Positioning & Satellite Links

01 / 10

An accuracy headline fails at the window

The receiver works on the bench. Why does its timing output move when the gateway goes beside a window?

The fictional condition-monitoring gateway needs a useful clock. The roof removes some directions; the window admits direct and reflected signals; the cable adds delay. A receiver accuracy headline usually describes a particular signal set, estimator and test population. The installed product has to establish those conditions again. Even a signal-in-space performance commitment excludes important user-side errors. [GPS-SPS]

This lesson asks how timing, geometry, propagation, antenna visibility and acquisition determine the result. GNSS means global navigation satellite system. Here it supplies ranging and timing information to a receiver. A satellite messaging terminal additionally transmits reports through an operator’s service. Keep those jobs separate from the beginning.

Two new illustrative variants; no earlier visit or saved workbook required
Variant	Chosen engineering question	Evidence still missing
Gateway GNSS timing	Can a calibrated 1 PPS output stay within ±100 ns for ≥99% of scheduled 1 s epochs in each 24 h placement stratum?	Receiver/firmware mode, independent time reference, cable/antenna delay, startup and holdover, sky view and bias population.
Remote satellite fallback	Can ≥99% of unique complete 32-byte reports reach host D0 within 900 s, one generated every 600 s?	Exact terminal/SBD interface, country/service conditions, visibility, queues, transmit/receive state energy and host acknowledgement.

These are chosen experimental goals, not measured results. The prior generic 2.450 GHz QPSK waveform remains a teaching baseline; it is neither a GNSS signal nor a satellite service mode. D0 names the application boundary. The fallback’s 900 s deadline is a named variant, not a revision of the earlier cellular 60 s report or 30 s command goals.

Think about itIf all five received signals are strong, is a good position/time estimate assured?

Answer

No. Their directions can still constrain the same combination of vertical position and clock error. Strong biased reflections can also mislead an estimator. Predict which problem changes when you move the antenna, versus when you add gain.

Prepare the evidence, then read the geometry

Review 07.1’s source hierarchy, 05.3’s noise/plane ledger, and the 06.1 antenna definitions. Installed-antenna evidence belongs to 06.3; channel models belong to 06.5. Measurement execution and market authorization remain later-path work.

Decision: replace the headline with a time-error statistic at a named output, plus a placement and state population. First find what the receiver must estimate.

02 / 10

Pseudorange contains an unknown clock

Why can a timing measurement contain both a distance and an unknown clock?

A receiver aligns a local code replica with the received navigation signal. The apparent travel time uses two clocks, so multiplying by propagation speed does not produce geometric distance alone. We call it pseudorange. The local model below assumes the stated satellite-clock, ephemeris and correction treatment has already been applied. [ESA-BOOK]

$\rho _{i} = \Vert r - r_{i}\Vert + c\delta t + \epsilon _{i}$ ρᵢ is corrected pseudorange in metres; r and rᵢ are receiver and satellite positions in metres in one frame; c = 299792458 m/s exactly. δt is residual receiver clock bias in seconds. εᵢ contains residual propagation, clock, measurement and model errors in metres; it is not automatically independent white noise.

Linearize near an approximate receiver position. Use a right-handed local East/North/Up (ENU) frame: East × North = Up. Azimuth A increases clockwise from North; elevation e is above the horizon. Controls use degrees; trigonometry uses radians. A ray’s unit vector points from receiver to satellite.

$u = [\cos (e) \sin (A), \cos (e) \cos (A), \sin (e)]$ uE, uN and uU are dimensionless. At zenith u = [0,0,1]; azimuth has no physical meaning.

$x = [\delta E, \delta N, \delta U, b]^{T}\qquad b = c\delta t\qquad H_{i} = [-u_{E}, -u_{N}, -u_{U}, 1]$ All four state coordinates are metres; all four H columns are dimensionless. Linearized range residuals satisfy Δρ ≈ Hx + ε. Moving toward a satellite shortens the path, explaining the minus signs. Increasing b adds the same range to every row.

Read two rows before solving anything
Ray	Unit ENU vector	Measurement row Hᵢ	Physical check
S01: zenith	[0, 0, 1]	[0, 0, −1, 1]	A +1 m Up displacement shortens range by 1 m; +1 m clock state restores it.
S02: North, 30°	[0, √3/2, 1/2]	[0, −√3/2, −1/2, 1]	The same +1 m Up displacement shortens range by only 0.5 m.

Common misconceptionThe clock column should contain c even when b is labelled metres.

That would mix state definitions. Use coefficient 1 for b in metres; use c only if the state is δt in seconds. The clock covariance converts back to time by dividing its standard deviation by c.

Go deeperWhat has already been removed from this simplified equation?

Real processing needs transmit-time satellite coordinates, satellite-clock and relativistic corrections, Earth rotation during flight, signal-specific group delays and propagation models. Combining constellations may require inter-system clock-bias states. The explorer has one common clock and a frozen synthetic sky; it performs no navigation solution or ephemeris propagation.

Decision: freeze the state and units before counting equations. Four unknowns need independent information, not merely four rows.

03 / 10

Geometry transforms ranging error

What does the fifth direction contribute that four satellites at one elevation cannot?

Freeze S01 at zenith and S02–S05 at 30° elevation, azimuths 0°, 90°, 180° and 270°. They are teaching IDs, not real constellation identifiers. The default mask is 15°, with no blockage and all five rows used. Assume independent, zero-mean range errors with common standard deviation σ = 3 m.

$Q = (H^{T}H)^{-1}\qquad \operatorname{Cov}(x) = \sigma ^{2} Q$ This covariance requires full column rank and the stated error model. It describes local uncertainty under those assumptions, not observed product error. The operational solver uses SVD of H, not an unchecked inverse of HᵀH. [ESA-ERROR]

Analytic five-ray HᵀH · dimensionless · E, N, U, b
Row / column	E	N	U	b
E	1.500000000	0.000000000	0.000000000	0.000000000
N	0.000000000	1.500000000	0.000000000	0.000000000
U	0.000000000	0.000000000	2.000000000	-3.000000000
b	0.000000000	0.000000000	-3.000000000	5.000000000

Analytic Q · dimensionless · E, N, U, b
Row / column	E	N	U	b
E	0.666666667	0.000000000	0.000000000	0.000000000
N	0.000000000	0.666666667	0.000000000	0.000000000
U	0.000000000	0.000000000	5.000000000	3.000000000
b	0.000000000	0.000000000	3.000000000	2.000000000

The Up/clock block of HᵀH has determinant 2×5 − (−3)² = 1. Its inverse is [[5,3],[3,2]]; the horizontal diagonal entries invert from 3/2 to 2/3. This provides an algebraic check independent of the explorer.

$\mathrm{HDOP} = \sqrt{Q_{\mathrm{EE}} + Q_{\mathrm{NN}}}\qquad \mathrm{VDOP} = \sqrt{Q_{\mathrm{UU}}}\qquad \mathrm{PDOP} = \sqrt{Q_{\mathrm{EE}} + Q_{\mathrm{NN}} + Q_{\mathrm{UU}}}$ HDOP, VDOP and PDOP are dimensionless horizontal, vertical and position dilution of precision. TDOP = √Qbb; GDOP = √(PDOP² + TDOP²). Clock-time standard deviation = σ√Qbb/c seconds.

