Please note: This PhD seminar will take place in E5 4047.
Murray
Dunne,
PhD
candidate
David
R.
Cheriton
School
of
Computer
Science
Supervisor: Professor Sebastian Fischmeister
Due to a lack of instrumentation techniques, black-box fuzzing in embedded systems lags behind software fuzzing. We can leverage the power side-channel to detect the conditions of branches in a target program to provide a substitute for software instrumentation. By analyzing the power consumption recorded in response to a system input, we can guide a fuzzer to exploit execution paths that would otherwise require source code or machine code access to the target.
We will discuss topics and challenges facing side-channel instrumentation for fuzzing in embedded systems, including data collection and noise reduction, input mutation, and branch condition recovery. We will present our optimization approach to branch condition recovery and discuss its integration with modern fuzzing techniques.