion and Future Work

3
Pervasive Computing Environments and SOA
e-learning
Multimedia streaming
Heterogeneity
e-shopping
Travel reservation
4
Example Scenario
5
Pervasive Computing Environments

• Mobile nodes host services providing various
  functionalities
• Mobile users need to perform tasks anytime
  anywhere, using functionalities of the pervasive
  environment
• Challenges
• Allow users to perform tasks, by composing on
  the fly available networked services
• Enable autonomy and adaptation to the
  environments changes

6
Issues Requirements

• Existing approaches to ad hoc service
  composition in pervasive computing environments
• Commonly assume that services have been
  pre-developed to integrate
• Poorly deal with the environments dynamics and
  heterogeneity
• Requirements
• Deal with the environments heterogeneity
• Semantic Web services
• Deal with the environments dynamics
• Automating service discovery and composition
• User tasks realized through dynamic composition
  of networked services

7
Semantic Web services

• Web Services
• A software component developed using any
  language, deployed on any platform, described
  using WSDL, accessible via remote calls using
  SOAP on top of internet protocols
• Web Services in pervasive computing environments
  (e.g. WSAmI ISTS04)
• Semantic Web Services
• Semantically annotating Web services descriptions
  using Ontologies
• Semantic annotation on top of WSDL, e.g., WSDL-S
• Web Service Ontologies, e.g., OWL-S, WSMO,FLOWS,
  ROWS

8
OWL-S based Model for Services and Tasks

• Service
• Workflow of high level provided capabilities
• A set of non-functional properties
• Task
• Workflow of high level required capabilities
  (abstract capabilities)
• A set of non-functional requirements
• Matching
• High level capabilities (Service discovery and
  selection)
• Workflows (Service integration)

Networked Services
User Task
Browse
Browse
Browse
Required Capability
Provided Capabilities
9
Service Discovery Composition-COCOA
? Service Discovery
?
?
?
?
?
?
?
?
Service Composition
Abstract user task conversation
10
COCOA-SD Semantic Matching of Service
Capabilities

• Service Advertised capability (Adv)
• required inputs (Adv.In)
• provided outputs (Adv.Out)
• provided properties (Adv.P)

• Service Requested capability (Req)
• provided inputs (Req.In),
• required outputs (Req.Out)
• required properties (Req.P)

Client
Service
Ontologies
11
COCOA-CI Modelling conversations as finite state
automata

• A model to map OWL-S conversations to finite
  state automata
• A mapping rule for each OWL-S control construct
  (e.g. Choice, Sequence, While, Split)
• Translate the problem of dynamic conversation
  integration to an automata analysis problem

12
COCOA-CI Conversation Integration-Example
3
1
1
2
3
4
5
5
4
2
13
COCOA-CI Binding to Concrete Services

• Each composed capability corresponds to either
  an atomic or a composite process of a networked
  service

14
Properties of COCOA-CI

• Workflow integration
• Interleaving
• The number of services to be composed is dynamic
• The degree of involvement of each composed
  service is dynamic

COCOA-CI
Networked services
1
2
3
4
Service 2
Service 1
1
3
2
4
1
2
3
4
COCOA-CI
Networked services
Service 1
Service 2
COCOA-CI
COCOA-CI
Networked services-Env 1
Networked services-Env 2
15
COCOA in action
COCOA-SD
COCOA-SD
COCOA-CI
16
Prototype Implementation and Evaluation

• Implementation of COCOA-CI
• Runtime environment
• Java, Linux platform running on an Intel Pentium
  4, 2.80 GHz CPU with 512 MB of memory
• Services from 1 to 100 of semantically
  equivalent capabilities
• Evaluation of the runtime overhead produced by
  integrating conversations
• Comparison with the XML parsing time of services
  and tasks (mandatory)

17
Prototype Implementation and Evaluation

• Task composed of a single required capability
• The runtime overhead of COCOA-CI is negligible
  compared to the XML parsing time of services and
  task descriptions

18
Prototype Implementation and Evaluation (2)

• Task from 1 to 20 of semantically equivalent
  capabilities in sequence
• The runtime overhead of COCOA-CI depends on the
  number intermediate paths assessed during the
  composition
• Here for a task composed of i capabilities
  there are 100 i possible composition schemes

19
Prototype Implementation and Evaluation (3)

• Realistic scenario
• 20 required capabilities
• 30 available capabilities
• Various control constructs
• 32 ms for finding the 2 possible compositions
  among 36 intermediate compositions
• 152 ms for XML parsing of service and task
  descriptions
• In most realistic cases the runtime overhead of
  COCOA-CI is negligible compared to XML parsing
  time

20
Conclusion Future work

• COCOA Conversation-based Service Composition in
  Pervasive Computing Environments
• A flexible approach for dynamically composing
  services to perform a user task
• Built upon Web services and Semantic Web
  paradigms
• Based on conversation integration
• Ongoing work
• Advanced prototype implementation (including QoS-
  and Context-awareness) underway as a part of the
  European IST AMIGO project
• Integration of COCOA into a semi-distributed
  discovery protocol adapted for pervasive
  computing environments Sailhan,Percom05
• Efficient matching of semantic Web service
  capabilities
• Reconfiguration of the composed user task at
  runtime

21
Questions ?
Allez Les Bleus !