This brief presents an effective way to design high-throughput and low-power True Random Number Generators (TRNGs) for Field Programmable Gate Array (FPGA)-based digital systems. The proposed design makes an unconventional usage of the Digital Signal Processing (DSP) slice embedded within the AMD-Xilinx FPGA devices to implement high jitter ring oscillators as entropy sources for efficient TRNG designs. Thanks to its wide bit-width output, several configurations can be enabled to group multiple oscillators within a single DSP slice. As a result, a TRNG designed through the proposed scheme outputs up to 4 random bits per clock cycle, thus leading to a considerably high-throughput, while exploiting an ultra-compact architecture. When implemented on the AMD-Xilinx Zynq XC7Z020 System on Chip (SoC), the new architecture achieves a throughput of 800W106 bit/sec and an energy consumption of only 22 pJ/bit. When compared to state-of-the-art competitors it achieves a throughput rate up to 2.6W higher and an energy consumption up to 8W lower. The new TRNG has been validated by means of the NIST SP 800-22, the NIST 800 90B and the AIS statistical tests.
A High-Speed and Low-Power DSP-Based TRNG for FPGA Implementations
Frustaci F.;Spagnolo F.
;Corsonello P.;Perri S.
2024-01-01
Abstract
This brief presents an effective way to design high-throughput and low-power True Random Number Generators (TRNGs) for Field Programmable Gate Array (FPGA)-based digital systems. The proposed design makes an unconventional usage of the Digital Signal Processing (DSP) slice embedded within the AMD-Xilinx FPGA devices to implement high jitter ring oscillators as entropy sources for efficient TRNG designs. Thanks to its wide bit-width output, several configurations can be enabled to group multiple oscillators within a single DSP slice. As a result, a TRNG designed through the proposed scheme outputs up to 4 random bits per clock cycle, thus leading to a considerably high-throughput, while exploiting an ultra-compact architecture. When implemented on the AMD-Xilinx Zynq XC7Z020 System on Chip (SoC), the new architecture achieves a throughput of 800W106 bit/sec and an energy consumption of only 22 pJ/bit. When compared to state-of-the-art competitors it achieves a throughput rate up to 2.6W higher and an energy consumption up to 8W lower. The new TRNG has been validated by means of the NIST SP 800-22, the NIST 800 90B and the AIS statistical tests.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.