Receiver protocols read by receiver-trust (0.38)
Read only, throughout: nothing here writes a byte to a receiver, and there is no transmit path of any kind. Advisory software, not type-approved navigation equipment (IEC 61108, IEC 61162): the operator remains responsible.
The receiver's own spoofing and jamming flags (UBX)#
| message | field | read as |
|---|---|---|
| UBX-NAV-STATUS (0x01 0x03), 16 bytes | flags2 byte at offset 7, bits 4..3: spoofDetState |
0 unknown or deactivated, 1 no spoofing indicated, 2 spoofing indicated, 3 multiple spoofing indications (protocol 18 and later; u-blox F9 HPG 1.32 interface description, section 3.15.20) |
| UBX-SEC-SIG (0x27 0x09) version 1, 12 bytes | jamFlags at 4 (bit 0 jamDetEnabled, bits 2..1 jammingState); spfFlags at 8 (bit 0 spfDetEnabled, bits 3..1 spoofingState) |
states are read only where the detector is enabled; 0 unknown, 1 none, 2 warning or indicated, 3 critical or affirmed |
| UBX-SEC-SIG version 2 | sigSecFlags at 1 (bit 0 jamDetEnabled, bits 2..1 jammingState, bit 3 spfDetEnabled, bits 6..4 spoofingState); jamNumCentFreqs at 3 and that many 4-byte groups (not used) |
as above |
Source of the SEC-SIG layout. The u-blox F9 HPG 1.32 and M10 SPG 5.10 interface descriptions that
could be downloaded when this was written name UBX-SEC-SIG but do not define it. The layouts above are
those of the later u-blox interface descriptions as transcribed in the pyubx2 library
(BSD-3-Clause, SEC-SIG-V1 and SEC-SIG-V2); the fixture tests in
tests/receiver_trust_self_report.rs are built from that layout, not from a real receiver's output.
Layout from open transcription, not checked against the current vendor guide: check it against the current u-blox interface description before relying on SEC-SIG.
A SEC-SIG has no time of week; it belongs to the current navigation epoch. A state already read at an epoch is kept unless a later one is worse.
The receiver-self-report comparison#
Where a log carries these flags, every epoch of the result has self_report (the two states, whether the
receiver flags, whether any Kshana monitor alarmed, and agreement: both, receiver-only,
kshana-only, neither, or calibrating), and the result has receiver_self_report with the counts and
the first epoch of each. The receiver "flags" at a state of 2 or more. It raises no alarm and changes no
score, and neither side is ground truth: agreement is not validation, and a disagreement says one
detector saw something the other did not, nothing about which was right. Studio's strip reads
epochs[].self_report.
Septentrio SBF (format = "sbf")#
receiver_trust::ingest::read_sbf decodes with the sbf-tools crate (0.4, MIT; its only dependency
is thiserror; a permissive licence that deny.toml allows). Blocks read: PVTGeodetic (4007) for the fix,
MeasEpoch (4027) for C/N0 per satellite and signal (band is the signal's name, such as L1CA), and
RFStatus (4092) for the number of RF bands whose Info byte is non-zero (rf_bands_reported, carried and
not interpreted: it never counts as a flag). Epochs are keyed by time of week and week number and
labelled in GPS time (2024-09-05T00:00:00.000 GPST). A block failing its CRC or parse counts as a
skipped record. This reader does not read AGC or a jamming indicator, so the C/N0-drop and loss-of-lock monitors run
and the AGC and jamming-indicator monitors do not.
Source of the layouts. Septentrio's reference guide could not be downloaded when this was written (the
server refused the request), so the layouts were taken from the sbf-tools source, which cites the guide,
and the fixtures in tests/receiver_trust_sbf.rs are synthetic blocks built to that layout and its
CRC-16-CCITT. No real SBF log was used. Layout from the open-source crate, not checked against the current vendor guide: check against Septentrio's reference guide before relying on it.
Direct serial and USB input (receiver-trust live --serial)#
kshana receiver-trust live session.toml --serial /dev/ttyUSB0 --baud 4800 [--gate ...]Unix only. It opens the device for reading, sets the local line to raw mode at --baud (default
4800, the NMEA 0183 rate), and reads NMEA lines like any other source. It never writes to the receiver.
--configure is refused with an explanation: Kshana does not configure receivers; set the receiver to
output NMEA with the maker's own tool, then read it. The feed is NMEA, as for every live source (UBX
and SBF are read from files). A device that goes away ends the command as a fault (exit 1), for a
supervisor to restart; the layer does not reconnect. Tested on a pseudo-terminal loopback
(tests/receiver_trust_live_serial.rs), including that the master side receives no byte from the command.