Beamforming

[oturn]

I have an RT-AC87R running 378.55.

Beamforming 101 please:

What's the quick and dirty explanation for Beamforming being off by default?
Is it off by default on the official Asus firmware?
Are the issues with the implementation of Beamforming due to limitations in the router, or with the clients?
What is the outlook for the future of Beamforming? It seemed so promising.

 

[eddiez]

On by default if I remember correctly. As for issues, those might always be present depending on the connected devices. Hard to say if any of those issues can blamed on the connected devices or the beamforming though. Working fine here with 20 wireless devices.

 

[cosmoxl]

When I lived in a masonry house (wifi signals attenuated a lot) beamforming was needed to reach certain areas. I have an AC68 and none of my devices are AC devices. The universal beamforming option obviously does make a difference and I guess some of my 2.4ghz devices may support explicit beamforming. I don't know. Perhaps the AC beamforming in 5ghz range would be even more effective.

 

[RMerlin]

Asus recently turned it off by default as quite a few devices are having compatibility issues with it. Best to have better out-of-the-box compatibility than out-of-the-box performance.

 

[Haggster]

Got Beamforming enabled for 2.4GHz wifi and kept it that way since roughly a year ago... gives me no issues at all and I got loads of different types of devices.

My router is an RT-AC87 by the way.

Beamforming for 5GHz is however a safe recipe for connection issues, so I always keep it turned off. Some devices won't last for an hour without a disconnect otherwise. Hope that helped.

 

[snb008]

For the RT-AC87U, has the beamforming feature always had compatibility issue since the router's debut? or, it only becomes an issue recently?

I have only been using it for about a month. Currently, I have disabled it (using Merlin 378.55).
May I ask, if there is a problem connected with beamforming, what is/are the usual symptom(s)?

 

[sfx2000]

Asus recently turned it off by default as quite a few devices are having compatibility issues with it. Best to have better out-of-the-box compatibility than out-of-the-box performance.

Interesting... I could see disabling beam-forming for 2.4GHz, as 11n was a bit unsorted, but 802.11ac this is inherent property that many devices support. Vendors have been backporting some 11ac items, but if one just does 11g/b/n there, one should be fine..

5Ghz - Not a "mandatory" feature, but if supported, all 11ac clients should be fine as this is standards based - but some 11n clients there might have some issues..

Again, with Quantenna, who knows?

 

[RMerlin]

Interesting... I could see disabling beam-forming for 2.4GHz, as 11n was a bit unsorted, but 802.11ac this is inherent property that many devices support. Vendors have been backporting some 11ac items, but if one just does 11g/b/n there, one should be fine..

5Ghz - Not a "mandatory" feature, but if supported, all 11ac clients should be fine as this is standards based - but some 11n clients there might have some issues..

Again, with Quantenna, who knows?

5 GHz can also be used by 802.11n devices, where you will end up with non-native beamforming. My Nexus 4 throughput drops to 1-2 Mbps if I enable beamforming on the 5 GHz band. Whether it's broken or simply experiencing compatibility issues is anyone's guess... A lot of iDevice owners also stabilized their 5 GHz connectivity by disabling beamforming.

Meanwhile, my former laptop (Intel 7260AC) had no issue at all with beamforming enabled, same with my 2x2 802.11ac Nexus 9.

 

[ryzhov_al]

I wonder RT-N56U/65U/66U boxes are labeled with AiRadar ads? It's a lie. The beamforming is disabled in drivers (RT-N56U/65U sources are available) and it can be reached without special antennas like this ones.

 

[RMerlin]

I wonder RT-N56U/65U/66U boxes are labeled with AiRadar ads? It's a lie. The beamforming is disabled in drivers (RT-N56U/65U sources are available) and it can be reached without special antennas like this ones.

AiRadar is a marketing name for various techniques, such as Antenna Diversity AFAIK, not just beamforming alone.

Sent from my Nexus 9 using Tapatalk

 

[sfx2000]

5 GHz can also be used by 802.11n devices, where you will end up with non-native beamforming. My Nexus 4 throughput drops to 1-2 Mbps if I enable beamforming on the 5 GHz band. Whether it's broken or simply experiencing compatibility issues is anyone's guess... A lot of iDevice owners also stabilized their 5 GHz connectivity by disabling beamforming.

Meanwhile, my former laptop (Intel 7260AC) had no issue at all with beamforming enabled, same with my 2x2 802.11ac Nexus 9.

Hmmm... I wonder if it's trying to do implicit beamforming for HT mode clients (HT == 802.11n) - implicit mode doesn't need sounding, so one won't see it in a capture.

