GNSS Simulator
Talk to an engineer
GNSS Simulator

GNSS Receiver Testing Guide

GNSS Multipath in Urban Canyons: Causes, Symptoms, and 9 Receiver Tests

A receiver can report a plausible position in a city even when the position is wrong. Buildings obstruct direct satellite signals and reflect others, so a useful validation plan must examine more than a single accuracy number.

Quick answer: Urban-canyon GNSS errors usually combine sky masking (too few usable direct signals), multipath (direct and reflected signals arriving together), and non-line-of-sight reception (the direct path is blocked and a reflection is tracked). Diagnose them with time-aligned signal, geometry, position-error, availability, and recovery data. Then test the receiver under controlled scenarios, recorded field RF where appropriate, and structured field trials.

These nine tests are a receiver-validation framework, not a universal pass/fail standard. Define thresholds from the application requirement before running them. Where an RF simulator does not support a particular impairment model, use an appropriately captured field recording or a controlled field test instead.

Why urban canyons degrade GNSS positioning

An urban canyon is a street or operating area where buildings restrict the view of the sky and create strong reflections. Satellite geometry changes along a route, and the receiver may move repeatedly between open, masked, and reflected-signal conditions. GPS.gov identifies building blockage and multipath as factors that degrade received GPS accuracy.[1]

Effect What happens What to watch
Sky masking Buildings block direct signals, reducing usable satellite visibility and sometimes weakening geometry. Tracked-satellite count, geometry indicators such as DOP, fix availability, and dropouts.
Multipath A direct signal and one or more delayed reflections reach the antenna together, potentially biasing code or carrier measurements.[2] Position-error tails, per-satellite measurement behavior where available, and changes near reflective structures.
NLOS reception The direct path is blocked and the receiver may track a reflected path instead. A fix can remain available while its error grows. Plausible but wrong positions, integrity or confidence outputs, and recovery when direct signals return.[3]

The effects often overlap. A high DOP can indicate weak geometry, but a reasonable DOP does not prove that the measurements are free of reflection bias. Likewise, C/N0 is a signal-strength indicator, not a direct accuracy or line-of-sight guarantee. Use these indicators alongside position error and event timing, rather than treating any one as a verdict.

How to read urban-canyon failure signatures

Before changing the receiver or the test setup, identify the behavior you need to explain. Use a reliable reference position or trajectory when claiming measured accuracy; without one, describe availability, continuity, signal, and receiver-diagnostic behavior instead of calling it position error.

Observed signature Likely question to investigate Useful measurements
Satellite count falls, DOP rises, then the fix disappears. Is sky masking or geometry limiting the solution? Visible and tracked satellites, DOP, fix availability, outage duration, and reacquisition time.
C/N0 looks acceptable, but the position jumps or drifts. Are reflected or NLOS measurements biasing the solution? Reference-position error, per-satellite diagnostics if exposed, C/N0 history, and event markers.
Median error looks acceptable but the worst intervals are poor. Are short NLOS or transition events being hidden by the average? 95th-percentile error, maximum error with context, time within the required error bound, and event timing.
The fix returns slowly after a building shadow or turn. Is recovery behavior the real application risk? Outage count, time to first valid fix, time to stable performance, and stale-output behavior.

For definitions of DOP, C/N0, TTFF, CEP, RMS, and percentile error, see the GNSS accuracy metrics guide. For the wider receiver pipeline, see how GNSS receivers work.

C/N₀ looks normal, but position jumps: a worked example

A receiver continues reporting a position while moving through a built-up street. Signal-level and geometry indicators change only slightly, but its horizontal error increases. What can you conclude, and what evidence should you collect next?

Synthetic teaching example. Every number below is invented to explain a diagnostic workflow. This is not a customer recording, an equipment benchmark, or measured NSG8000 performance. These values are not acceptance limits.

Assume an independently validated reference trajectory, aligned with receiver outputs by measurement epoch, coordinate frame, and antenna reference point. The C/N₀ column follows one unchanged signal identity: GPS G12 L1 C/A. It is not an average over changing satellites.

