BeiDou PRN Reassignment (April 2026)¶

Between 2026-04-10 and 2026-04-21 BeiDou retired most BDS-2 satellites and moved BDS-3 satellites onto the PRNs they freed. Since then, PRN 1–18 is no longer a reliable marker of BDS-2: most of C01–C14 are now BDS-3, and two BDS-2 satellites are still in service on C09 and C10.

MRTKLIB (like MADOCALIB 2.2) cannot tell the generation of a satellite on PRN 1–18 from the data. You tell it with two configuration keys:

Which value to use¶

[positioning]
systems = ["GPS", "Galileo", "QZSS", "BeiDou"]
bds2_sats = ["C09", "C10"]       # data after 2026-04-21

[ambiguity_resolution]
# bds3_igso_sats not set: C06-C08

[positioning]
systems = ["GPS", "Galileo", "QZSS", "BeiDou"]
bds2_sats = ["C01-C18"]          # data before 2026-04-10

[ambiguity_resolution]
bds3_igso_sats = ["C38-C40"]

Entries are BeiDou satellite IDs ("C09") or ranges ("C01-C18"), in any mix.

• bds2_sats only accepts PRN 1–18. PRN 19 and above are always BDS-3.
• An empty list bds2_sats = [] means "no BDS-2". It is the same as the default, but it counts as set, so it silences the warning. Use it only when you know the data has no BDS-2 satellite.
• An empty or absent bds3_igso_sats means C06-C08.
• An invalid list falls back to the default. The run prints a warning.

Why the generation matters¶

MRTKLIB keeps BDS-2 and BDS-3 apart wherever they behave differently:

• Signals and frequencies. BDS-2 uses B1I / B3I / B2I. BDS-3 uses B1I / B3I plus B1C / B2a / B2b, and does not transmit B2I. The [signals] bds2 and bds3 presets select the frequency slots per generation.
• Receiver clock and biases. SPP and PPP estimate a separate BDS-2 clock offset / inter-system bias.
• RINEX reading. Observation codes are mapped to frequency slots per generation.
• PPP-AR. Ambiguities are differenced within each generation. IGSO and BDS-2 satellites are never taken as reference satellites, and are dropped first in partial AR.
• MADOCA-PPP (IS-QZSS-MDC-005). A BDS-3 satellite on PRN 1–18 may be sent in GNSS ID 3, and its signal mask is then read with the BDS-3 signal table. A wrong bds2_sats gives its code and phase biases the wrong signal codes.

A satellite classified as the wrong generation either loses its observations (no matching frequency) or contaminates the filter with a wrong signal model.

What changed in April 2026¶

PRN assignments from the IGS satellite metadata (see also the IGSMAIL notices, e.g. IGSMAIL-8693):

Dates are the day the new satellite took the PRN; "was" gives its previous PRN.

The IGSO change is why bds3_igso_sats exists. Before April the BDS-3 IGSO satellites were C38–C40; today they are C06–C08, and C38–C40 are MEO.

Data from the transition window¶

Between 2026-04-10 and 2026-04-21 PRNs changed generation one by one, and some changed in the middle of a day. A single bds2_sats list cannot describe such a day. Two options:

• Process with the list that matches the end of your session. Then add the PRNs that changed during the session to [positioning] excluded_sats.
• Leave BeiDou out. Drop it from systems for those days.

Bundled configurations¶

When you adapt a bundled configuration to your own data, check these two keys against the table above.

Legacy .conf files and MADOCALIB¶

The keys correspond to MADOCALIB 2.2's options:

The defaults are the same as MADOCALIB 2.2: no BDS-2 and IGSO C06-C08. MADOCALIB 2.1 and earlier classified PRN 1–18 as BDS-2 and C38–C40 as IGSO unconditionally. To reproduce them, use ["C01-C18"] / ["C38-C40"].

The setting is process-wide. mrtk post applies it before reading the input files; mrtk run applies it when the server starts.

mrtk bias reads no configuration file. Give the list on the command line with -bds2prn (--bds2-sats), e.g. -bds2prn C09,C10, the same option name as MADOCALIB 2.2's cssr2ssr. Without it, every PRN 1–18 satellite is BDS-3 and a warning is printed; -bds2prn "" states "no BDS-2".