test: real-data PVT integration test (Fraunhofer L125, GPS L1 C/A)#44
Merged
Conversation
Opt-in integration test that computes a full PVT fix end to end from a real GNSS recording, the real-signal counterpart to the synthetic pvt.jl tests: Acquisition.jl acquires GPS L1 C/A, Tracking.jl tracks, GNSSDecoder.jl decodes the navigation message, and calc_pvt produces a position that is asserted to land at the recording site. - Capture: Fraunhofer IIS "L125" static roof recording (Nuremberg, 2014-09-23, CC BY-NC). GPS L1 C/A only, as it is the sole constellation with >=4 satellites that decode a complete ephemeris (calc_pvt needs >=4 of the same system); the III-7a capture used by Tracking/GNSSDecoder is too short (~15.7 s) to decode a full ephemeris. - Streaming: range-download only a bounded compressed prefix of the zip (cached on CI via PVT_TESTDATA_DIR), then raw-inflate and demux one 0.2 s window of int8 I/Q at a time straight into track!, mirroring JuliaGNSS/Tracking.jl#158 so peak memory is a single window rather than the whole decoded signal. - Asserts the fix lands within 100 m of the (deterministic) decoded Fraunhofer-roof position and that the decoded time is GPS week 1811 (2014-09-23). - Gated on ENV["PVT_RUN_INTEGRATION_TEST"]; a plain `]test` skips it (no download). CI runs it on a dedicated job that caches the prefix. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #44 +/- ##
=======================================
Coverage 92.77% 92.77%
=======================================
Files 7 7
Lines 360 360
=======================================
Hits 334 334
Misses 26 26 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
Contributor
Benchmark Results (Julia v1)Time benchmarks
Memory benchmarks
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Adds an opt-in real-data integration test that computes a full PVT fix end to end from a real-sky GNSS recording — the real-signal counterpart to the synthetic
pvt.jl/pvt_iono_tropo.jltests. It exercises the whole JuliaGNSS stack the way a receiver does: Acquisition.jl acquires GPS L1 C/A, Tracking.jl tracks, GNSSDecoder.jl decodes the navigation message, andcalc_pvtproduces a position that is asserted to land at the recording site.Modelled on the GNSSDecoder.jl / Tracking.jl Flexiband integration tests, but taken all the way to a position fix.
Capture & why GPS L1 C/A only
.bin(zero-IF complex baseband, fs = 20 MHz, interleaved int8 I/Q — parameters from Fraunhofer's own GNSS-SDR config).calc_pvtneeds ≥4 healthy satellites of the same GNSS. GPS L1 C/A is the only constellation here with ≥4 satellites that decode a complete ephemeris. (The III-7a capture that Tracking/GNSSDecoder use is only ~15.7 s — too short to decode a full ephemeris, so it can't yield a fix.)Streaming demux (constant memory)
Mirrors JuliaGNSS/Tracking.jl#158: the full archive is ~3.45 GB, so we range-download only a bounded compressed prefix (~2 GB) and cache that (never the decoded signal). During tracking it raw-inflates the prefix and demuxes one 0.2 s window of int8 I/Q → ComplexF32 at a time, feeding each straight into
track!. Peak memory is a single window (~16 MB), not the ~16 GB a whole-span decode would need.Assertions
Run against the real capture, the pipeline decodes 4 satellites (PRNs 13, 17, 20, 23) and fixes at 49.4873°N, 11.1255°E, 421 m (the Fraunhofer roof). The fix is fully deterministic (bit-identical single- vs multi-threaded), and the satellite set is pinned so the geometry is fixed too, so the test asserts:
Opt-in & CI
Gated on
ENV["PVT_RUN_INTEGRATION_TEST"]— a plain]testskips it (no download). A dedicated CI job enables it and caches the compressed prefix (actions/cache, fixed key) so repeat runs skip the download.Verified locally: default suite passes with the integration test skipped; the integration test passes 8/8 with the fix above.
🤖 Generated with Claude Code