| Abstract [eng] |
Summary Paper addresses the problem of accessible authentication for people with disabilities and presents a proposed adaptive identification and authentication method aimed at improving access to digital services. The relevance of the topic is determined by the fact that traditional authentication methods, such as text-based passwords, PIN codes, or actions requiring precise input, are not equally suitable for users with visual, motor, or text processing impairments. Therefore, authentication solutions must be evaluated not only in terms of security, but also in terms of practical accessibility, flexibility of use, and the availability of fallback authentication methods. The aim of the thesis is to develop a secure and user friendly adaptive authentication model oriented toward greater accessibility for people with disabilities. To achieve this aim, the problems caused by traditional authentication methods were analyzed, existing biometric, multi method, and adaptive authentication approaches were examined, the requirements for the proposed model were defined, its conceptual model was developed, and a prototype was implemented, including voice, gesture based, biometric, and one time password authentication. In addition, a qualitative comparison of the proposed model with the analyzed authentication solution approaches and an experimental evaluation were carried out. First part of the thesis analyzes authentication challenges faced by different user groups and reviews existing authentication solution approaches. The analysis showed that a single authentication method cannot be considered a universal solution for all users, therefore, accessible authentication must combine multiple methods, context awareness, and fallback authentication options. Second part presents the proposed adaptive authentication model, defines its requirements, operating principles, and conceptual architecture. The model is based on a combination of active and passive adaptivity: active adaptivity responds to environmental changes, while passive adaptivity forms a user trust profile based on login context. Third part describes the implementation of the developed prototype on the Android platform, presents the system architecture, the main modules, and the implementation of individual authentication methods. The prototype includes voice, gesture based, biometric, and one time password authentication, as well as active and passive adaptivity logic that 6 allows the system to respond to environmental changes and accumulate the user’s trust profile. Fourth part provides a qualitative and experimental evaluation of the proposed model. The qualitative comparison allows the position of the model among the analyzed authentication solution approaches to be assessed, while the experimental evaluation shows how the prototype operates under different usage conditions. The evaluation analyzes profile formation, trust level stability, response to context changes, adaptation to behavioral changes, time based decay, and the dependence of voice authentication on environmental noise. The conducted research leads to the conclusion that the proposed adaptive authentication model can be regarded as a practically justified solution aimed at balancing security, accessibility, and usability. The value of the model lies not only in the choice of individual authentication methods, but also in their combination with context aware evaluation and decision logic. |
