Abstract
Today’s societal challenges, such as sustainable urban living and public safety and security require monitoring and control solutions for large-scale complex and dynamical systems. The distinguishing features of these systems are serious resource constraints, demanding non-functional requirements such as robustness, timeliness, lifetime and the capability of handling system evolution through runtime reconfiguration. In this chapter, a multi-aspect modeling language is introduced that allows system designers to model the architecture of large scale networked systems from different aspects. This modeling language introduces innovative concepts to model runtime reconfiguration at design-time. The proposed architecture for modeling runtime reconfiguration consists of primary tasks in one layer and secondary management tasks in another layer. Special reconfiguration primitives allow the description of four types of reconfiguration: re-parameterisation, re-instantiation, rewiring and relocation. The modeling language is accompanied by a modeling and design methodology (inspired by the MAPE-K technique [1]) and uses feedback loops in the system model to realize runtime reconfiguration. This chapter also proposes Key Performance Indicators (KPIs) that allow designers to quantify the “quality” of the system designs and pick the most promising one. Special attention is paid to the fact that the availability of a runtime reconfiguration (i.e. re-design capability) in a system requires KPIs to be derived and evaluated at runtime as a precondition for guiding the reconfiguration process.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
For large and complex systems, the model can now also be used as a specification for the independent realization of system components in parallel by different teams and possibly at different locations. After the realization of the components is complete, the model can be used to verify their construction and properties and subsequently, the components can be integrated into the final system.
- 2.
Obviously, the everything, which is beyond is not the whole world. Only those elements (incl. humans, eventually) are to be considered, which are connected to/influenced by the system to be designed. Usually these relevant elements are identified in the use-case models. Use case models are not considered here, they are assumed to be well-defined and stable.
- 3.
Deriving KPIs for runtime reconfigurable systems requires evaluation tools allowing model changes during the evaluation cycle. This is merely a tool implementation issue and will be detailed in Chap. 3. Note that the KPI calculation processes should be part of the implemented system itself to guide the runtime reconfiguration.
- 4.
The underlying models for determining the instantaneous supply are typically very complex and the construction of these models go beyond the competence of the system designer (e.g. deriving channel models for wireless communication). Consequently the system designers work should be supported with parameterizable model libraries. In this case the designer just has to identify the matching model classes and has to set the parameters according to the scenario to be investigated. Many times determining the instantaneous supply is a computationally demanding process. The system designer has to find the balance between the fidelity and complexity.
- 5.
In reality computing nodes run schedulers to control access to the processor and other physical resources. The scheduler is typically a part of the runtime environment (operating system) managing the nodes operation. The proper execution of the system model requires the model of the scheduler also because the scheduler has the primary control on the local (in-node) resource access. The model of the scheduler is used by the EXECUTION block of the Fig. 1.10.
- 6.
In order to preserve memory in practical implementations only the state changes are stored (which is a much smaller set than the full system state as typically only a few components change states in response to an event). Conceptually it is the same as listing the complete system state. For the sake of simplicity we assume direct access to the full system state.
- 7.
The system design evaluation process should be supported by tools providing unified (standard) execution trace representation and post-processing libraries for filtering and calculating frequently used KPIs (e.g. energy consumption of components, utilization of resources, availability of functionalities, etc.). See Chap. 3 for details.
- 8.
References
J.O. Kephart, D.M. Chess, Computer 36(1), 41 (2003)
ISO/IEC: International standard ISO/IEC 10746-3 (1996)
ISO/IEC: International standard ISO/IEC 10746-2 (1996)
ISO/IEC: International standard ISO/IEC 10746-4 (1998)
ISO/IEC: International standard ISO/IEC 10746-1 (1998)
Feiler, Gluch, in Model-Based Engineering with AADL: An Introduction to the SAE Architecture Analysis and Design Language (2012)
O.M. Group: OMG Systems modeling language (2012)
G. Karsai, F. Massacci, L. Osterweil, I. Schieferdecker, Computer 43(5), 34 (2010)
T. Vogel, in DEMANES (2014)
O.M. Group: OMG Unified Modeling Language Superstructure Specification (2007)
B. Brown, Model-based systems engineering: revolution or evolution? (2011). http://www-01.ibm.com/common/ssi/cgi-bin/ssialias?infotype=SA&subtype=WH&htmlfid=RAW14289USEN
B. Morin, O. Barais, J.M. Jzquel, F. Fleurey, A. Solberg, IEEE Comput. 46–53 (2009). http://www.irisa.fr/triskell/publis/2009/Morin09f.pdf
G. Karsai, F. Massacci, L. Osterweil, I. Schieferdecker, Computer 43(5), 34 (2010). doi:10.1109/MC.2010.135
T. Streichert, D. Koch, C. Haubelt, J. Teich, EURASIP J. Embed. Syst. 2006(1), 042168 (2006). doi:10.1155/ES/2006/42168, http://jes.eurasipjournals.com/content/2006/1/042168
T. Gjerlufsen, M. Ingstrup, J. Olsen, Computer 42(10), 61 (2009). doi:10.1109/MC.2009.325
C. van Leeuwen, J. Sijs, Z. Papp, in Fusion (2013)
J. Teich, M. Köster, in Proceedings of the conference on Design, automation and test in Europe (IEEE Computer Society, 2002), pp. 559–566
J.C. Georgas, A. van der Hoek, R.N. Taylor, Computer 42(10), 52 (2009)
M. Schmill, T. Oates, M. Anderson, S. Fults, D. Josyula, D. Perlis, S. Wilson, in AAAI-08 Workshop on Metareasoning, (Chicago, IL, 2008)
J. Kephart, D. Chess, Computer 36(1), 41 (2003)
T. Streichert, D. Koch, C. Haubelt, J. Teich, EURASIP J. Embed. Syst. 2006(1), 9 (2006)
G. Karsai, J. Sztipanovits, IEEE Intell. Syst. 14(3), 46 (1999)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2016 Springer Science+Business Media Singapore
About this chapter
Cite this chapter
van Leeuwen, C., Rieter-Barrell, Y., Papp, Z., Pruteanu, A., Vogel, T. (2016). Model-Based Engineering of Runtime Reconfigurable Networked Embedded Systems. In: Papp, Z., Exarchakos, G. (eds) Runtime Reconfiguration in Networked Embedded Systems. Internet of Things. Springer, Singapore. https://doi.org/10.1007/978-981-10-0715-6_1
Download citation
DOI: https://doi.org/10.1007/978-981-10-0715-6_1
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-10-0714-9
Online ISBN: 978-981-10-0715-6
eBook Packages: EngineeringEngineering (R0)