Management of evaluation processes and creation of authentication metrics: Artificial intelligence-based fusion framework

Abstract

While the literature extensively covers various authentication systems, management of evaluation processes and creation of authentication metrics remain significant information challenges for researchers. To overcome this complex challenge, we present a taxonomy of research processes based on fusion and fuzzy strategies and give an overview and comparison of related studies. Specifically, we develop an artificial intelligence-based fusion framework (f_f) incorporating Mamdani-type fuzzy rules and key user factors: security, privacy, and trust. Its uniqueness and innovation lie in the application of trapezoidal functions to describe these factors as key input metric values. Moreover, we are the first to incorporate trust as an independent comparative factor and provide a comparison of traditional and modern authentication methods, including artificial intelligence (AI), electroencephalogram (EEG), electrocardiographic (ECG), and photoplethysmogram (PPG) methods. Also, we use a workflow diagram to define the topological relationships among user factors and authentication factors, clarifying the role of fusion in multi-factor authentication (MFA) approaches. In comparison to other similar frameworks implemented solely for traditional methods, the proposed f_f yields better and more realistic quantification metric results. In addition, we present and discuss the key mathematical differences between one-factor authentication (1FA) and MFA, aiming to shed light on issues such as complexity and bias. Lastly, the developed f_f not only advances MFA metrics by introducing modern authentication methods such as AI, EEG, ECG, and PPG but also paves the way for future research on how and why AI algorithms need to be incorporated into information processing and the creation of strong MFA solutions.