Question about streams


ok so i know "the more streams, the more bandwidth".

but lets say we only have 2x2 or 3x3 devices.

does it hurt performance to set a router in 4x4 mode when no client is capable of 4x4? that is, would you get better performance by setting the router's wifi streams to the capabilities of the client devices?

There is no knob to change these settings anyways (aside from debugfs, but that's a different story and hardly enduser compatible). The router defaults are good the way they are - and aside from rather academic uses or very, very specific circumstances, it doesn't make the slightest sense to mess with these settings.

Just in general.

In a one-to-one situation, you can only achieve the speed of worst device involved. In general, the router is expected to be better (3x3 or 4x4, up to 8x8 with contemporary 802.11ax routers) than the client (usually 1x1 or 2x2, maybe 3x3 on good notebooks). It doesn't hurt if the router could do better than the client connecting to it. But even in this situation, you can't determine which 'stream' (antenna) would be the 'least useful' one - it depends on the signal propagation, changes with your location relative to the router, the obstacles inbetween, wandering external interferences and time. In a pre-802.11n world one would have called this antenna diversity, with routers despite only supporting 2x2 streams shipping with 3 antennas and picking the 'best' two of them for the given circumstances - in a >802.11n world this is a bit more complex in reality (and with >802.11ac MIMO and Mu-MIMO join the party), but it's still a valid (simplified) explanation.

Reality isn't a one-to-one situation though, you won't find many households with exactly one router (o.k., this you can find) and exactly one client (with an exact match of supported concurrent streams) - that's where Mu-MIMO (>=802.11ac/ wave2) plays its hand. Where you're sitting at the breakfast table, 'antennas' 1 and 3 might provide the best results, but your wife's smartphone on the other side of the table would be better served with 'antennas' 2 and 4 - while uncle Bob in the guest bedroom would get best results from 'antennas' 1 and 4, with the 3x3 gaming console in the living room wanting to use 'antennas' 1, 3 and 4. The concept of Mu-MIMO would know (under ideal circumstances) to assign this dynamically, and in combination with beam-forming provide the best results for everything involved (reducing interference from each other, virtually achieving directional-antenna-like behaviour, using the available spectrum and time slicing to best effect). As a result, the whole is greater than the sum of its parts.

In practice, wireless after 802.11g involves a lot of black magic and fairy dust. While the explanations above are valid in principle (and its approximation even more so in reality), they're massively simplified. Still, you'd only be hurting yourself by artificially crippling your router to do less than it could do - you don't have the data to decide which strategy is best for the current situation at this very time. The underlying situation is fluctuating rapidly and constantly re-evaluated by the wireless chipset - this is where a large chunk of the performance is found by modern wireless standards. Especially 802.11ac/ wave2 and 802.11ax have gained massively in this regard.


ahh i see.

we rely on the device's hardware to automatically determine the optimal antennas.

i didn't know the streams were directly related to the antennas, but that makes sense.

with the RT_WIFI driver you can set the streams manually, but this would be a bad idea from what you've shared above.

in-effect i'd be disabling an antenna that would otherwise be available. i thought maybe 'concentrating' the traffic to less streams would improve throughput, but it seems that is the wrong idea.

Those settings are usually only there for developer-side debugging in a hf lab - or to remedy 'broken' situations, akin to a notebook with only two physical antennas, but a 3x3 wireless card inserted and the wifi calibration data stored on this card incorrectly being configured to use all three (it happens…).

yeah must be something like that.

i would imagine if the device is limited to 3 streams, then anything higher (say, 4) will automatically just get set to the maximum supported by the device.

Advantages: Round robin, best reception, least multi path issues.

You can just look at the MIMO Wiki page.

I've never heard of adjusting the streams - believing they would help with bandwidth...especially since altering a transmitter/antenna is illegal in most nations. :wink:

1 Like

lol had no idea it was illegal, thanks for the tip.

just trying to optimise performance, man.

every setting counts!

learned a bit in this. didn't know MUMIMO's job was to pretty much figure this out

1 Like

This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.