Selected epochs from a hypothetical urban-route test
Elapsed time Route segment G12 L1 C/A C/N₀ (dB-Hz) HDOP Satellites used Horizontal error (m)
00:00 Open-sky approach 42 1.2 12 0.9
00:10 Approaching buildings 41 1.3 11 1.3
00:20 Built-up street 40 1.4 10 12.8
00:30 Built-up street 41 1.4 10 18.6
00:40 Leaving the street 42 1.2 11 4.2
00:50 Open-sky exit 42 1.2 12 1.1

Highlighting marks two selected epochs with elevated error, not failure against a universal threshold. One signal does not describe the entire constellation. Six selected samples cannot establish run-wide percentiles, continuous fix availability, or an exact recovery time.

1. Separate the observation from the explanation

The selected 00:20 and 00:30 samples show larger horizontal errors than the open-sky samples, while the monitored C/N₀ and reported HDOP remain similar. That is the observation. It does not identify a faulty satellite or establish that the monitored signal arrived along a direct path.

Multipath and NLOS are plausible hypotheses. Reference errors, timestamp misalignment, correction changes, and receiver configuration also require investigation. This table alone cannot distinguish the cause.

2. Verify time alignment and reference quality first

Compare positions at the measurement epoch, not automatically at host-message arrival. Check the coordinate frame, antenna offsets, and reference uncertainty through the difficult interval. A second GNSS receiver exposed to the same environment is not automatically reliable ground truth.

If the reference becomes unreliable, mark the interval accordingly. You can still examine output continuity and signal diagnostics, but should not claim measured position accuracy for that interval.

3. Inspect the complete receiver timeline

  • Solution evidence: validity flags, solution mode, measurement age, missing epochs, and correction status where relevant.
  • Signal evidence: satellite and signal identities, tracked-versus-used status, C/N₀, lock time, cycle slips, observables, and residuals where exposed and documented.
  • Environmental evidence: time-aligned route markers, imagery, turns, and obstruction observations. A nearby building alone does not establish which signal was reflected.

NMEA can help describe PVT and selected status information, depending on the messages enabled. Do not assume it exposes the raw observables or internal diagnostics needed to confirm a multipath mechanism.

4. Design a repeatable follow-up

Repeat the baseline, then compare configurations using the same suitable RF recording or validated simulator scenario where supported. Define the required error bound, usable-output criteria, and recovery endpoint in advance. Preserve failed, invalid, and missing outputs alongside successful fixes.

A reproducible error is useful regression evidence; reproducing it does not, by itself, prove its physical cause. Continue with the nine receiver tests and record the plan in the editable worksheet.

Nine practical tests for an urban-canyon receiver

Hold the receiver configuration, antenna arrangement, logging, and reference method constant when comparing runs. Record the constellation and frequency configuration, firmware version, start state, scenario ID, and event times. Use conducted or properly shielded RF testing where required, and follow applicable local rules for any field setup.

  1. Establish an open-sky baseline

    Purpose: Confirm that the receiver, reference system, and logging pipeline work before introducing urban stress. Method: Run a defined open-sky or lightly masked scenario with the same receiver configuration planned for later tests. Measure: Fix availability, median and 95th-percentile position error when ground truth is available, DOP, C/N0 trends, and startup or reacquisition definitions.

  2. Reduce sky visibility in controlled steps

    Purpose: Separate loss of direct satellite visibility from reflection effects. Method: Apply progressively restricted visibility or use well-characterized route segments with different sky view. Change one condition at a time. Measure: Tracked satellites, DOP, percentage of time with a usable fix, dropout rate, and the error distribution.

  3. Compare geometry at similar satellite counts

    Purpose: Show why “number of satellites” alone is not a sufficient quality metric. Method: Compare scenarios or intervals with similar usable-satellite counts but different spatial geometry, where the test platform and dataset permit this. Measure: DOP, solution availability, median error, and 95th-percentile error. Keep signal quality and receiver settings as comparable as possible.

  4. Isolate multipath where the setup supports it

    Purpose: Observe the receiver response to a direct signal plus reflections. Method: Use a validated multipath-capable scenario or a characterized recording; do not assume that every GNSS simulator can generate the required delayed paths. Measure: Changes in reference-position error and tail error, tracking continuity, and per-satellite measurement diagnostics if the receiver exposes them.

