Jiri HARAZIM
harazim@mail.muni.cz
Mediation Chain GenerationMediation Chain Generation
in Pervasive Environmentsin Pervasive Environments
Supervisors:
RNDr. Radek Ošlejšek, Ph. D.
Prof. Philippe Lalanda
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 2
AgendaAgenda
Introduction
State of the Art
Problematic Challenges and Related Work
Background: Creating Pervasive Applications
Proposition
Implementation
Validation
Conclusion and Future Work
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 3
AgendaAgenda
Introduction
State of the Art
Problematic Challenges and Related Work
Background: Creating Pervasive Applications
Proposition
Implementation
Validation
Conclusion and Future Work
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 4
Pervasive (ubiquitous) ComputingPervasive (ubiquitous) Computing
„Ubiquitous computing names the third
wave in computing, just now beginning.
First were mainframes, each shared by
lots of people. Now we are in the personal
computing era, person and machine
staring uneasily at each other across the
desktop. Next comes ubiquitous
computing, or the age of calm technology,
when technology recedes into the
background of our lives.“ Mark Weiser (1952-1999)
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 5
Pervasive (ubiquitous) ComputingPervasive (ubiquitous) Computing
Thre are three main categories of devices: small, hand-size and large
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 6
Pervasive EnvironmentPervasive Environment
Freely shared, multi-purpose devices integrated into
everyday objects used by users in different contexts.
Pervasive Environment consists of computing
devices integrated into everyday objects
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 7
Mediation & Mediation ChainMediation & Mediation Chain
Mediation consists of
Data Collecting
Synchronization
Mediation Logic Execution (i.e. data transformations)
Routing and Delivery via Mediation Chain
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 8
MotivationMotivation
When building pervasive applications, it is necessary
to exploit sensing devices in pervasive environment,
but:
Sensing devices have heterogeneous interfaces
Devices often produce different data
(different unit, time interval, format, ...)
Pervasive Environment
?Pervasive Application
As a result, an integration solution is needed
How to exploit
sensing devices?
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 9
AgendaAgenda
Introduction
State of the Art
Problematic Challenges and Related Work
Background: Creating Pervasive Applications
Proposition
Implementation
Validation
Conclusion and Future Work
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 10
State of the ArtState of the Art
New challenges...
Complexity
Dynamism
Heterogenity
… bring new solutions:
Component-Based Software Engineering
Service-Oriented Computing
Mediation
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 11
Service-Oriented ComputingService-Oriented Computing
Schema
Advantages
Loose-coupling: only service contract is shared
Late-binding: on-demand binding
Evolution: a provider can be replaced
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 12
Service-Oriented ComponentsService-Oriented Components
Provider
Component
Provider
Component Consumer
Component
Consumer
Component
Provided service
(specification)
Consumed service (dependency)
Binding
(performed at runtime by framework)
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 13
IntegrationIntegration
Proxy based integration
Advantages
Separation of concerns
Disadvantages
Lack of modularity
Lack of composition
AA ProxyProxy BB AA BB
Platform-based integration
Advantages
Composition of
transformations
Modular
Disadvantages
Complex, hard to introspect
Integration platform
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 14
AgendaAgenda
Introduction
State of the Art
Problematic Challenges and Related Work
Background: Creating Pervasive Applications
Proposition
Implementation
Validation
Conclusion and Future Work
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 15
Integration issues of devices in pervasive environment:
Data Representation (i.e. different units)
Dynamism (devices may appear or disappear anytime)
Configuration (amount of manual configuration)
Timing aspects (data aggregation, sync vs async)
Problematic ChallengesProblematic Challenges
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 16
PervML
Generation of code from several models
Doesn't address heterogeneity
SStream
Considers data from sensors as tuples
Interpretation of data is delegated on the client
EEML
Extensible Environment Markup Language
Flexible scheme for representation of data from sensing devices
Used in Cosm project
DynaMo
Dynamic Multimodal Interface
Integration based on proxies, semantic and type of data
Related WorkRelated Work
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 17
AgendaAgenda
Introduction
State of the Art
Problematic Challenges and Related Work
Background: Creating Pervasive Applications
Proposition
Implementation
Validation
Conclusion and Future Work
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 18
Rondo FrameworkRondo Framework
Global scheme: Model-Driven approach of Rondo
Framework for defining and dynamically assembling
service-oriented components
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 19
Background: Creating PervasiveBackground: Creating Pervasive
ApplicationsApplications
Aspects of pervasive applications addressed by Rondo:
Discoverability
Context (i.e. User's location)
Dynamic Adaptability
Composability
Rondo is essentially functional today
Components need to express high-level dependencies to
the data produced by devices.
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 20
Goal of this workGoal of this work
Propose smarter integration of sensing devices in the
Rondo approach by generating mediation flow from
high-level description that:
Hides the complexity of heterogeneous devices
Provides unified access to underlying devices
Provides unified data
Provides additional capabilities (transformations)
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 21
AgendaAgenda
Introduction
State of the Art
Problematic Challenges and Related Work
Background: Creating Pervasive Applications
Proposition
Implementation
Validation
Conclusion and Future Work
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 22
ProposalProposal
Incorporate concept of Semantic Tag
Human-understandable semantic expression associated with
various devices: humidity, presence, temperature, illuminance etc.
Extend the Rondo language by mediation concept
Uses expressive filter (RFC#1960) applicable on data
Contract of Mediator: tries to translate data before filtering (i.e.
