Applications of Resistive Switching Memory as Hardware Security Primitive

Abstract

With the widespread diffusion of ubiquitous mobile computing and internet of things (IoT), secured communication and chip authentication become key requirements. Hardware-based security concepts generally provide the best performance in terms of good security standard, low power consumption, and large area density. In these concepts, the stochastic properties of the device, such as the physical and geometrical variations of the process, are harnessed to generate random bits and functions. This is the basis for the true-random number generator (TRNG), where true random numbers are generated by exploiting the physics and randomness of nanoscale devices. The same random variations can also be used to implement physical unclonable function (PUF) for the authentication of individual hardware chips. Emerging memory devices rely on unique physical mechanisms for transport and switching, thus appear as the ideal source of entropy for hardware TRNG and PUF. These novel memory concepts include resistive switching memory (RRAM), phase change memory (PCM) and spin-transfer torque magnetic memory (STT-MRAM) devices. As these devices are increasingly adopted as memory and computing elements in several applications, exploiting their intrinsic stochastic variations for TRNG and PUF becomes an attractive solution for low cost, high performance security primitives. This chapter provides an overview of TRNG and PUF adopting emerging memory devices as the fundamental entropy source. TRNG concepts are classified by the microscopic stochastic variation that is adopted as entropy source, namely, current noise, switching delay time, or switching voltage. While most TRNG concepts rely on RRAM devices, we also show novel concepts using STT-MRAM devices which take advantage of the excellent endurance and speed of switching. The TRNG schemes are discussed in terms of the simplicity of the design, e.g., the ability to generate random bits without a probability tracking by adopting a

Hardware implementation of a true random number generator integrating a hexagonal boron nitride memristor with a commercial microcontroller

“…Memristors, usually implemented as metal/insulator/metal (MIM) nanocells with a conductance that can be adjusted to two or more levels by applying sequences of electrical stresses 10 , could be an efficient entropy source for TRNG circuits because they can produce random variations of different magnitudes (e.g., state resistance, switching voltage/time) during operation while consuming little energy, in some cases down to ~0.1 pJ per state transition [11][12][13][14] . References [15][16][17][18][19][20] subjected metal-oxide memristors to sequences of ramped voltage stresses (RVS) of different polarities or pulsed write/verify schemes 21,22 to induce cyclical switching between a high resistive state (HRS) and low resistive state (LRS), and used the switching voltages, cycle-to-cycle resistance variability or stochastic time-dependent relaxation as entropy source in a TRNG circuit -as the switching is related to ionic movement in the MIM nanocell, these values show some degree of variability in every cycle, and they cannot be accurately predicted.…”

“…, state resistance and switching voltage/time) during operation while consuming little energy, in some cases down to ∼0.1 pJ per state transition. 11–14 Ref. 15–20 subjected metal oxide memristors to sequences of ramped voltage stresses (RVS) of different polarities or pulsed write/verify schemes 21,22 to induce cyclical switching between a high resistive state (HRS) and a low resistive state (LRS), and used the switching voltages, cycle-to-cycle resistance variability or stochastic time-dependent relaxation as an entropy source in a TRNG circuit—as the switching is related to ionic movement in the MIM nanocell, these values show some degree of variability in every cycle, and they cannot be accurately predicted.…”

Hardware Security of Fog End-Devices for the Internet of Things

“…Strong PUFs are resistant to brute force attacks. The main characteristics of PUFs are reliability (it should always provide same CRPs), unpredictability (PUF CRPs should not be predicted based on other CRPs), unclonability (CRP mappings should be physically unclonable), and physical unbreakability (physical PUF modification should result of malfunction or permanent damage to chip as protection of chip integrity) [ 60 ]. Advantages of the PUFs are resistant to invasive attacks and have no need for extra programming, testing, and processing power [ 61 ].…”

Intrinsically Secure Non-Volatile Memory Using ReRAM Devices

“…A large applications space for these devices has accordingly emerged, ranging from high-density storage class memory [8], [9] to multiply and accumulate (MAC) units for neuromorphic computing [10], [11]. ReRAM devices can also be used to realize various security primitives for such architectures due to their reconfigurability and inter-device variability [12], including physical unclonable functions (PUFs) for device authentication [13], [14], truerandom number generators (TRNGs) [15], reconfigurable vias enabling split manufacturing [16], switch boxes for non-volatile FPGAs [17], and programmable "memory fingerprints" for provable destruction of digital keys after use [18].…”