الفهرس | Only 14 pages are availabe for public view |
Abstract Nowadays, the popularization of internet services and modern communication technologies leads to much research efforts in the field of data security and privacy protection. Against such a problem, a number of authentication and data encryption techniques have got much attention to guarantee safeguards for data from unauthorized access and the validity of the person. Identity-based authentication techniques are extensively used in several applications and services to truly validate the identity of a person. Traditionally, user specific passwords and/or tokens which for years have been the most widely used tools to secure systems are susceptible to many user inconvenience and unreliability issues that can be easily led to compromised systems and made information vulnerable. For instance, tokens and cards may be lost or stolen; Passwords and PINs may be forgotten, easily guessed or even broken by fraudulent attacks and long passwords are difficult to remember as well as non-certainty of who is the actual user. Biometric technologies, on the other hand, identify individuals by means of linking a person with his normally unique, permanent and hard to reproduce body parts such as fingerprint, iris, voice, palmprint, face and signature. These physical and behavioral traits are unique across individuals and thereby can’t be lost, stolen, guessed, borrowed or forgotten. Traditional biometric systems are unimodal as they rely on a single biometric modality for authentication. By using unibiometric system so as to have poor accurateness with unacceptable error rates and may not be sufficient to guarantee security against spoof attacks. Multibiometric systems, on the other hand, integrate different types of biometric traits. There are a number of benefits inherent to multimodal biometrics, the most prominent being heightened levels of security and accuracy either by reducing the False Reject Rate (FRR) or False Accept Rate (FAR) and greater levels of reliability/flexibility. However, if a multibiometric system is compromised; multiple biometric traits are disclosed to adversaries. As a result, protecting biometric templates stored in centralized databases of multibiometric systems without sacrificing the recognition accuracy of the underlying biometric recognition systems has become a necessary prerequisite to allow wide-spread deployment of these systems, since such templates can’t be revoked or reissued, like passwords and tokens. |