Antenna diversity: selection, switching, and combining

Worth recording for the next person who hits it.

1) Selection diversity needs a fast, low-loss switch and a reliable metric.
2) Switching diversity is simple but a switching transient corrupts data unless timed during a guard interval.
3) Combining gives the best performance but needs two complete receive chains.
4) Correlation between branches sets the ceiling on the diversity gain.

That is as far as I got.

Useful diversity checklist — selection vs switching vs combining is really a cost / transient / RF-chain budget decision, not a single "diversity gain" slide.

Editorial keepers:
1. Selection: needs a fast, low-loss switch and a metric that tracks the link you care about (SNR / RSSI / CRC), not a slow average that lags fades.
2. Switching: simple, but the transient corrupts symbols unless gated into a guard / blanking window; freeze switch timing next to BER.
3. Combining: best theoretical gain, but two full receive chains — freeze correlation between branches; high correlation caps diversity gain no matter how fancy the combiner.
4. Log measured branch correlation and the diversity scheme on the same traveler as any "diversity improved X dB" claim.

Open: at what branch correlation do you stop expecting useful diversity gain — and do you require a measured correlation number before freezing the scheme?

— rfforum 编辑部 / Editorial Team

From a field perspective: make sure monitoring is in place before deployment.