Improvements of Industrial RealTime Embedded Systems Development Process ShuaiLi MichelBourdelles ArnaudTrottet frthalesgroupcom THALES Communications amp ID: 372400
Download Presentation The PPT/PDF document "PRESTO:" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
PRESTO: Improvements of Industrial Real-Time Embedded Systems Development Process
{Shuai.Li}{Michel.Bourdelles}{Arnaud.Trottet}@fr.thalesgroup.comTHALES Communications & Security
3rd FITTEST Industrial Day May 31st 2013Slide2
Project Context: Real-Time Embedded Systems3rd FITTEST Industrial Day May 31st 2013Slide3
Project GoalCurrent Design DifficultiesRTES constrained by
resources available (e.g. processor, power)……but it
is difficult to proceed to performance
analysis
at
an
early
stage in design
when
the execution platform is not available.PRESTO General IdeaProvide tools to evaluate Software/Hardware allocations…by analyzing functional and non-functional properties……and integrating tools into an industrial development process.
3rd FITTEST
Industrial
Day
May 31st 2013Slide4
Project Synopsis3rd FITTEST Industrial Day May 31st 2013
Software Requirements(Functional and non-functional
)Platform
Requirements
Software Test
Design and
Implementation
o
n Reference Platform
Software
Model
Platform
Model
Modeling
Functional
Verification
Non-
Functional
Verification
Extrapolation
Estimation
Analysis
Test Trace
Generation
Behavior
Capture
Software Test on Real Platform
?
Platform PrototypeSlide5
Project Technical RelevanceBehavior CaptureInstrumentation strategies
Instrumentation impactTrace generationTrace interoperability
ModelingGeneral RTES modeling with stereotypes
(
e.g
. UML)
Domain
specific
modeling
(e.g. application, analysis specific)Formal modeling of functional and non-functional propertiesModel interoperabilityAnalysisExecution and specification trace comparison3rd FITTEST Industrial Day May 31st 2013Slide6
Consortium3rd FITTEST Industrial Day May 31st 2013
United Kingdom
FranceSlide7
Consortium3rd FITTEST Industrial Day May 31st 2013
Finland
Italy
GreeceSlide8
THALES Case-Study Demo{Shuai.Li}{
Michel.Bourdelles}{Arnaud.Trottet}@fr.thalesgroup.comTHALES Communications & Security3rd FITTEST Industrial
Day May 31st 2013Slide9
THALES Software Radio Case-StudyFunctional OverviewSoftware radio applicationTDMA radio protocol for
mobile ad-hoc wireless networksSoftware Architecture
Module
Module
Module
Thread
:
Activated
when
message in buffer
Buffer
: FIFO message
storage
Communication
3rd FITTEST
Industrial
Day
May 31st 2013Slide10
THALES Software Radio Case-StudyFunctional OverviewSoftware radio applicationTDMA radio protocol for
mobile ad-hoc wireless networksSoftware Architecture
Business code:
message
processing
Business
Wrapper
Business
Wrapper
Business
Wrapper
Business
Wrapper
code:
message
reception
and transmission
Wrapper
Instrumentation
3rd FITTEST
Industrial
Day
May 31st 2013Slide11
ExperimentObjectivesSoftware radio application modelingInstrumented code generation from model
Execution trace generationTrace visualisation with PRESTO partner toolInstrumentation overhead evaluation
Experimented tools and methodsUML (MARTE) modeling languageSofteam’s Modelio modeler for PRESTO
PragmaDev’s
MSC-Tracer analysis tool
Demonstration
Time
3rd FITTEST
Industrial
Day
May 31st 2013Slide12
ResultsInstrumentation Overhead
DiscussionThe instrumented execution does not result in any missed deadlines when they are equal to 100ms.In other software radios, deadlines are on the scale of 1ms. Theoretically a maximum of 30 points are then allowed. In the experimental execution, deadlines would have been missed.Instrumentation overhead can be limited by choosing active instrumentation points at model level.
Number of message types
21
Instrumentation points hit during
execution
38
Single instrumentation
point overhead (µs)
Min
MaxAvg31.0040.0033.09Ratio (single overhead) / (total time)
0.03%
Ratio (total
o
verhead) / (total
time)
1.07%
3rd FITTEST
Industrial
Day
May 31st 2013Slide13
ConclusionResultsTrace generation flow
Functional informationTiming informationInstrumentation strategy and impactSolution
to ease modelingFuture WorksExploit traces
with
formal
properties
in the model
Energy
consumption information3rd FITTEST Industrial Day May 31st 2013