Requirements, Design and Data Repositories

779 views

Published on

Romina Torrees, Raise'13,

Published in: Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
779
On SlideShare
0
From Embeds
0
Number of Embeds
5
Actions
Shares
0
Downloads
7
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Requirements, Design and Data Repositories

  1. 1. Addressing the QoS drift inSpecification Models of Self-adaptiveservice-based systemsRomina Torres (1), Nelly Bencomo(2), Hernán Astudillo(1)(1)Universidad Técnica Federico Santa María(2)INRIA Paris-Rocquencourt2nd International Workshop on Realizing Artificial IntelligenceSynergies in Software Engineering (RAISE – Mayo 2012)1
  2. 2. CONTEXTSelf-adaptive Service based Systems2
  3. 3. Introduction• Dynamic adaptive systems must be able toadapt themselves during runtime to cope withthe uncertainty associated with changes intheir goals or in the environment in whichthey operate WITH NO (OR LIMITED) HUMANINTERVENTION.3
  4. 4. Requirements (R):Send the weather prediction for the weekend tothe user wherever he is as fast as possible.Functional goals and non functional constraintsSpecifications (S)M. Mannion, B.Keepence. 1995. SMART requirements. SIGSOFT Softw. Eng. Notes, or GQM4
  5. 5. Specifications (S)Architectural Configuration (C) f(S)=C’During Runtimeproactivereactive 5P. Zaveand M. Jackson, “Fourdarkcornersofrequirementsengineering,” 1997 ACM Txson SE &MethoAdaptation
  6. 6. PROBLEMPerceptual Systems under changing environments6
  7. 7. Challenge7uncertainty
  8. 8. Challenge - uncertaintyt0t1t2t3Perceptual System8
  9. 9. Problem• Cause:– Perceptual systems may be degraded whenenvironments are highly changing• Implication– Specification models (S) may be degraded due tothe highly changing service market• Self-adaptive Service-based systems may missrequirements violations because S is notanymore a valid representation of R.D. Hall, “Automaticparameterregulationofperceptualsystems,”ImageandVisionComputing, vol. 24, no. 8, pp. 870 – 881, 2006.9
  10. 10. The natural solution to the challenge• Humans should regularly and manuallymaintain their specification models in order toensure they are a valid representation of theirrequirements during runtime10
  11. 11. But…• Dynamic adaptive systems must be able toadapt themselves during runtime to cope withthe uncertainty associated with changes intheir goals or in the environment in whichthey operate WITH NO (OR LIMITED) HUMANINTERVENTION.• Humans cannot be aware of a huge and highlydynamic operating environment (partialviews)11
  12. 12. The real required solution• Specification models must be automaticallymaintained during runtime in order tomaintain them as valid representations of thehumans requirements12
  13. 13. RELATED WORKKeeping alive models13
  14. 14. Related WorkI. Epifani, C. Ghezzi, R. Mirandola, G. Tamburrelli. “Modelevolution by run-timeparameteradaptation”. ICSE 09[Firelli et al. 2011] [Cardenilli et al 2011] [Calinescu et al. 2011] [Calinescu et al. 2012]• [Epifani et al. 2009]– Humans estimations are seldom correct + indynamic environments, the value of parametersmay change over time• keeping models alive during runtime and maintainingupdated the parameters by feeding a bayesianestimator with data collected from the running system• Parameters always are average values– Problem: humans cannot constraint properly theirfunctional goals14
  15. 15. PROPROSALComputing with words + strategy to detect drifts on words15
  16. 16. Our proposal• A constraint language for humans to specify the NFCsin the specification models using concepts instead ofnumerical ranges.• Necessarily we also need– An architecture capable to assess “abstract specificationmodels”– An infrastructure capable to obtain from the NFPs of theofferings of the service market the numerical meaning ofthe concepts.• Initial values• Each time there is enough evidence a drift has occurred in theNFPs16
  17. 17. Preliminaries• Computing with words (CWW)• Words instead of numbers for computing and reasoning• Words constrain linguistic variables– Fuzzy setsLoftiZadeh. “FuzzyLogic = ComputingwithWords”. IEEE TxsonFuzzy Systems 1996Language to define NFCs17
  18. 18. Proposal (1/3)• Language18
  19. 19. Preliminaries• Specification models become linguisticdecisions models• How do we assess them?– Perceptual Computer – an architecture for CWWJ. M. Mendel, D. Wu. “ChallengesforPerceptualComputerApplicationsand How TheyWereOvercome”. IEEE ComputationalIntelligence Magazine, 2012.L. Martínez, D. Ruan, F. Herrera, E. Herrera-Viedma, P. P. Wang: Linguistic decision making:Tools and applications. Inf. Sci. 179(14): 2297-2298 (2009) 19
  20. 20. Proposal(2/3)20
  21. 21. Preliminaries• Concept drift occurswhenthecontextshiftsinduce changes in thetargetconcept• Functionally-equivalentservices arecontinuouslycompeting intermsofQoStoachievetheirownsurvivalgoal: tobe selected• Perceptionsaboutwhatmeanseachconceptofeach NFPofeachfunctionalgroupofserviceschange21
  22. 22. PreliminariesG. Leng, X.-J. Zeng, and J. A. Keane, “A hybridlearningalgorithmwith a similarity-basedpruningstrategyforself-adaptiveneuro-fuzzysystems,” AppliedSoftComputing 2009.– randomnoise,– randomtrends (gradualchanges),– randomsubstitutions(abruptchanges),– orsystematictrendsWe record also the trend of thedrift and we base our decisionon the trend history of the past“potential” drifts22Outliers?
  23. 23. Proposal (3/3)23
  24. 24. 24
  25. 25. VALIDATION25
  26. 26. Validation Strategy• Prototype Implementation of Adaptive• Dataset from programmableweb.com– 10 functional categories, 2 non-functional properties,over 150 services (3000 “words”).– 5 SBSs clients subscribing their adaptation needs– 3 QoS drifts synthetically simulated in a window of 10new NFP certifications.• We were able to detect requirements violations whereconfigurations become relative violators because othersfunctionally-equivalent competitors improve their QoS. =>reduction of false negative error26
  27. 27. CONCLUSIONS & FUTURE WORK27
  28. 28. Conclusions• Current systems are being released too fragile– Lack of self-monitoring and self-repairingcapabilities– Lack of awareness– Self-adaptive Systems demands systems“maintainability” be performed automatically– But even manually it is performed poorly• So, still there is a lot of work on the SE side28
  29. 29. Ongoing and future work• Tuning Adaptive• False positives?• Stress testing - Scalability• Sensitivity analysis - Parameters tradeoff• New techniques to mitigate the obsolescence ofmodels during runtime• Release dataset of measurements of NFPs ofservices during a frame of time where severalQoS drifts may occur.29
  30. 30. AcknowledgementsThis work was partially funded by UTFSM DGIP(PIIC and 241167), BASAL FB0821, ConicytChile, the EU Connect project and the EUMarie Curie Requirements@runtime.30

×