Decomposition of Process Control Algorithms for Parallel Computing Systems Using Automata Models

Autor: Dmitry V. Pashchenko, Dmitry A. Trokoz, Alexey I. Martyshkin
Rok vydání: 2020
Předmět:
Zdroj: 2020 International Russian Automation Conference (RusAutoCon).
DOI: 10.1109/rusautocon49822.2020.9208165
Popis: The paper considers the decomposition of process control algorithms for parallel computing systems using automata models. When designing parallel processing systems, an important problem is the formal representation of process control algorithms since they allow integrated solving the tasks of development, implementation, and analysis of complex control systems, including the control over interacting processes and resources in parallel systems. One of the control algorithm formal description techniques is based on the use of nondeterministic automaton (NDA) logic; it allows representing the data processing control algorithms as canonical equation systems describing all private events implemented in the algorithm. This language’s advantage is describing all the control system transitions in not the system state terms but private event ones, the simultaneous existence of which determines all the system states and transitions, which allows avoiding a ‘combinatorial explosion’ in the state space at the currently available verification capabilities. The paper objective is studying the process control algorithms for parallel computing systems using the NDA tool. Herein, the development and research object is the parallel decomposition of control algorithms for parallel computing systems using automata models. An automata model has been obtained that describes the synchronization of parallel processes based on the finite NDA logic, the correctness of which has been proved by simulation in the VHDL language. The experimental hardware implementation of synchronization device using FPGAs and the resulting time-charts of its operation completely confirm its correct functioning. Conclusions have been drawn on the correctness of the basic results obtained in the study.
Databáze: OpenAIRE