True Random Number Generator (TRNG)

The core of a TRNG is a non-deterministic physical process that cannot be algorithmically predicted. Common entropy sources include:

• Thermal noise (Johnson-Nyquist noise): Amplifying the random voltage fluctuations across a resistor.
• Avalanche noise: Exploiting the random breakdown of a reverse-biased P-N junction.
• Ring oscillator jitter: Sampling the phase drift between multiple free-running oscillators caused by thermal and shot noise.
• Radioactive decay: Measuring the time interval between decay events from a weak radioactive source. The raw analog signal from these sources is the foundation of true randomness.

Raw physical entropy is often biased and requires post-processing to produce statistically uniform bits. This process involves:

• Analog-to-Digital Conversion (ADC): Sampling the noisy analog signal to produce a raw digital bitstream.
• Health Tests: Continuous online checks (e.g., repetition count test, adaptive proportion test) mandated by standards like NIST SP 800-90B to detect catastrophic entropy source failures.
• Cryptographic Conditioning: Using an approved algorithm (e.g., a cryptographic sponge or a vetted conditioning component) to distill the raw entropy into a full-entropy output string, removing bias and correlation.

The fundamental distinction between a TRNG and a PRNG lies in determinism:

• TRNG: The output is non-deterministic and unpredictable, even with complete knowledge of the device's prior state and internal design. The next bit is a function of a physical process, not a mathematical formula.
• PRNG (Deterministic Random Bit Generator - DRBG): The output is deterministic. Given the initial seed value and the algorithm, the entire sequence of 'random' numbers can be perfectly reproduced. A PRNG is a state machine that stretches a finite seed into a longer bitstream.

TRNGs are critical for generating high-quality seeds and non-repeating values in cryptographic systems:

• Seed Generation: A TRNG provides the initial, unpredictable seed for a cryptographically secure PRNG (CSPRNG). The security of the entire CSPRNG collapses if the seed is guessable.
• Nonce and Initialization Vector (IV) Creation: Protocols like TLS require unique, unpredictable nonces to prevent replay attacks. A TRNG ensures these values are never repeated in a predictable way.
• Ephemeral Key Pair Generation: For protocols like ECDHE (Elliptic Curve Diffie-Hellman Ephemeral), a TRNG generates the private key for a single session, ensuring compromise of one session's key does not reveal others.

The security of a TRNG is quantified by its min-entropy rate, a measure of the worst-case unpredictability per bit. Key standards for validation include:

• NIST SP 800-90B: Provides a framework for designing and testing the entropy source, requiring a detailed analysis of the noise source and conditioning components.
• AIS 20/31: The German BSI standard for evaluating deterministic and physical random number generators, defining functionality classes (e.g., PTG.2 for a physical TRNG with post-processing).
• FIPS 140-3: Mandates that an approved random bit generator (RBG) be used, which often chains a TRNG seed source to a DRBG.

TRNGs are integrated into modern computing hardware as dedicated peripherals:

• CPU Instruction Set Extensions: Intel's RDRAND and AMD's equivalent instructions provide a hardware TRNG directly accessible to software via a single instruction, backed by an on-die entropy source.
• Trusted Platform Module (TPM): A TPM 2.0 chip contains a mandatory TRNG used for generating attestation keys and sealing data.
• Secure Enclaves: Dedicated security processors, like a Secure Enclave or an integrated Hardware Security Module (HSM), use a private, on-die TRNG to generate keys that never leave the secure boundary.