Quantum Random Number Generation (QRNG)

• QRNG is defined by using inherent quantum phenomena such as wavefunction collapse and vacuum fluctuations to generate truly unpredictable random numbers.
• Key architectures include laser phase noise, vacuum-state homodyne detection, and photon counting, achieving rates from Mbps to multi-Gbps.
• Advanced extraction methods like Toeplitz hashing and Trevisan’s extractor ensure high entropy and security by mitigating classical noise influences.

Quantum Random Number Generation (QRNG) refers to the generation of random numbers by exploiting the fundamental indeterminacy inherent in quantum processes. Unlike pseudo-random number generators, QRNGs extract entropy from quantum phenomena whose outcomes are theoretically unpredictable even with perfect classical information. This ensures unbiased and provable randomness suitable for cryptographic, computational, and foundational applications.

1. Quantum Physical Principles Underlying QRNG

The irreducible unpredictability in QRNGs directly arises from quantum mechanics, embodied in phenomena such as measurement-induced wavefunction collapse, phase diffusion due to spontaneous emission, and vacuum field fluctuations. Formally, if a quantum system is prepared in state $|\psi\rangle = \sum_i c_i |i\rangle$ and measured in the $\{|i\rangle\}$ basis, the outcome probabilities $|c_i|^2$ are not reproducible by any classical hidden-variable model constrained by no-signaling and locality (Ma et al., 2015).

Key entropy sources include:

• Phase Fluctuations in Lasers: Quantum spontaneous emission introduces random increments to the phase $\theta(t)$ of the field $E(t) = E_0 e^{i(\omega t + \theta(t))}$, with variance $\langle [\Delta \theta(t)]^2 \rangle$ dominated by quantum noise near threshold (Xu et al., 2011).
• Vacuum State Fluctuations: In balanced homodyne detection, the vacuum quadrature exhibits pure quantum noise, providing a Gaussian random variable as a direct entropy source (Bai et al., 2021, Qiao et al., 16 Sep 2025, Bruynsteen et al., 2022).
• Photon Counting and Arrival Times: Single-photon events in beam splitters, photon arrival times in attenuated pulses, and Poisson counting statistics from spontaneous emission or photon detection provide sequence-level quantum entropy (Wang et al., 2014, Stipčević et al., 2014).
• Solid-State Quantum Emitters: On-demand single-photon sources such as defect centers in hBN coupled with integrated photonic circuits offer room-temperature deterministic quantum state collapse (White et al., 2020).

2. QRNG Architectures and Implementation Modalities

The diversity of QRNGs is reflected in the architectures and quantum entropy sources exploited.

QRNG Architecture	Quantum Entropy Source	Characteristic Rate/Advantages
Phase noise in semiconductors (Xu et al., 2011)	Laser spontaneous emission	>6 Gbps, telecom-grade simplicity
Vacuum-state homodyne (Bai et al., 2021Qiao et al., 16 Sep 2025Bruynsteen et al., 2022)	Optical field quadrature	5 Mbps – 100 Gbps, chip-scale integration
Beam splitter + single-photon detection (1407.46020807.4111)	Path measurement	Up to 20 MHz, ultra-low latency, bias-free
Photon arrival-time interpolation (Wang et al., 2014)	Poisson process	22–45 Mbps bias/correlation-free, no postproc.
Embedded LEDs and SPAD arrays (Regazzoni et al., 2021Moeini et al., 2023Stipčević et al., 2021)	Spontaneous emission, photoeffect	1 Mbps–400 Mbps, CMOS integration, scalability

Most modern high-speed QRNGs utilize either phase-noise-to-amplitude mapping (via interferometers) or balanced homodyne detection, owing to their scalability to Gbps and compatibility with integrated optics (Xu et al., 2011, Bai et al., 2021).

Post-processing architectures are commonly implemented in FPGAs, microcontrollers, or dedicated logic to enable real-time throughput (Zhang et al., 2016, Qiao et al., 16 Sep 2025).

3. Randomness Quantification, Security, and Extraction

Statistical quantification of quantum randomness employs min-entropy as a metric:

$H_\infty(X) = -\log_2 \max_x \Pr[X = x]$

For trusted-device QRNGs, side-information about classical noise (electrical, technical, or laser instabilities) is conservatively modeled as available to an adversary; only the quantum portion contributes to $H_\infty$ (Xu et al., 2011, Bai et al., 2021).

Extractor Algorithms:

• Toeplitz Matrix Hashing: Implements information-theoretically secure extraction, mapping $n$ bits of input to $m < n$ nearly uniform bits given min-entropy $H_\infty$ per input block. Output $m \leq H_\infty n - 2\log_2(1/\epsilon)$ for error parameter $\epsilon$ (Xu et al., 2011, Qiao et al., 16 Sep 2025, Bruynsteen et al., 2022).
• Trevisan’s Extractor: Uses error-correcting codes and combinatorial design for small-seed, quantum-resilient extraction, outputting $m \approx H_\infty n - O(\log^3 n)$ (Xu et al., 2011).
• FIR/whitening filters: In some Gaussian-like sources or distribution-transforming post-processing, simple FIR filtering achieves uniformity but not compression of entropy; must be used with care regarding cryptographic security (Marangon et al., 2018, Moeini et al., 2023).

