MU-MIMO, 160MHz and >1Gbps Wi-Fi testing

[Cosmin Cazan]

Seems like MU-MIMO and 160Mhz both are potentially exceeding Gigabit Ethernet, but unfortunately, I have not seen much data on this. Even on SNB, which does a pretty great job for testing Routers (good methodology, etc), has not really been able to step over this challenge (as far as I can tell). The only testing I've seen for >1Gbps is data/measurements collected through simulators of various sorts (veriWave, etc), rather than the classic max TPT testing.

Does anyone know of a good methodology for evaluating this? For example, the classic setup:

Wired desktop <----> Switch (optional) <-------> Access Point / Router tested <- - - - (Wi-Fi) - - - - > Wi-Fi Laptop/Device

The challenge is finding >1G (2.5/5/10 G) support for the: wired desktop, switch, and often, even the Router interface (Routers claiming to support 5300 or 7200 Mbps, but come with 1G ports .. ).

Has anybody looked into what this setup might look like? And would it be affordable? (especially the switch).

 

[thiggins]

I tested 160 MHz bandwidth with a second WRT3200ACM in bridge mode, 6 feet away in open air. Can't get any simpler than that.

I have generally found Gigabit ports to not limit measured performance.

MU-MIMO doesn't come anywhere near maxing out a Gigabit Ethernet port.

 

[Cosmin Cazan]

I tested 160 MHz bandwidth with a second WRT3200ACM in bridge mode, 6 feet away in open air. Can't get any simpler than that.

I have generally found Gigabit ports to not limit measured performance.

MU-MIMO doesn't come anywhere near maxing out a Gigabit Ethernet port.

Thanks Tim, how do you test the routers in bridge mode? (where does the data go? don't you still need clients attached to the Router?). Can you share a link?

I guess MU-MIMO with 160MHz should definitely get us over 1G

 

[thiggins]

http://www.smallnetbuilder.com/wire...igabit-wi-fi-router-reviewed?showall=&start=3

I guess MU-MIMO with 160MHz should definitely get us over 1G

I don't know if that combination makes any sense or even works.

 

[sfx2000]

http://www.smallnetbuilder.com/wire...igabit-wi-fi-router-reviewed?showall=&start=3

I don't know if that combination makes any sense or even works.

I would agree - even 160MHz is a bit of a stretch...

From an AP perspective - 3*3:3 SU QAM256 seems to be the sweet spot... (e.g. AC1900 class)

Most clients in 2016 are 2*2:2 SU QAM256...

(which is reflected in the SNB test processes)

MU and 80+80 or 160 is digging deep, and diminishing returns here...

 

[L&LD]

I have generally found Gigabit ports to not limit measured performance.

What did you test it against to come to that conclusion? (2.5Gbps/5Gbps ports)?

 

[sfx2000]

What did you test it against to come to that conclusion? (2.5Gbps/5Gbps ports)?

I wouldn't speak for Tim - but I'm thinking that we're running up against the noise wall on the WiFi interfaces...

It's a bit complicated - but 160MHz throws twice as much ambient noise into the link - much like 40MHz on 2.4GHz - in my testing in 2.4GHz, there's little difference, but that's 11n, not 11ac, but the physics behind the numbers makes sense...

FWIW - Most of the current AC1900 class and higher internal switches are non-blocking and have a lot of BW across that switch... and if I recall, the Marvell WiFi solution on the WRT3200acm is over PCI-e 1x lanes, which again, is more than enough BW there.

 

[Cosmin Cazan]

Thanks for link Tim, for the bridge testing of WRT3200ACM, but as I figured, you are using 2 wired clients connected at 1Gbps. So even though the performance doesn't seem to be bottlenecked by the wire in this instace (it reaches only ~830Mbps), if it got just a 15% improvement (which I'm sure it will once mature), it will exceed 1G.

 

[thiggins]

(which I'm sure it will once mature), it will exceed 1G.

It may. But you'll need a very strong signal to reach the maximum MCS rate. So don't expect > 1 Gbps when going any significant distance.

 

[SledgeHammer01]

I tested 160 MHz bandwidth with a second WRT3200ACM in bridge mode, 6 feet away in open air. Can't get any simpler than that.

I have generally found Gigabit ports to not limit measured performance.

MU-MIMO doesn't come anywhere near maxing out a Gigabit Ethernet port.

I'm thinking of getting this setup. Did you test any further then 6 ft? Obstacles (interior 2x4 / 2x6 wall)?

