Receiver Regression · Signal Combination Testing
Multi-Constellation GNSS Receiver Testing: Signal Combinations and Failure Analysis
Multi-frequency GNSS testing checks how a receiver acquires, tracks and uses named signals across multiple carrier frequencies. Multi-constellation testing adds another axis: simultaneous operation across satellite systems. A useful regression campaign tests both axes and preserves enough evidence to explain a failure.
This guide gives receiver-validation engineers a worked combination matrix, a signal-level evidence record and a method for investigating cases that meet their PVT criteria separately but fail when combined. It also separates receiver reconfiguration from selective RF-loss testing.
Part of GNSS Simulation 101. Use GNSS signals, bands and codes for signal definitions, and GNSS constellation comparison for system-level context.
Define Signals, Components and Simultaneous Combinations
A case named “GPS + Galileo, dual frequency” is incomplete. State the exact signals, components, satellite sets, navigation-data profile and supported simultaneous combination. Record the source configuration and DUT hardware/firmware readback alongside the case ID.
The worked example below uses GPS L1 C/A and L5 I5/Q5, plus Galileo E1-B/E1-C and E5a-I/E5a-Q. L1/E1 share 1575.42 MHz; L5/E5a share 1176.45 MHz. GPS and Galileo retain different signal structures. I/Q and B/C components do not represent additional carrier frequencies or additional satellites.[1][2]
Freeze the applicable signal-document version and generated navigation data. A frequency label does not establish support for every component or newer service. Consult the signal requirements guide when preparing the detailed manifest.
Example scope: all profiles, counts and limits in this article are hypothetical. Assume a generic DUT and source have been confirmed to support the selected combinations. These examples are not measurements or default configurations of NSG8000, u-blox or another product.
Worked Matrix: Separate the Constellation and Frequency Axes
Start with a static, healthy-signal scenario. Select eight GPS satellites and eight Galileo satellites, each with the two signals required for its dual-frequency case. Use the same satellites within each system when moving between single- and dual-frequency profiles. Keep the corresponding scenario epoch, truth, trajectory and propagation definitions consistent.
For each profile, declare preparation and acquisition limits before execution, then score a fixed 300 s window at 1 Hz: 300 planned PVT epochs. The illustrative PVT rule requires a valid 3D solution, horizontal error ≤3.0 m against epoch-aligned independent truth, and output age from 0 to 0.50 s. The illustrative availability threshold is 99%, or at least 297 qualified epochs. Use a predeclared scoring start, not a window selected after the first successful fix.
Output age is completed-record arrival time minus its measurement epoch on the verified common time basis. Future epochs, unknown age and unverified time mappings cannot pass this freshness check. No external corrections or assistance are used in this example. All limits must be replaced with project requirements.
| Case | 1575.42 MHz signals | 1176.45 MHz signals | Comparison role | Qualified PVT epochs |
|---|---|---|---|---|
| M01 | GPS L1 C/A | None | GPS single-frequency baseline | 298/300 · 99.33% |
| M02 | Galileo E1-B/E1-C | None | Galileo single-frequency baseline | 297/300 · 99.00% |
| M03 | GPS L1 C/A | GPS L5 I5/Q5 | GPS frequency extension | 299/300 · 99.67% |
| M04 | Galileo E1-B/E1-C | Galileo E5a-I/E5a-Q | Galileo frequency extension | 299/300 · 99.67% |
| M05 | GPS L1 C/A + Galileo E1-B/E1-C | None | Two constellations at one carrier center | 298/300 · 99.33% |
| M06 | GPS L1 C/A + Galileo E1-B/E1-C | GPS L5 I5/Q5 + Galileo E5a-I/E5a-Q | Combined dual-frequency operation | 270/300 · 90.00% |
M01–M05 meet the example PVT threshold; M06 fails it. These counts score PVT only. Full profile acceptance also needs separately declared acquisition, tracking, navigation-data and signal-use checks. A good position produced entirely from a subset of the intended signals cannot establish the remaining coverage.
Keep the denominator visible
In the hypothetical M06 record, 270 epochs meet all PVT conditions, 20 return valid fresh positions outside the error limit, and 10 have confirmed missing DUT output. Capture is healthy throughout.
Qualified PVT availability: 270 / 300 = 90.00%
Returned-output-only ratio: 270 / 290 = 93.10%, which excludes the missing epochs and is a different metric.
Retain every planned epoch. Do not fill missing coordinates with zero error or carry forward the last good position as a new measurement. If capture itself is incomplete, mark the affected evidence unverified rather than assigning a proven receiver dropout.
