WDS 802.11n bridge- what kind of speed you can expect?

[liukuohao]

Hi all,

Currently, I have my wireless WDS ptp 802.11n bridge set up like this:

Local network -----WDS router(A)~~~802.11n~~~WDS router(B)--PC(2)
|
PC(1)

I would like to know does anyone in the forum have set up a
dedicated point to point indoor WDS 802.11n wirless bridge
(not used as a repeater or access point, so no wireless station
can be associated) by using 2 x Engenius ESR-7750 dual band router
or any brand- Netgear/Buffalo.

I would like to know, what kind of speed that you are experiencing?
Approx. 60Mbps????? measure at PC(2) using 2.4GHz 40Mhz /5GHz 40Mhz
band(using iperf to meaure the speed)???

I would like to compare mine, because my current speed is about 30Mbps
measured at PC(2) which I think is very slow when you compare to the
real 802.11n speed.

Thanks.

Regards,
Marcus

 

[liukuohao]

Hi all,

I measured the speed with inSSIDer and found out it is at 65 mbps.
Since 802.11n is like 802.11g where the communication is always half duplex.
Meaning the radio chip cannot be
transmitter and receiver at the same time.
Otherwise it will be using full duplex.

The expected speed is around about 30 Mbps.

Just like 802.11g 's data rate = 54 mbps, but since it is half duplex, then it
will around about 20 Mbps.

Am I right here?

Thanks,

Regards,
Marcus

 

[thiggins]

inSSIDer doesnt' measure actual throughput. The "Speed" it reports is link rate, which is very indirectly related to actual usable bandwidth.

The difference between the link rate and actual bandwidth isn't a full / half duplex issue, but more due to protocol overhead.

iperf will measure actual bandwidth, but you need to be careful with the settings. LAN Speed Test or NetStress are easier to use. See How Fast Is Your Network? Five Ways To Measure Network Speed.

To your original question, a point to point wireless bridge will deliver the same throughput from Ethernet client to Ethernet client as you would get from an Ethernet client to a wireless client. The far end bridge partner is just acting as a bridge and there is no retransmission throughput hit.

 

[liukuohao]

Hi Tim,

Thanks for your reply.

I agreed that the speed reported when inSSIDer was running is in fact the
link speed or the theoritical speed. Just like the theoritical speed of using
802.11g = 54Mbps.

I have attached of the diagram that I cut and pasted from your
webpage link:

http://www.smallnetbuilder.com/wireless/wireless-howto/31191-everything-you-need-to-know-about-wireless-bridging-and-repeating-part-1-wds?start=4

I am referring the "wireless how to" section on the subject:
"Everything You Need To About Wireless Bridging and Repeating - Part 1:
WDS"

You did a test on the bandwidth performance on WDS between two
EnGenius routers: ESR-7750 and ESR-9850.

I would like to confirm the measurement is correct or not.
Please see the attached jpeg. I have posted a question in the diagram.

If the test on the bandwidth performance is true, then what is the distance
across the WDS link? If the distance is substantial, then I might want to
change my existing WDS routers to EnGenius.

Thanks in advance.

Regards,
Marcus

 

[thiggins]

The test was done with both ends of the bridge within 10 feet of each other and is real throughput.

 

[stevech]

WiFi (802.11) is Half duplex, yes, but that doesn't mean half the air link rate.
If it was half, that means there is an equal amount of to-client and from-client traffic. This is not the usual situation.

For predominant one-way traffic, like Internet browsing, file transfers, video streaming, where the reverse direction has mostly TCP ACKs, a good rule of thumb is that the IP layer net yield is about 70% of the air link rate - IF there is not a lot of competition for air time with other networks.

 

[liukuohao]

The test was done with both ends of the bridge within 10 feet of each other and is real throughput.

Hi Tim,

Thanks for your confirmation and distance provided.

Just 1 question more, if you don't mind.

I guess if the WDS routers are more 10 feet apart, say 20 or even 30 feet,
the bandwidth will be less than 87Mbps since the signal is weaker at
20 feet location(assuming there is no obstruction in path and the WDS routers
"Line of Sight" with each other), right????

Thank you.

Regards,
Marcus

 

[thiggins]

I guess if the WDS routers are more 10 feet apart, say 20 or even 30 feet,
the bandwidth will be less than 87Mbps since the signal is weaker at
20 feet location(assuming there is no obstruction in path and the WDS routers "Line of Sight" with each other), right????

Of course. Throughput drops off sharply with 11n once signal level drops.

 

[liukuohao]

Of course. Throughput drops off sharply with 11n once signal level drops.

Ok, Tim thanks for all your support.

 

[liukuohao]

Hi Tim,

Thanks for your confirmation and distance provided.

Just 1 question more, if you don't mind.

I guess if the WDS routers are more 10 feet apart, say 20 or even 30 feet,
the bandwidth will be less than 87Mbps since the signal is weaker at
20 feet location(assuming there is no obstruction in path and the WDS routers
"Line of Sight" with each other), right????

Thank you.

Regards,
Marcus

Hi Tim,

I bought 2 x ESR-7750 routers and tested WDS link (802.11n only)bandwidth using
IPERF.

I was very disappointed with the bandwidth. I am no way near the 87Mbps!!!

Please see the attachment.

Is there something that I have done wrong with my settings?

Thank you.

 

[stevech]

'11n is really over-sold, over-promised.

But because the manufacturers saturated the WiFi market with 11g, they had to over-hype the potential of 11n to keep the revenue coming in.

