Faults that Persist: Fault-Injection Attacks and ASCON-Based Defense in Multi-Tenant FPGAs

FPGAs are increasingly used in cloud computing infrastructures and reconfigurable system-on-chip, particularly for AI acceleration. The availability of FPGAs in cloud data centers has opened up new opportunities for users to improve application performance by implementing customizable hardware accelerators directly on the FPGA fabric. To improve resource utilization and flexibility, FPGAs allow multi-tenancy, in which multiple users share a single FPGA through dynamic partial reconfiguration. However, the virtualization and sharing of FPGA resources among multiple users open up new security vulnerabilities during reconfiguration. For instance, an adversary can inject faults into partial bitstreams during the reconfiguration process. Unlike prior runtime fault attacks, the injected faults become embedded in the bitstream and persist after configuration, continuing to corrupt FPGA operation even after the attack ceases. Cyclic redundancy check (CRC) is used in commercial FPGA toolchains for bitstream integrity verification, but CRC is insufficient against reconfiguration-time fault attacks. In particular, CRC performs coarse-grained, post-load bitstream verification, which prevents early detection of bitstream manipulation and can induce denial-of-service by forcing the configuration interface into an error state that blocks subsequent benign bitstreams. To address these limitations, we present a lightweight countermeasure that integrates ASCON, a standardized cryptographic authentication primitive, directly into the reconfiguration flow to detect and block fault-injected bitstreams from configuration. Experimental results on a Xilinx Pynq FPGA platform demonstrate preemptive detection and fine-grained localization of fault-injected regions within the bitstream, while incurring low overhead.