Iv Robinson.

For the last two years, I have been the lead instructor of MIT's BWSI Embedded Security and Hardware Hacking course. This is a month-long summer program for high schoolers that covers C and Python programming, cryptography, and software and hardware exploitation basics. The main focus is the design challenge, where teams of four build secure firmware update systems. Teams who finish enter the attack phase, where they try to hack other teams' designs. The course is very intense, for the students, instructors, and TAs. I hope silly details like running 110 feet of LED gamer lightstrips around the classroom made it even more fun and memorable.

MITRE's Embedded Capture the Flag (eCTF) is an attack/defense style competition spanning about 3 months. Teams of college students attempt to design a secure embedded system, then they try to exploit vulnerabilities in other school's designs. eCTF is the basis for the BWSI Embedded Security design challenge. Pictured: an automation harness I and a teammate created for our 2023 eCTF design: a secure car/key fob pairing system.

My advisor, Prof. Robert Walls. and I. As the capstone project for my Bachelor's degree, I worked with MTIRE to do research on how effectively neural networks can overcome existing protections against power-based sidechannel attacks on embedded systems. That is a lot of words. I could write it in fewer if it weren't 2AM. By monitoring how much power an embedded system draws while doing computations, we can get a "power trace". These power traces can then be analyzed and used to determine exactly what data the device was processing. Problematic if, for example, the device was encrypting something with a secret key. Software protections exist that make this strategy nearly impossible, but incur a large performance hit. My research found that using a more recent attack strategy involving training a neural network to recognize the right patterns, these protections effectiveness is greatly reduced.