A Frequency-Based Leakage Assessment Methodology for Side-Channel Evaluations

Abstract

With the constant improvement of side-channel attack technology, people pay more attention to the security of chips and many new assessment technology arise at the historic moment. Currently, one of the most prevailing evaluation technology is TVLA (Test Vector Leakage Assessment), which does not depend on the specific implementation of cryptographic algorithms and the underlying architecture of cryptographic devices, but just utilizes t test to evaluate the possibility of leakage in cryptographic algorithms. Existing TVLA methods carry out t tests on power traces at every sampling point, and evaluate whether or not the algorithms exist leakages in each cryptographic operation, which firstly requires the traces must be aligned strictly. But when the power traces are badly aligned or can't be aligned, the TVLA assessment results are meaningless. In order to resolve this issue, we propose a new leakage assessment methodology. First, we carry out frequency spectrum analysis on the power traces collected in time domain; then, evaluate whether exists a leakage by conducting t tests; finally, we verify the validity and effectiveness of this frequency-based TVLA method.