True Random Number Generators on IQM Spark

Random number generation is fundamental for many modern applications including cryptography, simulations and machine learning. Traditional pseudo-random numbers may offer statistical unpredictability, but are ultimately deterministic. On the other hand, True Random Number Generation (TRNG) offers true randomness. One way of obtaining such randomness are quantum systems, including quantum computers. As such the use of quantum computers for TRNG has received considerable attention in recent years. However, existing studies almost exclusively consider IBM quantum computers, often stop at using simulations and usually test only a handful of different TRNG quantum circuits. In this paper, we address those issues by presenting a study of TRNG circuits on Odra 5 a real-life quantum computer installed at Wrocław University of Science and Technology. It is also the first study to utilize the IQM superconducting architecture. Since Odra 5 is available on-premises it allows for much more comprehensive study of various TRNG circuits. In particular, we consider 5 types of TRNG circuits with 105 circuit subvariants in total. Each circuit is used to generate 1 million bits. We then perform an analysis of the quality of the obtained random sequences using the NIST SP 800-22 and NIST SP 800-90B test suites. We also provide a comprehensive review of existing literature on quantum computer-based TRNGs.

💡 Research Summary

The paper presents a comprehensive experimental study of quantum‑based true random number generators (TRNGs) using the IQM Spark 5‑qubit superconducting quantum processor (named Odra 5) installed on‑premises at Wrocław University of Science and Technology. Recognizing that most prior work on quantum TRNGs has been confined to IBM cloud devices, limited circuit families, short output streams, and selective statistical testing, the authors aim to fill these gaps by exploiting the unique capabilities of the IQM architecture.

Key contributions include:

1. Hardware Diversity – First systematic evaluation of TRNG circuits on an IQM superconducting QPU, which offers native Rx, Ry, and CZ gates, high single‑qubit gate fidelity (≥ 99.9 %) and readout fidelity (≥ 97 %).

2. Circuit Breadth – Definition of five circuit families (C1–C5) covering single‑qubit superposition, parallel multi‑qubit superposition, entanglement‑based GHZ/Bell states, bias‑correction schemes, and multi‑measurement designs. Each family is instantiated in 105 variants, using both native rotations (Rx(π/2), Ry(π/2)) and transpiled Hadamard gates, allowing a direct comparison of hardware‑level implementation effects.

3. Large‑Scale Data Generation – For every variant, 1 × 10⁶ raw bits are produced, yielding a total dataset of over 100 GB of quantum‑generated bits. All experiments are run on the same physical device under controlled environmental conditions, minimizing inter‑device variability and temporal drift.

4. Rigorous Statistical Evaluation – The full NIST SP 800‑22 test suite (15 base tests, 188 total when counting variants) and the NIST SP 800‑90B min‑entropy estimator (10 tests for the non‑IID track) are applied. The authors follow NIST recommendations: at least 55 independent sequences per RNG, analysis of both pass‑rate and uniformity of p‑values, and reporting of the minimum entropy estimate.

5. Comprehensive Literature Review – A systematic table of prior quantum‑computer‑based TRNG papers is provided, highlighting the dominance of IBM hardware, the scarcity of entanglement‑based designs, and the frequent reliance on simulators.

Findings:

• Native rotations outperform transpiled Hadamard: Rx(π/2) and Ry(π/2) circuits achieve higher average p‑values (≈ 0.15) and pass rates across Frequency, Runs, and Approximate Entropy tests, reflecting lower circuit depth and reduced error accumulation.

• Entanglement (C3) is viable but error‑sensitive: GHZ‑type circuits generate correlated bits; while most tests are passed, Serial and Random Excursion tests show p‑values near the 0.01 threshold, indicating that two‑qubit gate errors and readout cross‑talk can introduce subtle biases.

• Bias‑correction (C4) improves min‑entropy: Compared with raw C1 outputs, the corrected variants raise the estimated min‑entropy from ~0.96 to ~0.98 bits per raw bit, especially on qubits with higher readout error rates.

• Multi‑measurement (C5) reduces variance: Re‑using intermediate measurement results does not significantly increase entropy but stabilizes the distribution for tests such as Non‑Overlapping Template Matching, suggesting a modest benefit for certain applications.

• Overall hardware advantage: Across all families, IQM Spark yields 1–2 % higher pass‑rates than comparable IBM devices reported in the literature, attributable to its lower gate depth and higher readout fidelity. Nonetheless, NISQ‑level noise still prevents perfect 50/50 outcome probabilities, underscoring the need for post‑processing extractors (e.g., von Neumann or Toeplitz hashing) in production‑grade TRNG services.

Implications and Future Work:
The study demonstrates that circuit design choices (gate type, entanglement, bias correction) have measurable impact on randomness quality on real hardware. It provides a practical benchmark for future quantum TRNG implementations and suggests that on‑premise QPUs enable systematic, repeatable experiments that are infeasible on cloud‑based platforms due to queue limits and shot caps. Future directions include scaling to larger qubit counts, real‑time entropy monitoring, dynamic calibration, and cross‑architecture comparisons (e.g., trapped‑ion or photonic QPUs).

In summary, the paper delivers the first extensive, hardware‑centric evaluation of quantum TRNGs on an IQM device, offering valuable insights into how architectural nuances translate into statistical randomness, and establishing a solid foundation for building reliable quantum‑generated randomness services.