On the identifiability in the limit of timed automata
Year: 2006
Type: conference-paper
Abstract: We are interested in identifying a model for discrete event systems from observations. A common way to model discrete event systems is by using deterministic finite state automata (DFA). When observing a sys-tem, however, there often is information in addition to the system events, namely, their times of occurrence. If this time information is important, a DFA is too limited. For example, it is impossible to distinguish between events that occur quickly after each other, and events that occur after each other with a signifi-cant delay between them. Consequently, we would like a model that can also deal with time information. A well-known model for this purpose is the timed au-tomaton (TA) (Alur & Dill, 1994). In this model the time information is represented by a finite set of clocks, which can be reset by state transitions. The values of these clocks are then used in the guards of transitions. A clock guard is a Boolean constraint on the values of the clocks. A state transition can only occur when the guard of the transition is satisfied. We use intervals to intervals to denote a constraint for a single clock. Such a constraint is satisfied when the clock value is an element of this interval. Formally a timed automaton (TA) is a tuple A = 〈Q,X,Σ, T, q0, F 〉, where Q is a finite set of states, Σ is a finite set of symbols, X is a finite set of clocks, T is a finite set of transitions, F ⊆ Q is a subset of final states. A transition d ∈ T in this automaton is a tuple 〈q, q′, a, g, R〉, where q, q ′ ∈ Q are the source and target states, a ∈ Σ is the transition label, g is a clock guard, and R ⊆ X denotes the set of clocks reset by this transition. Each clock of a TA increases synchronously with time. Thus when at some point in time a clock valuates to 1, it will valuate to 2 one time unit later. This is true unless the clock has been reset in the meantime. A reset of clock x at time t sets the time value of x to 0, after which it immediately continues increasing. We study the problem of identifying TAs from a data
Institution: Delft University of Technology
Language: English
Subfield: Computational Theory and Mathematics
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W1544348681
