MOSAIC Alpha

Alpha contract

What works today

This matrix is the alpha contract. Rows are independent: a green UHF result does not imply an HF contact.

ClaimStatusNotes
UHF shared-channel multi-user, radiatedYesTwo independent SDR transmitters; up to 12 catalogue identities recovered CRC-valid on air.
Adaptive UHF Basic/Fast/Turbo, radiatedYes3/3 CRC-valid with 100% measured active-span occupancy and 53.4% two-or-more-user overlap.
HF Basic/Chat/Fast waveform over UHF carrierYes, bounded claim3/3 CRC-valid at 431.200 MHz and 100% measured occupancy; not an HF-band or ionospheric result.
UHF bidirectional keyboard QSOYesLab antennas at 431.200 MHz; KISS + operator panel.
UHF continuous demod keep-upYesuhf-live path; backlog-capped workers; offline keep-up <1× wall.
Realtime operator consoleYes, RF provenLive spectrum/waterfall, decoder diagnostics, station selection, directed QSO flow, actual RF completion, and fail-closed TX via --ui-port.
HF single-user via sound cardYesReal DAC/ADC path; intended HF deployment model.
HF browser/KISS sound-card serviceImplementation provenContinuous receive, bounded peer catalogue, browser QSO flow, local KISS TCP, and fail-closed audio/PTT are packaged. No on-air HF QSO is claimed.
HF realtime continuous RXYesMOSAIC_HF_AUDIO narrow CFO search; keeps up with live audio.
HF multipath single-user (simulated ITU-R F.1487)YesMeasured Eb/N0 cliff; fading costs ~3 dB near threshold.
HF multi-user with ionospheric fading (sim)Limitedhf-congested + residual rebuild: 6-user ≈5.8/6; 8-user mean 6.5/8. Seed-21 pure-MAI clears 8/8 when uncoordinated arrivals span ~40 s (≈ several packet durations). Default HF TX jitter ≈ one packet. Not on-air HF.
HF bounded CW QRM + impulsive QRN (sim)BoundedFour users, two seeds: robust 48/48 vs baseline 30/48, clean unchanged, zero false deliveries. Covers persistent CW, one keyed schedule, +0.75 Hz/s drift, and impulsive samples; not general QRM rejection or on-air HF.
UHF multipath FIR proxy (cabled + radiated)Yes2 ms / 0.7 FIR on composite TX; both users CRC-valid on pad path and on antenna path at 431.200 MHz under HF congested decode.
UHF multi-user under HF-equivalent fade (sim)Limited → improveduhf-multipath congested stack; 4-user 15/16 at 8 s arrival span (offline). Tight 1.5 s span still limited.
HF waveform on free-running UHF SDR LOYes (cabled + radiated)CRC-valid HF profile decode at 431.200 MHz on pad path and antenna path with local CFO/drift refine. Sound-card audio remains the intended HF deployment model.
On-air HF QSONot demonstratedRequires a real HF transceiver in the loop.
Sub-second chatNoAirtime is the floor (~2.6 s UHF / ~40 s HF packets).

Evidence snapshots

HF limits summary
HF alpha limits at a glance.
Multi-user recovery
On-air UHF multi-user proof versus HF fade ceiling.
Adaptive mixed-rate three-radio RF evidence
Measured occupancy, waterfalls, and CRC-gated cancellation for all three rates in both waveform families.
HF QRM and QRN recovery simulation evidence
Four-user bounded QRM/QRN front end: 48/48 robust recovery versus 30/48 baseline across persistent, keyed, drifting, impulsive, and combined cells. Simulation only.

What alpha testers should try

  • UHF uncoordinated multi-station decode with independent radios.
  • Keyboard QSO over UHF SDR using the operator panel.
  • HF audio loopback and, with care, dummy-load transceiver tests for interface/ALC only — not as multi-user fading proof.