dtmb-sdr

A minimal, vendor-neutral DTMB receiver. It accepts interleaved signed 8-bit complex samples (CI8) from a file or standard input and emits MPEG transport stream packets to a file or standard output.

CI8 file/stdin
  -> sample-rate conversion
  -> PN420/PN595/PN945 synchronization and C=3780 equalization
  -> 16QAM/32QAM/64QAM deinterleaving and soft demapping
  -> LDPC, BCH, and descrambling
  -> MPEG-TS file/stdout

Hardware acquisition is intentionally outside this repository. Any application that can produce CI8 bytes at a known sample rate can feed the receiver.

Build

Requirements are CMake 3.20+, a C++20 compiler, and Python 3.10+.

For a released platform wheel, installation is simply:

pip install dtmb-sdr

The wheel contains the native receiver executables and required LDPC data; no separate source checkout or CMake build is required.

The new 16QAM, 32QAM, PN420 and PN595 paths require this source revision or a development wheel built from it. They are not yet included in a published release.

To build from a source checkout:

python -m pip install -e ".[dev]"
cmake -S core/cpp -B build/core-cpp -DDTMB_CORE_BUILD_TESTS=ON
cmake --build build/core-cpp --config Release
ctest --test-dir build/core-cpp -C Release --output-on-failure
pytest

Build a distributable wheel with:

python -m build --wheel

Prebuilt wheels use the portable CPU decoder. CUDA requires a source build with a supported NVIDIA toolkit.

Decode a CI8 file

dtmb-decode \
  --input capture.ci8 \
  --input-rate 16000000 \
  --output recovered.ts \
  --acceleration cpu \
  --error-policy continue

The default profile is QAM64, FEC rate 0.6, interleaver mode 2 (--system-info-index 22). Select the modulation, FEC rate and interleaver with an explicit system-information profile:

FEC rate	Interleaver	16QAM profile	32QAM profile	64QAM profile
0.4 (rate 1)	mode1	11	—	19
0.4 (rate 1)	mode2	12	—	20
0.6 (rate 2)	mode1	13	—	21
0.6 (rate 2)	mode2	14	—	22
0.8 (rate 3)	mode1	15	17	23
0.8 (rate 3)	mode2	16	18	24

Each 16QAM, 32QAM, and 64QAM symbol carries 4, 5, or 6 soft bits. The native frontend and demappers accept --qam 16qam|32qam|64qam, and --normalization qam uses the selected alphabet. The existing 64QAM API and dtmb_core_deinterleave_qam64 executable remain compatible; the generic executable is dtmb_core_deinterleave_qam.

Synthetic regressions cover all six 16QAM profiles, continuous interleaving, noisy natural LDPC/BCH recovery and exact transport bytes. They do not establish over-the-air validation.

32QAM profiles 17/18 group two signal frames into five rate-3 LDPC codewords.

--qam32-frame-phase auto detects packing alignment from two naturally LDPC/BCH-clean codewords at exactly one of two possible phases in a bounded prefix. The prefix is replayed and any one-frame startup discard is reported. Use 0 for a known group boundary or 1 to skip the first signal frame. An ambiguous prefix or incomplete terminal group is an error. FEC frame counters and --decode-batch-frames count two-signal-frame groups for 32QAM; diagnostics also report signal-frame counts and the grouping factor. All five codewords must pass the clean gates before the group is emitted. Synthetic tests cover both packing phases, both interleavers, noise and exact CI8-to-TS recovery.

Select the known frame header explicitly with --pn-mode pn420|pn595|pn945. The default remains pn945; PN mode is independent of the system-information profile. Automatic PN-mode selection is not implemented.

