A performance comparison study of ECC and AES in commercial and research sensor nodes

Abstract

In the last years, a number of research sensor nodes based on the combination of low-power microcontrollers and FPGAs have been proposed. In these platforms, the FPGAs serve as accelerators of complex algorithms, making it possible to process a vast amount of sensing data in real time. In medical applications that rely on sensors, both privacy and security are of utmost importance. However, the execution of cryptographic primitives on nodes solely based upon microcontrollers clocked at a low frequency can affect the overall power consumption of the platform. In this manuscript, we present the design of a cryptographic accelerator based on FPGA and aimed to be coupled to a microcontroller. It provides the most secure implementation of the IEEE 802.15.4 security suite together with an Elliptic Curve Cryptography (ECC) engine to perform key establishment. Furthermore, we provide a comparison on performance and energy consumption of cryptographic primitives in commercial nodes and in the proposed design. Our results suggest that the controlled execution of cryptographic algorithms on FPGAs clocked at a low frequency improves the performance and energy consumption of the state-of-the-art implementations based on the MICA and Tmote nodes. However, our results also reflect two inherent limitations of this type of design: the very nature of the bus logic can undermine the expected improvement on performance as well as the communication link between the microcontroller and the FPGA can make the platform vulnerable to physical attacks.