ti itiming m d llim odelling with autosaradt.cs.upb.de/timmo-2-use/pdf/date2012_1.pdf ·...

Ti i M d lli ith AUTOSARTiming Modelling with AUTOSARCurrent State and Future Directions

Stefan Kuntz

Continental Automotive GmbH

Marie Agnes Peraldi Frati INRIA CNRS FranceMarie‐Agnes Peraldi‐Frati, INRIA, CNRS, France

Hans Blom and Daniel Karlsson, Volvo Group Trucks Technology

Co‐authors

• Marie‐Agnès Peraldi‐Frati, INRIA, Nice Sophia‐Antipolis, France

• Hans Blom, Volvo Group Trucks Technology, p gyGothenburg, Sweden

• Daniel Karlsson, Volvo Group Trucks Technology,Daniel Karlsson, Volvo Group Trucks Technology, Gothenburg, Sweden

14‐Mar‐12 1Stefan Kuntz / Continental Automotive GmbH

Overview

• Introduction

• Current State

• Future DirectionsFuture Directions

• Conclusion

Q ti d A• Questions and Answers


Introduction

• The reflections presented have been primarily conducted in the ITEA TIMMO‐2‐USE project where the scope are different levels of abstraction.

• Purpose of Timing Modelling:

– Support the construction of systems that satisfySupport the construction of systems that satisfy the given timing requirements

Timing analyses of those systems for example– Timing analyses of those systems, for example scheduling analysis


Current State ‐ AUTOSAR

• Automotive Open System Architecture (AUTOSAR)

• Scope is implementation level with different views

• Primary focus is software and describes hardwarePrimary focus is software and describes hardware from the software point of view

• Component based architecture• Component based architecture

– Basic Elements: Component, Port and Interfaces, C tConnectors

– Application Software Layer

– Software Infrastructure Layer

– Basic Software LayerBasic Software Layer


Current State ‐ AUTOSARCSWC_001

SWC_003RE4

RE1

RE2CSWC_002

SWC 001 SWC 002RE3

SWC_001RE1

SWC_002RE2

RE1

RE2

RE3

CSWC Composition Software Component RE Runnable Entity


CSWC Composition Software Component RE Runnable EntitySWC Software Component

Current State ‐ AUTOSAR Architecture

C AUTOSAR D l P hi


Courtesy AUTOSAR Development Partnership

Current State ‐ AUTOSAR TIMEX

AUTOSAR Timing Extensions (TIMEX):

• Event: Specifies an event and refers to a location in a system where occurrences of such event are yobserved

• Event Chain: Specifies a causal relationship betweenEvent Chain: Specifies a causal relationship between events and their temporal occurrences (stimulus and response)and response)

• Event Triggering Constraints are imposed on events

L t d S h i ti C t i t• Latency and Synchronization Constraints are imposed on event chains


Current State ‐ AUTOSAR TIMEXCSWC_001

SWC_003LC 1RE4

RE1

Event Chain 1Stimulus Response

RE2CSWC_002

SWC 001 SWC 002RE3

SWC_001RE1

SWC_002RE2

RE1

Event 1

RE2

RE3

ETC 1Event 2

CSWC Composition Software Component RE Runnable Entity LC Latency Constraint

ETC 2


CSWC Composition Software Component RE Runnable Entity LC Latency ConstraintSWC Software Component ETC Event Triggering Constraint

Current State ‐ Case Study

Third presentation of this Hot Topic Session: Timing Modelling and Analysis in the Automotive Development Process ‐ An Industrial Case Study

14‐Mar‐12 Stefan Kuntz / Continental Automotive GmbH 9

Current State ‐ Deficiencies

Although AUTOSAR TIMEX already provides a rich set of elements to model timing, there are some deficiencies:

• Traceability

• Probabilistic TimingProbabilistic Timing

• Higher Levels of Abstraction (EAST‐ADL)

M lti f Ti i• Multi‐form Timing

• Symbolic Timing Expressions


Future Directions ‐ Traceability

Seamless traceability between timing information on different levels of abstraction and AUTOSAR views

• Supports reasoning about timing requirements and pp g g qconstraints while constructing systems

• Improves the validation of higher level timingImproves the validation of higher level timing requirements while considering specific properties on the implementation levelon the implementation level


Future Directions ‐ Probabilistic Timing

Functionalities may tolerate for example infrequent deadline misses, and infrequent data loss, etc.

• Capability to specify probabilistic timing constraints p y p y p gand properties for events and event chains

• Existing methods and tools for timing analyses mustExisting methods and tools for timing analyses must be adapted accordingly

Second presentation of this Hot Topic Session: Ch ll d N T d i P b bili ti Ti iChallenges and New Trends in Probabilistic Timing Analysis


Future Directions ‐ EAST‐ADL

The scope of AUTOSAR is implementation: software and hardware architectures

• Decision regarding timing are taken on higher levels g g g gof abstraction respectively phases in the development processp p

• Tracing between such higher levels of abstraction and the implementation level are crucial for aand the implementation level are crucial for a consistent and seamless timing modelling approach

• EAST‐ADL already provides the framework• EAST‐ADL already provides the framework supporting such an approach


Future Directions ‐Multi‐form Time

Timing modelling must provide means to express time in any unit – multi‐form of time – and not only constant values – expressions

• Not only chronometric [s], but also in any physical measure, like temperature, distance in meters, etc. pIn engine management systems time is often specified in degrees (crank‐ and camshaft)

• Time values must be expressed by formulas

• Relation between different multi‐form time bases• Relation between different multi‐form time bases


Conclusion

• AUTOSAR gained momentum in the automotive industry

• Subsequent releases of the AUTOSAR Timing q gExtensions may follow the presented direction

• Tool Support is key to timing modelling andTool Support is key to timing modelling and analysis: Fourth presentation of this Hot Topic SessionSession


Questions and Answers

Thank you very much for your attention!


ti itiming m d llim odelling with autosaradt.cs.upb.de/timmo-2-use/pdf/date2012_1.pdf ·...

Documents