Quantitative analysis of sliding window attack for the RSA timing channel

The timing channel residing in RSA hardware implementations imposes a major threat to hardware security. In this work, we propose a sliding window timing attack method with significant improvement in the attack success rate and employ an information theoretic model to quantitatively measure the effectiveness of the proposed attack. Experimental results show that the sliding window attack method increases the attack success rate from 73％ to 96％ as compared to Kocher's timing attack, which uses a single bit window. Our results also reveal a strong correlation between attack success rates and the mutual information measurements of the key and encryption time. This indicates that information theoretic metrics can serve as an approach for quantifying the effectiveness of the sliding window timing attack method and evaluating timing channel threat of the RSA hardware architecture.