Same thing for 802.16e (mobile WiMax, a la Clearwire/Sprint). It's even more aggressive about spectral efficiency. But alas, in non-line-of-sight, you just can't beat mother nature.

 

[thiggins]

Hi Tim,

I bought 2 x ESR-7750 routers and tested WDS link (802.11n only)bandwidth using
IPERF.

I was very disappointed with the bandwidth. I am no way near the 87Mbps!!!

Please see the attachment.

Is there something that I have done wrong with my settings?

Thank you.

The plots you attached are showing actual throughput.
If I remember correctly, the previous "87 Mbps" came from inSSIDer, which reports link rate, not actual bandwidth.

You also need to be careful with iPERF, since the various settings can produce very different throughput results. I would use LAN Speed Test or some other banwidth metering program, which doesn't require messing with settings.

At any rate, 35 Mbps isn't too bad, depending on how far apart the two routers are. Also is the bandwidth measurement being made between wired clients attached to each router?

 

[liukuohao]

At any rate, 35 Mbps isn't too bad, depending on how far apart the two routers are. Also is the bandwidth measurement being made between wired clients attached to each router?

To answer your question:

How far apart? approx. 1 metre apart.

Yes, the bandwidth test was done on 2 wired attached clients- laptop and a desktop PC. However, the PC's on board gigabit NIC is hooked up to a gigabit network switch.

For my next test, I will test it on using 2 wired laptops instead using a desktop PC and
see if there is any difference or not.

The plots you attached are showing actual throughput.
If I remember correctly, the previous "87 Mbps" came from inSSIDer, which reports link rate, not actual bandwidth.

H....mmmmm.....
I thought you did the test on using IxChariot not inSSIDer???? please see the attached jpg on my previous thread.

 

[liukuohao]

Hi,

I did some further testings with the freeware of testing the speed of my WDS
network. That is using LAN Speed Test v1.1.

Please see the attachments.

So, I still cannot get the speed up to 87Mbps.
Maybe there is some kind of setting need to be done??

Any feedback is welcome.

Thanks.

 

[thiggins]

You should have WMM enabled on both routers. That should be the default. If it isn't, you should still enable them.

Separate the two routers by at least 3 meters. You could be overloading the receivers.

The "instability" you are seeing it typical of 802.11n. N operates with a higher packet loss than 11g, which causes the througput variation.

It is also normal to see higher throughput in one direction vs. another. The difference in Netmeter and Lan Speed test is that LAN speed test is just giving you an average.

Again, the 87 Mbps reading you saw from inSSIDer is a LINK rate, not actual throughput.

Are the two test laptops connected via Ethernet to the ESR7750s or wireless?

The performance you are seeing is typical; not the best, but also not the worse among products that I've tested.

 

[liukuohao]

Hi Tim,

Just to answer your question:

Are the two test laptops connected via Ethernet to the ESR7750s or wireless?

Ever since I started using the Lan Speed Test v1.1 software to test the WDS speed, 2 laptops are used as it is showed on all the test. Both laptops are connected to
the 1 of the routers's internal switch of 10/100Mbps ethernet ports. And not using wireless.

You should have WMM enabled on both routers. That should be the default. If it isn't, you should still enable them

From experience, WMM is only helpful if you are watching video streaming over the
Internet, right? That is using WAN port only(in my case, I am not using the WAN port at
all because WDS transmission is only happening on 1 of the ethernet ports). WMM is
about priotizing the video and voice packets which are more important and other data
packets.

Therefore, you have smoother video streaming over the net, I don't it will be much help if WMM is enabled in this kind of testing enviroment?

Thanks.

 

[liukuohao]

Again, the 87 Mbps reading you saw from inSSIDer is a LINK rate, not actual throughput.

I think having advertised data /link rate at 300Mbps which in reality is only approx.
30Mbps - 40Mbps is a real hidden drawback. This drawback/hype is still misleading ordinary
consumers who drawn in to purchase or chasing after these kind of fantastic 802.11N
300Mbps routers. Alas, most of them does not know the real fact that such speed is not
obtainable at all in real life.

 

[thiggins]

802.11n depends on WMM being enabled to enable link rates higher than 54 Mbps.
http://www.smallnetbuilder.com/wireless/wireless-features/30938-dont-mess-with-wmm

 

[liukuohao]

Hi Tim,

802.11n depends on WMM being enabled to enable link rates higher than 54 Mbps.

It may be true when the chipset is Atheros, but this did not happen to be
true with my case when testing WDS bridge using Ralink chipset.

When I turned on WMM for both my EnGenius ESR7750 routers running at
5GHz band WDS 802.11n link (with Channel Bonding enabled). The overall speed for:

Writing(uploading) for 2GB file to a shared out folder is 32 Mbps (WMM on) compared to
36 Mbps (WMM off). Difference by 4Mbps

Reading(downloading) for 2GB file from a shared out folder is 39 Mbps (WMM on)
compared to 40 Mbps (WMM off). Diffrence by 1Mbps

The result is not a Hugh difference in speed gain if WMM is turn on. In fact there is a
slight decrease in speed.

 

[thiggins]

WMM enabled is an 802.11n spec. If the Engenius products are not limiting link rates to 54Mbps with WMM off, then they are not meeting spec. I have seen spec compliance issues before with Ralink and/or Engenius products.

The ESR7750 was not Wi-Fi certified when I tested it. Don't know if they have had it certified since. Engenius is not diligent about having it's products certified.