Evaluation of resistance of ECC designs protected by different randomization countermeasures against horizontal DPA attacks

In this paper we investigate how different randomization approaches influence the success of horizontal DPA attacks. We use our own ECC design to run the experiments. We applied the following randomization techniques: EC point blinding, key randomization and a combination of both. Our experiments demonstrate the fact that these well-known randomizations of processed data are not effective against horizontal DPA attacks. In addition we investigated randomized calculation sequences of partial multiplications, i.e. a kind of intermediate data randomization, as potential countermeasures. We simulated power traces of elliptic curve point multiplication kP for all applied countermeasures and performed our low-cost horizontal DPA attack. We applied a difference of means test to reveal the scalar k. Our experiments show that the randomization of the processing sequence of the partial multiplications improves the resistance of the design against horizontal attacks significantly but is not sufficient as the only means against DPA attacks.