Abstract
The paper considers a method of user identification based on dynamic characteristics of mobile device hardware. The characteristics include battery state, battery charge level, battery charging status, memory utilization and device orientation. All indicators can change their values over time and allow to form data sets from them for processing and analyzing in the task of user identification. The correctness of identification is determined by testing statistical hypotheses based on three criteria: one-sample Student's t-criterion, one-sample criterion for testing the hypothesis about the numerical value of variance, Kolmogorov-Smirnov criterion, as well as calculating the resulting probability, which depends on the probabilities of correct identification for each of the criteria and selected significance levels. The application of the method can improve the efficiency of user identification, which is confirmed in the experiment. For generated 100 samples of 3600 values for each investigated dynamic indicator and one data set, user identification with high value of the resulting probability was successfully performed. The results of the work are useful for their analysis and practical application, as well as the development of new methods of identification and authentication.