Checked five-ray anchors · σ = 3 m · not an accuracy specification
Quantity	Derived value
HDOP / VDOP	1.1547005383792515 / 2.23606797749979
PDOP / TDOP / GDOP	2.516611478423583 / 1.4142135623730951 / 2.886751345948129
Horizontal root-sum-variance	3.4641016151377544 m
Vertical standard deviation	6.708203932499369 m
Clock-time standard deviation	14.151926020498106 ns

Think about itRemove S01. Four satellites remain: is the four-state solution still observable?

Answer

No. Every remaining Up coefficient is −0.5; every clock coefficient is 1. The perturbation [0,0,2,1] metres changes no predicted range because −0.5×2 + 1 = 0. Rank is 3. Four-state DOP and error scales are unavailable, not zero or an enormous invented finite number.

Think about itWhat changes if those same four low rays have an independently exact receiver clock?

Answer

The clock column disappears. The spatial normal matrix is diag(3/2,3/2,1), full rank 3; spatial Q is diag(2/3,2/3,1). HDOP remains 1.154700538, VDOP becomes 1, and PDOP is 1.527525232. Four-state TDOP/GDOP do not apply. A stable oscillator does not provide this exact clock constraint.

Common misconceptionDOP × sigma is a 95% accuracy guarantee.

σ×HDOP is the square root of the sum of two variances. It is not one axis’s standard deviation, circular error probable (CEP), a 95% radius or a protection level. The distribution, anisotropy, biases, probability definition and population matter; multiplying every number by two does not solve that problem.

Go deeperRank, conditioning and additional information

The explorer retains singular values strictly greater than 10⁻¹⁰ of the largest. A full-rank condition number above 10⁸ is labelled near singular and finite error scales are withheld. These are numerical teaching policies, not integrity criteria. A finite independent positive-information row cannot worsen a full-rank covariance with fixed states and unchanged noise assumptions. Adding a state or changing the noise law is a different comparison. A finite clock prior needs a separate covariance model and is deferred.

Decision: request a wider spread of directions or justify additional independent information when geometry fails. Then ask which source says what about the real receiver.

04 / 10

Read signal-in-space and receiver evidence separately

Which document could support a signal definition, and which could support the window installation?

A signal interface tells a receiver designer what arrives from space. A performance document sets a service boundary and conditions. A receiver specification describes one implementation. Only a condition-matched installed test addresses the product. Follow all four layers without transferring a number across them.

Source chain checked on 2026-09-08; signal definitions versus product evidence
Layer	Pinned source / scope	What it does not establish
GPS signal interface	GPS L1 C/A: IS-GPS-200N, PDF dated 2022-08-01 (register 2022-08-22), §3.3.1.1; nominal L1 1575.42 MHz. IRN-003: 2024-01-19. [GPS-IS] [GPS-IRN]	No product acquisition time, antenna placement or timing-error statistic.
GPS amendment status	Register also lists IRN-IS-200N-004, 2026-06-16, civil ISM formats. Its detailed text could not be retrieved; implementation impact is unknown. [GPS-INDEX]	An amendment listing proves existence/scope, not an unexamined clause or a safety-integrity design.
GPS service performance	SPS Performance Standard, 5th ed., April 2020, listed in effect; §2.4.5 excludes important user errors. [GPS-SPS]	No automatic inclusion of receiver, multipath, user antenna or residual propagation errors.
Galileo comparison	E1-B/E1-C; OS SIS ICD 2.2 (Nov 2025), OS SDD 1.3 (Nov 2023), plus SNGU 2026002 issue 1.0 (22 Jan 2026). [GAL-INDEX] [GAL-ICD] [GAL-SDD] [GAL-NOTICE]	No silent GPS/Galileo time-scale combination. The SNGU updates NAGU notice definitions, not this toy geometry.
Receiver / installed gateway	Exact SKU, firmware, tracking/aiding/timing mode, antenna and cable calibration, PPS population and independent reference.	Currently unknown. No measured receiver or field result is embedded in this lesson.

For any “95%” claim, ask: 95% of which epochs, sites, satellites, users or trials; over what interval; conditional on which healthy/valid state; and with which excluded errors? A global-average signal-in-space statistic is not the same population as the gateway’s signed PPS error during 24 hours under a roof.

Common misconceptionA newer interface document proves that my receiver supports every new feature.

Implementation, firmware and service availability each need evidence. A superseded document can still explain a deliberately selected legacy mode, but a new product needs its applicable amendment chain reviewed. An overdue review means recheck; it does not automatically erase the old fact.

Go deeperTurn the amendment into an engineering action

For GPS, the source owner must obtain IRN-004, identify changes affecting the chosen signal and firmware, and record explicit applicability before closing the product trace. For Galileo, record the revised notice-category interpretation from SNGU 2026002 and the actual health/status inputs used. Do not draw S01–S20 as today’s sky.

Decision: accept the sourced definition within its scope and leave receiver/installation claims open. Next separate signal quality from the receiver’s ability to find and retain it.

05 / 10

C/N₀, acquisition and tracking answer different questions

A receiver tracks a faint signal after startup. Does that prove it could acquire it from a cold start?

Acquisition searches code delay and frequency uncertainty; tracking follows an already located signal. Time to first fix (TTFF) additionally depends on the needed navigation data, initial position/time knowledge and receiver state. For a timing application, record time to valid time separately. Longer integration can help, but data transitions and phase uncertainty limit coherence. [ESA-BASEBAND]

$\frac{C}{N_{0}} [\mathrm{dB}-\mathrm{Hz}] = 10 \log _{10}[\frac{\frac{C}{N_{0}}}{1 \mathrm{Hz}}]$ C is carrier/signal power in watts; N₀ is noise power density in W/Hz at the same reference plane. The ratio has units hertz; dB-Hz is its logarithmic representation. It is not carrier power in dBm. Here the synthetic annotation is at R1-RX after a declared reference antenna/front end.

$\frac{C}{N} [\mathrm{dB}] = \frac{C}{N_{0}} [\mathrm{dB}-\mathrm{Hz}] - 10 \log _{10}(\frac{B_{n}}{1 \mathrm{Hz}})$ Bn is the specified effective noise bandwidth in hertz. For 45 dB-Hz and an assumed 1 MHz Bn, C/N = −15 dB before the relevant processing. Do not substitute channel width or an acquisition threshold for Bn.

$\mathrm{SNR}_{\mathrm{int}}=(C/N_0)_{\mathrm{linear}}T_{\mathrm{coh}}$ Normalized coherent matched-integration convention: Tcoh seconds, constant phase/data, white noise. At 45 dB-Hz, Tcoh = 0.001 s gives 15 dB; 0.010 s gives 25 dB. This is not an unspecified integrate-and-dump bandwidth convention.

Three different receiver questions
Question	Evidence required
Can it acquire?	Cold/aided state; code/Doppler search uncertainty, integration/detection policy, false alarm and acquisition probability, data age, temperature and dynamics.
Can it track?	Loop/estimator mode, dynamics, oscillator quality, outages, integration and data/secondary-code handling.
When is time valid?	Enough decoded/corrected time information, time scale, PPS validity flag, receiver configuration and calibrated output delay.

The signal fixture starts each ray at an illustrative 45 dB-Hz. Flat relative gain is g = 1. The optional shape g(e) = sin²(e) is a power factor, so its adjustment is 10log₁₀g dB. It is not an actual GNSS antenna pattern. With extra offset Δ, C/N₀ᵢ = 45 + 10log₁₀gᵢ + Δ. A null has no finite C/N₀; the model never hides a floor.

