Graphical Based Authentication Method Combined with City Block Distance for Electronic Payment System

Abstract

User authentication is an essential component in nearly all electronic payment systems. Authentication provides the foundation for a user, legal access control and user accountability. The foremost used authentication methods are textual and alphanumeric passwords. However, the alphanumeric password suffers some drawbacks such as easy guessing and hard to remember the password, which can make it vulnerable to attacks such as social engineering and dictionary attacks thereby exposing sensitive information and thus compromising the security of a system. To overcome these challenges associated with alphanumeric password, we propose a graphical-based authentication method combined City block distance to measure the similarity score between the image passwords at the points of registration and login using the KNN algorithm to compute the distance. In this paper, similarity measure was found as an important task for text matching, image processing, and retrieval of images from the database. To achieve optimal performance of the system and make it robust, an experiment was conducted during the login session using the city block distance and other different distance measures that include Euclidean, Cosine similarity, and Jaccard utilizing the KNN algorithm. The experimental results show that the proposed city block distance method has the fastest execution time of 0.0318 ms, minimal matching error of 1.55231, and an acceptable login session rate of 64%, compared to when the graphical-based password is combined with other similarity score distance measures. This paper concludes that the proposed method would be the most reliable authentication for e-payment systems.