Field-Programmable Gate Arrays (FPGAs) are reprogrammable hardware whose
functionality is specified by a binary configuration file, or bitstream.
However, FPGAs are vulnerable to malicious bitstream modification, where an
attacker modifies the original bitstream to insert a trojan or other unwanted
functionality.
Modern bitstreams are signed and encrypted, but have still proven vulnerable to
modifications.
Our research is focused on ensuring the integrity of the bitstream through verification and defensive measures. We've designed FPGA bitstreams that include thermal watermarks, or that dynamically overwrite any malicious modifications.
FPGA bitstreams often contain valuable Intellectual Property (IP), creating incentive for IP theft. This research project is focused on automatically detecting pirated IP in FPGA using only extracted bitstreams.
{ PAINE'20 }
90% of the FPGAs sold today are based on SRAM, a volatile memory that loses all state if powered down. A new wave of emerging non-volatile technologies are competing to replace SRAM for FPGAs, i.e. ReRAM, MRAM, etc. Our group is studying how these new non-volatile memory technologies can be used to improve the performance and security of FPGA architectures.