(location=livingroom) but device may produce location=living-room
or room-04 although they denote the same location)
Use Cilia as underlying component model in the
generation phase
Enables dynamism, modularity, follows EAI patterns
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 23
Environment Interaction FrameworkEnvironment Interaction Framework
Rondo
1. Data described by
description language
2. Generation of mediation chain
5. Data delivery
Available devices
4. Mediation
Mediation Chain
Environment Interaction Framework
3. Data retrieval
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 24
Description LanguageDescription Language
Consists of tag & filter
Tag denotes the semantic meaning of data that is
collected in environment.
Emphasis on semantic aspect instead of form of data
For example: ”temperature”, ”humidity”, ”illuminance”,
”sensed-presence-of-human” etc.
Filter influences generation of mediation chain
Incorporates mediators to enable transformation
By chaining mediators it is possible to obtain quite
complex behaviour
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 25
AgendaAgenda
Introduction
State of the Art
Problematic Challenges and Related Work
Background: Creating Pervasive Applications
Proposition
Implementation
Validation
Conclusion and Future Work
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 26
ImplementationImplementation
EIF
Available as a service with simple and intuitive API
3 OSGi bundles (API, Controller and Cilia-Components)
20 Cilia components for additional transformations
Independent from underlying technologies
Used technologies
Cilia Mediation Framework
Apache iPOJO
Apache Maven
JScience library
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 27
AgendaAgenda
Introduction
State of the Art
Problematic Challenges and Related Work
Background: Creating Pervasive Applications
Proposition
Implementation
Validation
Conclusion and Future Work
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 28
ValidationValidation
Validation
Scenario based on real use-case brought by ADELE
tested on iCASA: Digital House Simulator
Scenario
Human's presence in the house triggers actions
(switches on/off light, air-condition etc.) based on certain
conditions that are retrieved from sensing devices via EIF
(such as time, illuminance and temperature)
Out-adapterTrigger
Starts mediation as
soon as data from all
adapters are available
In-adapter
Unit
converter
Average
Aggreg.
Delta
filter
Converts Fahrenheit
and Kelvin to Celsius
Calculates arithmetic
mean (average)
Filters out data if difference
comparing to last propagated
value is lower than 0.2
In-adapter
ºF
ºK
ºC
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 29
iCASA House SimulatoriCASA House Simulator
Results
 Passed succesfully
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 30
ValidationValidation
Validation in practice
Verified against validation scenario
Tested with various combinations of mediators
Exploited advanced features such as unit conversions
Full integration with all sensing devices in iCASA such as
thermometers, presence sensors, noise sensors, illuminance
sensors and power-consumption sensors
iCASA: Digital House Simulator
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 31
Conclusion and Future WorkConclusion and Future Work
Conclusion
Generates mediation chain and provides additional
transformations of sensed data
Simplifies the development of pervasive applications (the
simplest mediation chain can be created and started in
only 2 lines of code)
EIF will be integrated in Rondo Framework
Future Work
Integration into Domain-Specific Language in Rondo
Extend description language and enable hierarchical data
Two-way interaction with environment
More components (triggers & mediators)
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 32
Hansmann, Uwe. Pervasive Computing: The Mobile World.
Ebling R. Maria, Baker Mary. Pervasive Tabs, Pads, and Boards: Are We There Yet?
Krakowiak Sacha. Middleware Architecture with Patterns and Frameworks.
Crnkovic Ivica, Vulgarakis Aneta, Chaudron R. V. Michel. A Classification Framework for Software
Component Models.
Gugen Levent, Labbe Cyril, Roncancio Claudia, Olive Vincent. Sstream: A Model for Representing Sensor
Data and Sensor Queries.
Haque Design + Research Ltd.. Extensible Environments Markup Language: EEML.
Avouac Pierre-Alain, Lalanda Philippe, Nigay Laurence. Service-Oriented Autonomic Multimodal
Interaction in a Pervasive Environment.
Hall S. Richard, Pauls Karl, McCulloch Stuart, Savage David. OSGi in action. Creating modular applications
in Java.
Isbell Douglas, Hardin Mary, Underwood Joan. MARS CLIMATE ORBITER TEAM FINDS LIKELY CAUSE OF
LOSS.
Cervantes Humberto, Donsez Didier, Touseau Lionel. An Architecture Description Language for Dynamic
Sensor-Based Applications.
Solis Ignacio, Obraczka Katia. The Impact of Timing in Data Aggregation for Sensor Networks.
ReferencesReferences
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 33
Questions?Questions?
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 34
The EndThe End
Thank you for your attention!
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 35
Implemented TransformationsImplemented Transformations
Implemented Cilia components for mediation can be
combined and chained to obtain desired behaviour.
3 triggers
10 processors (aggregators, statistical ops., converter ..):
Average Aggregator, Delta filter, Forward processor, Max
Aggregator, Min Aggregator, Min filter, Max filter,
OlderThanFilter, Sum Aggregator filter, Unit converter
6 + 1 adapters
22. 11. 2012 Jiri HARAZIM: Mediation Chain Generation in Pervasive Environments 36
SourcesSources
Pics:
http://www.google.cz/search?um=1&hl=cs&client=opera&rls=cs
http://www.google.cz/search?
um=1&hl=cs&client=opera&rls=cs&channel=suggest&biw=
1440&bih=729&tbm=isch&sa=1&q=smartphone&oq=smart
phone&gs_l=img.3..0l10.23774.25655.0.26360.10.6.0.4.4.
0.130.479.5j1.6.0...0.0...1c.1.AeK0LEiyUoU
http://en.wikipedia.org/wiki/Mark_Weiser