Wire-level CW=8 negotiation + D8PSK gate SNR>=22 (throughput push)#27

Merged

secup merged 13 commits intomainfrom

experimental/throughput-push

May 5, 2026

Commits on May 4, 2026

EXPERIMENTAL: re-enable D8PSK gate on OFDM_CHIRP wide

secup
and
claude
committed
narrow D8PSK gate: tighten R2/3 thresholds after file-transfer stress

secup
and
claude
committed
EXPERIMENTAL: D8PSK now uses high-throughput ARQ window (=16)

secup
and
claude
committed
docs: throughput push results — D8PSK ladder + window=16

secup
and
claude
committed
HARDWARE-VALIDATED: tighten D8PSK gates to real Mac↔Pi5 measurements

secup
and
claude
committed
docs: hardware-validation results for throughput push

secup
and
claude
committed
docs: window-size + higher-rate investigation after pushback

secup
and
claude
committed
docs: direct main vs experimental hardware A/B

secup
and
claude
committed
docs: throughput investigation — CW=8 is the missing lever (+50 %)

secup
and
claude
committed
docs: auto-CW-bump implementation attempt — rolled back

secup
and
claude
committed
Negotiate fixed-frame CW count on the wire (CONNECT_ACK + MODE_CHANGE)

secup
and
claude
committed
Preserve --cw-count override across mid-transfer MODE_CHANGE

secup
and
claude
committed

Commits on May 5, 2026

Tighten D8PSK gate SNR>=20 -> SNR>=22 (good fading)

secup
and
claude
committed