Ranging-Piggyback Synchronization

Mechanism

A two-way ranging exchange between endpoints A and B records four timestamps: A's transmit time of the request, B's receive time, B's transmit time of the response, and A's receive time. The standard symmetric double-sided two-way ranging algorithm uses these four timestamps to estimate the round-trip delay and thus the distance, and it cancels the clock-offset term so that the distance estimate is robust to drift between the two endpoints. The clock-offset term that the ranging algorithm cancels is itself a measurement: it is the offset between B's clock and A's clock at the midpoint of the exchange, observable from the same four timestamps with the same precision.

Ranging-piggyback synchronization extracts that offset as a first-class credentialed observation. Each ranging exchange therefore emits a tuple — a range estimate, a clock-offset estimate, the four contributing timestamps, the credentialed identities of the two endpoints, and the lineage of the exchange — into the mesh observation log. A joint solver consumes the tuples from all exchanges in the mesh's local neighborhood and estimates a spacetime state for each endpoint: a position and a clock parameterization, jointly. The solver is structurally a least-squares or filter-bank estimator whose residuals span both the spatial and temporal axes; a single ranging exchange contributes a constraint to both, and the consistency of the joint estimate is the consistency of the same physical events viewed through both lenses.

The drift of any individual endpoint's clock is bounded by the rate at which it participates in ranging exchanges with credentialed neighbors. An endpoint that ranges frequently against a stable neighborhood holds tight time, regardless of whether it can hear an external time reference; an endpoint that loses contact accumulates drift at its oscillator's free-running rate until it next ranges, at which point the next exchange supplies a fresh offset observation and the joint solver re-integrates it. The architecture is accordingly resilient to GNSS outage, to network partition, and to deliberate jamming of conventional time-distribution channels: as long as ranging continues, time consensus continues, because they are the same observations.

The joint solution structurally enforces a consistency that separated solutions cannot. Range and time-offset observations from a single exchange are not independent measurements: they are two projections of the same physical event, and any inconsistency between them indicates either a fault in the exchange itself, a model violation such as multipath, or an attempted spoof. When the joint solver finds that the range-only and time-offset-only residuals from the same exchange cannot be reconciled within the noise model, it surfaces a diagnostic event naming the exchange, the contributing endpoints, and the residual signature, and either down-weights or excludes the exchange from the current update under a policy declared by the deploying operator. Conventional architectures running separate solvers cannot perform this cross-check because the two solvers do not share the underlying timestamp tuples; they observe only the post-processed range and post-processed time and have no way to attribute a discrepancy to a specific exchange.

Operating Parameters

Ranging modalities differ in their inherent timestamp precision, which sets a floor on the achievable per-exchange clock-offset uncertainty. Impulse-radio UWB at 3.5–6.5 GHz typically resolves timestamps to the order of one hundred picoseconds, yielding centimeter-class range and sub-nanosecond clock-offset observations per exchange. Chirp-spread-spectrum ranging at 2.4 GHz operates in the few-nanosecond regime. Acoustic ranging operates in the microsecond regime but at correspondingly shorter distances. The architecture is parameterized over the modality's timestamp precision and over the ranging cadence, which together determine the effective bandwidth of the time-consensus service: a UWB mesh ranging at twenty hertz per pair holds endpoints within a sub-nanosecond joint estimate continuously, while a sparser acoustic mesh holds microsecond-class consensus suitable for industrial coordination.

The joint solver is parameterized over the mesh topology and the credentialing policy. A densely connected mesh allows a hard cross-check of every observation against many alternative paths; a sparser mesh accepts a longer latency before drift can be re-bounded but still benefits from the unification of the two synchronization paths. The credentialing policy determines which endpoints' observations are admitted to the solver: a defense mesh may admit only platforms with current credential and a recent neighbor-confirmed presence, while a civil mesh may admit any credentialed sensor in good standing. The solver emits per-endpoint spacetime estimates with associated covariances, and the covariances are the basis on which downstream consumers — coordination admissibility, sensor fusion, event timestamping — decide whether the present estimate is fit for their purpose.

Oscillator quality at the endpoints sets the second-order parameter that governs how gracefully an endpoint coasts between ranging exchanges. A temperature-compensated crystal oscillator typical of consumer-grade UWB hardware drifts at a rate of roughly one part in a million, which corresponds to a microsecond per second of free-running drift; an oven-controlled crystal oscillator improves this by two orders of magnitude; a chip-scale atomic clock improves it by another two. The architecture admits all three regimes by tuning the freshness requirement and the solver's process-noise model to the deployed oscillator class, so the same mechanism delivers nanosecond-class consensus at consumer cost in a dense mesh and nanosecond-class consensus through extended outages in a sparse mesh equipped with better holdover hardware.

