TECHNOLOGY FOR ANALYZING THE QUALITY OF HASH FUNCTION CHARACTERISTICS

Abstract

The technology for analyzing the quality of a hash function is proposed. To test it, a specially developed hash function, which is a simplified analog of MD5, is used. It is shown that the traditional quality assessment of the obtained hash values based on the number of collisions has a significant drawback - high computational complexity. Therefore, the technology of quality assessment is reduced to the approximation of the analytical expression of the function, which is amenable to mathematical analysis by traditional methods applicable to polynomials. A better quality hash function is a hash function whose histogram is as close as possible to the graph of the function y=x. This will correspond to the Gaussian distribution of values. When using the proposed technology, it is necessary to develop criteria for classifying functions according to deviations from the ideal distribution of values by one or more of the following indicators: variance; mathematical expectation; arithmetic mean; geometric mean, etc.