|
This article is cited in 1 scientific paper (total in 1 paper)
Network-based models in Control
Algorithms of resource management in generalized stochastic networks
N. N. Ivanov V.A. Trapeznikov Institute of Control Sciences of RAS, Moscow
Abstract:
Generalized stochastic network contains vertices corresponding to events, the fulfillment of which can occur in the discipline "AND" and "OR". These networks can be used for simulation of real-time control processes using parallel computing systems. The weight of edges in these networks are considered random variables distributed according to given laws. The main purpose of modeling of such networks is monitoring and ensuring of the specified directive execution time of the management process. The time management of the network schedule can be managed by utilization of free resources (processors, channels of parallel computing systems, etc.). For a generalized stochastic network, three methods of managing free resources are considered, aimed at accelerating the execution of a simulated control process in real time. Evaluation of the feasibility and verification of the effectiveness of the proposed methods is carried out on the basis of the state tree of the network modeling the management process. The state tree defines those states of the network process that require fewer resources than what is called the degree of parallelism. The degree of parallelism determines the maximum number of resources that ensures the absence of queues. Simulation tools are discussed that allow you to obtain comparative data on each of the proposed methods in the tree of states.
Keywords:
the generalized stochastic network, the state tree of the network, the execution time of the network, simulation.
Received: June 15, 2017 Published: May 31, 2018
Citation:
N. N. Ivanov, “Algorithms of resource management in generalized stochastic networks”, UBS, 73 (2018), 95–107
Linking options:
https://www.mathnet.ru/eng/ubs955 https://www.mathnet.ru/eng/ubs/v73/p95
|
Statistics & downloads: |
Abstract page: | 146 | Full-text PDF : | 57 | References: | 27 |
|