 

[sfx2000]

It may. But you'll need a very strong signal to reach the maximum MCS rate. So don't expect > 1 Gbps when going any significant distance.

I have some "bulls-eye" charts that indicate MCS values at a given range to set expectation for RF planning - I'll dig around and see if I can pull them up...

 

[thiggins]

I'm thinking of getting this setup. Did you test any further then 6 ft? Obstacles (interior 2x4 / 2x6 wall)?

No.

 

[SledgeHammer01]

No.

Thiggins, two questions:
1) Do I need all 4 antenna on the secondary router to achieve 160mhz? Just asking because it fits with 3/4, but 4/4 means moving some stuff out of alignment.
2) Is there any way to verify the 160mhz connection without expensive test equipment?

 

[L&LD]

Do not run a WiFi router with the radios on and without their antennae attached (all or some of them).

Yes, all four antennae are needed to achieve the potential maximum.

 

[sfx2000]

1) Do I need all 4 antenna on the secondary router to achieve 160mhz? Just asking because it fits with 3/4, but 4/4 means moving some stuff out of alignment.
2) Is there any way to verify the 160mhz connection without expensive test equipment?

1) Yes, you need all the antennas - running the WRT with an antenna disconnected may damage the radio it was connected to

Note that the WRT is fairly large, akin to a 11-inch laptop like a MacBook Air 11" or an HP Stream 11" on the width/depth perspective...

2) Not really, as there are no 160MHz capable client chipsets out there right now...

For a sizing perspective... WRT vs. HP Stream 11...

 

[SledgeHammer01]

1) Yes, you need all the antennas - running the WRT with an antenna disconnected may damage the radio it was connected to

Note that the WRT is fairly large, akin to a 11-inch laptop like a MacBook Air 11" or an HP Stream 11" on the width/depth perspective...

Yeah, its a big boy. Guess I'll have to live with my speaker being slightly off-center. Pretty weird that running an antenna disconnected can damage a radio though.

Clarification for your statement though... running the twins in 160mhz mode requires you to run in AC only mode, so I disabled the 2.4Ghz network. Are the antennas tied to a specific network? I.e. 2.4Ghz = side antennas / 5Ghz = rear antennas or something like that? In which case disabling the 2.4Ghz network would disable some of the antennas, no? Just curious / grasping at straws here LOL .

I DID check Linksys's web site for dimensions, but they didn't include the measurements with the side antennas on . Bastards!

Thanks for the info.

 

[sfx2000]

Clarification for your statement though... running the twins in 160mhz mode requires you to run in AC only mode, so I disabled the 2.4Ghz network. Are the antennas tied to a specific network? I.e. 2.4Ghz = side antennas / 5Ghz = rear antennas or something like that? In which case disabling the 2.4Ghz network would disable some of the antennas, no? Just curious / grasping at straws here LOL .

All four antennas are shared between the 5GHz side and the 2.4GHz side, so it really doesn't matter if the 2.4GHz is on or off...

 

[L&LD]

The antennae are used for both 2.4GHz and 5GHz bands. The router is a 4x4:3 design.

http://www.smallnetbuilder.com/wire...-ac3200-mu-mimo-gigabit-wi-fi-router-reviewed

The 88W8864 data sheet states "4x4 MIMO 3-spatial Stream Dual-band 802.11ac offering 1.3Gbps WLAN PHY rate" (emphasis mine). That is pretty clear. But the 88W8964 data sheet doesn't state the number of spatial streams, instead opting for the vaguer "4x4 MIMO Dual-band 802.11ac Wave-2 solution enabling 2.6 Gbps WLAN PHY rate". But the 2.6 Gbps doesn't refer to the total link rate of a standard 4x4, four-stream design. Instead it refers to only the maximum 5 GHz link rate made possible when the contiguous 160 MHz bandwidth mode is used. Got it?

 

[sfx2000]

The antennae are used for both 2.4GHz and 5GHz bands. The router is a 4x4:3 design.

Exactly - the WRT1900/3200's are all the same here in that regard...

That extra radio does help with range on both the Tx and Rx side, and if one looks carefully, the spacing of those antennas is quite smart in the analog RF domain...

Always appreciated the HW design of the WRT's - it was the SW that was the big let down...

 

[thiggins]

Moved posts more specific about WRT3200ACM pros/cons to the article discussion thread. Please discuss product-specifics there.