Modeling Context and Dynamic Adaptations with Feature Models
Click here to load reader
description
Transcript of Modeling Context and Dynamic Adaptations with Feature Models
Modeling Context and Dynamic Adaptations
with Feature ModelsMathieu Acher
1, Philippe Collet
1 , Franck Fleurey
2 , Philippe Lahire
1 , Sabine Moisan
3 , and Jean-Paul Rigault
3
1 University of Nice Sophia Antipolis, CNRS, France
{acher,collet,lahire}@i3s.unice.fr
2 SINTEF, Oslo, Norway
3 INRIA Sophia Antipolis Mediterranée, France,
{moisan,jpr}@sophia.inria.fr
Results
Video Surveillance Case Study
Future Work
Modeling Context
Modeling Software Variants
o Dynamic Adaptive Systems (DAS) are software systems
which have to dynamically adapt their behavior in order
to cope with a changing environment.
SPL of
Segmentation
Segmentation ClassificationFrame to Frame
Analysis
Task
DependentAcquisition
Variants
Base
o Consider DAS as a Software Product Line (SPL)
From common assets, different programs of a domain can be assembled
o Model also the context as an SPL
o Issues:
o Large number of software configurations
o Large number of contexts
TraversalAlgorithm
KernelFunction
Edge
Region
GridStep
WithMask
WithWindow
Classification
Contour
Density
Segmentation
VSSystem
ShadowElimination
LightingAnalysis
HeadLightDetection
DetectRapidChanges
LightingConditions
TimeOfDay
Night
Day
NaturalLight
ArtificialLight
Indoors
Outdoors
LightingNoise
Shadows
HeadLight
Flashes
Scene
VSContext
TraversalAlgorithm
KernelFunction
Edge
Region
GridStep
WithMask
WithWindow
Classification
Contour
Density
Segmentation
VSSystem
ShadowElimination
LightingAnalysis
HeadLightDetection
DetectRapidChanges
LightingConditions
TimeOfDay
Night
Day
NaturalLight
ArtificialLight
Indoors
Outdoors
LightingNoise
Shadows
HeadLight
Flashes
Scene
VSContext
o Leverage the expressiveness of FMs (e.g. attributes).
o Achieve an automatic translation between DSML and FMs.
o Update automatically contextual information.
o Connect state-of-the-art adaption engines to our models.
o The concept of configuration is naturally present and defined by the
semantics of FM.
o Uniform representation of the context model and the software system
makes possible to express relations between the two models.
o DSML and FM-based approaches can complement each other.
GridStep or WithWindow excludes Edge (C1)
GridStep excludes Ellipse (C2)
Edge excludes Density (C3)
initial context initial system
new context SPL after reconfiguration
Modeling Adaptation
Night and HeadLight implies HeadLightDetection (AR0)
not LightingNoise implies Region (AR1)
LightingNoise implies Edge (AR2)
ArtificialLight implies DetectRapidChanges (AR3)
Flashes or HeadLight implies Contour (AR4)
Revisiting the Approach with Feature Models
Problem Statement DSML Approach
o Variants
o Constraints
o Context
o Rules
3
1
2
1 Initial deployment: configuration of the system from the context
Dynamic update of the context happens
Dynamic reconfiguration of the system from the updated context
2
3
SPL of
Classification
SPL of Frame to
Frame Analysis
SPL of Task
Dependent
VariabilityModel
-context0..*
Rule
-rule0..* -priority : Integer
PropertyPriority
-priority 0..* ContextConstraint-context
1..1
-value : Integer
PropertyValue
-upper : Integer
-lower : Integer
DimensionVariant
VariantConstraint
-property0..*
-direction : Integer
Property-property
1..1
Variable
BooleanVariable
EnumVariable
-property
0..*
-available
-required
0..1
0..1
0..*-propertyValue
0..*
1..1
-type
-variant-dimension
0..*
0..1
-dependency
-property
1..1
VSSystem
LightingAnalyses
Acquisition
TraversalAlgorithm
GridStepWithWindow
KernelFunction
Color EdgeGrey
Classification
Model
Ellipse OmegaParallelepiped
Density HeadlightDetectContour
Segmentation
threshold: integer
Region
And-Group
Optional
Mandatory
Xor-Group
Or-Group
VSContext
ObjectOfInterestScene
LightingConditions
NaturalLight
OutdoorsArtificialLight Indoors
Camera
Resolution DepthOfField
TimeOfDay{night, day}: enum
LightingNoise{flashes,headlight,shadows}: enum
Sort
PersonVehicle