True random number generators (TRNGs) ideally produce unbiased, uncorrelated, and incompressible bits of information by extracting randomness from a stochastic process. These circuits help in secure communication, user authentication, and user identification protocols. Some TRNGs employ a bistable ring (BR), a digital logic circuit made up of an even number of inverters connected in a loop, as their core. When powered on, the BR oscillates and may settle into one of two states. We introduce concepts from nonlinear dynamic system analysis to determine whether the BR’s trajectories are random enough to be considered a promising entropy source in TRNGs. Our example BR simulations and Monte Carlo process variation experiments in Ngspice show that the BR’s trajectories are chaotic in the best-case scenario. We also study the FPGA realizations of BRs and observe periodic behavior in those implementations. Following these observations, we evolved instances of BRs using a genetic algorithm (GA) to determine whether one could surpass the chaotic and periodic characteristics of simulated and implemented BRs. According to our results for optimizing the BR trajectories’ complexity (measured by permutation entropy), some instances created by the GA could exhibit stochastic behavior.

Creator

• Mercola, Victor

Contributors

• Ganji, Fatemeh
• Tajik, Shahin

Degree

• MS

Unit

• Electrical & Computer Engineering

Publisher

• Worcester Polytechnic Institute

Identifier

• etd-105531

Keyword

• nonlinear dynamics
• chaos theory
• random number generation
• simulation
• FPGA
• genetic algorithm

Advisor

• Tajik, Shahin
• Ganji, Fatemeh

Orcid

• https://orcid.org/0009-0000-8823-5274

Defense date

• 2023-04-24

Year

• 2023

Date created

• 2023-04-26

Resource type

• Thesis

Source

• etd-105531

Rights statement

• In Copyright

Last modified

• 2023-06-02