ﻻ يوجد ملخص باللغة العربية
The paper addresses the problem of computing maximal expected time to termination of probabilistic timed automata (PTA) models, under the condition that the system will, eventually, terminate. This problem can exhibit high computational complexity, in particular when the automaton under analysis contains cycles that may be repeated very often (due to very high probabilities, e.g. p =0.999). Such cycles can degrade the performance of typical model checking algorithms, as the likelihood of repeating the cycle converges to zero arbitrarily slowly. We introduce an acceleration technique that can be applied to improve the execution of such cycles by collapsing their iterations. The acceleration process of a cyclic PTA consists of several formal steps necessary to handle the cumulative timing and probability information that result from successive executions of a cycle. The advantages of acceleration are twofold. First, it helps to reduce the computational complexity of the problem without adversely affecting the outcome of the analysis. Second, it can bring the worst case execution time problem of PTAs within the bounds of feasibility for model checking techniques. To our knowledge, this is the first work that addresses the problem of accelerating execution of cycles that exhibit both timing and probabilistic behavior.
The paper presents a novel algorithm for computing best and worst case execution times (BCET/WCET) of timed automata models with cyclic behaviour. The algorithms can work on any arbitrary diagonal-free TA and can handle more cases than previously exi
The probabilistic bisimilarity distance of Deng et al. has been proposed as a robust quantitative generalization of Segala and Lynchs probabilistic bisimilarity for probabilistic automata. In this paper, we present a characterization of the bisimilar
Timed automata are a convenient mathematical model for modelling and reasoning about real-time systems. While they provide a powerful way of representing timing aspects of such systems, timed automata assume arbitrary precision and zero-delay actions
We consider a notion of non-interference for timed automata (TAs) that allows to quantify the frequency of an attack; that is, we infer values of the minimal time between two consecutive actions of the attacker, so that (s)he disturbs the set of reac
Model checking timed automata becomes increasingly complex with the increase in the number of clocks. Hence it is desirable that one constructs an automaton with the minimum number of clocks possible. The problem of checking whether there exists a ti