Performance evaluation in stochastic process algebra dtsdPBC (Q2058309)

The stochastic process algebra dtsdPBC (discrete-time stochastic and deterministic Petri box calculus) was proposed in [\textit{I. V. Tarasyuk}, Sib. Èlektron. Mat. Izv. 17, 1598--1679 (2020; Zbl 1448.68352)]. This algebra is a stochastic extension of the Petri box calculus (PBC) presented in [\textit{E. Best} et al., Lect. Notes Comput. Sci. 609, 21--69 (1992); Petri net algebra. Berlin: Springer (2001; Zbl 0968.68110); Inf. Comput. 178, No. 1, 44--100 (2002; Zbl 1012.68117)]. In [\textit{I. V. Tarasyuk} et al., Programm. Comput. Softw. 40, No. 5, 229--249 (2014; Zbl 1339.68033); Sib. Èlectron. Mat. Izv. 15, 1743--1812 (2018; Zbl 1414.60062)] a stochastic extension of the Petri box calculus, discrete-time stochastic and immediate Petri box calculus (dtsiPBC), was presented. The stochastic process algebra dtsdPBC is an extension of dtsiPBC with deterministic multiactions. It enhances the expressiveness of dtsiPBC and extends the application area of the associated specification and analysis techniques. From text: ``The main result of this paper is the performance analysis methods in the framework of dtsdPBC. To evaluate performance, we construct and solve the underlying stochastic process, which is a semi-Markov chain (SMC). The obtained stationary probability masses in the states of the SMC are used to calculate the performance measures (indices) of interest. We call that approach \textit{embedding}, since the SMC is described by the embedded DTMC (EDTMC) specifying the state change probabilities, together with the probability distribution functions (PDFs) of the residence times in the states. The alternative solution techniques are also developed, based on the corresponding discrete time Markov chain (DTMC) and its reduction (RDTMC) by eliminating vanishing states, i.e those with zero sojourn (residence) times. The approach based on the DTMC allows one to avoid the costly intermediate stages of building the EDTMC, weighting the probability masses in the states by their average sojourn times (rescaling) and final normalization. We call that approach \textit{abstraction}, since we abstract from all vanishing states by taking into account only the (normalized) DTMC-based stationary probabilities of the tangible states, i.e. those with positive sojourn times. The approach based on the RDTMC simplifies performance analysis of large systems due to eliminating the non-stop transit (vanishing) states where only instantaneous activities can be executed, resulting in a smaller model having only tangible states that can be solved directly with less efforts. We call that approach \textit{elimination}, since we eliminate all vanishing states. Thus, the main contributions of the paper are the following. \begin{itemize} \item Performance analysis in dtsdPBC via semi-Markov chains (\textit{embedding}). \item The solution technique via discrete time Markov chains (\textit{abstraction}). \item The solution method via reduced discrete time Markov chains (\textit{elimination}).'' \end{itemize}