Think about itAttenuate every ray by 6 dB without excluding any. Does HDOP change?

Answer

No. H and unweighted DOP are unchanged. Under the deliberately assumed law σᵢ = σref×10^((45−C/N₀ᵢ)/20), range standard deviations multiply by 10^0.3 = 1.99526231496888 and weighted covariance by 10^0.6 = 3.9810717055349722. The equal-σ geometric result remains a separate comparison.

Common misconceptionCrossing 30 dB-Hz means this receiver loses acquisition.

The explorer’s 30 dB-Hz marker is an authored risk note. All visible rows remain usable by default. Only the explicitly enabled quality exclusion changes the used set; it remains a hypothesis, not a real acquisition or tracking limit.

Go deeperWhy a pilot is not unlimited coherent integration

Removing navigation data bits does not remove secondary codes, residual Doppler, dynamics or oscillator phase error. The signal/receiver combination must establish which transitions are wiped off or synchronized. Aided and cold-start results need separate trials.

Decision: request the receiver’s acquisition/tracking conditions and average C/N₀ over a named interval in the field log. Now combine the antenna hypothesis with the frozen geometry.

06 / 10

The installed antenna changes signal and timing

The antenna has low return loss. What else can change the time that reaches the application?

Low reflection does not prove efficient reception or useful sky coverage. Directional gain and efficiency select which rays contribute. Polarization mismatch changes received power. The antenna phase centre locates the effective spatial reference; group delay describes envelope delay and can vary with frequency, direction and hardware. The code/timing observable is sensitive to delay, not merely match. Reuse the installed-antenna and propagation evidence from Paths 06.3 and 06.5.

Galileo’s interface specifies RHCP transmission. The installed receiving antenna’s polarization and orientation response still need checking; a screen drawing cannot establish handedness without its propagation/view convention. This explorer does not calculate a polarization pattern. [GAL-ICD]

Antenna-to-timestamp plane ledger
Boundary	Required transformation / observation
S0 → R2 antenna-feed plane	Direction/polarization, effective phase centre, gain/efficiency and mismatch convention. Record mounting, roof/window and nearby reflectors.
R2 → active front end → R1-RX	LNA/filter/cable gain/loss and added noise at one plane; group delay and thermal variation separately. The 45 dB-Hz fixture already includes the declared reference front end.
R1-RX → R3 decisions	Acquisition/tracking/estimator processing and validity. R3 is never an antenna input.
Timing output → D0 timestamp	PPS cable/instrument reference, time scale and output calibration; application timestamp formation is a further transformation.

$\Delta R = c\Delta t = 299792458 \times 10^{-8} = 2.99792458 m$ A 10 ns uncompensated antenna/cable/system timing bias has this one-way range-equivalent value. It does not mean the inferred horizontal position is displaced by exactly 2.99792458 m; the estimator and bias pattern determine that mapping.

Think about itWith the elevation-gain shape, S01 is strongest. What happens when the overhead path is blocked?

Answer

The four 30° rays are 6.020599913 dB below S01 before extra offset. Removing S01 also removes the distinct elevation: the four-state model becomes rank deficient. Restoring amplitude to those four rays cannot separate Up from clock. Placement must restore information, not just power.

Interactive · local synthetic snapshot · Class 1

Geometry, C/N₀ & Visibility Explorer

Predict which information changes, then apply one controlled experiment. The signal marker, geometric visibility and chosen measurement rows are separate. No live satellites or receiver test results are loaded.

Guided comparison · restores every model input

Sequence: reveal the analytic five-row check → rotate → remove S01 → restore and attenuate → inspect the risk marker → explicitly exclude weak rows → compare an independently exact clock. Presets reset selection, masks, sectors, gain, noise and clock together.

Geometry and range-noise assumption

Global azimuth rotation (°)0–359; step 1; default 0. Clockwise from North; exact pool angles remain internal.

Elevation mask (°)0–60; step 1; default 15. Equality is visible; no atmospheric model is inferred.

Reference range standard deviation σref (m)0.3–30; step 0.1; default 3. Independent zero-mean errors only; not a product specification.

Independently exact receiver clock?Yes removes b from the state. A stable oscillator is insufficient; finite priors are deferred.

Simplified obstructions

Azimuth intervals are [start,end), including wrap-around. Equal endpoints mean empty. Zenith ignores azimuth-only blockage. Up to three enabled sectors; defaults all off, 0°/0°.

Sector 1

Use sector 1?

Sector 1 start azimuth (°)0–359°; step 1°; default 0°.

Sector 1 end azimuth (°)0–359°; step 1°; default 0°.

Sector 2

Use sector 2?

Sector 2 start azimuth (°)0–359°; step 1°; default 0°.

Sector 2 end azimuth (°)0–359°; step 1°; default 0°.

Sector 3

Use sector 3?

Sector 3 start azimuth (°)0–359°; step 1°; default 0°.

Sector 3 end azimuth (°)0–359°; step 1°; default 0°.

Block all azimuths?Disable individual sectors before enabling this full-azimuth alternative.

Block the zenith / overhead ray?The explicit overhead control can remove S01; independent of sector rules.

Signal annotations and explicit exclusion

Relative elevation power shapeRelative analytic factor only; no absolute gain or real antenna pattern.

Additional C/N₀ offset Δ (dB)-20–20; step 0.1; default 0. Baseline 45 dB-Hz at R1-RX after the reference front end.

Exclude below 30 dB-Hz?Illustrative hypothesis only; not an acquisition limit. Default keeps visible rows regardless of this marker.

Canonical five-satellite fixture · committed · evaluation 2026-09-08 UTC

Unknown receiver clock · four states: available

Bounded geometry hypothesis only; test installed timing bias, acquisition and holdover.

Rank 4/4; condition 6.8541e+0. Rank threshold s > 10⁻¹⁰ sₘₐₓ; withhold finite error scales above condition 10⁸. These are numerical teaching policies.

Selected: 5
Visible: 5
Used: 5
Blocked: 0

Committed inputs: 0° azimuth rotation; mask 15°; σref 3.0 m; flat relative gain; Δ 0.0 dB; quality exclusion disabled (risk is annotation); overhead clear; full azimuth off. Sectors: 1: off; 2: off; 3: off. All results are synthetic.

Unweighted geometry and equal-σ error scales
Quantity	Calculated value	Interpretation
HDOP	1.154700538	Dimensionless; √(QEE + QNN)
VDOP	2.236067977	Dimensionless; √QUU
PDOP	2.516611478	Dimensionless; √(QEE + QNN + QUU)
TDOP	1.414213562	Four-state clock range geometry only
GDOP	2.886751346	Four-state position + clock geometry
Horizontal root-sum-variance	3.464101615 m	√(CovEE + CovNN), not CEP or a 95% radius
Vertical standard deviation	6.708203932 m	√CovUU under independent zero-mean equal-σ errors
Clock-time standard deviation	14.151926020 ns	√Covbb / c; b is metres

Optional weighted range-noise hypothesis

σᵢ = σref · 10^((45 − C/N₀ᵢ)/20) m, fixed receiver/estimator/integration, independent white-noise regime. This is an assumed law, not a universal GNSS conversion. 5 rows carry finite positive information. Weighted design status: available, rank 4/4; condition 6.8541e+0.

