lrpt-encoder

Channel 1 (APID 64)	Channel 2 (APID 65)	Channel 3 (APID 66)

Three MSU-MR channels decoded by SatDump from an RF transmission generated by this encoder.

SatDump capturing and decoding the live RF signal (OQPSK demodulator, Viterbi, Deframer, and Reed-Solomon all synced).

Encode images into Meteor-M N2-4 LRPT (Low Rate Picture Transmission) digital signals in Rust.

Takes an image (or camera feed), runs it through the full CCSDS digital transmission chain used by the Meteor-M2-4 weather satellite, and emits baseband I/Q samples ready for SDR transmission. Output is decodable by SatDump and any standards-compliant LRPT decoder.

Features

• Full satellite-faithful digital chain: JPEG-like DCT/Huffman compression → CCSDS Space Packets → VCDU multiplexing → Reed-Solomon(255,223) interleave-4 → CCSDS scrambler → NRZ-M differential encoding → rate-1/2 K=7 convolutional coding → OQPSK with RRC pulse shaping
• 3-channel MSU-MR emulation (APIDs 64/65/66) with telemetry packet (APID 70) cadence
• Output formats: 2-channel WAV (16-bit I/Q at 288 kHz), raw float32 I/Q, stdout streaming (int16/int8/float32)
• Direct HackRF support via hackrf_stream.py — upsamples + pipes to hackrf_transfer (no GNU Radio / gr-osmosdr needed)
• Camera capture mode (optional, V4L2 via nokhwa)
• Configurable: SCID, VCID, APIDs, JPEG quality, CCSDS timestamp, telemetry payload
• Debug dumps for raw pre-modulation CADU inspection

Build

cargo build --release

With camera support:

cargo build --release --features input-v4l

The binary is produced at target/release/lrpt-encoder.

Quick start

Encode a single image to a WAV file containing baseband I/Q:

./target/release/lrpt-encoder a.png -o output.wav

Open the resulting output.wav in SatDump:

1. Offline processing → METEOR M2-x LRPT 72k
2. Input format: cs16 (or wav), sample rate 288000
3. Satellite: M2-4

Usage

lrpt-encoder [OPTIONS] [INPUT_IMAGE]

Common options

Flag	Default	Description
-o, --output <FILE>	—	Output file (.wav, .raw, .bin, .iq)
--stdout	off	Stream raw I/Q to stdout (for piping to SDR tools)
--stdout-format <FMT>	int16	Stdout sample format: int16, int8, or float32
-r, --repeat <N>	1	Repeat the image N times
--quality <1-100>	80	JPEG compression quality
--rgb	off	Split a color image: R → ch1, G → ch2, B → ch3
--input-ch2 <FILE>	—	Separate image for channel 2
--input-ch3 <FILE>	—	Separate image for channel 3

CCSDS / telemetry options

Flag	Default	Description
--scid <N>	57	Spacecraft ID
--vcid <N>	5	Virtual Channel ID
--apid-ch1/2/3 <N>	64/65/66	APID per MSU-MR channel
--msumr-id <0-15>	4	MSU-MR instrument ID in telemetry
--ccsds-day <N>	24938	CCSDS day (days since 1958-01-01)
--ccsds-ms <N>	43200000	Time-of-day in milliseconds
--telemetry-payload <HEX>	—	Custom APID-70 payload (AABBCC, AA BB CC, 0xAA,0xBB)

Camera capture (requires --features input-v4l)

Flag	Description
--capture	Enable camera capture
--continuous	Keep capturing until Ctrl+C
--cam-idx <N>	V4L2 device index (default 0)

Debug

Flag	Description
--dump-first-cadu <FILE>	Write first 1024-byte raw CADU before NRZ-M/Conv
--dump-cadu-stream <FILE>	Write entire raw CADU stream before NRZ-M/Conv

Examples

Single-image encode (same image used for all 3 MSU-MR channels):

./target/release/lrpt-encoder a.png -o output.wav

Three separate channel images (ch1, ch2, ch3), repeated 10 times:

./target/release/lrpt-encoder a.png --input-ch2 b.png --input-ch3 c.png -r 10 -o output.wav

