2021/05/19 by Ataberk Olgun, Minesh Patel, Olgun, Ataberk +15 · 3 citations
Computer Science · #Chaos-based Image/Signal Encryption #Cryptography and Security (cs.CR) #Digital Media Forensic Detection #FOS: Computer and information sciences #Hardware Architecture (cs.AR) #Physical Unclonable Functions (PUFs) and Hardware Security
paper · pdf · doi:10.48550/arxiv.2105.08955
openalex publication_date 2021/05/19 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
True random number generators (TRNG) sample random physical processes to\ncreate large amounts of random numbers for various use cases, including\nsecurity-critical cryptographic primitives, scientific simulations, machine\nlearning applications, and even recreational entertainment. Unfortunately, not\nevery computing system is equipped with dedicated TRNG hardware, limiting the\napplication space and security guarantees for such systems. To open the\napplication space and enable security guarantees for the overwhelming majority\nof computing systems that do not necessarily have dedicated TRNG hardware, we\ndevelop QUAC-TRNG.\n QUAC-TRNG exploits the new observation that a carefully-engineered sequence\nof DRAM commands activates four consecutive DRAM rows in rapid succession. This\nQUadruple ACtivation (QUAC) causes the bitline sense amplifiers to\nnon-deterministically converge to random values when we activate four rows that\nstore conflicting data because the net deviation in bitline voltage fails to\nmeet reliable sensing margins.\n We experimentally demonstrate that QUAC reliably generates random values\nacross 136 commodity DDR4 DRAM chips from one major DRAM manufacturer. We\ndescribe how to develop an effective TRNG (QUAC-TRNG) based on QUAC. We\nevaluate the quality of our TRNG using NIST STS and find that QUAC-TRNG\nsuccessfully passes each test. Our experimental evaluations show that QUAC-TRNG\ngenerates true random numbers with a throughput of 3.44 Gb/s (per DRAM\nchannel), outperforming the state-of-the-art DRAM-based TRNG by 15.08x and\n1.41x for basic and throughput-optimized versions, respectively. We show that\nQUAC-TRNG utilizes DRAM bandwidth better than the state-of-the-art, achieving\nup to 2.03x the throughput of a throughput-optimized baseline when scaling bus\nfrequencies to 12 GT/s.\n