Weighted error scales · separate from geometric DOP
Quantity	Calculated value
Horizontal root-sum-variance	3.464101615 m
Vertical standard deviation	6.708203932 m
Clock-time standard deviation	14.151926020 ns

Illustrative S0 directions. ● used; ◇ excluded. Dashed circle: 15° mask. Zenith azimuth is undefined; only the overhead control blocks S01. Sector obstruction is a simplified azimuth rule, not a roof model. All plotted values appear in the sky table.

Selected rows · degrees, dimensionless ENU, R1-RX dB-Hz and assumed range σ in metres. Unselected rows remain inspectable in the pool view.
ID	Az ° / El °	uE / uN / uU	Selected / visible / used	Blocked / reason	C/N₀ dB-Hz	σᵢ m / quality
S01	undefined (stored 0°) / 90	0.000000 / 0.000000 / 1.000000	yes / yes / yes	no · used	45.000000	3.000000 · at/above illustrative marker
S02	0.000000 / 30	0.000000 / 0.866025 / 0.500000	yes / yes / yes	no · used	45.000000	3.000000 · at/above illustrative marker
S03	90.000000 / 30	0.866025 / 0.000000 / 0.500000	yes / yes / yes	no · used	45.000000	3.000000 · at/above illustrative marker
S04	180.000000 / 30	0.000000 / -0.866025 / 0.500000	yes / yes / yes	no · used	45.000000	3.000000 · at/above illustrative marker
S05	270.000000 / 30	-0.866025 / 0.000000 / 0.500000	yes / yes / yes	no · used	45.000000	3.000000 · at/above illustrative marker

Inspect design rows, singular values and covariance matrices

H columns are −uE, −uN, −uU, 1. HᵀH is shown for the analytic check; the solver uses one-sided Jacobi SVD directly on H. Singular values: 2.618033989e+0, 1.224744871e+0, 1.224744871e+0, 3.819660113e-1. Numerical convergence tolerance 2e-15; all display rounding is separate from decisions.

Used measurement design rows H
Row	δE	δN	δU	b
S01	0.000000000	0.000000000	-1.000000000	1.000000000
S02	0.000000000	-0.866025404	-0.500000000	1.000000000
S03	-0.866025404	0.000000000	-0.500000000	1.000000000
S04	0.000000000	0.866025404	-0.500000000	1.000000000
S05	0.866025404	0.000000000	-0.500000000	1.000000000

HᵀH · dimensionless
Row / column	E	N	U	b
E	1.500000000	0.000000000	0.000000000	0.000000000
N	0.000000000	1.500000000	0.000000000	0.000000000
U	0.000000000	0.000000000	2.000000000	-3.000000000
b	0.000000000	0.000000000	-3.000000000	5.000000000

Q · unweighted dimensionless covariance factor
Row / column	E	N	U	b
E	0.666666667	0.000000000	0.000000000	0.000000000
N	0.000000000	0.666666667	0.000000000	0.000000000
U	0.000000000	0.000000000	5.000000000	3.000000000
b	0.000000000	0.000000000	3.000000000	2.000000000

Equal-σ covariance · all state coordinates in metres, entries m²
Row / column	E	N	U	b
E	6.000000000	0.000000000	0.000000000	0.000000000
N	0.000000000	6.000000000	0.000000000	0.000000000
U	0.000000000	0.000000000	45.000000000	27.000000000
b	0.000000000	0.000000000	27.000000000	18.000000000

Weighted covariance · state entries m²
Row / column	E	N	U	b
E	6.000000000	0.000000000	0.000000000	0.000000000
N	0.000000000	6.000000000	0.000000000	0.000000000
U	0.000000000	0.000000000	45.000000000	27.000000000
b	0.000000000	0.000000000	27.000000000	18.000000000

Next experiment · timing/antenna owner

Bounded geometry hypothesis only; test installed timing bias, acquisition and holdover. Compare calibrated PPS against an independent clock at outdoor, window and roof placements. Document biases, signal averaging, acquisition and holdover; evidence due 2026-09-22. Satellite messaging needs a separate service trial and symbolic link budget.

No orbit propagation, time series, live constellation, receiver accuracy guarantee, 95% protection level or safety-integrity conclusion. geometry-cn0-visibility-explorer/2.0; p07-m06-sky-snapshot-v1; p07-m06-cn0-v1. Review due 2026-12-07.

The complete canonical inputs and results above are available without JavaScript. Input changes require JavaScript; the lesson below includes the singular, attenuation and clock-known counterexamples.

Canonical counterexamples retained for print and no-script reading
Change from canonical	Geometric result	Signal / evidence result
Rotate all azimuths by 37°	HDOP, VDOP, PDOP and GDOP invariant.	ENU covariance transforms under rotation; no geometry improvement from a relabelled heading.
Remove S01	Four-state rank 3; DOP/error scales unavailable.	All remaining rays at 30°; good signal cannot repair dependence.
Uniform −6 dB; no exclusion	Five used rows; unweighted DOP unchanged.	C/N₀ = 39 dB-Hz; σᵢ = 5.98578694490664 m; weighted covariance = 10^0.6 times default covariance.
Uniform −20 dB; no exclusion	Five rows; same geometry.	25 dB-Hz: illustrative risk only.
Uniform −20 dB; explicit quality exclusion	Five visible; zero used; results unavailable.	A new measurement-choice hypothesis changed H.
Four low rays + independently exact clock	Three-state rank 3; HDOP 1.154700538, VDOP 1.	Ideal clock-known comparison. TDOP/GDOP not applicable; a stable oscillator is not this constraint.

Go deeperThe optional pool and signal-null boundary

S06–S13 have elevation 60° and azimuth 22.5°+45°k for k=0…7. S14–S20 have elevation 20° and azimuth (15°+360°k/7) modulo 360°, k=0…6. Only S01–S05 start selected. The horizon null of sin²(e) is tested independently; the default pool avoids it. A null contributes no weighted measurement information, while geometric visibility remains separately defined.

Decision: investigate placement, antenna/front-end response and calibrated delay independently. Do not add a feed loss twice or assume a higher LNA gain removes an already accumulated timing bias.

07 / 10

Biases can defeat a good geometry number

A receiver reports excellent geometry, but its PPS is consistently late. Which assumption failed?

Covariance was calculated for zero-mean errors. A reproducible offset violates that assumption. Ionospheric and tropospheric delays require their own treatment; multipath combines direct and reflected signals; non-line-of-sight (NLOS) reception can replace the direct path with an excess path. A small covariance cannot bound an omitted bias. [ESA-BOOK] [GPS-SPS]

$\begin{aligned}\mathrm E[\hat x-x]&=(H^{\mathrm T}H)^{-1}H^{\mathrm T}\mu\\\mathrm{MSE}&=\operatorname{Cov}+\mathrm{bias}\,\mathrm{bias}^{\mathrm T}\end{aligned}$ In this unweighted full-rank linear model, μ is a nonzero mean range-error vector in metres. Mean-square error includes both variance and squared bias. If every ray has a common 10 ns delay, μ = H[0,0,0,c×10 ns]ᵀ: the model can absorb it into the clock state while residuals look small.

