Quantitative analysis of interacting systems: foundations and algorithms
Research » QAIS » FctForm » T5

T5 - Application: a QoS-robust composition calculus for services

This task aims at integrating the theoretical results obtained in T1 ad T2 with the team previous work in formal modeling of service-oriented architectures, to define a QoS-aware architectural model for services, an associated modeling language and new techniques for analyzing and verifying service composition taking into account explicit quantitative properties.

This is a crucial application area for the theory and techniques envisaged in this project. Actually, to a large extent, existing formal models for service composition lack sound, pragmatic means for reasoning about real-time and performance requirements in service oriented design. Such inadequacy is explained by systems complexity, inappropriate simplifications (e.g., the assumption that QoS parameters are independent of each other) and the inability to deal with insufficient data about service behavior and/or QoS characteristics.

This task's envisaged contributions will have direct application in the area of dynamic, i.e. runtime, reconfiguration of service-oriented architectures in which the team has current work.The novelty concerns the introduction of quantitative reasoning to assess reconfiguration scenarios and ensuring that quantitative behavioral properties are maintained, within certain limits of invariance, along the system’s life.

The development of the new, QoS-aware architectural model will be framed in the context of Reo, a model for exogenous coordination of software with strict service decoupling to support loose inter-component dependency, under development at CWI, by a group with which elements of our team have collaborated for the last 4 years and which is lead by one of the experts proposed as consultant for this project. This will ensure a larger impact of the Task results: a prototype of the calculus will be developed as a plug-in for ReoTools.

Central to this Task is the characterization of a notion of QoS-robustness is intended to measure the quality of response to unpredictable change and avoidance of failure. Robustness is studied in several contexts. An approach to be explored relies on game-theoretic semantics, resorting to a notion of adversary to model environmental factors beyond the system control. Game-theoretic approaches enable a fine-grained understanding of the logical principles underlying interaction based on the system’s structure, which may provide an interesting alternative to model endurance to failure. The suitability of this notion will be tested with data from a number of service-based application and assessed through the collected empirical evidence.

Once defined QoS-robustness, the theory of behavioral distance and metric bisimulation, from T1 and T2, will be in order to classify service composition patterns and structure the envisaged calculus.

Main contributions

  • a formal characterization of QoS-robust service composition
  • a QoS-aware architectural model for services and an associated modeling language
  • a calculus of QoS-robust service composition

Research team: (BI) Bolseiro de Investigação (Mestrado) 2; Nuno Oliveira; Luís Soares Barbosa; José Nuno Oliveira

Deliverables: We envisage the publication of two research articles in mainstream conferences (such as COORDINATION, WS-FM, WICSA), one journal paper (such as SCP, JSSM, Acta Informatica, one workshop paper plus the development of prototype of the calculus simulator and one PhD? thesis (Nuno Oliveira).

r4 - 06 Jan 2012 - 13:08:12 - AlexandraSilva
This site is powered by the TWiki collaboration platform Copyright © by the contributing authors. Ideas, requests, problems? Send feedback.
Syndicate this site RSSATOM