A Morphable Physically Unclonable Function and True Random Number Generator using a Commercial Magnetic Memory

Mohammad Nasim Imtiaz Khan1, Chak Yuen Cheng2, Sung Hao Lin2, Abdullah Ash- Saki1, Swaroop Ghosh1
1Pennsylvania State University, 2The Pennsylvania State University


Abstract

We propose morphable security primitive using commercial magnetic memory which can be utilized as a Physically Unclonable Function (PUF) and as a True Random Number Generator (TRNG) by manipulating the write time and the number of write pulses respectively. Our analysis indicates that more than 75% bits in the PUF are unusable without any correction due to their inability to exhibit any randomness. We exploit temporal randomness of working columns to fix the unusable columns and write latency control to fix the unusable rows during the enrollment. The intra-HD, inter-HD, energy, bandwidth and area of the proposed PUF is found to be 0, 46.25%, 0.14pJ/bit, 0.34Gbit/s and 0.385µm2/bit (including peripherals) respectively. The proposed TRNG provides all possible outcomes with a standard deviation of 0.0062, correlation coefficient of 0.05 and an entropy of 0.95. The energy, bandwidth and area of the proposed TRNG is found to be 0.41pJ/bit, 0.12Gbit/s and 0.769µm2/bit (including peripherals). The performance of the proposed TRNG has also been tested with NIST test suite. The proposed designs are compared with other magnetic PUFs and TRNGs from other literature.