Siddharth Shankar Swain*
Department of Computer Sciences, BITS Pilani, Pilani Campus, India
*Corresponding Author: Siddharth Shankar Swain, Department of Computer Sciences, BITS Pilani, Pilani Campus, India.
Received: July 25, 2021; Published: August 05, 2021
Differential power analysis side channel attacks uses the power consumed by different instructions when executed on a processor. Repeated experimentation by trying to encrypt the data with different keys. CPU instructions consume varying power, executes in varying lengths of time. Basic operation for any program executing on a processor involves switching of semiconductors. Semiconductors consume current while switching. Shape of this power consumption profile reveals activity. Comparison of these power profiles reveals processes happening and data consumed. There is also vulnerabilities in branch prediction  and speculative execution which is exploited by another class of side channel attacks. Our proposed solution directly acts at assembly instructions level, where we use the method of outlining to replace only the repeated sets of instructions with Turing complete instructions . We replace the actual instructions with different Turing complete instructions to introduce randomness and also to eliminate the explicit jump or branch instructions. At the end we present a game theoretic approach to model such side channel attacks , with the aim to have a cost benefit analysis of these attacks and what should be the right approach for the user/defender in this zero sum-two player game.
Keywords: Differential Power Analysis; Branch Prediction Attacks; Turing Complete Instructions; Two Player Game
Citation: Siddharth Shankar Swain. “Methods for DPA and Branch Prediction Side Channel Attack Mitigation". Acta Scientific Computer Sciences 3.9 (2021): 07-15.
Copyright: © 2021 Siddharth Shankar Swain. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.