Mode	Header symbols	Frame symbols	Frames per superframe	Header/body power
PN420	420	4200	225	2
PN595	595	4375	216	1
PN945	945	4725	200	2

For example, a known PN595/16QAM rate-2, mode2 signal uses:

dtmb-decode --input input.ci8 --input-rate 7560000 --output recovered.ts \
  --pn-mode pn595 --system-info-index 14 --acceleration cpu

Use the actual input sample rate. Synthetic CI8 tests cover all three PN modes, all six 16QAM profiles, multipath, CFO, sample displacement and exact TS recovery. PN420/PN595 and 32QAM over-the-air validation is still pending. PN595/32QAM is supported by the same native stages and has synthetic coverage. For a known rate-3, mode2 transmission:

dtmb-decode --input input.ci8 --input-rate 7560000 --output recovered.ts \
  --pn-mode pn595 --system-info-index 18 --acceleration cpu

Strict mode is the default. --error-policy continue omits unclean frames and adds MPEG-TS discontinuity indications before later clean output. The runtime fails when no transport bytes are recovered. Validate nonempty output with dtmb-ts-analyze; video claims additionally require an actual codec decode, not only stream identification.

Optional CUDA LDPC acceleration

CPU decoding is the portable baseline. Build the optional CUDA LDPC backend when a supported NVIDIA toolchain is available:

cmake -S core/cpp -B build/core-cpp-cuda \
  -DDTMB_CORE_ENABLE_CUDA_LDPC=ON \
  -DDTMB_CORE_BUILD_TESTS=ON
cmake --build build/core-cpp-cuda --config Release

Then select it explicitly:

dtmb-decode \
  --bin-dir build/core-cpp-cuda \
  --input capture.ci8 \
  --input-rate 16000000 \
  --output recovered.ts \
  --acceleration cuda

The CUDA backend batches LDPC codewords and reports host-to-device, kernel, device-to-host, and total timing metrics. CPU remains available as the correctness and portability reference.

Decode a pipe

Use - for stdin or stdout:

ci8-producing-command |
  dtmb-decode --input - --input-rate 16000000 --output recovered.ts

The source command is deliberately unspecified: the receiver depends only on the byte-stream contract, not an SDR vendor or device API.

Play while decoding

With ffplay:

ci8-producing-command |
  dtmb-decode --input - --input-rate 16000000 --output - --error-policy continue |
  ffplay -fflags nobuffer -flags low_delay -f mpegts -

With VLC:

ci8-producing-command |
  dtmb-decode --input - --input-rate 16000000 --output - \
    --error-policy continue |
  vlc - --demux=ts

The producer owns the SDR device, tuning and capture policy. dtmb-decode only consumes CI8 bytes, so the same receiver command works with a hardware adapter, a vendor utility or another SDR framework on any platform. Keep diagnostics on stderr; stdout is binary MPEG-TS.

Long-running receiver controls are explicit and off by default. A deployment that has independently validated them for its RF environment can enable:

ci8-producing-command |
  dtmb-decode --input - --input-rate 16000000 --output - \
    --pn-schedule-tracking \
    --source-frame-confidence inverse-mse \
    --resample-headroom-db 6 \
    --error-policy continue |
  vlc - --demux=ts

The receiver inserts bounded 64 MiB queues after bulk stages by default so capture, frontend, demapping and FEC can progress concurrently. Use --pipeline-buffer-mib 0 only for controlled diagnostics. These options are not universal RF defaults and do not replace transport or codec validation.

PN schedule options apply to PN420 and PN945. PN595 has a fixed header and uses direct correlation for timing tracking; cyclic schedule and cyclic header averaging options are rejected. In the native frontend, --pn-estimator wideband selects the appropriate cyclic or direct channel estimator for the chosen mode. For PN595, use --pn-wideband-header-observation direct (or core). Its default linear fit spans 149 signed delays; --pn-wideband-max-span-symbols controls that span, up to 298 taps. The existing PN945 API and command defaults remain available.

Playback quality depends on RF quality and FEC cleanliness. A player detecting a service is not evidence that the complete stream is error-free.

Inspect MPEG-TS output

dtmb-ts-analyze recovered.ts
ffprobe -v error -show_programs -show_streams recovered.ts

Scope

Included:

• portable C++20 receive stages;
• CI8 file and stdin integration;
• MPEG-TS file and stdout output;
• required LDPC matrices;
• core and pipeline-construction tests.

Not included:

• SDR drivers or hardware-control code;
• capture recipes, device identifiers, frequencies, gains, or locations;
• signal-generation, scanning, UI, visualization, sweep, or research tooling;
• real broadcast captures or derived analysis artifacts.

License

MIT. See LICENSE and THIRD_PARTY.md.