A Novel and Efficient Multilayered Approach to CAPTCHA: Design, Performance and Usability Evaluation

Abstract

With the wide usage of the World Wide Web getting more and more popular nowadays, the impact of the Internet on all user aspects of computer applications is huge. Almost all the web applications focus on the authentication and security preserving factors in order to overcome breach of data, spamming and to prevent bots from accessing them. To prevent the misuse of online servicing platforms and to provide a good level of reliable, secure and authenticated service, CAPTCHAs (Completely Automated Public Turing test to tell Computers and Human Apart) play a vital role. Usability aspect of CAPTCHA is defined as the ease of humans dealing with its challenges and overcoming them. CAPTCHAs should not be too easy for a human to identify while maintaining high difficulty for the bots to recognize the solution. Many of the existing text and image CAPTCHAs do not provide an acceptable level of usability for various groups of people. In this paper, we have designed a novel, usable, robust and reliable text CAPTCHA schemes framework named MACS-TCHA (Manipulation of Alphanumeric characters, Colors and Shapes - Turing test to tell Computers and Human Apart). A total of 4 primary variations of CAPTCHAs were designed and implemented by incorporating multilayered randomness approach, involving alphanumeric characters, shapes and colors for obtaining better usability. A large scale user study involving more than 200 distinct users was conducted in order to compare and evaluate the MACS-TCHA’s accuracy and usability. A thorough investigation has been made on the reliability, robustness and performance of the proposed MACS-TCHA schemes with the existing methods in order to thwart recognition attacks from the bots. We also identified various usability factors for our evaluation and the results show that the users achieved high accuracy and consider our framework to be highly usable, reliable, fun and efficient.