Satellite corrections that come and go: a positioning service that adapts in seconds

…

…

The distinction between the four service tiers described in the study maps directly onto the physical content of the correction products that a PPP-RTK service can broadcast. In the absence of any service-side products, a receiver operates in standard single point positioning mode, relying only on broadcast ephemerides and code observations, which bounds its accuracy at the meter level. Once precise satellite orbit and clock products become available, the same receiver can upgrade to PPP without any change to its hardware, because PPP fundamentally depends on replacing coarse broadcast ephemeris information with network-derived estimates of satellite state. When fractional-cycle biases are additionally supplied so that ambiguities can be fixed to integers, the PPP-AR tier becomes accessible, collapsing the ambiguity-related uncertainty that otherwise limits PPP accuracy. Finally, when atmospheric corrections covering ionospheric and tropospheric delays are delivered, the PPP-RTK tier can be realized, in which the atmosphere acts as the spatial bridge that allows network-referenced ambiguity information to be exploited at a roaming user location. This layered structure explains why the study frames positioning capability as a function of correction-product availability rather than as a property of the receiver alone.

The undifferenced and uncombined formulation underlying the method deserves particular attention because it determines how gracefully a user terminal can move between tiers. Earlier double-difference RTK implementations required reference receivers and differencing operations that tied the user solution to a specific baseline, whereas the UDUC model retains all original observation equations and treats each observable as an independent measurement with its own stochastic properties. In such a formulation, atmospheric corrections and satellite-side corrections enter as external constraints on specific parameters rather than as transformations of the observation set. Consequently, when a particular product becomes unavailable, the corresponding parameters can simply be estimated or constrained differently, and the filter continues operating in a degraded but coherent mode. This property is what makes a genuine multi-tier architecture possible on a single embedded device, since the same estimation engine can represent SPP, PPP, PPP-AR, and PPP-RTK as different constraint configurations rather than as separate software stacks.

The convergence hierarchy reported in the experiments reflects well-understood information content differences among the tiers. PPP converges slowly, on the order of the 15.98 minutes observed, because the receiver must separate position, clock, tropospheric delay, and real-valued ambiguity parameters from noisy code and carrier-phase observations alone, and the geometry of the satellite constellation changes only gradually. PPP-AR shortens this to 4.27 minutes because fixing ambiguities to integers removes a dominant source of uncertainty and effectively sharpens the information available from the phase measurements. PPP-RTK achieves convergence in 7.08 seconds because the atmospheric corrections supplied by the reference network pre-resolve most of the spatially correlated error budget, leaving only local effects and receiver-side quantities to be estimated. The roughly two-order-of-magnitude difference between the PPP and PPP-RTK convergence times quantifies how much of the classical PPP convergence problem is attributable to atmospheric uncertainty rather than to satellite orbit and clock error.

The hot-restart experiment carries practical significance beyond its headline number. In many intelligent applications, such as automated machinery operating in ports or agricultural equipment moving between field sections, brief signal outages caused by obstructions, canopy, or infrastructure are routine rather than exceptional. A system that requires full re-convergence after each interruption would spend substantial fractions of its operating time in a degraded state. The reported 4.53 second average recovery time across 40 forced restarts indicates that the fixed PPP-RTK solution can be re-established almost immediately when the receiver retains recent state information, which is the defining characteristic of a hot restart. This recovery behavior, combined with the tiered fallback structure, suggests a service design in which temporary loss of atmospheric products would drop the user to PPP-AR or PPP operation rather than to a cold start, preserving continuity of centimeter- to decimeter-level service during partial outages.

The choice of an embedded GNSS device as the user terminal strengthens the engineering relevance of the results. Real-time PPP-RTK performance is often demonstrated on desktop-class platforms with generous computational resources, which can mask the constraints faced by fielded receivers. Embedded processors impose limits on filter complexity, memory footprint, and communication bandwidth, and the fact that the full tier hierarchy was realized within such constraints indicates that the UDUC-based multi-tier model is computationally tractable for practical deployment. The end-to-end character of the experiment, spanning reference-network processing, product generation, dissemination, and user-side estimation, also matters because latency and formatting in the correction stream directly affect achievable accuracy, particularly for the atmosphere-dependent PPP-RTK tier, where corrections degrade as they age.

