Methods for determining the identity of the user based on the individual characteristics of computer handwriting

Abstract

In this article the hypothesis is considered that typing on the keyboard by each person has individual characteristics. In the future the development of this technology will help prevent attempts of unauthorized access to personal data, bank accounts and trade secrets. Among the existing methods of biometric authentication, the proposed approach belongs to the category of dynamic methods that undergo changes over time. This feature prevents an attacker from stealing, copying, or forging a user’s handwriting template through network access. While working on the keyboard, a person uses more than 20 different muscles, which makes the typing style unique. The typing speed, key holding time, time to find the next key, periodic typos during typing, and much more were taken as the main characteristics for carrying out identity authentication. Computer handwriting can be recorded in the form of various metrics and analyzed by statistical methods. The author of the article reveals the methodology and conditions of the experiment. The system counts the number of clicks per unit of time, sets time stamps, and collects statistical data to build histograms. The experiment is carried out at different times of the day using different types of keyboards. It is hypothesized that the sample data obey a normal distribution, which is confirmed by the analysis of the results obtained using Pearson’s nonparametric criterion. To determine the differences between the recruiting styles of the subject from his own, the percentage of coincidences of indicators is found according to the integral formula of Moivre — Laplace for normal distributions, the value of which is about 90 %. In a similar way, a comparative analysis of the results obtained with different users is carried out. In this case, this figure will be much lower and does not exceed 60 %. Comparative analysis makes it possible to authenticate a person and is a sufficient information indicator to prevent unauthorized access attempts.