Research Topics

This research area covers security mechanisms, along with the associated compositional aspects, to protect the entire compute stack. We are dealing with strong, nation state adversaries, and our solutions must withstand the most sophisticated attacks. Thus, Draper is interested in collaborations that span a wide range of topics including:

• Secure processor design that includes methods to (formally) verify the (generated) hardware and its security properties (i.e., lack of side channel leakage, integrity and confidentiality of the computation, and reverse engineering and Fault Injection protections).
• Secure software stack design that includes secure firmware, operating systems and languages, property-based fuzzing, compiler transformations to enforce security policies, etc.

Draper applies a wide range of formal methods to understand and then modify programs (in source code or binary form). Our analyses include static, dynamic, and hybrid approaches. We are interested not only in scaling and extending existing approaches, but also creating languages and tool interfaces to make these analyses useful for others. Research topics of interest include:

• Specification composition/synthesis, proof automation (i.e., in Coq, Agda, etc.), counterexample guided inductive synthesis.
• Secure compilation, sound decomplilation, weakest precondition analysis, abduction inference, abstract interpretation.
• Mathematical topics (e.g., type theory, homotopy, category theory, program logics), hyperproperties, datalog/e-graphs.

This research domain covers a broad area of offensive techniques at both the hardware and software layers. Specific areas of collaboration include:

• Reverse engineering and vulnerability research approaches and tool development, focused on cyber-physical systems.
• Novel hypervisor development for code protection, instrumentation, and code obfuscation techniques.
• Research into defeating hardware-based software protections within IoT/Embedded systems.
• Compiler-based techniques including automatic generation of exploits based on X-oriented programming, transformations to increase diversity and obfuscation (i.e., static and dynamic opaque predicates, etc.), taint analysis, control follow analysis, etc.
• Operating systems exploitation including via process injection, packer techniques, networking stack, etc.
• Analog-based attacks (i.e., RF, acoustic, power, etc.) and physical attacks AI-driven exploit campaigns, AI poisoning, etc.

This recently established research domain takes a multidisciplinary approach to solving tough problems in cyber by harnessing the power of machine learning and artificial intelligence. Research topics plan to include:

• Building machine learning models that have the power to identify adversarial techniques tactics and procedures (TTPs) in network traffic, executables, and source code.
• Delivering cyber security to the edge via AI; this includes techniques for hardening IoT devices, military hardware and software, and space systems that automatically adapt to protect the high value assets.
• Leveraging LLMs to aid in the analysis of firmware, binaries, and malware, helping to alleviate some of the manual labor for reverse engineers.