Viewed in the broader context of positioning, navigation, and timing services, the study addresses a gap that has grown as high-precision GNSS has moved from surveying into mass-market and safety-relevant applications. Traditional RTK and network RTK services were typically designed around a single performance contract, and users experiencing degraded conditions simply lost service. A tiered architecture instead treats accuracy as a graded, availability-dependent quantity, which aligns better with the heterogeneous requirements of modern applications where different task stages tolerate different accuracy levels. The empirical characterization of each tier provides the quantitative basis needed for such adaptive service design, allowing a system to make informed decisions about which tier to request or fall back to under given communication and correction-product conditions.

Several aspects of the reported performance merit interpretation. The 0.72 centimeter horizontal and 1.78 centimeter three-dimensional RMS accuracy at the PPP-RTK tier is consistent with the expectation that fixing ambiguities with network atmospheric support yields accuracy comparable to short-baseline RTK while retaining wide-area coverage. The meter-level SPP result serves as a useful baseline, confirming that the tier hierarchy spans the full range from code-only positioning to phase-fixed network-aided positioning within one coherent framework. The ordering of convergence times across tiers, with PPP-AR faster than PPP yet far slower than PPP-RTK, matches the theoretical information hierarchy in which atmospheric corrections contribute more to rapid convergence than integer ambiguity resolution alone.

The service-oriented framing also has implications for how correction-product infrastructure might evolve. Because the user-side tiers are defined by which products are received and utilized, a service provider could offer differentiated products over separate channels, allowing terminals with different communication capabilities or subscription levels to access correspondingly different tiers. The same structure supports graceful degradation: if a dissemination link for atmospheric products fails, affected users fall back to PPP-AR rather than losing high-precision service entirely. This availability-aware perspective reframes GNSS service reliability as a layered property, in which continuity is preserved by stepping down through tiers rather than by binary availability of a single premium product.

It is worth noting that the multi-tier organization does not alter the classical definitions of SPP, PPP, PPP-AR, or PPP-RTK as positioning methods; rather, it provides a service-level interpretation in which these methods correspond to specific subsets of available correction products. This distinction matters for interoperability, because terminals implementing the multi-tier model remain compatible with conventional product streams and can participate in existing service ecosystems while gaining the ability to adapt their configuration to whatever products are actually available at a given moment. The formal basis established for this adaptive configuration could inform future standardization efforts for high-precision GNSS services, particularly as correction delivery over mobile networks and satellite links continues to diversify.

The study’s emphasis on near-Earth PNT applications situates the work within a growing recognition that positioning is becoming a continuous sensing service rather than an occasional measurement task. In such a service model, metrics beyond static accuracy, including convergence time, recovery time after interruption, and behavior under varying product availability, become primary design targets. The empirical results reported here, spanning all of these dimensions within a single real-time system, offer a template for how continuity-oriented GNSS services can be specified, implemented, and validated, and they demonstrate that correction-product availability can serve as a practical organizing principle for the next generation of high-precision positioning architectures.

Subject of Research: A multi-tier PPP-RTK positioning method based on correction-product availability

Article Title: A multi-tier PPP-RTK positioning method based on correction-product availability

Article References: Hu, Y., Cheng, F., Shen, P., Gao, X., Sun, W., Hui, Z., & Lu, X. (2026). A multi-tier PPP-RTK positioning method based on correction-product availability. Frontiers in Astronomy and Space Sciences, 13, Article 1973781. https://doi.org/10.3389/fspas.2026.1973781

Image Credits: AI Generated

DOI: 10.3389/fspas.2026.1973781

Keywords: multi-tier, PPP-RTK, positioning, method, based, correction-product, availability, scientific research