Enhancing Authentication Models Characteristic Metrics via Probability Modeling

Abstract

Many choices exist for authenticating users into secure computing systems, yet users continue to lose confidence in the services of online and offline systems because of the increasing activities of identity attackers to circumvent in-built authentication models. Researchers and security experts are becoming uncertain in their opinion of what actually constitute the characteristic metrics that should be evident in invulnerable  authentication models. In this work, we derive the universal characteristic metrics set for authentication models based on security, usability and design issues. We then compute the probability of the occurrence of each characteristic metrics in some single factor and multifactor authentication models in order to determine the effectiveness of these models. Our result show that single factor models do not have enough strength to counter identity attacks., for our findings revealed that textual passwords had efficiency of 30.0%, graphical passwords (63.3%), tokens (43.3%) and biometric systems (90.0%). Conversely, multifactor models proved to be more efficient and offer a more robust characteristic metric and by adding
two unique characteristic metrics: reusability and randomization, we proffer an authentication solution, HUn3 I, which is 6.7% more effective. Using Bayes' theorem, we verify that authenticating users with HUn3 I
reduces the chances of false positivity by 0.7%.