Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

ICSE 2011: Q-ImPrESS - An Industrial Case Study on Quality Impact Prediction

2,404 views

Published on

Published in: Technology, Business
  • Be the first to comment

  • Be the first to like this

ICSE 2011: Q-ImPrESS - An Industrial Case Study on Quality Impact Prediction

  1. 1. © ABB Group <br />May 26, 2011 | Slide 1<br />Q-ImPrESSAn Industrial Case Study on Quality Impact Prediction for Evolving Service-oriented Software<br />Heiko Koziolek, ABB Corporate Research, GermanyBastian Schlich, Carlos Bilich, Roland Weiss, Steffen Becker, Klaus Krogmann, Mircea Trifu, Raffaela Mirandola, Anne Koziolek<br />
  2. 2. Industrial automation: Process Control Systems<br />© ABB Group <br />May 26, 2011 | Slide 2<br />
  3. 3. MotivationRelease History for an ABB Process Control System<br />© ABB Group <br />May 26, 2011 | Slide 3<br />Version A<br /> First version release with complete system concept<br /> Single environment from independent solutions<br /> Outstanding Operations Offering<br /> Function based Engineering<br /> Redundant Controllers and I/O capabilities<br /> Connectivity for Harmony and Melody<br /> FF, Redundant Profibus, HART, ABB Drives<br />Version C3<br /> Windows 7 support<br />Alarm Analysis and Alarm Shelving<br />WirelessHART Integration<br />Profinet, Ethernet IP, DeviceNet<br />New Controller PM891 (2x PM866)<br />Engineering efficiency improvements<br />Detailed difference reporting<br />Foundation Fieldbus improvements<br />2004 2005 2006 2007 2008 2009 2010<br />Version B<br />Increased system size<br />SIL 2 Integrated Safety<br />Connectivity for DCI and MOD 300 <br />Alarm and Event Improvements<br />Remote Clients via MS Terminal Services<br />Version C1<br /> Increased system size<br /> Multi-system Integration<br />SPI Integration (PETI)<br /> MODBUS TCP<br />Version C2<br />Virtualization support<br />MS WPF Graphics<br />SIL3 Safety<br />IEC 61850 (Intel Elect Devices)<br />New PM866 controller (2x PM864)<br />New S800 I/O (non-red HART)<br />New Power Supplies with smaller footprint<br />Evolution Libraries MOD300 and Infi90<br />Version C<br />Online Upgrade Capability<br />Multi-User / Distributed Engineering<br />Large screen / Multi-screen enhancements<br />Digital Security Improvements<br />
  4. 4. MotivationProblems of software evolution at ABB<br />Continuous evolution of ABB software systems<br />New requirements, technologies, failure reports<br /> Software maintenance and evolution are a large cost factor for ABB software development<br />Current practice<br />Experience to rationalize design decisions<br />Prototyping for new technologies, performance impacts<br />Unknown change impacts on performance/reliability<br />Apply model-based prediction methods for systematic decision support to save costs and achieve higher quality?<br />© ABB Group <br />May 26, 2011 | Slide 4<br />
  5. 5. Q-ImPrESS MethodOverview<br />
  6. 6. Q-ImPrESS Workbench<br />
  7. 7. Case Study<br />© ABB Group <br />May 26, 2011 | Slide 7<br />
  8. 8. Manual ModelingSteps executed to create a Q-ImPrESS model<br />© ABB Group <br />May 26, 2011 | Slide 8<br />~2.5 person months<br />
  9. 9. Manual ModelingQ-ImPrESS model of the ABB process control system<br />© ABB Group <br />May 26, 2011 | Slide 9<br />
  10. 10. Performance PredictionSteps executed to determined resource demands<br />© ABB Group <br />May 26, 2011 | Slide 10<br />~1 person month<br />
  11. 11. Performance PredictionSample predictions for different design alternatives<br />© ABB Group <br />May 26, 2011 | Slide 11<br />
  12. 12. Performance PredictionResults: Measurements vs. Simulation Results<br />© ABB Group <br />May 26, 2011 | Slide 12<br />
  13. 13. Reliability PredictionSteps executed to determine failure probabilities<br />© ABB Group <br />May 26, 2011 | Slide 13<br />~1 person month<br />
  14. 14. Each line shows how the system reliability changes if we change onesubsystem reliability (8 subsystems in total)<br />Reliability PredictionSample sensitivity analysis<br />© ABB Group <br />May 26, 2011 | Slide 14<br />
  15. 15. Reliability PredictionResults<br />© ABB Group <br />May 26, 2011 | Slide 15<br /> More research and tool development needed<br />
  16. 16. ConclusionsSummary<br />Q-ImPrESS <br />provides a structured method and useful tool support<br />is best used for evolutionary changes, not full redesigns<br />still needs to demonstrate costs/benefits<br />© ABB Group <br />May 26, 2011 | Slide 16<br />
  17. 17. ConclusionsFuture Work<br />Future work desired by ABB:<br />More robust reverse engineering tools<br />Model transformations from UML to Q-ImPrESS<br />Tools and best practices for data collection<br />© ABB Group <br />May 26, 2011 | Slide 17<br />
  18. 18. © ABB Group <br />May 26, 2011 | Slide 18<br />

×