The demand for data exchange is ever growing. Internet of Things (IoT), industry 4.0, smart city and next-generation interconnected vehicles are some examples of scenarios in which a high volume of nodes share data across networks. Hence, the data protection plays a fundamental aspect to avoid disclosure or manipulation of sensitive information and disruption of services, particularly in safety critical applications. On the other hand, also the compute power at disposal of possible attackers and hackers is growing, and next-future post-quantum capabilities will require the usage of longer keys, certificates and digital signatures, to preserve the security level offered by cryptographic functions. This will affect not only the amount of exchange data, but also the computational resources to secure data, increasing processing time, latencies and power consumption, and lowering data rates. In this work, we investigate different implementation strategies to overcome such performance limitations. This work provides a comparison among pure software approach (both on 32b and 64b processors) and hardware-based solutions we developed for FPGA and ASIC System-on-Chip platforms, for the most common symmetric-key and public-key cryptographic algorithms. The proposed hardware accelerators feature one order of magnitude higher throughput (and lower latency) and more than two orders lower power consumption than their software counterparts. A highly configurable cryptographic suite is proposed that can be customized according to the application requirements and thus able to increase as much as possible the efficiency in terms of energy per enciphered bits per second.
Keywords: Cryptographic hardware accelerators, energy efficiency, ARM 32-bit, ARM 64-bit
