Countering power analysis attacks by exploiting characteristics of multicore processors

Abstract

Power analysis attacks are major concerns among all kinds of side channel attacks. This paper proposes three kinds of countermeasures for Differential Power Analysis (DPA) attacks by fully exploiting characteristics of a many-core processor: (1) Data-level parallelism, (2) Random Dynamic Task Scheduling (RDTS), and (3) Random Dynamic Adjustment (RDA), which combines RDTS, Random Dynamic Frequency Scaling (RDFS) and Random Dynamic Phase Adjustment (RDPA). For the first time these techniques are applied to a multicore processor from the perspective of secure system design. AES algorithm with mentioned countermeasures is implemented on an 8-core processor. Simulation results show that these countermeasures considerably improve the system security with little performance overhead.