We consider query optimization techniques for data intensive P2P applications. We show how to adapt an old technique from deductive databases, namely Query-Sub-Query (QSQ), to a setting where autonomous and distributed peers share large volumes of interelated data. We illustrate the technique with an important telecommunication problem, the diagnosis of distributed telecom systems. We show that (i) the problem can be modeled using Datalog programs, and (ii) it can benefit from the large battery of optimization techniques developed for Datalog. In particular, we show that a simple generic use of the extension of QSQ achieves an optimization as good as that previously provided by dedicated diagnosis algorithms. Furthermore, we show that it allows solving efficiently a much larger class of system analysis problems.

In this paper we study the diagnosis of distributed asynchronous systems with concurrency. Diagnosis is performed by a peer-to-peer distributed architecture of supervisors. Our approach relies on Petri net unfoldings and event structures, as means to manipulate trajectories of systems with concurrency. This article is an extended version of the paper with same title, which appeared as a plenary address in the Proceedings of CONCUR’2003.

This paper studies on-line fault detection and isolation of modular dynamic systems modeled as sets of place-bordered Petri nets. The common places among the set of Petri nets modeling a system capture coupling of various system components. The transitions are labeled by events, some of which are unobservable, i.e., not directly recorded by the sensors attached to the system. The events whose occurrence must be diagnosed have unobservable transition labels. These events model faults or other significant changes in the system state. The existing theory of diagnosis of discrete-event systems is extended in the context of the above model. The modular structure of the system is exploited by a distributed algorithm for fault diagnosis. A Petri net diagnoser is associated to every Petri net and the diagnosers communicate in real-time during

Some recent Petri net-based approaches to fault diagnosis of distributed systems suggest to factor the problem into local diagnoses based on the unfoldings of local views of the system, which are then correlated with diagnoses from neighbouring supervisors. In this paper we propose a notion of system factorisation expressed in terms of pullback decomposition. To ensure coherence of the local views and completeness of the diagnosis, data exchange among the unfolders needs to be specified with care. We introduce interleaving structures as a format for data exchange between unfolders and we propose a distributed algorithm for computing local views of the unfolding for each system component. The theory of interleaving structures is developed to prove correctness of the distributed unfolding algorithm.

Monitoring or diagnosis of large scale distributed Discrete Event Systems with asynchronous communication is a demanding task. Ensuring that the methods developed for Discrete Event Systems properly scale up to such systems is a challenge. In this paper we explain why the use of partial orders cannot be avoided in order to achieve this objective. To support this claim, we try to push classical techniques (parallel composition of automata and languages) to their limits and we eventually discover that partial order models arise at some point. We focus on on-line techniques, where a key difficulty is the choice of proper data structures to represent the set of all runs of a distributed system, in a modular way. We discuss the use of previously known structures such as execution trees and unfoldings. We propose a novel and more compact data structure called “trellis”. Then, we show how all the above data structures can be used in performing distributed monitoring and diagnosis. The techniques reported here were used in an industrial context for fault management and alarm correlation in telecommunications networks. This paper is an extended and improved version of the plenary address that was given by the second author at WODES’2006.

Abstract not found

Abstract: A new distributed algorithm for on-line fault detection and isolation of discreteevent systems modeled by Petri nets is presented. The algorithm is applicable to systems modeled in a modular manner by means of place-bordered Petri nets, i.e., Petri nets with common places but distinct transitions. These Petri nets have transition labeled with events; fault events are modeled as transitions labeled with unobservable events. It is assumed that the diagnoser modules are able to communicate in real-time during the diagnostic process. A merge function is defined to combine the individual diagnoser states and recover the complete diagnoser state that would be obtained under a monolithic approach. Copyright c ○ 2005 IFAC

To engineers, scientists, and mathematicians with double X factor ii ACKNOWLEDGEMENTS This thesis reports on work performed while the author was in under the super-vision of Professor Stéphane Lafortune at the University of Michigan. The financial

Abstract not found

Abstract. We define an efficient online method to monitor the execution of asynchronous distributed systems. The code of such systems has been instrumented to record and output, during execution, some pertinent events. This output can be abstracted as a trace, i.e. a partially ordered set of events. During the execution, the online monitoring system collects the trace and checks on the fly that it satifies a requirement, given by any Ltl property on finite sequences. The monitor checks that any execution sequence compatible with the partial order induced by the trace satisfies the property. This problem is NP-complete in the number of concurrent processes. Therefore, to provide an online monitor which, in practice, can cope in real-time with the workload, our method explores the possible configurations symbolically, as it handles sets of configurations. Moreover, it uses techniques similar to the partial order reduction, to avoid exploring as many execution interleavings as possible. It works very well in practice, compared to the standard non symbolic monitoring method.