ASCENS at end of Year 1
1 like382 views
The document discusses the ASCENS project, which aims to build autonomic service-component ensembles through modeling languages, analysis techniques, and case studies. It summarizes results from the first year, including modeling languages for specifying ensemble knowledge and behavior, a system model for adaptive ensembles, and verification techniques. It also discusses challenges in ensuring reliability and confidence in large, heterogeneous ensembles operating in uncertain environments. Formal methods are proposed to specify reliability and confidentiality requirements and validate them through compositional verification and runtime monitoring techniques.
ASCENS at end of Year 1
- 1. ASCENS Autonomic Service Component Ensembles Martin Wirsing and Matthias Hölzl LMU Munich, Germany AWARENESS Inter-Project Workshop Bologna, 23-24 January, 2012 Future Emerging Technologies www.ascens-ist.eu
- 2. Ensembles Massive numbers of nodes Extremely heterogeneous Complex interactions between nodes Complex interactions with humans or other systems Operating in open and non-deterministic environments Dynamic adaptation to new requirements technologies and environmental conditions Seite 2 Martin Wirsing 2
- 3. Approach Service Component Ensemble (SCE) Engineering Emergence reliable, predictable, self-adaptive, balances efficient execution and Design and runtime control flexible behaviour via of intended and dynamic self-expression emergent behaviours, static and dynamic Service Component (SC) support from self-aware, adaptive formal methods secure and safe based on knowledge Seite 3 Martin Wirsing
- 4. Approach Service Component Ensemble (SCE) Case Studies Science Clouds Engineering Emergence Swarm Robotics Collaborating E-Vehicles Tool Integr. Platform Engineering SCEL SCE Language (SCEL) Knowledge Correctness Self-Awareness of SCs and SCEs Adaptation & Foundational Models Self-Expression Seite 4 Martin Wirsing
- 5. Aim of ASCENS Build a coherent and integrated suite of models and techniques for constructing Autonomic Service-Component Ensembles Seite 5 Martin Wirsing
- 6. Summary of Results Year 1 Languages SCEL V1.0, BIP, KnowLang for modeling, analyzing and specifying knowledge in ensembles System model GEM for adaptive and self-aware ensembles SOTA approach for specifying and analyzing adaptation requirements Verification techniques for qualitative and quantitative properties of systems in uncertain environments Application requirements and system needs for case studies Robot ensembles, science cloud, collaborative e-vehicles Dissemination Website with blog 50 publications 2 summer schools (co-organized) ASCENS results taught in > 15 courses and tutorials Seite 6 Martin Wirsing 6
- 7. Confidence and Reliability: Difficulties Ensembles are infinite state systems with varying members and environments concurrent systems suffering the state space explosion problem Uncertain environment and changing requirements Heterostatic ensembles (Klopf ~1975) maximize performance instead of simple goal satisfaction Confidence vs. confidentiality Confidence in reliability of a syteme = subjective reliabiity of the system Confidentiality = notion of information security Seite 7 Martin Wirsing 7
- 8. Confidence and Reliability Ideas: A Formal Methods Approach Specify confidentiality and reliability requirements in a (continuous/discrete time stochastic) temporal logic Validate and verify requirements for SC ensembles A posteriori verification Compositional techniques for specific properties (e.g. deadlock) specific architectures (e.g. controllers, from synchronous to asynchronous communication, …) Correctness by supervised construction Correctness preserving architectural principles for system construction and adaptation Runtime monitoring of global/emergent properties Predictive model analysis (at design and runtime) Seite 8 Martin Wirsing 8
- 9. Confidence and Reliability Advice from other projects Trust models for measuring confidence and reliability Measures for confidence of/in (self-) aware components Designing emergence Challenges for measuring confidence Monitoring global and emergent properties Specifying and controlling adaptation Combining formal reasoning techniques with models of trust „Pervasive formal methods“ for enhancing confidence Interaction of global and local confidence Seite 9 Martin Wirsing 9