An approach for the synthesis of decentralised supervisors for distributed adaptive systems

  • Authors:

  • Amine Belhaj Seboui;Nejib Ben Hadj-Alouane;Gwenael Delaval;Eric Rutten;Moez Yeddes

  • Affiliations:

  • The OASIS Laboratory, National School of Computer Sciences, University of Manouba, Campus Manouba, 2040 Manouba, Tunisia.;The OASIS Laboratory, National School of Computer Sciences, University of Manouba, Campus Manouba, 2040 Manouba, Tunisia.;INRIA Rhone-Alpes, Inovallee – 655, avenue de l'Europe, Montbonnot, 38334 Saint-Ismier cedex, France.;INRIA Rhone-Alpes, Inovallee – 655, avenue de l'Europe, Montbonnot, 38334 Saint-Ismier cedex, France.;National School of Computer Sciences, University of Manouba, Campus Manouba, 2040 Manouba, Tunisia

  • Venue:
  • International Journal of Critical Computer-Based Systems
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, a decentralised supervisory control approach for distributed adaptive systems is proposed. The adaptive systems are modelled, with their adaptation managers, as synchronous reactive systems using finite state machines with inputs/outputs and transition guards. A formal automated control synthesis technique is developed, for supervising the method of coupling adaptation managers with the system, for the purpose of ensuring important safety properties. Our approach is based on existing techniques for discrete controller synthesis, and develops controllers for decentralised systems, consisting of multiple subsystems spread across remote sites. At each site, abstracted behaviour models for each remote subsystem, is computed. In order to compensate for the lack of information caused by the distributed nature of the system, the model used at each site, consists of the abstracted models of the remote subsystems coupled with the full model of the local subsystem. Discrete controller synthesis is then applied at each site, to obtain its local controller. The site controllers then act collectively and independently, on their corresponding subsystems, to ensure that the total global system behaviour respects the given global property.