Residual classes based mathematical model of the computer system's reliability

Abstract

It is also known, and it is practically shown that the use of a unposition number system in residual classes (RNS) allows you to drastically increase the speed of the computer system (CS). There are various approaches and difficulties regarding to the assessment of the reliability of the CS in RNS. So, the existing options of mathematical models do not always satisfy a comprehensive assessment of the reliability of the CS in RNS. The variant of mathematical model of reliability of CS, that functions in RNS, is considered in the article, based on the use of a mathematical model of structural sliding backup in the case of a loaded mode of operation of the backup elements, which is used in the positional binary number system (PBNS). A calculation and comparative security of the trooped computing structure (TCS) analysis was carried out in PBNS with an ideal majority element and the CS in RNS with an ideal reliability machine. The results of analysis showed the following. At the initial stage of the functioning of computational systems, reliability, by probability of trouble-free operation, the CS in RNS with two control bases above the reliability of the trooked positional computing structure, widely used in the PDS and with a smaller addition of an additional equipment insertion. It supposes the effective use of RNS for the increase of reliability of CS and computing devices on the initial stage of their functioning. For example, for the increase of reliability the practical use of unposition code structures is possible in RNS in the side digital calculable complexes of ballistic rockets and in the computing devices of pilotless aircrafts of brief action.