Independence as an Architectural Property

Abstract

True random number generators that seed cryptographic systems are evaluated on two properties: the entropy density of the physical noise being sampled and the independence of the samples produced. NIST SP 80090B specifies that a multisource entropy generator may combine multiple noise sources only when the sources are independent that is, when the joint distribution of source outputs is the product of the marginals. In current practice, independence is achieved by substrate engineering: chaotic laser arrays and microcombbased parallel chaos sources are designed to be independent at the optical substrate, and the independence claim is supported by interchannel crosscorrelation measurements at the device level. This paper argues that the next architectural direction for multisource entropy generation in criticalinfrastructure cryptography is to design for verifiable independence at the jointobservation level to build entropy architectures in which the independence of sampled noise is a measurable architectural property rather than an assumed substrate property. The position is framed as a research direction, not a specific implementation: this paper argues that the architectural problem is worth pursuing, not which apparatus solves it. The argument is positioned within the NIST SP 80090B framework, distinguished from existing singlesource thermal, singlecamera opticalchaos, and paralleloptical TRNG architectures, and connected at the architectural level to MILO's sensorreceptor subsystem.

Summary

Plain Language Summary. Cryptographic systems that protect critical infrastructure power grids, secure communications, financial transactions, defense commandandcontrol depend on a continuous supply of highquality random numbers seeded by physical noise sources. When multiple noise sources are combined, the resulting random numbers are only as strong as the independence between sources. Today, independence is achieved by building sources that are independent at the substrate level (for example, separate laser chaos channels). This paper argues that the next architectural step is to design entropy systems where independence is not assumed but measured at the jointobservation level built in, verifiable, and auditable. The argument is a researchdirection position paper, not a claim about a specific apparatus.

Key takeaways

• Treat independence as a measurable architectural property, not only a substrate assumption.
• Position NIST SP 80090B health testing as a designtime target for multisource entropy systems.
• Use MILO's sensorreceptor framing as an auditfirst pattern for future entropysource evaluation.

Concept map

Sources to follow

Use these official references as starting points for the standards context in the full paper.