About

The Automated Security Analysis lab is headed by Prof. Flavio Toffalini. More info at flaviotoffalini.info

Research topics

We build techniques that find and contain software flaws without a human in the loop. Our work spans the whole path from discovering a bug automatically, to understanding how far an attacker could push it, to keeping the rest of the system intact when a component fails. We test on real targets: browsers, language interpreters, and code that runs on constrained or hardware-protected platforms.

• Testing browsers and interpreters Browsers and language runtimes accept untrusted input all day long, and their optimising compilers make them hard to test. We design generators and feedback mechanisms that reach deep engine states instead of bouncing off the parser.
• Fuzzing and automated software testing Better coverage signals, input structure, and target instrumentation, so that campaigns spend their time in code that actually matters.
• Threat mitigation and exploit containment When a bug cannot be removed, it can still be made useless. We study memory-safety defences, compartmentalisation, and runtime policies, along with what they cost.
• Trusted computing Trusted execution environments such as SGX and TrustZone promise isolation. We examine what that promise holds under, and how to test code inside the enclave.
• Security of embedded and system software Firmware and low-level components often escape both testing and defences. We look at how to analyse them when source code, symbols, or a shell are missing.