MRTKLIB v0.4.3 Release Notes¶

Release date: 2026-03-09 Type: Feature (real-time CLAS PPP-RTK via rtkrcv) Branch: feat/rtkrcv-clas-realtime

Overview¶

v0.4.3 enables real-time CLAS PPP-RTK positioning via rtkrcv using BINEX/SBF observation streams and CLAS L6D correction streams. The same PPP-RTK engine used by rnx2rtkp (post-processing) now runs in the rtksvr real-time pipeline, achieving 97.7% fix rate (3517/3599 Q=4 epochs) on the 2019/239 reference dataset — matching the post-processing engine's 99.86% steady-state fix rate after convergence.

Highlights¶

• Real-time CLAS PPP-RTK — rtkrcv processes BINEX or SBF observations alongside CLAS L6D corrections to produce centimetre-level PPP-RTK solutions in real time.
• L6 rate limiter — Prevents the L6 correction stream from overrunning observations during file replay, protecting the CLAS bank ring buffer from premature overwrite.
• Time-tag generators — New gen_bnx_tag.py and gen_l6_tag.py scripts create RTKLIB-compatible .tag files for synchronised ::T::xN file replay.
• CTest rtkrcv_rt_clas — Automated regression test replaying 1 hour of BINEX+L6 data at 10x speed (~370s wall time), verifying >=90% output line coverage.
• Documentation — docs/rtkrcv-clas-realtime.md with setup guide, PP vs RT performance comparison, architecture overview, and troubleshooting notes.

Post-Processing vs Real-Time Performance¶

Both modes use the same CLAS PPP-RTK engine and the same 2019/239 dataset (0627 station, 2019-08-27 16:00-17:00 UTC, Trimble NetR9).

The 77 Q=1 epochs represent the initial convergence period (~77 seconds). After convergence, the steady-state fix rate is identical to post-processing.

What Changed¶

L6 Rate Limiter (mrtk_rtksvr.c)¶

During file replay, the L6 correction stream can deliver data faster than the observation stream consumes it. Without throttling, the CLAS bank ring buffer (32 entries, ~16 minutes of corrections) is overwritten before the positioning engine can use the data.

The rate limiter pauses L6 byte processing when the L6 decode time exceeds the observation time by more than 60 seconds. Unconsumed bytes are shifted to the start of the buffer for processing in the next server cycle.

gtime_t l6t = svr->clas->l6buf[ch].time;
gtime_t obst = svr->raw[0].time;
if (l6t.time > 0 && obst.time > 0 && timediff(l6t, obst) > 60.0) {
    /* pause L6 processing; shift remaining bytes */
}

CLAS Real-Time Integration (mrtk_rtksvr.c)¶

The rtksvr thread now: 1. Initialises the CLAS context (clas_ctx_t) when the L6 stream format is detected 2. Wires the CLAS context into nav->clas_ctx for PPP-RTK engine access 3. Initialises week_ref[] from the first valid observation time 4. On each CSSR epoch boundary (clas_decode_msg() == 10), performs bank lookup and grid status checks for the detected network

Time-Tag Generators¶

gen_bnx_tag.py¶

Parses BINEX 0x7F-05 observation records to extract epoch timestamps (minutes since GPS epoch + milliseconds). Creates one tag entry per observation epoch with accurate file position mapping.

gen_l6_tag.py¶

Generates tags for L6 data at 1-second intervals (250 bytes/frame). With --sync-tag, synchronises to a master tag file by computing the GNSS-time offset and adjusting tick_f and tick_n scaling. Automatically detects compressed master tags (recorded faster than real-time) and adjusts accordingly.

Test Script Parameterisation (run_rtkrcv_test.sh)¶

The rtkrcv test script now accepts optional parameters:

bash run_rtkrcv_test.sh <binary> <project_root> <reference> \
     [playback_speed] [conf_file] [port] [max_timeout]

This enables reuse for both the MADOCA rtkrcv_rt test (default conf) and the CLAS rtkrcv_rt_clas test (custom conf, port, and timeout). Both tests use RESOURCE_LOCK rtkrcv_port to prevent parallel execution conflicts.

Debug Cleanup¶

Removed temporary debug fprintf statements from five source files: - mrtk_clas.c — orbit/pbias/trop lookup diagnostics - mrtk_clas_grid.c — grid pickup and trop interpolation diagnostics - mrtk_clas_osr.c — OSR trop grid status - mrtk_ppp_rtk.c — filter/zdres/ddres stage diagnostics

Configurations¶

Three rtkrcv configurations for different input stream combinations:

Test Suite (58 tests)¶

58 total tests (57 from v0.4.2 + 1 new rtkrcv_rt_clas).

claslib_ppp_rtk_2ch_check remains a marginal failure (0.202 m vs 0.200 m tolerance) unrelated to this release.

Files Changed¶