Does the official release of openwrt support mu-mimo technology? The only thing I could find online was this: https://forum.archive.openwrt.org/viewtopic.php?id=72019
But that was back from 2017.
OpenWRT 'supports' MU-MIMO by way of hostapd, but the driver and firmware for the wifi chipset also have to support MU-MIMO for it to work.
It is not supported on the WRT3200ACM because the opensource driver and firmware bundle for the Marvell wifi chip was never completed.
It appears the project has been abandoned.
Might be interesting now that Marvell has sold the WiFi assets to NXP... not better/worse, and contributors from Marvell that shipped FOSS code might be looking for a job (and this always happens...)
https://www.marvell.com/company/news/pressDetail.do?releaseID=10959
Considering what happened with Broadcom and Cypress - Cypress opened up a lot of the datasheets that Broadcom used to keep hidden under NDA...
1 Like
FWIW - MU is really hard to trigger to get real MU-MIMO...
Needs the right client STA's, and the right traffic that is suited for MU delivery, and the right RF considerations (beamforming needs to be just right, and the common MCS) to bundle the MU packets across a common VHT MU set.
When going cross vendor - e.g. QCA MU AP, and clients also QCA, it might work - going cross STA vendor with Intel, Realtek, Broadcom, QCA, and different groupings, I never observed it to work consistently, and performance compared to SU-MIMO was either similar or negatively impacted (e.g. early Broadcom MU capable STA's where they would drop from 2 SU streams to 1 stream if MU beamformer was enabled).
With Marvell - would perhaps be better just to keep things stable with Wave1 features, and then perhaps look at a sub-set of Wave2 items that are really beneficial - 160MHz might be better than trying to sort out the mathematical mess that MU has to offer.