Estimation of timing characteristics in real-time computer systems using Petri nets

Abstract

The work is devoted to the problem of determining the time characteristics of tasks in real-time systems, the success of which depends not only on their logical correctness, but also on the time for which they receive the result. Determining such time characteristics of the system at the design stage is a rather difficult problem. Its solution is currently based on the use of two main methods: theoretical calculations related to the calculation of the so-called feasibility criteria and modeling of the system's operation on models. Among the models, statistical models of mass service systems are the most widespread. However, in both the first and second options, it is impossible to obtain a guaranteed result, which significantly complicates the design process. Recently, it has been proposed to use the methods of researching models based on the use of the Petri net apparatus. The paper proposes a method for estimating the time characteristics of tasks in real-time systems by means of data analysis. Obtained by modeling the distribution of processor time between tasks according to the selected algorithms of the scheduler by using a model of the software system in the form of a Petri net. The paper proposes a method for estimating the time characteristics of tasks in real-time systems by means of data analysis. Obtained by modeling the distribution of processor time between tasks according to the selected algorithms of the scheduler by using a model of the software system in the form of a Petri net. The method guarantees obtaining the time characteristics of the tasks when choosing specific types of processor and scheduler, which is required to start the technical design of the real-time system. The use of the proposed mathematical model and a package of application programs allow, at the early stages of the development of a real-time system, to determine the real terms of tasks and choose the optimal option for the implementation of technical and lost support. The proposed mathematical models are based on determining the time characteristics of program execution by modeling the distribution of processor time between tasks, subject to the prior selection of planning algorithms and the characteristics of a complex of technical means. The research is based on the use of simulated statistical models and Petri net models, The use of the proposed tools allows you to significantly reduce the time of analysis of possible options for real-time system implementation, increases the quality of the project and significantly reduces the overall time and cost of the entire development. The paper examines in detail the process of modeling a multitasking computer system using a Petri net device. A concrete example shows the simulation of the operation of a complex of programs, including the development of simulation algorithms, the listing of the simulation program, and the analysis of the obtained results.