Microwave Spintronic System with Tunable Stochastic Properties

True random number generators (TRNG) are of great interest in a broad range of scientific fields, such as cryptography or hardware security, neuromorphic systems or stochastic computing, as well as in large-scale physics or climate simulations Emerging post-silicon CMOS or hybrid solutions are competitive TRNG solutions.Our overall purpose is, to experimentally study a bi-stable, resonant cavity, spintronic microwave oscillator configuration that can provide true random bi-level pulse trains with a GHz attempt rate.The pulse train dwell times and bi-level frequency spacing are tuned by a DC-current and an applied external magnetic field, to obtain bias-free generation of uniformly distributed random states.

Our initial aim is to prove that the sampled bit-streams meet the accepted criteria for TRGNs, such as being pattern free with no memory effects., As our second aim, a system level implementation, including a control-loop, will be investigated in the later stage of the project.

The system will be designed and implemented using a custom analog/RF integrated circuit (ASIC) featuring a modified phase-locked-loop concept and assembled on a PCB for system level studies.In summary, we propose a comprehensive system level study that would result in a bias-free TRNG source, based on thermal entropy, with resilience to ambient condition variations, accomplished by dedicated tuning circuitry and system level architecture.