Accessibility setting

Section: New Results

Analysis and Verification of Quantitative Systems

Verification of Concurrent Timed Systems

Participants : Éric Fabre, Loïc Hélouët, Karim Kecir

Combining Free Choice and Time in Petri Nets

Time Petri nets (TPNs) are a classical extension of Petri nets with timing constraints attached to transitions, for which most verification problems are undecidable. In [3], We consider TPNs under a strong semantics with multiple enablings of transitions. We focus on a structural subclass of unbounded TPNs, where the underlying untimed net is free choice, and show that it enjoys nice properties in the timed setting under a multi-enabling semantics. In particular, we show that the questions of firability (whether a chosen transition can fire), and termination (whether the net has a non-terminating run) are decidable for this class. Next, we consider the problem of robustness under guard enlargement and guard shrinking, i.e., whether a given property is preserved even if the system is implemented on an architecture with imprecise time measurement. For unbounded free choice TPNs with a multi-enabling semantics, we show decidability of robustness of firability and of termination under both guard enlargement and shrinking.

Production Systems with Concurrent Tasks

The work in [7] considers the realizability of expected schedules by production systems with concurrent tasks, bounded resources that have to be shared among tasks, and random behaviors and durations. Schedules are high level views of desired executions of systems represented as partial orders decorated with timing constraints. Production systems (production cells, train networks... ) are modeled as stochastic time Petri nets STPNs with an elementary (1-bounded) semantics. We first propose a notion of time processes to give a partial order semantics to STPNs. We then consider boolean realizability: a schedule S is realizable by a net N if S embeds in a time process of N that satisfies all its constraints. However, with continuous time domains, the probability of a time process with exact dates is null. We hence consider probabilistic realizability up to $a$ time units, that holds if the probability that N realizes S with constraints enlarged by $a$ is strictly positive. Upon a sensible restriction guaranteeing time progress, boolean and probabilistic realizability of a schedule can be checked on the finite set of symbolic prefixes extracted from a bounded unfolding of the net. We give a construction technique for these prefixes and show that they represent all time processes of a net occurring up to a given maximal date. We then show how to verify existence of an embedding and compute the probability of its realization.

Testing of Timed Systems

Participants : Léo Henry, Thierry Jéron, Nicolas Markey

Partial observability and controllability are two well-known issues in test-case synthesis for interactive systems. In [25], we address the problem of partial control in the synthesis of test cases from timed-automata specifications. Building on the tioco timed testing framework, we extend a previous game interpretation of the test-synthesis problem from the untimed to the timed setting. This extension requires a deep reworking of the models, game interpretation and test-synthesis algorithms. We exhibit strategies of a game that tries to minimize both control losses and distance to the satisfaction of a test purpose, and prove they are winning under some fairness assumptions. This entails that when turning those strategies into test cases, we get properties such as soundness and exhaustiveness of the test synthesis method.

Analysis of Stochastic Systems

Participants : Nathalie Bertrand

A decade ago, Abdulla, Ben Henda and Mayr introduced the elegant concept of decisiveness for denumerable Markov chains. Roughly speaking, decisiveness allows one to lift most good properties from finite Markov chains to denumerable ones, and therefore to adapt existing verification algorithms to infinite-state models. Decisive Markov chains however do not encompass stochastic real-time systems, and general stochastic transition systems (STSs for short) are needed. In [4], we provide a framework to perform both the qualitative and the quantitative analysis of STSs. First, we define various notions of decisiveness, notions of fairness and of attractors for STSs, and make explicit the relationships between them. Then, we define a notion of abstraction, together with natural concepts of soundness and completeness, and we give general transfer properties, which will be central to several verification algorithms on STSs. We further design a generic construction which will be useful for the analysis of $\omega$-regular properties, when a finite attractor exists, either in the system (if it is denumerable), or in a sound denumerable abstraction of the system. We next provide algorithms for qualitative model-checking, and generic approximation procedures for quantitative model-checking. Finally, we instantiate our framework with stochastic timed automata (STA), generalized semi-Markov processes (GSMPs) and stochastic time Petri nets (STPNs), three models combining dense-time and probabilities. This allows us to derive decidability and approximability results for the verification of these models. Some of these results were known from the literature, but our generic approach permits to view them in a unified framework, and to obtain them with less effort. We also derive interesting new approximability results for STA, GSMPs and STPNs.

Opacity for Quantitative Systems

Participants : Loïc Hélouët, Hervé Marchand

Quantitative Opacity

The work in [26] considers quantitative approaches for opacity. A system satisfies opacity if its secret behaviors cannot be detected by any user of the system. Opacity of distributed systems was originally set as a boolean predicate before being quantified as measures in a probabilistic setting. This paper considers a different quantitative approach that measures the efforts that a malicious user has to make to detect a secret. This effort is measured as a distance w.r.t a regular profile specifying a normal behavior. This leads to several notions of quantitative opacity. When attackers are passive that is, when they just observe the system, quantitative opacity is brought back to a language inclusion problem, and is PSPACE-complete. When attackers are active, that is, interact with the system in order to detect secret behaviors within a finite depth observation, quantitative opacity turns out to be a two-player finite-state quantitative game of partial observation. A winning strategy for an attacker is a sequence of interactions with the system leading to a secret detection without exceeding some profile deviation measure threshold. In this active setting, the complexity of opacity is EXPTIME-complete.

Opacity with Powerful Attackers

In [27], we consider state-based opacity in a setting where attackers of a secret have additional observation capabilities allowing them to know which inputs are allowed by a system. This capability allows attackers of a system to partially disambiguate the possible set of states the system might be in, and increases the power of an attacker. We show that regular opacity (opacity of a property described by a regular language) is decidable in this setting. We then address the question of controlling a system so that it becomes opaque, and solve this question by recasting the problem in a game setting.

Diagnosis of Quantitative Systems

Participants : Blaise Genest, Éric Fabre, Hugo Bazille, Nicolas Markey

Diagnosis for Timed Automata

In [20], we consider the problems of efficiently diagnosing and predicting what did (or will) happen in a partially-observable one-clock timed automaton. We introduce timed sets as a formalism to keep track of the evolution of the reachable configurations over time, and build a candidate diagnoser for our timed automaton. We report on our implementation of this approach compared to the algorithm of Tripakis, Fault diagnosis for timed automata, 2002.

Quantitative Diagnosis for Stochastic Systems

For stochastic systems, several diagnosability properties have been defined. The simplest one, also called A-diagnosability, characterizes the fact that after each fault, detection will almost surely occur. We have considered quantitative versions of the problem in [17]. We are interested in quantifying how fast the diagnosability can be performed. For that, we give an algorithm to compute in polynomial time any moment of the distribution of the detection delay. This allows one to approximate the distribution of detection delay, and to provide lower bounds on the probability that detection takes place at most T events after the fault.

One problem with A-diagnosability is that in the worst case, a subset construction needs to be performed, leading to an exponential blow-up in the number of states. To mitigate this, we proposed in [16] different techniques that avoid this blow-up in a large number of cases.