TELECOMMUNICATION SYSTEMS AND NETWORK TECHNOLOGIES

Use of Petri nets for modeling the methods of parallelization of information security algorithms in the systems with massive parallel coprocessors

A. B. Sizonenko^a, V. V. Menshikh<sup>b

a</sup> Krasnodar University of the Ministry of Internal Affairs of the Russian Federation
^b Voronezh Institute of the Ministry of Internal Affairs of the Russian Federation

Abstract: The purpose of the work is to develop the formalized way of realization of the algorithms of information security in the multiprocessor computing systems and an assessment of their efficiency. The analysis of the architecture of the modern computing systems showed that in the majority of the computing systems, except central processing unit, there are the coprocessors intended for the performance of the specific tasks. It can be the graphic or arithmetic coprocessors, the resources of which are not always used completely. It is offered to use the unused resources of the coprocessors to increase the productivity of software of the information security, functioning of which is associated with the performance of the logical calculations. For this purpose, the algorithm providing the analysis of the architecture of the computing system, realization of the logical calculations by the system of commands of the calculator, allocation of the independent branches in the algorithm, determination of the labor input of the calculations of each functional block and the algorithm as a whole, is worked out. Even if the system of commands does not support the logical calculations, there is a possibility of their realization in such a calculator. It is possible while using the mathematical apparatus of representation of the logical functions by arithmetic polynomials. It is offered to determine the performance of software for information protection by simulation using timed Petri nets. An example of creation of Petri network for modeling of realization of the enciphering algorithm of State Standard 28147-89 in the system with the central processor and the arithmetic coprocessor is reviewed. The complexity of the computing functions of the individual blocks of the algorithm of State Standard 28147-89 is defined at their realization by arithmetic polynomials. These values are the initial data for the creation of the model and define time of imprimitive events. As the modeling environment the CPN Tools system is used. The results of modeling showed that the productivity increases by 4 percent when using the resources of the single-core arithmetic coprocessor.

Keywords: Boolean functions, system of Boolean functions, parallel logical computing, algorithms, Petri nets, CPN Tools.

UDC: 004.02

Received: 27.05.2014