Our research mission
Our research focuses on system security, network security, and applied cryptography. We analyze widely-used standards and implementations from different areas. We search for novel attacks and protection techniques to make Internet communication more secure. We mainly concentrate on the security of Web applications, Web services, Single Sign-On (e.g., SAML and OpenID Connect), Transport Layer Security (TLS), and other cryptographic standards.
Systematic analysis of well-used standards
With our systematic analyses in the relevant areas, we were able to find attacks affecting high-profile implementations and well-used standards. For example, we found attacks on XML Encryption, OpenPGP and S/MIME, TLS (DROWN, DROWN), SAML, or even different printers.
Two of our attacks were awarded Pwnie awards for best cryptographic attacks: DROWN and ROBOT.
Development of open source tools for practical analyses
Along with the new attacks, we implement practical tools that help developers and security experts to analyze the security of their implementations. One of the most sophisticated tools is TLS-Attacker. This open-source tool allows for creating custom TLS flows and easy development of new TLS attacks. Besides, it can be used to analyze TLS implementations automatically.
Analysis of side-channel attacks
When analyzing new attacks, we consider different attacks and attacker models. One of our expertise fields is side-channel attacks, for example, timing attacks. Timing attacks exploit tiny timing differences in the behavior of an application. This tiny difference can lead to severe consequences and allow an attacker to retrieve confidential information, extract user data, or even cryptographic keys. In our research, we have shown how to adopt timing side-channel attacks to TLS or XML Encryption.
Usage of modern techniques for systematic analyses
Our analysis techniques are not only manual. We also rely on modern techniques involving machine learning and state learning libraries, which allows us to investigate the internal automata of different cryptographic implementations and detect new vulnerabilities and attacks. For this purpose, we mainly rely on the LearnLib library.
Evaluation of the real-world impact with systematic scanning
Large-scale Internet scanning has proved to be a useful tool to find a large diversity of vulnerability behaviors and detect novel attack techniques in web applications and TLS. We have used this technique to assess the impact of well-known attacks (e.g., padding oracles and nonce-reuse attacks), or our new attacks DROWN and ROBOT. The scans always reveal new attack insights and lead to vulnerability disclosures making well-established implementations more secure.
Prof. Dr.-Ing. Juraj Somorovsky