Separate error mechanisms before assigning a test
Mechanism	Observable / discriminating evidence	Limit of the toy model
Ionosphere / troposphere	Signal-frequency/correction configuration, elevation dependence and environmental/correction records.	No atmospheric residual covariance or correction algorithm.
Multipath / NLOS	Paired placement, reflecting surfaces, residual/time-error patterns and an independent reference.	No ray tracing; a strong reflection can still be biased.
Interference / jamming	Correlated C/N₀/receiver-status changes, receiver diagnostic flags and independently acquired RF evidence.	No waveform or source is identified from C/N₀ alone.
Spoofing / misleading time	Cross-check time/position with independent sources, plausibility and receiver-supported authenticity/status information.	Awareness only; no detection guarantee, offensive procedure or safety-integrity claim.

Common misconceptionLow DOP proves the timing is trustworthy.

DOP describes geometry under an error model. It does not authenticate a signal, discover a common timing offset, or supply a protection level. Independent evidence must challenge the failure mode rather than repeat the same dependence.

Go deeperIndependence is a physical requirement

Two receivers fed from the same antenna and affected by the same propagation or interference are not two independent references against that common fault. Record what each comparison shares: antenna, time source, corrections, power, firmware or network.

Decision: keep a separate bias/uncertainty budget and independent time reference. The remote node now poses another question: how does it send data when terrestrial links fail?

08 / 10

Satellite messaging is a separate link

Can the gateway’s GNSS antenna be treated as the node’s satellite fallback link?

GNSS reception estimates time and position from navigation broadcasts. Messaging needs a compatible transmitter/receiver, antenna, protocol and operator route to a host. Iridium SBD is one actual service candidate for that investigation; its public description establishes the service family, not our terminal’s deadline or indoor coverage. [IRIDIUM-SBD]

Original conceptual paths, not a network topology or band plan. GNSS navigation data do not carry the node’s telemetry. A two-way messaging transaction adds a terminal transmitter, service route and return-path contract.

For a fixed geometric length d in metres, vacuum propagation time is d/c. Slant distance is the actual straight-line leg; orbital altitude alone does not determine it. The following are separately assumed distances, not the current position of an operator’s satellite. At 2 GHz, the exact-c free-space loss follows the Path 06 equation and ITU-R P.525-5. [ITU-FS]

Checked slant-path examples · 2 GHz is an illustrative frequency, not an approved service band
Assumed slant distance	One vacuum leg d/c	Free-space basic loss 20log₁₀(4πdf/c)
600 km = 600000 m	2.0013845711889124 ms	154.03140814283586 dB
35786 km = 35786000 m	119.36924710761069 ms	189.54264628708657 dB

The shorter example resembles an overhead low-orbit scale only by assumption. The longer one uses a familiar geostationary scale, but a real ground-to-satellite slant changes with location/elevation. A bent-pipe ground→satellite→ground route sums both slants. Equal 35786 km legs give 238.73849421522138 ms before other delays; four equal legs for an application return path give 477.47698843044276 ms before processing/access/routing.

$\Delta f=-\frac{v_rf}{c}=-\frac{7500\times2\times10^9}{299792458}=-50034.61427972281\,\mathrm{Hz}$ First-order Doppler; positive radial speed vr means recession, so received frequency shifts below the nominal carrier. Radial speed is not full orbital speed and is not constant over a pass. A +7500 m/s value is an assumed local radial component, not a propagated orbit. Phasor convention remains e^(+jωt).

Think about itIf one vacuum leg takes about 2 ms, can a store-and-forward report still take minutes?

Answer

Yes. The application clock includes when access is possible, pass/contact waits where applicable, retries, queues, routing and host processing. A hypothetical 300 s service wait dominates the 0.002001 s vacuum leg. That waiting fixture is not a claim that Iridium SBD uses this service schedule.

$P_{\mathrm{r,R1}} = P_{\mathrm{t,R1}} - L_{\mathrm{TX}} + G_{\mathrm{real,TX}} - L_{\mathrm{FS}} - L_{\mathrm{pol}} - L_{\mathrm{atm}} - L_{\mathrm{point}} + G_{\mathrm{real,RX}} - L_{\mathrm{RX}}$ Logarithmic budget: P in dBm; realized antenna gains in dBi; all L terms in dB. R1-TX → R2-TX feed loss is LTX; R2-RX → R1-RX feed loss is LRX. S0 includes slant path, polarization, atmosphere and pointing. Local R1/R2 aliases retain the portfolio meaning. R3 remains the receiver decision boundary.

Here gains use the declared reference pointing with a compatible real 50 Ω source/load and include antenna mismatch once. Lpoint is the additional off-reference pointing penalty; if actual-direction gain already includes it, do not subtract it again. Radiation efficiency is already in realized gain. Polarization loss belongs once in Lpol. A receiver criterion and noise model must then connect R1-RX to R3.

Annotated symbolic budget for the actual service candidate
Input / owner	Status	Next evidence
Terminal TX, antenna gain and DC states / terminal owner	unknown	Exact supported mode, RF plane/power, mounted pattern and current trace. RF output is not terminal DC energy.
Satellite EIRP or receive G/T / service owner	unknown	Direction, mode, conditions and reference plane. G/T is receive gain divided by system noise temperature; do not insert it as an ordinary gain term.
Slant, pointing, polarization and atmosphere / RF owner	unknown	Actual path and terminal orientation/elevation; apply an appropriate atmospheric model only with its inputs.
Receiver criterion, access, route and host / systems owner	unknown	Return link and acknowledgement contract; pass/queue/deadline evidence for the exact service.

Go deeperA terminal-energy ledger needs states, not a satellite label

Eevent = Σ∫V(t)Istate(t)dt in joules at the terminal supply. Measure startup/search, listening, TX, RX/acknowledgement, backoff and timeout; distinguish on-state current from period-average current. A long receive window or repeated no-contact attempt can dominate a small D0 payload. No state currents, satellite EIRP, G/T or completed link margin are invented here.

Decision: retain the checked geometric anchors but leave the service link budget symbolic. Now test availability at the endpoint that matters.

09 / 10

Availability and resilience need their own evidence

A constellation covers a region. What can still prevent this product from delivering useful time or a report?

Distinguish constellation geometry, service operation, terminal capability and installation visibility. A navigation signal may be visible but unusable for the selected state/health/quality rule. A messaging terminal may see sky but lack access, provisioning or a working host route. Satellite connectivity does not establish universal indoor availability.

Availability is a chain of separate claims
Layer	GNSS timing test	Messaging fallback test
Constellation / geometry	Usable signal directions and state observability over the observation interval.	Actual contact/slant/pointing conditions, not a generic altitude.
Service	Signal health, correction/status and time-scale treatment.	Provisioning, operator reachability, routing, queues and dated service scope.
Terminal	Acquisition, tracking, valid timing output and oscillator holdover.	Supported mode, access behavior, TX/RX states and acknowledgement semantics.
Installation	Sky mask, multipath, antenna/cable delay and interference.	Antenna view, orientation, mounting, local blockage and power supply.

$\Delta t_{\mathrm{holdover}} = y \times T = 10^{-6} \times 60 s = 60 \mu s$ Illustrative constant fractional frequency error y = 1 ppm over 60 s without time correction. The result is 600 times the chosen 100 ns timing goal, even with zero initial bias. Real drift, temperature, aging and stochastic clock errors require evidence; this is fixed bias only.

Think about itA stable oscillator carries the last time estimate through an outage. Is its clock now exactly known?

