Patent
What makes US 10,078,492 B2 unique
Our patent covers a genuinely novel approach to pseudo-random number generation for cryptographic use. Here’s what sets it apart.
The Core Innovation
The invention combines two independent Linear Feedback Shift Registers (LFSRs) as dynamic boundary inputs to a bounded one-dimensional Elementary Cellular Automaton (ECA) — then extracts randomness from only the center bit of the automaton’s output after it has evolved for a defined number of steps.
What Makes It Different
Traditional PRNGs vs. this patent
| Feature | Traditional PRNG / FSR | This Patent |
|---|---|---|
| Boundary conditions | Fixed (leads to poor randomization) | Dynamic — fed by two independent LFSRs |
| Output extraction | Full register output | Center bit only (most chaotic, least boundary-influenced) |
| Correlation risk | Single seed source | Two uncorrelated, independently running LFSRs |
| Randomness quality | Variable | Passes the Diehard statistical battery of tests |
| Span requirement | Varies | As low as 27 bits span achieves quality output |
Why Each Element Matters
Five design choices, one cryptographic-grade output.
Bounded ECA with fixed boundary = poor randomness
The patent solves this known weakness by replacing static boundaries with live LFSR outputs.
Two uncorrelated LFSRs
A pair of independent LFSRs as left and right boundary inputs prevents predictable correlation patterns that would weaken cryptographic strength.
Center bit extraction only
The center cell is furthest from both boundaries, making it the most chaotically evolved and statistically independent output point, maximizing unpredictability.
Evolution delay (T = 2^K steps)
Output is only sampled after the automaton evolves for T = 2^K steps (K = span length), ensuring sufficient diffusion of the initial state before any bit is used.
Rule 30 (Wolfram)
Leverages Wolfram's chaotic Rule 30, known for highly non-periodic, complex behavior from simple inputs.
Practical Significance
Where it fits in the real world.
Applicable to cryptography, Monte Carlo simulations, network security, and communications
Works on 1D and 2D cellular automata and with chaotic rules beyond Rule 30
Implementable in hardware (ASIC, microprocessor) or software
Computationally lightweight yet statistically robust — a meaningful advantage for embedded and IoT security systems
In Summary
The uniqueness lies in the elegant hybridization of two well-known components — LFSRs and cellular automata — in a way that specifically solves the boundary condition problem that historically prevented bounded ECAs from being viable PRNGs. The center-bit-only extraction and dual uncorrelated LFSR boundary design together produce cryptographic-grade randomness from a structurally simple system.
See the entropy for yourself.
Run the benchmark against weak comparators, or dive into the docs.