  5. Examine NLOS-dominant intervals

    Purpose: Find plausible-looking wrong fixes rather than only complete outages. Method: Select field intervals with documented blocked direct paths, or use a validated environment model. Label the interval and preserve the basis for that label. Measure: Error against reference truth, confidence or integrity outputs, time outside the required error bound, and behavior when direct reception returns.

  6. Hold position in a difficult location

    Purpose: Reveal drift and persistent bias that a moving-route average can hide. Method: Keep the antenna stationary for a defined dwell period at a characterized location or replay a comparable stationary capture. Measure: Error time series, median and 95th-percentile error, fix continuity, and whether a small average conceals long excursions.

  7. Test stop-turn-start transitions

    Purpose: Capture the unstable moments as the receiver enters or leaves a building shadow. Method: Use repeatable route segments or controlled trajectories with explicit transition markers. Measure: Position jumps, heading or velocity consistency where applicable, dropout count, and time to stable performance after each transition.

  8. Measure outage and recovery

    Purpose: Evaluate continuity, not just steady-state accuracy. Method: Introduce or select defined periods of signal loss or severe masking, then restore reception under consistent conditions. Measure: Outage duration, time to first valid fix, time to a stable solution, and whether stale or low-confidence positions are correctly flagged. Define “stable” from the application requirement before testing.

  9. Run a controlled receiver A/B and field-to-lab regression

    Purpose: Determine whether a configuration or firmware change improves the difficult intervals without damaging the baseline. Method: Replay the same characterized RF capture or repeat the same controlled scenario for each version; then confirm the result with structured field runs. Multi-constellation or multi-frequency comparisons are useful only when the receiver and dataset support them. Measure: Availability, 95th-percentile error, event-specific error, recovery time, and run-to-run variation.

Do not copy thresholds from another application. A consumer tracker, fleet device, survey receiver, and autonomous platform can have very different acceptable error, availability, and recovery requirements. Set pass/fail limits and the ground-truth method before comparing systems.

Urban-canyon GNSS test worksheet

Use one worksheet per test case and configuration. Edit the template below, then copy it into your test record or download a plain-text copy. Set acceptance criteria from your application before collecting results.

Record Include Why it matters
Configuration Receiver, firmware, signals, antenna, RF path, startup state, and corrections. Makes the setup and comparison boundary explicit.
Input and reference Scenario or RF-recording ID, route interval, event times, reference method, and uncertainty. Connects an observed failure to a documented input and defensible reference.
Acceptance and result Error definition and limit, usable-output fraction, outage/recovery criteria, repetitions, exclusions, and outcome. Prevents a best-case screenshot or successful-only sample from becoming the verdict.

Edits stay in this page's memory and may be lost when you leave or reload. Copy or download your work. These controls do not submit it to our team.

Choose the right test method: simulation, RF replay, and field

These methods answer different questions and work best together. A generated GNSS scenario is valuable for controlled parameter changes; a field RF recording preserves a real event for repeated receiver input; a field trial confirms the installed system in its operational context. None is a complete substitute for the others.

Method Best use Important limitation
GNSS simulation Controlled visibility, geometry, motion, and other supported scenario parameters; systematic A/B tests. Only claim a multipath or NLOS effect if the selected system and configuration actually model it.
Signal recording and playback Repeat a captured urban route or failure event across receiver and firmware versions. Replay reproduces what was captured; it does not independently vary every environmental factor.
Structured field testing Confirm antenna installation, system integration, and performance in a real operating environment. Satellite geometry and surrounding conditions change, so field runs need careful logging and comparison design.

Example workflow: compare two firmware versions on the same urban RF capture

  1. Qualify the dataset. Confirm captured bands, usable recording quality, duration, timing, and available reference data. A recording from one antenna arrangement does not represent every installation.
  2. Define configurations A and B. Record firmware versions and all enabled settings. Use a documented, equivalent startup/reset procedure and the same replay segment and calibrated input levels.
  3. Repeat each configuration. Use a declared run count and check that the playback chain is repeatable. Where practical, alternate run order to reduce confounding from drift or changing equipment conditions.
  4. Compare complete outcomes. Align outputs to the same measurement epochs. Include invalid and missing outputs, time within the error requirement, outage/recovery behavior, and the labeled difficult intervals.
  5. Confirm in the field. Check the actual installation and additional routes. A better result on one recorded dataset does not establish improvement in every operating environment.

