Highly efficient entropy extraction for true random number generators on FPGAs

True random number generators are essential components in cryptographic hardware. In this work, a novel entropy extraction method is used to improve throughput of jitter-based true random number generators on FPGA. By utilizing ultra-fast carry-logic primitives available on most commercial FPGAs, we have improved the efficiency of the entropy extraction, thereby increasing the throughput, while maintaining a compact implementation. Design steps and techniques are illustrated on an example of a ring-oscillator based true random number generator on Spartan-6 FPGA. In this design, the required accumulation time is reduced by 3 orders of magnitude compared to the most efficient oscillator-based TRNG on the same FPGA. The presented implementation occupies only 67 slices, achieves a throughput of 14.3 Mbps and it is provided with a formal evaluation of security.