Technische Universität (TU) Berlin
Building C, Straße des 17. Juni 115,
September 2, 2010
Organizers: A. Giua (University of Cagliari, Italy) and L. Ricker (Mt Allison University, Canada)
09:00-09:45 Stefan Haar (Ecole Normale Superieure de Cachan, France) Diagnosability in concurrent systems
09:45-10:30 Stephane Lafortune (University of Michigan Ann-Arbor, USA) Dynamic sensor activation and diagnosability of discrete event systems
10:30 - 11:00 Coffee-Break
11:00-11:45 Laurie Ricker (Mount Allison University, Canada) Structures for calculating properties of decentralized systems
11:45-12:30 John Thistle (University of Waterloo, Canada) Blocking in networks of arbitrary size
12:30 - 14:15 Lunch
14:15-15:00 Klaus Schmidt (Cankaya University, Turkey) Dynamically parametrizable communication protocols for the control of distributed discrete event systems
15:00-15:45 Jan van Schuppen (CWI-Amsterdam, The Netherlands) Control of distributed discrete-event systems - Overview of the literature of the DISC Project
15:45-15:50 Concluding remarks
Diagnosability in concurrent systems While event-based diagnosability is a classical topic in discrete event systems, recent years have seen an extension of the existing research from the synchronous automata framework to systems that exhibit concurrent behaviour. Petri nets (and some variants) have been used to model the local-state-local-transition structure of concurrent systems. It has been observed that the partial order representation of system behaviours has important computational advantages over the traditional approach through system languages, i.e. through interleaved words of transition names; and the unfolding semantics for Petri nets comes in handy for this. This talk will emphasize that the importance of partial order representations in diagnosis goes beyond state-space reduction. We will discuss weak diagnosability, a system property that is inaccessible through interleaving models, and both adequate and characteristic of truly concurrent systems. The talk will conclude with an outlook on adjacent topics on concurrent behavior.
Stephane Lafortune (University of Michigan Ann-Arbor, USA)
Dynamic sensor activation and diagnosability of discrete event systems We consider the problem of dynamic sensor activation for event diagnosis in partially-observed centralized or decentralized discrete event systems. Diagnostic agents are able to activate sensors dynamically during the evolution of the system. Sensor activation policies for diagnostic agents are functions that determine which sensors are to be activated after the occurrence of a trace of events. Sensor activation policies must satisfy the property of diagnosability (case of centralized systems) or codiagnosability (case of decentralized systems). A policy is said to be minimal if there is no other policy, with strictly less sensor activation, that achieves (co)diagnosability. We present a formulation of the sensor activation problem and associated algorithms for the synthesis of minimal policies.
(Joint work with Weilin Wang, Feng Lin, and Anouck Girard.)
Laurie Ricker (Mount Allison University, Canada)
Structures for calculating properties of decentralized systems
Several years ago, we introduced a structure (based on the synchronization vectors of Andre Arnold) that allowed us to pinpoint where communication might be introduced to resolve violations of "sequence-based" co-observability for the control of decentralized discrete-event systems. In this tutorial-style presentation, the applicability of this structure for identifying additional properties in decentralized systems will be explored.
Klaus Schmidt (Cankaya University, Turkey)
Dynamically parametrizable communication protocols for the control of distributed discrete event systems
The automation of today's large-scale industrial systems relies on the operation of distributed controller devices that perform local computations and exchange information via industrial communication networks. Such distributed controllers transmit real-time data for tasks such as closed-loop control and non-real-time data for tasks such as diagnosis or system configuration. The subject of this talk is the development of a family of shared-medium industrial communication protocols that support the transmission real-time and non-real-time data among distributed controller devices. Different from existing protocols, it is shown how information that is available from the control application, that is described by a discrete event system model, can be incorporated in the protocol definition. As a result, the developed protocols dynamically change the bandwidth allocation on the shared medium according to the instantaneous communication requirements while ensuring hard real-time guarantees. In the presented work, the modeling framework of timed I/O automata allows to formally prove the desired protocol properties. Following the recent trends in industrial automation, the protocols family can be realized by two software layers on top of low-cost conventional Ethernet.
John Thistle (University of Waterloo, Canada)
Blocking in networks of arbitrary size
Inspired by problems in development of telecommunications services and other distributed systems, we consider the analysis of blocking in networks composed of arbitrary numbers of similar finite-state subsystems. These are examples of so-called parameterized systems, in this instance the parameter being the number of subsystems in the network. Considering different mechanisms of interaction or communication between subsystems, we identify some decidable and some undecidable instances of the problem of checking for blocking. For undecidable instances, we discuss some semidecision procedures.
(Joint work with Siamak Nazari.)
Jan van Schuppen (CWI-Amsterdam, The Netherlands)
Control of distributed discrete-event systems
Overview of the literature and of the DISC Project