Answer

No. It carries initial time uncertainty and accumulates frequency error. The explorer’s clock-known switch is an independent exact constraint, not holdover. A timing receiver may use a surveyed-position mode, but that is another state/measurement model, not permission to delete this clock column.

For the messaging variant, deadline availability is the fraction of all generated unique reports delivered completely by the 900 s deadline. For timing, count all scheduled epochs and treat missing or invalid PPS separately from the signed error of valid outputs. Do not remove failed attempts from the denominator. Repeated samples within one pass or one multipath state may be correlated; describe the test population and confidence rather than claiming universal reliability.

Go deeperResilience needs a failure model

Two links can share the same obstructed mounting position, supply or timing dependency. List common causes and independent fallbacks; choose a discriminating observation for each. National operation, terminal approval and operator conditions belong to separately assigned compliance/service owners and current Path 09 review. This lesson makes no safety or legal authorization decision.

Decision: assign independent evidence to each layer, with a deadline and reversal criterion. The two plans below are ready to hand to the next module.

10 / 10

Specify two experiments and evidence plans

What can the team decide now, and which observation would make it change course?

Accept the synthetic geometry calculation within its stated conditions. Request evidence for real installed timing. Separately investigate one provisioned satellite messaging terminal and host path. Neither task can borrow the other’s C/N₀, accuracy, energy or availability claim.

The two complete snapshots below carry p07-technology-evidence-card-v1. Prior IDs are retained as lineage; each changed requirement has a named variant. Unknown state durations/current, source applicability and outcomes remain unknown, not zero or a measured failure. Evaluation is frozen at 2026-09-08 UTC; source access dates record the actual review rather than browser time.

Experiment A · time at the gateway

P07-M06-GNSS-TIMING-v1 · complete static technology evidence card
Field	Evidence and next action
ID / owner	p07-technology-evidence-card-v1; P07-M06-GNSS-TIMING-v1; Timing engineer + antenna engineer
Inherited fixture IDs	P07-M05-CELLULAR-OPTIONS-v1 · P07-M04-REMOTE-FLEET-v1 · P07-M03-HE-BACKHAUL-v1 · P07-M02-LE-CONN-v1 · P07-M02-THREAD-SED-v1 · P07-M01-LE-EXT-v1 · p06-evidence-map-v1 · M01-A · M01-B-LOSS · M01-INSTALL-UNKNOWN
Scenario	Illustrative condition-monitoring gateway: compare an outdoor reference placement, window and under-roof placement with the same receiver. The generic 2.450 GHz QPSK node remains unchanged.
Requirement	Chosen experiment goal: \|1 PPS time error\| ≤100 ns for ≥99% of all scheduled 1 s epochs in each 24 h placement stratum after a separately recorded startup interval. Missing/invalid PPS counts against availability. No measured achievement.
Family / exact mode	Receive-only GNSS timing; GPS L1 C/A candidate. Galileo E1-B/E1-C is a separate implementation comparison, not a silently combined constellation. GPS IS-GPS-200N + IRN-003; IRN-004 impact review open. SPS PS 5th ed. For the Galileo comparison: OS SIS ICD 2.2, OS SDD 1.3 + SNGU 2026002. Receiver SKU/firmware/timing mode and inter-system time-bias treatment unknown.
Band / region / class	Proposed fixed gateway in CH; final receive antenna/active front end and product EMC applicability unresolved. GNSS reception gives no satellite transmit permission.
D0 / period / deadline	Payload not applicable to this receive-timing use; 1 s observation/report period; payload-delivery deadline not applicable. D0 is gateway timestamp use, not a satellite payload uplink. Time error is signed receiver PPS minus calibrated independent reference PPS at named electrical timing connectors. Record time scale, cable delays and uncertainty. TTFF/time-to-valid-time is a separate release-to-valid-state measurement.
PHY / MAC resources	Code/delay/Doppler search, tracking and decoded time/ephemeris. D1 is navigation data; no user data PHY rate or MAC uplink allocation is implied.
State duration / current assumptions	cold acquisition: duration unknown s; current unknown A; aided acquisition (separate variant): duration unknown s; current unknown A; tracking: duration unknown s; current unknown A; reacquisition: duration unknown s; current unknown A; holdover: duration unknown s; current unknown A
RF / antenna interfaces	S0 antenna phase centre → R2 antenna feed → LNA/filter/cable → R1-RX input → R3 receiver decisions → electrical timing output → D0 timestamp. Gain, noise and group delay are separate; antenna/cable calibration referenced to the timing connectors.
Infrastructure	Healthy usable navigation signals/corrections, receiver configuration, independent reference clock and calibrated time-interval instrument. Aiding data age/source and surveyed antenna position must be explicit if used.
Qualification	Exact receiver timing-mode and antenna/component compatibility; EMC/host test scope needs a lab plan. No module qualification or integrity certification is asserted.
Regulatory questions	Compliance owner checks final fixed-product and active-antenna obligations for CH; Path 09 owns the current market determination.
Claim / sources / evidence status	TIME-INTERFACE [GPS-IS, GPS-IRN, GPS-INDEX, GPS-SPS, GAL-ICD, GAL-SDD, GAL-NOTICE]: Selected source text read; IRN-004 detailed impact and receiver applicability unknown. TIME-GEOMETRY [ESA-ERROR, ESA-BOOK, p07-m06-sky-snapshot-v1]: Derived local independent equal-σ model, not product accuracy. TIME-ACQUIRE [ESA-BASEBAND]: Receiver limits and TTFF distribution unknown.
Uncertainty	Reference clock/time scale, instrument calibration, PPS quantization, antenna/group delay, multipath/NLOS bias, sample correlation and thermal holdover drift need an uncertainty budget. All current/state durations and measured outcomes are unknown.
Decision	Proceed to paired placement experiments. Do not accept the window from DOP or signal level alone. Retain unknown clock unless an independent exact constraint is deliberately imposed in the toy comparison.
Next evidence / owner / deadline	Timing engineer + antenna engineer; By 2026-09-15 pin hardware/firmware, time scale, survey/aiding mode and reference uncertainty. Timing/antenna owners then log 24 h per placement: scheduled PPS validity, signed time error, C/N0 with averaging interval, az/el/used status, residuals, acquisition state and temperature; repeat paired order on another day. Calibrate cable/antenna delay, test 60 s loss/holdover and compare independent time during GNSS anomalies. Experiment due 2026-09-22.
Decision-reversal criteria	Reject the installation hypothesis if a placement misses the chosen valid-PPS/time-error goal, shows a reproducible unbudgeted timing bias, or fails the separately chosen holdover need. Move the antenna or obtain a separate timing/holdover design and new evidence.
Review date / trigger	2026-12-07; Evaluation 2026-09-08 UTC; review on new interface notice, firmware/mode/antenna/cable/roof change, time anomaly or due date.

