An effective transition to post-quantum cryptography mandates its deployment on embedded-class devices, guaranteeing adequate performance while satisfying their strict area constraints. This work accelerates BIKE, a QC-MDPC code-based post-quantum KEM, through HLS on embedded-class heterogeneous SoCs that couple a CPU with FPGA programmable logic. The proposed methodology implements HLS-generated accelerators to compute the most time-consuming operations of BIKE, identified by analyzing the software-only execution. The mix of accelerators instantiated in hardware and operations executed in software, as well as the configurable architectural parameters of the former, are then determined, depending on the resources available on the target SoC, to minimize BIKE’s execution time. Experiments on AMD Zynq-7000 SoCs highlight a speedup of up to 3.34 times compared to the reference software execution and up to 1.98 times over state-of-the-art HW/SW implementations targeting the same chips.
HLS-based acceleration of the BIKE post-quantum KEM on embedded-class heterogeneous SoCs
Galimberti, Andrea;Montanaro, Gabriele;Zoni, Davide
2023-01-01
Abstract
An effective transition to post-quantum cryptography mandates its deployment on embedded-class devices, guaranteeing adequate performance while satisfying their strict area constraints. This work accelerates BIKE, a QC-MDPC code-based post-quantum KEM, through HLS on embedded-class heterogeneous SoCs that couple a CPU with FPGA programmable logic. The proposed methodology implements HLS-generated accelerators to compute the most time-consuming operations of BIKE, identified by analyzing the software-only execution. The mix of accelerators instantiated in hardware and operations executed in software, as well as the configurable architectural parameters of the former, are then determined, depending on the resources available on the target SoC, to minimize BIKE’s execution time. Experiments on AMD Zynq-7000 SoCs highlight a speedup of up to 3.34 times compared to the reference software execution and up to 1.98 times over state-of-the-art HW/SW implementations targeting the same chips.File | Dimensione | Formato | |
---|---|---|---|
2023_ICECS.pdf
accesso aperto
:
Post-Print (DRAFT o Author’s Accepted Manuscript-AAM)
Dimensione
179.38 kB
Formato
Adobe PDF
|
179.38 kB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.