GRAPHICAL BASED AUTHENTICATION MODEL FOR AN ELECTRONIC PAYMENT (E-PAYMENT) SYSTEMS

Abstract

ABSTRACT Client authentication is an essential component in nearly all electronic payment systems. This provides foundation for client the legal access control and user liability. The most foremost used authentication technique is the textual or traditional alphanumeric password. However, this method suffers several setbacks that include password guessing, slow login, duration time of execution and hard to remember the password. To provide an easy, friendly user interface and more secure authentication technique, knowledge- based graphical password authentication is employed in this research work of graphical based authentication for an electronic payment system (model) which uses a click point image or set of images for authentication. In this research, the similarity measure is found as an important task for document retrieval, text matching and retrieval of images from the database that is similar to query image. In order to achieve an optimal performance of the system and make it robust in the face of many challenges, an experiment was conducted during the login session using different algorithms that include Euclidean distance, Cosine similarity, City block distance (Manhattan) and Jaccard distance. The registration and login time were utilized to test the reliability, efficiency and robustness of the graphical scheme. The performances of these algorithms were evaluated and compared using various metrics such as the duration of login (execution time), the login success rate and the matching error (image matching point). The city block distance showed the best results and with an outstanding performance of execution time of 0.0318 milliseconds with matching error of 1.55231 and an acceptable login success rate of 64% compared to Euclidean distance with execution time of 0.0482 milliseconds, matching error of 1.55233 and login success rate of 92%. Many authentication-based applications including electronic payment systems find the use of graphical password to be robust especially with regard to security and ease of use. Hence, in this research work, a comprehensive research studies are carried out on existing graphical authentication password techniques with keen emphases on their suitability for electronic payment systems. This study has shown that graphical based password technique would be the most reliable authentication technique for e-payment systems.