Color image split into RGB channels (R → ch1, G → ch2, B → ch3):

./target/release/lrpt-encoder a.png --rgb -o output.wav

Stream directly to HackRF via hackrf_stream.py (no GNU Radio needed — works on macOS/Linux):

pip install numpy scipy  # one-time setup
./target/release/lrpt-encoder a.png --stdout --stdout-format int8 2>/dev/null | \
    python3 hackrf_stream.py --freq 433000000 --gain 0

hackrf_stream.py upsamples 288 kHz → 2 MHz (hackrf_transfer's minimum) and pipes to hackrf_transfer automatically.

Stream to HackRF via GNU Radio companion script (Linux, requires gr-osmosdr):

./target/release/lrpt-encoder a.png --stdout | python3 qpsk_stream.py

Custom CCSDS timestamp, SCID, and MSU-MR instrument ID:

./target/release/lrpt-encoder a.png \
    --scid 57 --vcid 5 \
    --ccsds-day 24938 --ccsds-ms 43200000 \
    --msumr-id 4 \
    -o output.wav

Custom APID-70 telemetry payload (raw hex):

# 62-byte payload; day/ms/MSU-MR ID fields are overwritten from CLI flags
./target/release/lrpt-encoder a.png \
    --telemetry-payload "00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 40 DE AD BE EF" \
    --msumr-id 4 --ccsds-day 24938 --ccsds-ms 43200000 \
    -o output.wav

Three channels with custom APIDs and quality:

./target/release/lrpt-encoder a.png --input-ch2 b.png --input-ch3 c.png \
    --apid-ch1 64 --apid-ch2 65 --apid-ch3 66 \
    --quality 90 \
    -o output.wav

Debug dump of raw CADUs before NRZ-M/convolutional coding:

./target/release/lrpt-encoder a.png \
    --dump-first-cadu first_cadu.bin \
    --dump-cadu-stream all_cadus.bin \
    -o output.wav

Signal parameters (Meteor-M2-4)

Parameter	Value
Symbol rate	72 ksps
Samples/symbol	4
Sample rate	288 kHz
Modulation	OQPSK
Pulse shaping	Root Raised Cosine, α = 0.5, span 33 symbols
Frame sync (ASM)	0x1ACFFC1D
VCDU size	892 bytes
Reed-Solomon	(255,223) GF(2⁸), interleave 4, first root α^112
Scrambler	CCSDS PN, x⁸+x⁷+x⁵+x³+1, init 0xFF
Differential	NRZ-M (continuous across CADUs)
Convolutional	rate 1/2, K=7, G1=171₈, G2=133₈
Image width	1568 px (196 MCUs × 8 px)
Default frequency	433 MHz (in companion GRC script)

Project layout

src/
├── constants.rs       LRPT signal parameters, JPEG tables
├── imaging.rs         Image loading, MCU extraction
├── compression.rs     DCT + quantization + Huffman
├── ccsds.rs           Source packets, VCDU multiplexer, CADU builder
├── reed_solomon.rs    RS(255,223) interleave-4
├── scrambler.rs       CCSDS PN scrambler
├── differential.rs    NRZ-M encoder
├── convolutional.rs   Rate-1/2 K=7 convolutional encoder (stateful)
├── modulation.rs      OQPSK + RRC pulse shaping
├── output.rs          WAV / raw / stdout sinks
└── main.rs            CLI + encoding pipeline

hackrf_stream.py       HackRF TX via hackrf_transfer (macOS/Linux, no GNU Radio)
qpsk_stream.py         HackRF TX via GNU Radio + gr-osmosdr (Linux)

Testing

cargo test --release

License

MIT — see LICENSE.

References

• CCSDS 131.0-B-3 — TM Synchronization and Channel Coding
• CCSDS 132.0-B-3 — TM Space Data Link Protocol
• CCSDS 120.0-G-3 — TM Synchronization and Channel Coding Summary of Concept and Rationale
• J.-M. Friedt, Decoding digital weather satellite images: the LRPT protocol from Meteor-M2 (2019)
• SatDump — reference decoder