Securing Cardless Automated Teller Machine Transactions Using Bimodal Authentication System

Abstract

In today’s corporate environment, it is important to ensure that only authorized customers have access to offered services. With the availability of ready-to-use sniffers and access code hacking tools, the standard card and Personal Identification Number combination may no longer be sufficient to withstand the test of secure authentication. Additionally, the huge and recurrent card possession and repossession cost incurred by banks’ customers, occasioned by card expiry, loss, theft, and damage is, agreeably, undesirable. In this article, we present the development of a bimodal customer authentication system for a cardless Automated Teller Machine (ATM). The system employs the principle of eigenvectors and Euclidean distance for fingerprint verification. The Personal Identification Number (PIN), which serves as the second factor of authentication, is determined on account opening and hashed using the truncated SHA 512/256 Secure Hashing Algorithm. Analysis of the system performance shows genuine acceptance rates (1-FRR) from 98% and upwards, and equal error rates of 0.0065. A low standard deviation of 0.01 of the Average Matching Times (AMT) shows the consistency of the algorithm in processing the fingerprints. Therefore, the performance evaluation of the system using these metrics portrays the adequacy and suitability of the developed system for ATM user authentication.