VHT mode, it's definitely explicit, and one will see the sounding requests/responses which makes sense... and I can confirm does help with a variety of 11ac Clients (the USB-AC56 is handy for testing this, as one can enable/disable it in the latest driver kit with the BeamCap switch)

Aha... ok, I think this might be doable, but I don't have an AC68U to tinker with...

When the wifi chip is configured, there are flags sent to it for Legacy, HT, and VHT modes - so one would want to set the HT TxBf to false, this will disable implicit or explicit beamforming for HT mode clients, and then set VHT TxBf to true... and this should fix it.

This is a static unfortunately, so it's done on the initial config - it's not a dynamic input. This is valid for the generic broadcom BSP (board support package), I'm not seeing an analogue for it in the Asus SDK, but I'm sure it is there somewhere where the wl chips are sent their initial configs.​

again, I'm looking at the generic Broadcom config, and I'm not seeing a way to split those out (HT vs. VHT) for beamforming. I might have to do a git pull on your repo to check what Asus is doing on their config - maybe the AiRadar special sauce?

 

[sfx2000]

@RMerlin - interested to work on this item?

We'll have to find a few clients to sort this one out...

 

[RMerlin]

@RMerlin - interested to work on this item?

We'll have to find a few clients to sort this one out...

Not really. I have no documentation on Broadcom's SDK, so I leave that kind of wireless stuff to Asus.

 

[DaveLessnau]

I feel silly. Where are you seeing the option to turn on/off beamforming? I can't find it.

 

[hggomes]

Probably because you have a MIPS router.

 

[sfx2000]

Not really. I have no documentation on Broadcom's SDK, so I leave that kind of wireless stuff to Asus.

Well, it's not the SDK - that's more production oriented (the SDK can be reskinned and have a production ready package, it includes the WebGUI and associated tools/code needed) - what I was reviewing is a different package... it's more simplistic in many ways, but gives a lot more control over the chipset so folks can either roll new features in, or do alternate implementations away from Broadcom's suggested bill of materials..

If I can find the bits/flags to try, would you be willing to do a test build since you have the tool chain and a production quality image? If so, I'll dig into the Asus code and find the diffs...

 

[RMerlin]

Well, it's not the SDK - that's more production oriented (the SDK can be reskinned and have a production ready package, it includes the WebGUI and associated tools/code needed) - what I was reviewing is a different package... it's more simplistic in many ways, but gives a lot more control over the chipset so folks can either roll new features in, or do alternate implementations away from Broadcom's suggested bill of materials..

If I can find the bits/flags to try, would you be willing to do a test build since you have the tool chain and a production quality image? If so, I'll dig into the Asus code and find the diffs...

A good portion of the wireless-related code in the firmware is now closed source.

 

[sfx2000]

A good portion of the wireless-related code in the firmware is now closed source.

It's not a driver mod, but a config mod for what the WebGUI would pass to Linux to configure the chips...

The problem for me is two fold...

1) I don't have an AC68U to even try this with - so even if we were to proceed, it would likely only apply to the 68U and variants...

2) I don't have much experience building the firmware for this device, so even pulling down your git and building out a toolchain, I'd rather pass a patch to someone that knows that build-environment...

 

[sfx2000]

Not really. I have no documentation on Broadcom's SDK, so I leave that kind of wireless stuff to Asus.

Ok, pulled your GitHub, and I see where the problem is, and it's probably not worth the effort to try and patch considering how the Asus SDK is structured (Asus supports all chipset vendors in their SDK, not just Broadcom)

It appears that Beamforming is taken as a global input for both HT and VHT modes, and this is why enabling Beamforming can impact performance of HT mode clients (In HT mode, one either get SM or TxBF, not both, VHT changes all this)...

In other words - I would suggest not enabling Beamforming in the factory code (or variants based on the SDK), unless all clients are 11ac, as N clients will basically go to single stream operation which has a significant impact on 11n performance.

cool...

 

[Just Checking]

Ok, pulled your GitHub, and I see where the problem is, and it's probably not worth the effort to try and patch considering how the Asus SDK is structured (Asus supports all chipset vendors in their SDK, not just Broadcom)

It appears that Beamforming is taken as a global input for both HT and VHT modes, and this is why enabling Beamforming can impact performance of HT mode clients (In HT mode, one either get SM or TxBF, not both, VHT changes all this)...

In other words - I would suggest not enabling Beamforming in the factory code (or variants based on the SDK), unless all clients are 11ac, as N clients will basically go to single stream operation which has a significant impact on 11n performance.

cool...

Good to know. Probably 75+% of my client devices are N with the rest being AC. I have alway had beamforming enabled and have not experienced any dropped connection problems. I recently ran tests with a RT-AC87 against a RT-AC66 router. I tested link and transfer speeds with beamforming on and off. I really didn't see much difference between them.

Under what circumstances will beamforming have the most effect on signal quality?