الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 148 |  |  |
| | | from | 148 | | |
Abstract
Biometric technologies, referring to those that identify or verify the identity of a person using physiological (e.g. face and fingerprint) or behavioral characteristics (e.g., signature and voice), have the potential to solve many of the security problems. Traditional biometrics, such as facial patterns, fingerprints, eye irises, hand geometry and voice patterns, are well known for person authentication or identification purposes. Despite their widely used, such biometrics have certain limitations. For example, most of them are prone to forgery. This motivated researchers to study alternative biometric traits. It has been shown in previous studies that the brain-wave pattern for each individual is unique and thus can be used as a biometric. The advantage of biometry from electroencephalography (EEG) is that it is almost impossible to duplicate human brain activity. Also, such electrophysiological biometric traits naturally allow aliveness detection to enhance the security of a traditional fingerprint-biometric-based system. Some potential application of EEG-biometry include building access control, secure information or multimedia access control.
EEG signals are brain activities recorded from electrodes mounted on the scalp. Compared to other means for monitoring brain activities, such as magnetoencephalography (MEG) and functional magnetic resonance imaging (fMRI), electroencephalography is the most recipient and practical one. Up to the present, EEG signals have been successfully applied to the research and development of brain-computer interfaces whose main goal is to enhance the communication and control abilities of motor-disabled people. For biometrics, without regard to the somewhat cumbersome data recording process the modality of EEG signals has several advantages. It is confidential and hard to imitate, since EEG signals are a reflection of individual-dependent inner mental tasks. In addition, one cannot force a person to give ideal EEG signals as those recorded in normal situations, as brain activity is easy to be influenced by the stress and mood of a person. In this sense, EEG-based biometrics can protect personal safety of its users.
iii
An authentication (or verification) system involves confirming or denying the identity claimed by a person (one to-one matching). In contrast, an identification system attempts to establish the identity of a given person out of a closed pool of N people (one-to-N matching). Authentication and identification share the same preprocessing and feature extraction steps and a large part of the classifier design. However, both modes target distinct applications. In authentication mode, people are supposed to cooperate with the system (the claimant wants to be accepted). The main applications are access control systems (airport checking, monitoring, computer or mobile devices log-in), building gate control, digital multimedia access, transaction authentication (in telephone banking or remote credit card purchases for instance), voice mail, or secure teleworking. On the other hand, in identification mode, people are generally not concerned by the system and often even do not want to be identified. Potential applications includes video surveillance (public places, restricted areas) and information retrieval (police databases, video or photo album annotation/identification). Such authentication systems are based on the characteristics of a person, such as face, voice, fingerprint, iris, gait, hand geometry or signature.
Very little work has been done in this area and was focusing mainly on person identification but not on person authentication. Several techniques has been used for identification and authentication such as autoregressive (AR) models with Kohonen’s.Vector Quantizer (VQ) for classification, models based on the spectral power of the signal together with a fuzzy Neural Network for the classification. Moreover these techniques has been used, Decision Tree and Neural Network Classifier and hashing of MAR coefficients.
We have conducted an experiment for multimodal bio-signal authentication technique. During the experiment the user as asked to perform several tasks including reading, transcriptional writing and password hacking. Several bio-signals were captured during the experiment including EEG and ECG.
iv
A novel Artificial Immune System (AIS) based approach as developed for user authentication. The approach was tested on the image/motor dataset. The algorithm reached an accuracy of 40%. Then we developed a rough set approach and the accuracy reached 92.6%.