Experiment B · a report at the remote host

P07-M06-SBD-FALLBACK-v1 · complete static technology evidence card
Field	Evidence and next action
ID / owner	p07-technology-evidence-card-v1; P07-M06-SBD-FALLBACK-v1; Systems engineer + service/field-test owner
Inherited fixture IDs	P07-M05-CELLULAR-OPTIONS-v1 · P07-M04-REMOTE-FLEET-v1 · P07-M03-HE-BACKHAUL-v1 · P07-M02-LE-CONN-v1 · P07-M02-THREAD-SED-v1 · P07-M01-LE-EXT-v1 · p06-evidence-map-v1 · M01-A · M01-B-LOSS · M01-INSTALL-UNKNOWN
Scenario	Illustrative remote-node fallback after terrestrial service loss. Retain 32-byte reports every 600 s; allow a new 900 s fallback deadline. The preceding cellular 60 s deadline and urgent 30 s command requirement are not silently relaxed for that original mode.
Requirement	Chosen fallback goal: ≥99% of unique complete D0 reports delivered within 900 s, evaluated over every generated event in each declared site/orientation stratum. The 30 s urgent command remains an unresolved separate service requirement.
Family / exact mode	Satellite messaging, distinct from GNSS reception; Iridium SBD candidate for a scoped service investigation. Actual service: Iridium Short Burst Data (SBD), mobile-originated report to host and separate mobile-terminated command. Public service description accessed 2026-09-08; exact terminal/SKU/firmware, SBD interface revision, application acknowledgement mode and service contract unknown.
Band / region / class	CH candidate plus any remote deployment country to be declared. Approved service band, EIRP, antenna, device class and regional provisioning/authorization unknown; the 2 GHz arithmetic is not an Iridium band claim.
D0 / period / deadline	256 bits; 600 s observation/report period; 900 s delivery deadline. Node D0 report generation → unique complete host D0 receipt. Include search/access/pass wait if applicable, queue, radio transfer, routing and processing. A return application acknowledgement has its own endpoint and time; modem acceptance is not host delivery.
PHY / MAC resources	Selected terminal protocol framing/coding, retries, access opportunities and MT polling behavior require pinned documentation. No PHY headline becomes application goodput.
State duration / current assumptions	cold start/search: duration unknown s; current unknown A; access/listen: duration unknown s; current unknown A; TX burst: duration unknown s; current unknown A; RX/acknowledgement: duration unknown s; current unknown A; retry/backoff: duration unknown s; current unknown A; sleep: duration unknown s; current unknown A; no-contact timeout: duration unknown s; current unknown A
RF / antenna interfaces	D0 → D1 → R1-TX → feed → R2-TX → S0 pointing/polarization/slant path → R2-RX → receiver chain → R3 → service route → host D0. Reverse link budget separate. RF output and terminal DC energy are different ledgers.
Infrastructure	Provisioned terminal, antenna sky view, operator satellite/routing infrastructure, host gateway/endpoint and acknowledgement path. Operator reachability, actual geometry and scheduled/store-and-forward behavior must be verified for the selected service; not inferred from a LEO label.
Qualification	Service/terminal acceptance and host/antenna integration evidence requested separately from national rules; no approved part list is invented.
Regulatory questions	Compliance/service owners confirm country-specific service permission, spectrum/device/antenna conditions and final-host authorization before field transmission.
Claim / sources / evidence status	SAT-SERVICE [IRIDIUM-SBD]: Informative service identity verified; terminal/mode/deadline evidence unknown. SAT-PATH [ITU-FS, p07-m06-link-examples-v1]: Derived vacuum/LOS arithmetic at assumed slant distances; symbolic service budget.
Uncertainty	Satellite EIRP/G/T (receive gain divided by system noise temperature), terminal power/gain/current, pointing, atmosphere, pass/contact and queue distributions are unknown. No link margin, daily energy, availability or subscription recommendation is calculated.
Decision	Request one provisioned-terminal field trial at the chosen sites. Keep the link budget symbolic until the service inputs and country applicability are closed. No indoor or universal-coverage acceptance.
Next evidence / owner / deadline	Systems engineer + service/field-test owner; By 2026-09-15 service owner freezes terminal/mode/interface and country matrix. By field deadline, log complete D0 sequence IDs and host receipt, modem states/retries, DC current/voltage and elapsed time during open-sky versus obstructed placement; include cold starts, deliberate no-contact periods, two orientations and all failed/late reports over a declared 7-day population. Confirm actual slant/pass evidence and operator queue/MT behavior rather than assuming the 600 km example. Experiment due 2026-09-29.
Decision-reversal criteria	Reverse the fallback choice if field evidence misses the 900 s goal, the independently specified terminal energy budget, host acknowledgement semantics or regional applicability. Quantify uncertainty; repeated events in one pass are not independent availability trials.
Review date / trigger	2026-12-07; Evaluation 2026-09-08 UTC; recheck service/contract, firmware, country, antenna/host, operator notice or due date.

Common misconceptionOne successful outdoor fix or message closes the placement and availability question.

It establishes one event under its recorded conditions. Repeat the planned state/site/orientation population, keep failures and uncertain observations, and compare against a predeclared endpoint and statistic. A field hypothesis may be worth testing long before it becomes a product claim.

Go deeperWrite a decision that can be reversed

“Proceed with this specific placement/service experiment because the model supports its bounded hypothesis; change placement or link if the named timing, delivery, energy or applicability evidence fails.” An owner, deadline and reversal observation make uncertainty actionable without pretending it is resolved.

Handoff to 07.7 · Coexistence & Technology Selection: compare these two cards with the terrestrial candidates and their coexistence constraints. Keep an explicit unresolved urgent-command path; a 900 s fallback cannot silently satisfy a 30 s command requirement.

Ungraded review

Check your understanding

Answer each question in your own words, then reveal the model answer.

01Why does the clock column contain 1 rather than c?
Model answer
The fourth state is b=cδt in metres, so a one-metre clock-equivalent error adds one metre to every pseudorange. A seconds-valued state would need a c coefficient and different covariance units.
02There are four satellites at 30° elevation. Why is the 3D/clock solution unavailable?
Model answer
Every Up coefficient is −0.5 and every clock coefficient is 1. Up and clock columns are dependent; rank is 3. Satellite count alone cannot restore the missing information. An independently exact clock removes the clock state, while an ordinary stable oscillator does not.
03All C/N0 values fall by 6 dB. What changes before any quality exclusion?
Model answer
The same line-of-sight rows give identical unweighted DOP. Under the explicit assumed noise law, each range σ increases by 10^0.3 and weighted covariance by 10^0.6. Neither error scale is a 95% accuracy guarantee.
04Why can the 1 ms to 10 ms integration example fail in a receiver?
Model answer
The 15 dB to 25 dB result assumes a coherent matched integration, constant phase/data and white noise. Data or secondary-code transitions, Doppler uncertainty, oscillator error and dynamics can break coherence. C/N0 is carrier power divided by noise density, not carrier power.
05Does 600 km imply a 2 ms application delay, and which Doppler sign belongs to recession?
Model answer
The assumed 600 km slant gives 2.001384571 ms for one vacuum leg. The application adds access, other legs, routing and queues; a store-and-forward wait can dominate. Positive recession gives a negative received shift: −50.034614 kHz at 2 GHz and +7500 m/s radial speed.
06Which evidence best challenges a window timing claim or a remote messaging claim?
Model answer
For timing, compare calibrated PPS against an independent clock and log placement, sky view, C/N0, biases and acquisition/holdover state. For messaging, time every unique D0 event to host receipt while logging terminal DC states and visibility; keep losses/late events and the exact service/country identity. Neither a DOP readout nor a service map replaces these observations.