Record What Was Delivered, Tracked and Used
A configuration screen, a satellite counter and a PVT fix answer different questions. Maintain a per-signal record keyed by case, trial, measurement epoch, constellation, satellite and signal ID.
| Boundary | Evidence to preserve | Interpretation limit |
|---|---|---|
| Configured | Source manifest, DUT readback, firmware, licenses, active signal plan and antenna/RF path. | A requested setting does not prove it became active. |
| Delivered in RF | Verified signal content, timing and per-signal level at the DUT reference plane. | A source log or total-power reading alone does not prove every named waveform arrived correctly. |
| Acquired / tracked / decoded | Signal IDs, lock state, time synchronization, applicable navigation-data validity and C/N0. | A listed or searched signal is not necessarily locked; pilot tracking alone does not demonstrate data decoding. |
| Used by the receiver | Documented code, carrier or Doppler use indications, associated with the navigation epoch. | Tracking does not establish use; one observable’s use does not establish another’s. |
| PVT qualified | Validity, truth-aligned position error, output age, continuity and any required solution mode. | Accurate PVT can coexist with an untested or unused signal group. |
A concrete logging example
The u-blox X20 HPG 2.10 interface, protocol 50.11, provides UBX-NAV-SIG records with gnssId, svId, sigId, qualityInd and prUsed/crUsed/doUsed flags. A listed signal may still be searching. The use flags distinguish pseudorange, carrier range and Doppler; Doppler use alone does not prove that signal’s pseudorange contributed to position.[3]
Use the actual DUT’s message version and validity rules. Match status to the PVT measurement epoch. Define signal-group aggregation and required satellite coverage before scoring: a receiver may process data/pilot components jointly, so requiring every component row to show use can misclassify valid operation. If the needed use evidence is unavailable, report that limitation.
Control the Comparison Before Explaining a Difference
M01 versus M03 and M02 versus M04 compare frequency extensions for the same constellation and selected satellite set. M01 versus M05 adds Galileo and changes available geometry and processing load. GPS-only versus Galileo-only is a configuration comparison; it does not isolate constellation quality.
- Preserve common signals. Keep their satellite identities, epochs, navigation data, trajectory and declared power definition consistent. State whether power is per component or the combined signal power, such as L5 I+Q.
- Record aggregate changes. Adding signals changes total power and resource load. Do not normalize total signal power by silently reducing existing signals, or boost survivors when other signals are removed. Check noise, combining paths and headroom.
- Verify configuration transitions. Read back the active profile and apply the declared initialization. Some receiver signal-configuration changes trigger a GNSS subsystem reset, as documented in the X20 interface.[3] Do not interpret that reset as an RF-loss response.
- Repeat from controlled states. Record retained navigation/filter state, acquisition budget, scoring start, run order and trial count. Keep failed acquisition and timeout trials in the campaign record.
- Repeat meaningful contrasts. Re-run a failing profile and its controls. A firmware comparison needs the same input recipe and documented configuration differences.
The clean nominal matrix establishes a starting point. Add relevant dynamics, propagation, signal levels and channel loads in separate stages. If the objective is ionospheric mitigation, use a validated frequency-dependent propagation case; merely tracking two bands does not establish how the receiver uses them.
Separate Signal Loss from Band Loss and DUT Reconfiguration
For continuity testing, keep the DUT in the confirmed M06 configuration and change only the declared source input. In this illustrative schedule, remove the target signals at 60 s and restore them at 80 s while the remaining scene continues. Confirm the actual RF events and surviving signal levels.
| Case | Controlled source change | What to examine |
|---|---|---|
| L01: GPS L5 loss | Remove GPS L5 I5/Q5; retain Galileo E5a and all 1575.42 MHz signals. | GPS secondary-signal loss, remaining signal use, PVT continuity and reacquisition. |
| L02: 1176.45 MHz loss | Remove both GPS L5 and Galileo E5a; retain GPS L1 C/A and Galileo E1. | Transition to the remaining single-frequency inputs and subsequent recovery. |
| L03: Galileo loss | Remove Galileo E1 and E5a; retain GPS L1 C/A and L5. | Constellation loss with GPS dual-frequency inputs still present. |
L01 needs verified signal-selective control: ordinary filtering of the shared 1176.45 MHz band cannot remove GPS L5 while preserving Galileo E5a. Likewise, switching off a combined RF output may remove more signals than the intended case. Review source controls and routing before execution.
Declare continuity, error and status requirements for each phase, including any permitted transition interval. Define recovery start at verified signal restoration at the DUT plane and the endpoint at completion of the required qualified dwell. Track signal reacquisition, renewed observable use and PVT recovery separately; one may precede another.
A healthy fallback PVT can be the expected result. Do not require a navigation outage simply to demonstrate that the removed signal mattered. Report whether the receiver detected the change, used the survivors appropriately and met the application criteria.