Self-testing (device-independent) QRNGs bound $H_\infty$ per round via violation of Bell inequalities (e.g., CHSH parameter $S$), ensuring certified randomness irrespective of internal device models (Ma et al., 2015, Mongia et al., 2024). Source-independent QRNGs provide certified lower bounds on output entropy by bounding Eve’s information via the extremality of Gaussian states and the Holevo quantity (Xu et al., 2017).

4. Performance, Integration, and Statistical Validation

Performance metrics for QRNGs span bit rate, entropy/correctness per sample, stability, and environmental robustness:

• Maximum Bit Rate: Phase-fluctuation QRNGs routinely exceed 6–18.8 Gbps (Xu et al., 2011, Bai et al., 2021); vacuum-state homodyne QRNGs (integrated PIC, high-bandwidth TIA and ADC, device-dependent extractor): up to 100 Gbps (Bruynsteen et al., 2022).
• Entropy Certification: E.g., H$_\infty$ ≈ 6.7 bits/8-bit sample in phase-noise schemes (Xu et al., 2011); ≥5.8–7.7 bits/sample in vacuum homodyne with proper SNR (Bai et al., 2021, Qiao et al., 16 Sep 2025).
• Temperature and Power Robustness: Embedded chip designs sustain min-entropy >5.2 bits/sample over –40 °C to 85 °C (Qiao et al., 16 Sep 2025).
• Latency: On-demand photon SPAD-based QRNGs achieve ≤10 ns latency with 100% per-trigger efficiency (Stipčević et al., 2014).
• Bias and Correlation Removal: XOR-based schemes and edge-discarding in time-of-arrival QRNGs achieve bias <10⁻⁷ and autocorrelation <10⁻⁹; statistical independence is analytically and empirically verified (Stipčević et al., 2014, Wang et al., 2014).

Statistical validation universally employs NIST SP 800-22, Diehard/Dieharder, and TestU01 batteries; successful QRNGs achieve uniform p-value distributions and pass rates >98% in all individual tests (Bai et al., 2021, Qiao et al., 16 Sep 2025, Stipčević et al., 2014, 0807.4111). Long-term stability over petabit-scale runs (e.g., 8 Gbps QRNG running 71 days uninterrupted) demonstrates hardware/software integrity under unattended operation (Marangon et al., 2018).

5. Security Models: Trusted, Semi-Trusted, and Device-Independent QRNGs

QRNGs are grouped by their security and trust assumptions (Ma et al., 2015):

• Practical (Trusted-Device): Source and measurement devices are trusted and calibrated. Randomness is certified by physical modeling and subtraction of classical noise; secure extractors are required (Xu et al., 2011, Regazzoni et al., 2021).
• Self-Testing (Device-Independent): No trust is placed in device details. Bell-inequality violation, measured experimentally (e.g., Hilbert-space dimension tests, CHSH S > 2.5), directly certifies the unpredictability of outputs, robust against internal device failure or malicious behavior (Mongia et al., 2024). Certified min-entropy is bounded from below by nonlocality metrics; rates remain low due to current photonic and detection constraints.
• Semi-Self-Testing: Source- or measurement-independent schemes trust only one device. E.g., continuous-variable SI-QRNGs admit adversarial sources; min-entropy is bounded using covariance matrix estimation and Gaussian extremality (Xu et al., 2017).

Device-dependent QRNGs can achieve Gbps rates but require accurate calibration and runtime monitoring, whereas self-testing protocols trade speed for cryptographic assurance under minimal assumptions.

6. Practical Applications, Limitations, and Open Challenges

Applications:

• High-speed QKD ($\mathrm{GHz}$-clocked), one-time-pad encryption, cryptographic key generation, Monte Carlo simulations, randomness expansion in quantum networks, and certified sampling in foundational physics experiments (Xu et al., 2011, Ma et al., 2015, Chen, 24 Jul 2025).

Limitations:

• Bottlenecks arise in real-time post-processing (cryptographically secure extraction) for >10 Gbps raw rates; commercial adoption hinges on FPGA/ASIC acceleration (Xu et al., 2011, Zhang et al., 2016).
• Phase and temperature stability become critical at scale. Long-term robustness and self-calibration strategies are required for integrated or field-deployed devices (Marangon et al., 2018, Qiao et al., 16 Sep 2025).
• Device-independent rates are limited by photon pair brightness and detector efficiency; ongoing research into high-rate Bell-test and contextuality-based QRNGs is progressing (Mongia et al., 2024).

Open Challenges:

• Integrating real-time, composable-security (quantum-proof) extraction at full optical bandwidths.
• Tighter experimental bounds and protocols for min-entropy estimation under realistic (non-IID, correlated, or side-channel–vulnerable) conditions (Bruynsteen et al., 2022).
• Integration of QRNGs in post-quantum cryptosystems and HSMs with clear security and performance metrics, and robust standards for randomness certification (Chen, 24 Jul 2025).

7. Outlook and Future Directions

Quantum random number generation now spans low-cost mobile and embedded platforms (Sanguinetti et al., 2014, Moeini et al., 2023), Gbps-grade photonic integrated circuits (Bai et al., 2021, Bruynsteen et al., 2022, Qiao et al., 16 Sep 2025), and device-independent protocols (Mongia et al., 2024). Scalability, chip-level integration, and information-theoretic extraction have reached levels sufficient for widespread cryptographic and scientific deployment. Future research is directed at ultrafast composably secure randomness extraction, further reducing device trust requirements, and tight composable security proofs for integrated QRNGs. Continued convergence of physical implementation, theoretical modeling, and secure extraction is required to fully leverage quantum randomness for emerging cryptographic and computational paradigms.