massif the love child of
play

Massif - the love child of Matlab Simulink and Eclipse kos Horvth , - PowerPoint PPT Presentation

Massif - the love child of Matlab Simulink and Eclipse kos Horvth , Istvn Rth and Rodrigo Rizzi Starr (ahorvath@mit.bme.hu) Budapest University of Technology and Economics Embraer S.A. IncQuery Labs Ltd. Eclipsecon 2015 Department of


  1. Massif - the love child of Matlab Simulink and Eclipse Ákos Horváth , István Ráth and Rodrigo Rizzi Starr (ahorvath@mit.bme.hu) Budapest University of Technology and Economics Embraer S.A. IncQuery Labs Ltd. Eclipsecon 2015 Department of Measurement and Information Systems

  2. Outline of the talk  Main Contributors Motivation and background o Ábel Hegedüs Integration with Matlab Simulink o Rodrigo Rizzi Starr Massif o Márton Búr • Overview o Lincoln • Architecture Nascimento Feature set o Róbert Dóczi • Import/export o Samoel Mirachi • OSLC interface o István Ráth o Ákos Horváth Conclusion • Current status • Future work Eclipsecon 2015 2

  3. Background: Matlab Simulink  Matlab : multi-paradigm numerical computation o matrix operations, algorithm development, data visualization etc.  Simulink : dataflow modeling environment o Modeling and simulating dynamic systems o Block diagram based  Large user base > 1M o Automotive o Avionics o Railway o Etc. Eclipsecon 2015 3

  4. Motivation: yet another tool integration  Started in 2011 (as a second iteration) o Provide HW-SW allocation for early simulation of avionics systems o Input: Low-level Simulink libraries o Output: Integrated HW-SW Simulink model  Allocation defined on high abstraction level  Eclipse o Hierarchical graph based visualizations  yFiles + EMF-IncQuery Viewers o On-the-fly model validation  EMF-IncQuery o Generic bridge between Matlab Simulink and Eclipse  Massif Eclipsecon 2015 4

  5. Massif Integrating Eclipse and Matlab Simulink Eclipsecon 2015

  6. Massif – Eclipse-Matlab Simulink bridge  Generic Bridge between Matlab Simulink and Eclipse o Import/Export Matlab Simulink systems and libraries o Traceability for incremental processing o Multiple import options for different usage scenarios • e.g., analysis or allocation  User-friendly EMF representation o Manipulating systems/libraries directly on EMF level o Support for easier navigability  Works with Matlab license server o Can connect to a running Matlab instance  Pure Java realization, EPL license o Project specific constraint! Eclipsecon 2015 6

  7. Massif- architecture overview  Communicating with Matlab Matlab Simulink Commads/ results o Java RMI based Matlab communication Matlab Command System o Server initialized within Evaluator Server Matlab • Java Matlab Interface Commads/ results traceability o Works with license server  Generic EMF metamodel Matlab Comand Metamodel Processor o EMF level operations [Ecore] translated into instance High level Matlab commands operations Matlab o Same library mechanism as System in Matlab Simulink [EMF] Eclipse o FQN based traceability Eclipsecon 2015 7

  8. Feature set What Massif can do Eclipsecon 2015

  9. Import/export models DEMO  Tool configuration and initialization  Import and export of Matlab Simulink systems o Import strategies  Adding/deleting an element to/from the EMF model o Incremental refreshing Eclipsecon 2015 9

  10. Shallow import option  Only blocks within non-referred systems are imported hierarchically System A Library A System A Block Block 1 Block 1 Lib1 Model Model Block Block Reference Reference 2 2 block block Block Block 4 3 System B Matlab Simulink – system Eclipse - EMF Eclipsecon 2015 10

  11. Deep import option  Each block inside each subsystem is imported. Each referenced model is imported as an individual model with direct model referencing in the parent model System A Library A System A Block Block 1 Block 1 Lib1 Model Model Block Block Reference Reference 2 2 block block System B Block Block Block Block 4 3 4 3 System B Matlab Simulink – system Eclipse - EMF Eclipsecon 2015 12

  12. Flattening import option  Each model reference block is imported as though it was a subsystem System A Library A System A Block Block 1 Block 1 Lib1 Model Model Block Block Reference Reference 2 2 block block System B Block Block Block Block 4 3 4 3 System B Matlab Simulink – system Eclipse - EMF Eclipsecon 2015 13

  13. Flattening import option  Each model reference block is imported as though it was a subsystem System A Library A System A Block Block 1 Block 1 Lib1 Model Block Block Reference 2 2 block Block Block Block Block 4 3 4 3 System B Matlab Simulink – system Eclipse - EMF Eclipsecon 2015 14

  14. Referencing import option  For blocks with active library links, each source library is imported once as an individual model (but may be referenced multiple times) System A Library A System A Library A Block Block 1 Block 1 Block Lib1 Lib1 Model Model Block Block Reference Reference 2 2 block block System B Block Block Block Block 4 3 4 3 System B Matlab Simulink – system Eclipse - EMF Eclipsecon 2015 15

  15. Additional Features  Import Filters o For leaving out unwanted elements/details from imported models  Bus creator and selector configuration o Configured via string (fqn like) in Matlab Simulink o Use EMF EReferences to define and Massif automatically calculates configuration  Easier Block-to-Block navigability o Derived EReferences defined by EMF-IncQuery Eclipsecon 2015 16

  16. OSLC API DEMO  Open Services for Matlab Simulink Queries /results Lifecycle Collaboration Matlab (OSLC) Matlab Command System o API for tool integration Evaluator Server Queries/  Live OSLC API for Matlab results Simulink Comand o Dynamically executes Eclipse Lyo Evaluator queries directly on the Matlab side Queries/ results o Uses the same Command Matlab Evaluator Server System o Based on Eclipse Lyo resource Apache Tomcat Eclipsecon 2015 17

  17. Conclusions What we have done and where to move forward Eclipsecon 2015

  18. Conclusions  Initial version is out o Basic import/export features supported o EMF level handling of Simulink models  Experience o Avionics • Largest model took ~2 hours  o Automotive • 5k elements within minutes  o Typical application scenarios • Offline processing of EMF representations • Library import  instance model generation on EMF side Eclipsecon 2015 19

  19. Final points  The examples and more details are available form o https://github.com/FTSRG/massif  Version 0.4 is available o http://incquery.net/update/massif/release/site/ o Contributors: • Main: BME-FTSRG, Embraer S.A. • Auxilliary: IncQuery Labs Ltd. • Supporting projects: Trans-IMA, Concerto (EU-Artemis)  Your contributions (feedback, forum posts, ideas, bugzillas, patches) are very welcome! o To what direction should we enhance Massif? Eclipsecon 2015 20

  20. Evaluate the sessions Sign in: www.eclipsecon.org +1 0 -1 Eclipsecon 2015

Recommend


More recommend