Reduce a Failing Combination Without Guessing the Root Cause
The M06 result is an interaction symptom: its PVT availability is 90%, while the individual dual-frequency profiles and mixed single-frequency profile meet the example threshold. Investigate the earliest boundary where the execution record diverges from the recipe.
| Observed boundary | Checks to run next | Supported conclusion |
|---|---|---|
| RF content differs from the recipe | Inspect licenses, signal generation, shared-band routing, timing and level calibration. | The intended stimulus is not established; receiver performance remains unverified for that case. |
| RF verified; acquisition or decoding differs | Inspect DUT readback, active signal plan, signal identities, navigation data and processing load. | The symptom is at the acquisition/tracking/data boundary; the cause still needs isolation. |
| Locked signals show unexpected use | Check measurement quality, health, rejection masks, estimator selection and the meaning of the reported flags. | Tracking and observable use diverge; this alone does not prove a firmware resource defect. |
| Use evidence meets the recipe; PVT fails | Check truth/epoch alignment, frames, reference points, biases, measurement consistency and navigation processing. | A result-level failure is confirmed after the earlier evidence boundaries pass. |
Worked reduction of M06
Assume RF verification passes and the example trace shows intermittent secondary-band range-use indications despite lock. Run controlled reductions with the DUT configuration fixed, preserve the retained signals, and prepare each trial consistently:
- Remove only GPS L5. The illustrative PVT result becomes 298/300 qualified epochs.
- Restore M06, then remove only Galileo E5a. The illustrative result becomes 299/300.
- Restore the full M06 input and repeat. The illustrative result returns to 270/300.
These hypothetical results associate the failure with the joint input condition in this setup. They do not identify whether measurement rejection, resource allocation, signal-generation interaction or another factor is responsible. Preserve the failing case and controls for the next discriminating test.
Extend the campaign with additional named signals, BeiDou or GLONASS only after defining their own controls and evidence. Pairwise coverage can help prioritize a large matrix, but cannot demonstrate that every higher-order combination works. Record tested, unsupported, blocked, failed and unverified cases separately.
Choose a Source Configuration That Can Execute the Matrix
Compare the GNSS signal simulator systems against the exact simultaneous signal sets, satellite/processing load, RF routing, selective controls and required logs. A list of supported bands does not establish that every combination can run together.
Review NSG8000 for Combination Regression
The NSG8000 product page describes configurable multi-constellation, multi-frequency generation and channel power control.[4] Confirm exact components, concurrent signal sets, licenses, channel accounting and event controls for the proposed campaign.
Check that the selected configuration can remove one named signal group while retaining another at the same carrier center. Confirm the control/logging interfaces needed to reproduce and export the failing case.
Review NSG8000 ConfigurationUse the GNSS signal simulation solution for bench planning and the receiver validation test plan for the wider campaign. Signal-combination regression is one part of receiver validation.
Turn Your Signal List into a Reproducible Test Campaign
Share the DUT model/firmware, exact signal components and simultaneous combinations, a passing and failing case if available, RF path, signal-level logs and acceptance targets. These details support a configuration review covering source generation, selective events, measurement evidence and repeatable execution.
Multi-Frequency GNSS Testing FAQ
Is multi-constellation the same as multi-frequency?
They are separate test axes. GPS L1 C/A plus Galileo E1 combines two constellations at one carrier center. Adding GPS L5 and Galileo E5a introduces a second carrier center in this example.
Does a valid position prove all enabled signals were used?
Check signal-level tracking and documented observable-use evidence alongside PVT. The position may meet its error limit using only part of the intended input set.
Why can separate signal profiles work while their combination fails?
The combined condition changes processing load, total input power, measurement selection and available geometry. Reduce the input set and verify each evidence boundary before assigning a root cause.
Can a band filter isolate GPS L5 from Galileo E5a?
They share the same nominal carrier center. A band-removal test affects both; a signal-selective test requires suitable source control and verification of the retained signal.
How many combinations are enough?
Base coverage on supported configurations and application risk. Test required nominal combinations, relevant transitions and known failure conditions; document the combinations left untested. No small generic matrix establishes complete qualification.
Technical References
Signal names and logging examples follow the cited documents. Matrix results, thresholds and reduction outcomes are illustrative and do not describe measured product performance.
- GPS.gov: IS-GPS-200N and IS-GPS-705J — L1/L2 and L5 signal specifications. Consult the interface-document index for applicable revision notices.
- European Union: Galileo OS SIS ICD, Issue 2.2 — carrier frequencies and signal components.
- u-blox: X20 HPG 2.10 Interface Description, R02 — protocol 50.11 signal status, observable-use flags and configuration behavior.
- NSG8000 product page — published platform capabilities and configuration scope.