Random Number Generation by Differential Read of Stochastic Switching in Spin-Transfer Torque Memory

The true random number generator (TRNG) is a key enabling technology for cryptography and hardware authentication, which are becoming essential features in the era of the Internet of Things (IoT). Here, we present a novel TRNG concept based on the stochastic switching in spin-transfer torque magnetic random access memory (STT-MRAM). The new methodology relies on the STT-MRAM switching variations affecting the current response under applied rectangular or triangular pulses. Random numbers are extracted from the differential read of the integrated current across two stochastic switching cycles. The proposed concept passes all tests in the NIST SP-800-22 suite with no post-processing, thus supporting STT-MRAM as a promising technology for data/hardware security in the IoT.