Replay preserves the combined input that was captured. To isolate one environmental factor, use a validated scenario that actually supports the required control, or design a separate controlled experiment.

For a deeper method comparison, read Record & Replay vs GNSS Simulation vs Field Testing. To design controlled scenarios, use the scenario-design guide; for a complete KPI and reporting framework, use the receiver validation test plan.

If your program needs generated GNSS scenarios, explore the GNSS signal simulation solution and NSG8000. If the priority is preserving and replaying actual field RF, compare the signal recording and playback systems. Confirm the required bands, impairment models, RF paths, and test configuration with the engineering team before selecting equipment.

What an urban-canyon test report should show

Report the distribution and timing of failures, not only an average or a best screenshot. ESA Navipedia describes accuracy, availability, continuity, and integrity as distinct performance concepts; an application may need more than one of them.[4]

  • Test definition: receiver and antenna, firmware, frequency and constellation settings, scenario or route ID, start state, repetitions, and reference-truth method.
  • Accuracy: median and 95th-percentile error for the whole run and for labeled worst intervals, only where trustworthy reference truth exists.
  • Availability and continuity: proportion of time meeting the application requirement, dropouts, outage lengths, and recovery times.
  • Diagnostic context: tracked satellites, DOP, C/N0, per-satellite diagnostics where available, and timestamps for masking, turns, or NLOS-dominant intervals.
  • Decision: pass/fail against requirements set in advance, plus changes relative to the open-sky baseline and the previous receiver version.

If you cannot establish reliable ground truth for a field interval, label the limitation explicitly. Receiver-reported precision or confidence is not a substitute for measured position error.

Common mistakes to avoid

  • Treating multipath, NLOS, and masking as one effect. They can overlap, but each calls for different diagnostics.
  • Using DOP or C/N0 alone as an accuracy verdict. Geometry and signal strength do not fully describe measurement bias.
  • Reporting only mean error. Short wrong-fix events can dominate the user experience while barely moving the average.
  • Comparing unmatched field drives as a controlled A/B test. Use repeatable scenarios or the same RF recording for the direct comparison, then confirm in the field.
  • Claiming the simulator reproduces an impairment without checking its configuration. Verify the actual scenario and RF capabilities before promising a test outcome.

Frequently asked questions

Why does GPS drift or jump between buildings?

Buildings block some direct satellite signals and reflect others. The receiver may have poor geometry, biased measurements, or both. It can lose the fix, drift, or continue reporting a plausible but incorrect position.

Is NLOS the same as multipath?

No. Multipath commonly describes a direct signal arriving along with reflections. In an NLOS case, the direct path is blocked and a reflected path may be the only tracked signal. Both can create position error, but their test interpretation differs.

Can good C/N0 or low DOP prove that the position is correct?

No. C/N0 and DOP are useful diagnostics, but neither directly measures whether a tracked path is unbiased. Compare them with reference-position error and time-aligned events whenever possible.

Does adding more constellations solve the urban-canyon problem?

More usable satellites can improve availability and sometimes geometry under masking. They do not automatically remove reflection or NLOS bias. Measure the change on the same test conditions rather than assuming an improvement.[3]

Can RF replay replace field testing?

Replay improves repeatability for conditions already captured. Field testing remains necessary to confirm installation, integration, and behavior outside the recorded dataset.

Which metrics matter most?

Start with the application's required position-error bound and usable-fix definition. Then measure median and 95th-percentile error where reference truth is available, availability, outage and recovery behavior, and event-aligned signal diagnostics.

Technical references

  1. GPS.gov — GPS Accuracy . Overview of geometry, blockage, receiver design, and multipath effects.
  2. ESA Navipedia — Multipath . Signal-reflection mechanisms and measurement effects.
  3. ESA Navipedia — Receiver Characteristics . Urban-environment trade-offs, including multipath, NLOS, and availability.
  4. ESA Navipedia — GNSS Performances . Definitions of accuracy, availability, continuity, and integrity.
Scroll to Top