ImPAKT: Real-World Implementation and Human-Centered Design of PAKE Technologies

点击翻译按钮获取中文摘要

The Internet has become the foundation of modern society. On a daily basis, we consume services such as e-mail, ebanking, e-government, e-commerce, social media, and cloud storage. These services require mechanisms to securely establish the identity of the user and communicate securely with them. Accidental or even malicious mis-authentication runs the risk of security threats ranging from economic loss, information leaks to identity theft.Despite being declared "dead" on a regular basis, password-based authentication is still the predominant form of user authentication either as stand-alone or part of multi-factor authentication schemes when higher security guarantees are required. This is despite the fact that password-based authentication suffers from several security problems. A number of tools have been developed and deployed to address these problems: password strength meters to help users choosing stronger passwords, proactively checking leaked passwords to prevent harm from password re-use and leaked credentials, risk-based authentication to detect malicious sign-ins, and more. However, phishing attacks still present the most significant threat, even in the presence of state-of-the-art security mechanisms. Password-Authenticated Key Exchange (PAKE) protocols offer strong protection against phishing attacks and bring several other advantages such as the establishment of forward and backward secure cryptographic session keys. However, their use in practice has been very limited.The main goal of ImPAKT is to improve the state of the art of password-based authentication on the Internet. This will be done by taking a holistic view of PAKE protocols and their implementations in practice. The project will take extensive input from practitioners via surveys and interviews to better understand the exact nature and impact of blockers in practice, as well as the importance of alternative (non-phishing) security goals for PAKE protocols such as service providers never seeing plaintext passwords. We will synthesize this in a new set of technical requirements for PAKE protocols (such as enabling strength estimates, handling typos, or risk assessment) and provide cryptographic security models for these settings. The project will study in detail, through a series of user studies, how users are interacting with PAKE protocols and how protocols, as well as interfaces, need to be designed to facilitate use and prevent phishing attacks. Finally, the project will construct novel protocols to support the new requirements and usability constraints, and provide efficient and ready to use implementations as open-source, as well as, validating the usability of these.