Mechanism
The mesh-derived time primitive comprises a plurality of clock-maintaining mesh agents, each maintaining a local clock with governance-policy-characterized drift properties. A clock-model learning mechanism refines per-agent drift characterizations through governance-credentialed training. The structural commitment is that each agent carries a per-agent drift characterization that is refined under governance credentialing rather than a static, factory-fixed correction.
A drift-compensation mechanism continuously compensates local-clock drift through fresh synchronization exchanges. The governance-credentialed inter-agent time-synchronization mechanism produces time-synchronization observations between participating mesh agents, and the cooperative time-estimation engine determines agent time-offsets through combination of those synchronization observations and admitted temporal-anchor contributions. Where direct-anchor synchronization is insufficient, a transitive time-propagation extender produces agent time-offsets through neighbor references.
A time-uncertainty propagator propagates synchronization uncertainty through the temporal graph, producing per-agent time-uncertainty estimates. The cooperative time-estimation engine consumes these per-agent uncertainty estimates when combining synchronization observations and anchor contributions, so that better-characterized agents and lower-uncertainty observations carry correspondingly greater influence in the resolved frame. An adversarial-time rejection mechanism rejects spoofed, injected, or otherwise inadmissible time-synchronization observations.
Each synchronization exchange, anchor admission, time-estimation event, and rejection event is recorded in the governance-chain lineage field by a time-lineage recorder. Because the per-agent drift characterizations are refined under governance-credentialed training and the synchronization record is preserved in the lineage chain, a downstream audit can reconstruct the synchronization chain that produced an agent's time from the governance lineage.
Governance and Admissibility
The clock-model learning mechanism refines per-agent drift characterizations through governance-credentialed training. Training proceeds under the training-governance regime, so the refinement of an agent's drift characterization is a governed activity whose policy parameters are deployment choices rather than fixed constants. The disclosure states the structure of this refinement without committing to a particular cadence, horizon, or buffer size.
Admission of an agent's time-synchronization observations is governance-credentialed. The composite admissibility evaluator applies governance-policy-defined rules to admitted observations and anchor contributions, and the propagated per-agent time-uncertainty estimate accompanies each contribution so that the cooperative time-estimation engine can weight contributions according to their estimated uncertainty.
An adversarial-time rejection mechanism rejects spoofed, injected, or otherwise inadmissible time-synchronization observations. Rejection events, like admissions and time-estimation events, are recorded in the governance-chain lineage field, so that an admissibility determination against a contributing agent is reconstructible from the lineage record.
Alternative Embodiments
The synchronization modality is open. The governance-credentialed inter-agent time-synchronization mechanism produces time-synchronization observations through at least one of a plurality of synchronization modalities, and the cooperative time-estimation engine combines those observations with temporal-anchor contributions. The structural commitment is that synchronization is governance-credentialed and that drift is compensated continuously through fresh synchronization exchanges, not the use of any one modality.
Temporal-anchor contributions are admitted through a governance-credentialed temporal-anchor observation admission interface. An evidential-fusion mechanism combines mesh-derived time with externally-sourced time, including satellite time, network time, an atomic reference, or any external source, through the composite admissibility evaluator. Where no anchor observations are available, an anchor-less temporal bootstrap mechanism produces a relative-only temporal frame.
A time-frame federation mechanism aligns independently-maintained temporal frames, producing governance-chain-preserving federation with cross-authority translation. Independently-maintained frames can therefore be aligned without a shared master clock, each frame contributing under its own authority credential.
A relativistic-consistency evaluator applies where relativistic effects are significant. In the unified spacetime composition, this evaluator participates so that the four-dimensional observation emitter producing observations carrying (x, y, z, t) remains consistent under conditions where relativistic effects are not negligible.
Composition
Per-agent drift compensation composes directly with the anchor-less temporal bootstrap mechanism. When no anchor observations are available, the bootstrap mechanism produces a relative-only temporal frame; continuous drift compensation through fresh synchronization exchanges maintains per-agent offsets within that relative frame between synchronization events. When anchor contributions are admitted, the same drift-compensation mechanism continues to operate, so that the resolved frame is preserved across gaps between anchor observations.
Per-agent drift compensation also composes with the mesh-derived coordinate primitive through a joint admission interface that admits combined range-and-synchronization observations, wherein ranging exchanges produce jointly-optimized spatial and temporal estimates. A joint uncertainty propagator produces per-agent spacetime uncertainty, and a four-dimensional observation emitter produces observations carrying (x, y, z, t) with joint uncertainty within a single joint spatial-temporal graph.
A further composition arises with the governance-credentialed timestamp attestation interface. The attesting agent's mesh-derived time value and estimated time uncertainty, both shaped by its per-agent drift characterization, are carried in each timestamp observation alongside the agent's authority credential and cryptographic signature. A downstream audit can reconstruct the timestamp's lineage, the synchronization chain producing the attesting agent's time, the composite admissibility evidence, and the authority-credential chain from the governance lineage.
Prior-Art Boundary
The spec distinguishes the mesh-derived time primitive from prior time-distribution architectures. Prior satellite-derived time services operate through broadcast signals from centrally-operated constellations whose denial precludes timing, whereas the present primitive produces time bearings from cooperating mesh agents without dependence on satellite availability. Prior network-time-protocol systems are client-server hierarchical and depend on centralized stratum-1 time servers, and prior precision-time-protocol systems require hierarchical master-slave configuration with dedicated grandmaster clocks, whereas the present primitive is master-less and self-organizes through mesh agents.
Prior blockchain timestamp protocols timestamp at block-commit granularity, prior trusted-timestamp-authority systems centralize timestamp issuance at a single authority, and prior chip-scale atomic clocks provide high-precision time without distributed consensus, whereas the present primitive produces continuous governance-credentialed, multi-authority timestamps and combines precision clock sources with distributed mesh consensus. Within this architecture, the per-agent drift characterization is refined through governance-credentialed training rather than treated as a private, uncredentialed property of a local clock, and its propagated time-uncertainty estimate enters the cooperative time-estimation engine under composite admissibility.
Disclosure Scope
This article describes the per-agent drift characterization and clock-model learning mechanism disclosed in U.S. Provisional Application No. 64/049,409. Timekeeping under contested conditions gains time bearings from cooperating mesh agents without dependence on satellite availability, and master-less operation removes the dependence on a centralized stratum-1 server or dedicated grandmaster clock. Long-endurance deployments gain the continuous drift compensation that maintains per-agent offsets through fresh synchronization exchanges across gaps between anchor observations.
The scope of the disclosure encompasses the clock-model learning mechanism refining per-agent drift characterizations through governance-credentialed training, the drift-compensation mechanism continuously compensating local-clock drift through fresh synchronization exchanges, the cooperative time-estimation engine combining synchronization observations and anchor contributions, the transitive time-propagation extender producing offsets through neighbor references, the time-uncertainty propagator producing per-agent time-uncertainty estimates, the adversarial-time rejection mechanism, the anchor-less temporal bootstrap mechanism, the time-frame federation mechanism, and the time-lineage recorder. The disclosure further encompasses the composition with the mesh-derived coordinate primitive through combined range-and-synchronization observations and with the governance-credentialed timestamp attestation interface.
Embodied properties of the disclosed architecture include the refinement of per-agent drift characterizations under governance-credentialed training rather than a static factory correction, the propagation of per-agent time-uncertainty into the cooperative time-estimation engine, the master-less cooperative-consensus structure, and the reconstructibility of each timestamp's derivation chain from the governance lineage. These properties are presented as part of the architectural disclosure rather than as performance claims.
