Incorporating homomorphic encryption with hybrid numerical authentication for blind computer users

Abstract

Current research on “Usable Security” provides security solutions designed without giving impotance to human disabilities. This paper focus on filling this void in the realm of blind or visually impaired computer users. More specifically, this work is towards developing an authentication technology, which is free from shoulder surfing and key board hacking suitable for the visually impaired or blind computer users. To address this problem, we propose this new Hybrid Numerical Authentication System which basically is based on sounds. It also makes use of the memorizing and computing capability of the blind users. The blind user has to set numeric values for each sound clip at the time of registration. These values are homomorphically encrypted and stored in the server. User should remember this as their password. Once these values are set, a secret expression (a two variable algebraic expression) is provided as the second level security which completes the registration process. During login the system randomly plays the sound clips. The user should identify the value of each sound clip and evaluates their secret expression with values of consecutive sound clip pairs. This value is entered as their session passwords. The system validates this session password after encrypting homomorphically with the system evaluated password corresponding to encrypted data in the server. As the order of sound clips played varies with each session, along with the secure secret expression, it becomes difficult for a third person to hack this security system. This two level security system can hence be used effectively by blind or visually impaired computer users for their Personal Digital Assistant. Also the incorporation of homomorphic encryption enhances the server level security of the system.