We propose a new information assurance model which can dynamically track the information flow in circuit designs and hence protect sensitive data from malicious leakage. Relying on the Coq proof assistant platform, the new model maps register transfer level (RTL) codes written in hardware description languages (HDLs) into structural Coq representatives by assigning all input, output, and internal signal sensitivity levels. The signal sensitivity levels can be dynamically adjusted after each clock cycle based on proposed signal sensitivity transition rules. The development of data secrecy properties and theorem generation functions makes the translation process from security properties to Coq theorems independent of target circuits and, for the first time, makes it possible to construct a property library, facilitating (semi) automation of the proof. The proposed cycle accurate information assurance scheme is successfully demonstrated on cryptographic circuits with various complexities from a small-scale DES encryption core to a state-of-the-art AES encryption design prohibiting the leakage of sensitive information caused by hardware Trojans inserted in RTL codes.
展开▼
