Computer Security by Hardware-Intrinsic Authentication

The Internet of Things (IoT) has brought evident security concerns. New solutions in security for IoT will need to reduce the dependency on nonvolatile memory for key storage, promote easier means to uniquely identify billions of devices, etc. Physical Unclonable Functions (PUFs) have been adopted as the future for key derivation and hardware fingerprinting. This work presents CSHIA: a new computer architecture that takes into account limitations and strengths of PUFs to provide code and data integrity and authenticity in a seamless design that does not demand changes in processors microarchitecture or software. We describe and analyze a full-fledged FPGA deployment of the architecture and consider attack scenarios, including side-channel attacks on PUFs.