This dissertation is composed of two parts. In the first part, performance evaluation and verification of safety properties are provided for real-time actors. Recently, the actor-based language, Timed Rebeca, was introduced to model distributed and asynchronous systems with timing constraints and message passing communication. A toolset was developed for automated translation of Timed Rebeca models to Erlang. The translated code can be executed using a timed extension of McErlang for model checking and simulation. In the first part of this dissertation, we induce a new toolset that provides statistical model checking of Timed Rebeca models. Using statistical model checking, we are now able to verify larger models against safety properties comparing to McErlang model checking. We examine the typical case studies of elevators and ticket service to show the efficiency of statistical model checking and applicability of our toolset.
In the second part of this dissertation, we enhance our modeling ability and cover more properties by performance evaluation and model checking of probabilistic real-time actors. Distributed systems exhibit probabilistic and nondeterministic behaviors and may have time constraints. Probabilistic Timed Rebeca (PTRebeca) is introduced as a timed and probabilistic actor-based language for modeling distributed real-time systems with asynchronous message passing. The semantics of PTRebeca is a Timed Markov Decision Process (TMDP). We provide SOS rules for PTRebeca, and develop two toolsets for analyzing PTRebeca models. The first toolset automatically generates a TMDP model from a PTRebeca model in the form of the input language of the PRISM model checker. We use PRISM for performance analysis of PTRebeca models against expected reachability and probabilistic reachability properties. Additionally, we develop another toolset to automatically generate a Markov Automaton from a PTRebeca model in the form of the input language of the Interactive Markov Chain Analyzer (IMCA). The IMCA can be used as the back-end model checker for performance analysis of PTRebeca models against expected reachability and probabilistic reachability properties. We present the needed time for the analysis of different case studies using PRISM-based and IMCA-based approaches. The IMCA-based approach needs considerably less time, and so has the ability of analyzing significantly larger models. We show the applicability of both approaches and the efficiency of our tools by analyzing a few case studies and experimental results.
Þessi ritgerð er tvískipt. Í fyrri hlutanum er farið í mat og sannprófun á eiginleikum öryggis í rauntímalíkönum. Fyrir stuttu síðan var leikendabyggða málið, Timed Rebeca, notað við líkana dreifingu og ósamstillt kerfi með tímastillingu og samskipti í skilaboðum. Búið var til verkfærasett fyrir sjálfvirka þýðingu á Timed Rebeca líkön yfir í Erlang. Hægt er að nota þýdda kóðann með því að nota tímastillta framlengingu af McErlang fyrir líkanaprófun og hermun. Í fyrri hluta þessarar ritgerðar, ætlum við að kynna verkfærasettið sem veitir tölfræðilega prófun á líkön á Timed Rebeca líkön. Með því að nota tölfræðileg próf á líkön er núna hægt að sannreyna stærri líkön eins og í öryggiskröfum McErlang. Við rannsökum dæmigerðar ferilsathuganir af lyftum og miðasölu til að sýna fram á skilvirkni tölfræðilegra líkana og beitingu verkfærasettsins okkar.
Í seinni hluta þessarar ritgerðar aukum við við getu líkanagerðarinnar og við náum yfir fleiri eiginleika með mati á framkvæmd og prófunum á líkönum á líkinda rauntíma leikara. Dreifð kerfi sýna líkindi og brigðgenga hegðun sem kunna að hafa tímamörk. Probabilistic Timed Rebeca (PTRebeca) er kynnt sem tímastillt og líkinda leikarabyggt mál líkindadreifðra rauntímakerfa með ósamstillta sendingu skilaboða. Merkingarfræði PTRebeca er Timed Markov Decision Process (TMDP). Við verðum með SOS reglur fyrir PTRebeca, og þróum tvö verkfærasett til að greina PTRebeca líkön.
