Abstract
Diagnosability is an intrinsic property of the language generated by discrete event systems (DES) and the computational procedure to determine whether a language possesses or not this property is called diagnosability verification. For regular languages, diagnosability verification is carried out by building either diagnoser or verifier automata; the former is known to have worst-case exponential complexity whereas the latter has polynomial complexity in the size of state space of the automaton that generates the language. A question that has been asked for some time now is whether, in average, the state size of diagnosers is no longer exponential. This claim has been supported by the size of diagnoser automata usually obtained in practical and classroom examples, having, in some cases, state space size much smaller than that of verifiers. In an effort to clarify this matter, in this paper we carry out an experimental study on the average state size of diagnosers and verifiers by means of two experiments: (i) an exhaustive experiment, in which ten sets of automata with moderate cardinality were generated and for these sets of automata, diagnosers and verifiers were built, being the exact average state size for these specific instances calculated; (ii) an experiment with sampling, which considers 1660 sets of different instance sizes and, for each one, sample sets of 10,000 automata are randomly generated with uniform distribution and we compute sets of diagnosers and verifiers for each set of randomly generated automata, which have been used to estimate an asymptotic model for the average state sizes of diagnosers and verifiers.
Similar content being viewed by others
Notes
A set A possesses a total order relation, denoted as \(\preccurlyeq \), if for any a,b,c ∈ A the following conditions are satisfied: (i) \(a \preccurlyeq a\) for any a (reflexivity); (ii) if \(a \preccurlyeq b\) and \(b \preccurlyeq a\), then a = b (antisymmetry); (iii) if \(a \preccurlyeq b\) and \(b \preccurlyeq c\), then \(a \preccurlyeq c\) (transitivity); (iv) either \(a \preccurlyeq b\) or \(b \preccurlyeq a\) (totality).
The notation ⌊x⌋denotes the integer part of x, which is defined as\(\lfloor x \rfloor =\max \{ m \in \mathbb {N} : m\leq x\}\).
The reader may find useful to follow the explanation with the help of Example 1.
= (X ′,Σ,f ′,Γ′,𝜖) is an automaton isomorphic to G through ψ ′ : X → X ′ where ψ ′ : X → X ′ is the mapping X ′ = {ψ(x) : x ∈ X}, where ψ(x) is defined in Eq. 11.
Base two has been chosen since the worst case computational complexity for diagnoser is \(\mathcal {O}(2^{n})\).
A box plot is a graphical display that provides a visual representation of the five-number summary of a data set: first quartile Q1, median, third quartile Q3 and the maximum value. The box of the plot contains the central 50% of the distribution, from the first (Q 1) to the third (Q 3) quartile. A line inside the box marks the median. The lines extending from the box are called whiskers, and encompass the rest of the data, except for potential outliers (data that lie more than 1.5I Q R = 1.5(Q 3 − Q 1) below the Q 1 and above Q 3), which are shown separately (Hair et al. 2007).
The calculation was carried out using the open source statistic package R (The R foundation 2016).
References
Almeida M, Moreira N, Reis R (2007) Enumeration and generation with a string automata representation. Theor Comput Sci 387(2):93–102
Bassino F, David J, Nicaud C (2009) Enumeration and random generation of possibly incomplete deterministic automata. Pure Mathematics and Applications 19(2-3):1–16
Bassino F, Nicaud C (2007) Enumeration and random generation of accessible automata. Theor Comput Sci 381(1-3):86–104
Clavijo LB (2014) Aspectos computacionais associados á implementação de algoritmos para sistemas a eventos discretos. Universidade Federal de Rio de Janeiro, Tese de doutorado
Fox J, Weisberg S (2011) An R companion to applied regression SAGE publications. USA, Thousand Oaks
Gubner JA (2006) Probability and random processes for electrical and computer engineers. Cambridge University Press, New York
Hair J, Anderson R, Tatham R (2007) Análise Multivariada de Dados. Bookman, Porto Alegre
Jiang S, Huang Z, Chandra V, Kumar R (2001) A polynomial algorithm for testing diagnosability of discrete-event systems. IEEE Trans on Automatic Control 46(8):1318–1321
Jirásková G, Masopust T (2012) On a structural property in the state complexity of projected regular languages. Theor Comput Sci 449:93–105
McGeoch C (2012) A guide to experimental algorithmics. Cambridge University Press, New York
Moore F (1971) On the bounds for state-set size in the proofs of equivalence between deterministic, nondeterministic, and two-way finite automata. IEEE Trans Comput 20(10):1211–1214
Moreira MV, Basilio JC, Cabral FG (2016) “Polynomial time verification of decentralized diagnosability of discrete event systems” vs. “Decentralized failure diagnosis of discrete event systems”: a critical appraisal. IEEE Trans Autom Control 61(1):178–181
Moreira MV, Jesus TC, Basilio JC (2011) Polynomial time verification of decentralized diagnosability of discrete event systems. IEEE Trans Autom Control 56(7):1679–1684
Qiu W, Kumar R (2006) Decentralized failure diagnosis of discrete event systems. IEEE Trans on Systems, Man and Cybernetics, Part A 36(2):384–395
Ramadge PJ, Wonham WM (1989) The control of discrete-event systems. Proc of the IEEE 77:81– 98
Reis R, Moreira N, Almeida M (2005) On the representation of finite automata. In: Proceedings of the 7th international workshop on descriptional complexity of formal systems, pp 269–276
Sampath M, Sengupta R, Lafortune S, Sinnamohideen K, Teneketzis D (1996) Failure diagnosis using discrete event models. IEEE Trans on Control Systems Technology 4(2):105–124
Sampath M, Sengupta R, Lafortune S, Sinnamohideen K, Teneketziz D (1995) Diagnosability of discrete event systems. IEEE Trans Autom Control 40(9):1555–1574
The R foundation (2016) The R Project for Statistical Computing. Website. https://www.r-project.org/
Wong KC (1998) On the complexity of projections of discrete-event systems. In: In IEE workshop on discrete event systems, pp 201–208
Yoo T-S, Lafortune S (2002) Polynomial-time verification of diagnosability of partially observed discrete-event systems. IEEE Trans Autom Control 47(9):1491–1495
Acknowledgments
The authors are in debit with Prof. Stéphane Lafortune, University of Michigan, Ann Arbor, for the encouragement to submit this paper for publication and for suggesting its title. They also would like to thank the Brazilian Research Council (CNPq) for financial support.
Author information
Authors and Affiliations
Corresponding author
Additional information
This work was carried out while L. B. Clavijo was a D.Sc. student at the Electrical Engineering Post-graduation Program of the Federal University of Rio de Janeiro.
Rights and permissions
About this article
Cite this article
Clavijo, L.B., Basilio, J.C. Empirical studies in the size of diagnosers and verifiers for diagnosability analysis. Discrete Event Dyn Syst 27, 701–739 (2017). https://doi.org/10.1007/s10626-017-0260-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10626-017-0260-y