DVB-S2 Downlink + OPV 64-Channel Receiver ACI (haifuraiya_opv)

Transmit IP network traffic over a DVB-S2 RF link while simultaneously demodulating the Opulent Voice (OPV) 64-channel uplink and capturing a receive spectrum snapshot, with both the decoded OPV frames and the spectrum embedded in the downlink.

Description

haifuraiya_opv extends DVB-S2 Downlink + Spectrum Monitor ACI (haifuraiya_aci) by adding the full OPV folded 64-channel demodulator on the receive path. It depends on the ocpi.opulent-voice project, which supplies the OPV workers (channelizer, decimate3, bd_x16, cr_x16, ted_x16, dsd_x16, fs_x16, msk_decode_x16, frame_merge4).

Transmit (unchanged from haifuraiya_aci): IP packets arriving on a Linux TUN interface are GSE-encapsulated (ETSI TS 102 606), packed into DVB-S2 Baseband (BB) frames, and streamed to the FPGA dvbs2_encoder via the OpenCPI ACI, then transmitted by the DRC.

Receive: the ADC stream is decimated by two cascaded half-band decimators (÷4) and split. One branch runs the OPV channelizer and the folded demod chain (four ports of bd_x16 → cr_x16 → ted_x16 → dsd_x16 → fs_x16 → msk_decode_x16), merged by frame_merge4 into a single decoded-frame DMA. The other branch feeds the triggered-sampler → windower → FFT spectrum path.

Both receive products are sent out as UDP through the TUN interface, so they are themselves GSE-encapsulated and transmitted over the DVB-S2 carrier:

• OPV frames — demultiplexed by abs_bin (from each frame’s metadata message) and sent to opv_base_port + abs_bin (default base 6000), so a monitor binding 6000 + N receives channel N (one 134-byte frame per datagram). The host tool opv_udp_monitor.py (or the rx-dashboard / tile_monitor.py --udp) displays them.

• FFT spectrum — a 4096-point log-power snapshot serialised as JSON {"s":[dB,…]} and sent to the FFT destination at a configurable rate.

Thread architecture: a TUN reader thread (TX BB frames), an FFT UDP thread, an OPV UDP thread (reads opv_out, demuxes abs_bin, fans out per channel), and the main thread that pumps BB frames into the encoder (sending a GSE null-packet padding frame when idle to keep the encoder fed).

The FPGA application (haifuraiya_opv.xml) is built by the haifuraiya_opv_asm HDL assembly with the cfg_1rx_1tx container (both transmit and receive active).

Execution

Command(s)

haifuraiya_opv [fft_dest_ip [fft_dest_port [opv_base_port [fft_rate_hz]]]]

Arguments (all optional, positional):

• fft_dest_ip — IP reachable via tun0 to receive spectrum data (default 10.0.0.2)

• fft_dest_port — UDP port for spectrum JSON (default 5005)

• opv_base_port — OPV channel-0 UDP port; channel N → base + N (default 6000)

• fft_rate_hz — spectrum update rate in Hz (default 1.0)

Prerequisites

Build the ocpi.opulent-voice dependency, the haifuraiya_opv_asm assembly, and the bitstream for the target platform, then run the executable on the target hardware with a configured tun0 interface.

Verification

Decoded OPV frames appear on opv_base_port + N and the spectrum JSON on the FFT destination port; confirm with opv_udp_monitor.py and an FFT/spectrum viewer (or the unified rx-dashboard).