References and further study

Primary sources accessed/reviewed 2026-09-08; review due 2026-12-07 or the stated document, receiver, antenna, service or market trigger. These are selected-text readings, with exclusions below. The frozen illustrative evaluation date is separate from source lifecycle. No live sky, orbit, network, subscription or standards service participates in rendering.

GPS-INDEX · GPS.gov
Technical documentation and interface register · Register checked at access. Published: Rolling register; individual dates below. Consulted: IS-GPS-200 N and associated IRN rows. Official catalogue / status evidence. Lists 200N, IRN-003 (2024-01-19) and IRN-004 (2026-06-16; RFC-00519 civil ISM formats). IRN-004 download failed in this review. Its existence/scope is verified from the register; detailed amendment applicability remains open. No integrity algorithm or current sky status inferred.
GPS-IS · U.S. Space Force
IS-GPS-200N, Navstar GPS Space Segment / Navigation User Interfaces · Revision N. Published: PDF: 2022-08-01; register: 2022-08-22. Consulted: §3.3.1.1 frequency plan; §20.3.3.3.3.2 group-delay treatment. Normative interface; selected text read. GPS L1 C/A selected receive signal, 1575.42 MHz. Single-frequency user treatment includes satellite group delay. With IRN-003 and the unexamined IRN-004 dependency retained. No receiver TTFF, indoor operation, product timing bound or safety requirement is inferred.
GPS-IRN · U.S. Space Force
IRN-IS-200N-003 · IRN 003 / RFC-00502. Published: 2024-01-19. Consulted: Cover; changed clock/ephemeris parameter table (PDF pp.20–22); message schedule change (PDF p.46). Normative amendment; selected text read. Demonstrates why an interface baseline needs its revision notices and message dependencies. Not a complete implementation audit; June 2026 IRN-004 remains separately unresolved.
GPS-SPS · U.S. Government
GPS Standard Positioning Service Performance Standard · 5th edition; official register lists in effect. Published: 2020-04. Consulted: §2.4.3–2.4.5, especially Excluded Errors; §3.1 conditions. Normative service document; selected text read. Service performance has a defined boundary. User antenna, multipath, receiver and residual propagation errors are excluded from §3 commitments. No source percentile is imported into the illustrative σ=3 m model or the proposed 100 ns timing goal.
GAL-INDEX · European GNSS Service Centre
Galileo Open Service in-force reference documents · Register checked at access. Published: Rolling register. Consulted: OS SDD v1.3; OS SIS ICD v2.2; SNGU 2026002. Official catalogue / status evidence. Pins the current listed OS documents and service notice. Not a live satellite-health report; the page footer date is not the issue date of every listed document.
GAL-ICD · European Union
Galileo Open Service Signal-In-Space Interface Control Document · Issue 2.2. Published: 2025-11. Consulted: §1.1 scope; Chapter 2 frequency/polarisation and modulation; Chapter 3 code structure; §4.3 I/NAV. Normative interface; selected text read. E1-B/E1-C receive-signal orientation: RF and data/pilot structure, RHCP transmission. No acquisition sensitivity, product compatibility, legal terminal band or guaranteed timing performance follows from the interface.
GAL-SDD · European Union
Galileo Open Service Service Definition Document · 1.3; mutable in-force URL, version frozen here. Published: 2023-11. Consulted: §2.4 usage assumptions; Annex E as amended by SNGU 2026002. Normative service definition; selected text read. Minimum performance levels depend on receiver, signal-status and usage assumptions. A sky mask and aiding policy must match a claim. The explorer’s 15° mask is authored, not the SDD usage mask. No MPL is asserted for the roof/window installation.
GAL-NOTICE · European Commission
Galileo Service Notice #24 / SNGU 2026002: Updates on OS NAGUs · Issue 1.0. Published: 2026-01-22 15:45 UTC. Consulted: Page 1 precedence statement; pages 2–4 updated notice categories. Official service change notice. Updated NAGU categories/definitions take precedence over the corresponding OS SDD content. This is a notice-format change, not a measured outage or an automatic change to the synthetic geometry.
ESA-ERROR · ESA Navipedia / gAGE
Positioning Error · Article, 2011. Published: 2011. Consulted: Formal Accuracy Eq.(1); Predicted Accuracy Eq.(7)–(10); DOP definitions; Measured Accuracy. Informative theory. Covariance conditional on a linear model and range-error assumptions; DOP definition versus observed error. Local b=cδt state uses metres explicitly; no generic conversion to a 95% horizontal radius.
ESA-BOOK · ESA / Sanz Subirana, Juan Zornoza, Hernández-Pajares
GNSS Data Processing, Volume I: Fundamentals and Algorithms · ESA TM-23/1. Published: 2013. Consulted: Chapter 4 introduction / §4.1 observables; §4.2 receiver noise and multipath; §5.1 transmit-time geometry. Informative reference. Pseudorange is apparent range from clocks; real measurements carry corrections and biases. Selected explanatory passages only; no reproduction of measured figures and no navigation/filter implementation.
ESA-BASEBAND · ESA Navipedia
Baseband Processing · Article revision oldid 16696. Published: Publication date not established. Consulted: Acquisition; coherent/non-coherent integration; tracking transition. Informative receiver explanation. Code/Doppler search, data-boundary and clock/dynamics limits on integration. The lesson’s normalized 15/25 dB examples and 30 dB-Hz risk marker are authored assumptions, not receiver limits.
ITU-FS · ITU-R
P.525: Calculation of free-space attenuation · P.525-5, in force; approved 2024-11-03. Published: 2024-11. Consulted: Annex 1 §2.3, Eqs.(4)–(6). Recommendation; selected full-text formula read. Free-space basic transmission loss between isotropic reference antennas; exact-c form used here. No atmosphere, obstruction, pointing, terminal power or service availability supplied.
IRIDIUM-SBD · Iridium
Iridium Short Burst Data (SBD) · Public service page, snapshot at access. Published: Not stated. Consulted: Service description; Two-Way Connectivity; Global, Real-Time Availability. Informative operator description. Identifies an actual short-packet, two-way service between equipment and a host; crosslinked LEO network orientation. Marketing coverage language is not a field deadline or indoor guarantee. Terminal model/firmware, interface revision, approved band, qualification and regional service conditions are unresolved.

Model geometry-cn0-visibility-explorer/2.0; fixture p07-m06-sky-snapshot-v1; signal fixture p07-m06-cn0-v1; numerical policy p07-m06-geometry-policy-v1; static physical examples p07-m06-link-examples-v1. SI internally, exact c, full precision for decisions. Independent analytic and NumPy/LAPACK references check matrices to 10⁻¹⁰ absolute, DOP to 10⁻⁹ absolute and timing/frequency anchors to 10⁻⁹ relative. Authors chose all synthetic angles, σ, signal levels, masks, marker and experiment goals; no source supplies those as product evidence.

Positioning &Satellite Links

An accuracy headline fails at the window

Pseudorange contains an unknown clock

Geometry transforms ranging error

Read signal-in-space and receiver evidence separately

C/N₀, acquisition and tracking answer different questions

The installed antenna changes signal and timing

Geometry, C/N₀ & Visibility Explorer

Optional weighted range-noise hypothesis

Biases can defeat a good geometry number

Satellite messaging is a separate link

Availability and resilience need their own evidence

Specify two experiments and evidence plans

Experiment A · time at the gateway

Experiment B · a report at the remote host

Check your understanding

References and further study

Positioning &
Satellite Links