Alternative Embodiments

In a UWB-mesh embodiment, fixed and mobile UWB endpoints in a building, factory, or vehicle ranging mesh share both position and time consensus through their normal ranging traffic, with no separate NTP or PTP service deployed and no GNSS dependency required for time. In a tactical-mesh embodiment, ground vehicles, dismounts, and aerial platforms exchange ranging packets in a contested-RF environment where GNSS is denied or spoofed; the joint estimate degrades gracefully as connectivity thins rather than failing abruptly when the time service loses its reference.

In an industrial-automation embodiment, robotic cells and conveyor instruments range against fixed anchors at high cadence to provide both position and time at sub-microsecond precision, supporting tightly coupled motion control without a deterministic Ethernet time-distribution overlay. In a surgical-robotics embodiment, intra-room ranging among instrument tags, navigation markers, and the imaging system delivers a joint spacetime estimate fine enough for sub-millimeter motion correlation, with the time consensus carried entirely on the same UWB infrastructure that delivers the position.

In a hybrid embodiment, ranging-piggyback time consensus runs alongside an external time reference such as GNSS or PTP, with the external reference treated as an additional observation in the joint solver rather than as the authoritative source. The mesh accepts the external reference when it is healthy and credentialed and falls back to pure ranging-piggyback consensus when the reference fails or is rejected by its own admissibility check, with the transition a continuous adjustment of weights inside the solver rather than a service-level cutover.

Composition

Ranging-piggyback synchronization composes naturally with the mesh's coordinate consensus because they are produced by the same observations. It composes upward with proximity-grounded coordination, supplying both the position attestations and the time bounds that coordination admissibility requires from a single attested record. It composes with credentialed event logging by stamping events into the joint spacetime frame, so that the temporal ordering of events recorded by different endpoints is consistent with the geometric arrangement of those endpoints to within the joint estimate's covariance. It composes with conventional time services through the hybrid embodiment described above, presenting an external face compatible with NTP or PTP consumers while internally operating on the unified mesh fabric.

The architecture also composes with sensor-fusion pipelines that consume both position and time as inputs. A perception or navigation stack typically requires that observations from heterogeneous sensors be both spatially registered and temporally aligned before fusion; here, the same joint estimate provides both, with consistent covariance, eliminating the class of fusion failures that arise when independent position and time estimates disagree at the millisecond-and-decimeter level relevant to fast-moving platforms. The credentialed lineage of each contributing observation propagates into the fused product, so that downstream consumers that filter by source credential — a regulator that admits only certain sensor classes, a safety case that requires diversity in observation sources — apply their filters uniformly to the joint estimate.

Prior Art Distinction

IEEE 1588 (PTP) and NTP distribute time over packet networks but make no use of ranging information and produce no position estimate. GPS and GNSS deliver joint time and position from a satellite reference but require sky visibility and an unjammed signal, neither of which is reliable in many of the operational environments contemplated here. Symmetric double-sided two-way ranging in IEEE 802.15.4z and similar UWB standards is structured precisely to cancel the clock-offset term so that the position estimate is drift-robust, and conventional implementations discard the cancelled term. White Rabbit and similar precision time-transfer protocols carry time over fiber with picosecond precision but again do not produce position. The disclosed architecture differs in that the clock-offset term that ranging cancels is recovered as a credentialed observation, integrated with the range observation in a joint solver, and used to bound mesh time drift through the same exchanges that bound mesh position drift, with the resulting joint estimate carrying credentialed lineage end-to-end.

Disclosure Scope

The disclosure covers mesh architectures in which ranging exchanges produce both range and clock-offset observations as credentialed first-class records; in which a joint solver integrates both classes of observation into a unified spacetime estimate per endpoint with associated covariance; in which mesh time drift is bounded by ranging cadence and topology rather than by an external time reference; and in which the joint estimate is consumed by upstream services such as coordination admissibility and event logging. The disclosure includes UWB, chirp-spread-spectrum, and acoustic ranging modalities; tactical, industrial, surgical, and civil embodiments; and hybrid operation alongside GNSS, NTP, and PTP. It includes the graceful degradation behavior under loss of external reference, the credentialing policy for admitted observations, and the structural composition with proximity-grounded coordination and credentialed event logging.