PS5 setup help / speeds

[timd]

We just got a set of PS5 and set them up at about 75 or 100 feet apart and the max throughput we are seeing on either a transmit or a receive test is only about 15 megs. We have tried almost every combination of settings but currently have it set to:

Channel width: 40 mhz
Channel shifting: enabled
Output Power: varied from 8 dBm to 22 dBm
Data Rate: 54 auto (tried other settings with auto on / off)
Security: tried with both none and WEP
Rate Algorithm: tried all of them but have it set currently to: EWMA
Noise Immunity: tried all of them but have it set currently to: For 802.11 Traffic
ACK Timeout: set to auto at 21
SuperG features: all enabled here

Also for whatever reason when we would enable WPA2 the radios would lose the link completely and sometimes just the client would connect but the AP would never recognize it - the weird thing is with PS2 we did not have the same problem. So I guess our two issues with PS5s are:
1) Speeds - we were hoping for 25 to 30 megs
2) WPA2 - we want to use WPA2 but for whatever reasons it does not seem to be working.

Thanks,

Tim.

[Frotihngdog.ca]

Come to think of it...we couldn't get WPA2 to work either.

But are speeds were decent even at 8km. I can't remember off hand what the speeds were though.

1.
Try DISABLING the CHANNEL SHIFTING.

2.
What Freq are you runnging it in? 5.2 or 5.7? or have you tried both?

[UBNT-Robert]

Hi guys --

I think the max we have seen in the lab with a 40MHz channel is ~27Mbps.

For this performance, you really should have a signal at -60dBm or better and a clean environment.

I recommend using 20MHz channel, throughput should get close to 25Mbps.

The WPA-2 bug is interesting -- defenitely check with support@ubnt.com; they might have some ideas.

Robert

[WHT]

I recommend using 20MHz channel, throughput should get close to 25Mbps.
Regardless of the actual throughput, what end-point speed are you seeing after you subtract all the wireless overhead. With 25 Mbps throughout I would expect like 12 Mbps to the subscriber.

[UBNT-keba]

WHT: it is real 25Mbps TCP/IP traffic. It is not 25Mbps wireless rate which is usualy = ~12Mbps TCP-IP traffic.

[timd]

We tried both 5.2 and 5.8 frequencies. The throughput was pretty much the same. After we got the link setup on the building out rates were:
down: 18 megs
up: 6 to 8 megs
we used speakeasy.net for the speed tests.

The AP says the rssi at for the subscriber unit is: -54 dbm
The subscriber unit says its' rssi is: -36 dbm

We have tried using a 20 mhz channel and the speeds seemed pretty much the same.

We were hoping to achieve at least 25 megs down - is that even possible?

Here are my config options:
users.status=enabled
telnetd.status=enabled
resolv.host.1.name=wap-bbank-01
resolv.host.1.status=enabled
igmpproxy.status=disabled
netmode=bridge
resolv.status=enabled
resolv.nameserver.2.status=disabled
resolv.nameserver.1.status=enabled
resolv.nameserver.1.ip=0.0.0.0
iptables.status=disabled
ppp.1.status=disabled
iptables.3.status=disabled
ppp.status=disabled
ppp.1.password=
users.1.status=enabled
users.1.name=ubnt
users.1.password=Aa.jIbGUBFLaA
httpd.status=enabled
httpd.port.http=80
radio.status=enabled
radio.countrycode=840
radio.ratemodule=ath_rate_minstrel
radio.1.status=enabled
radio.1.devname=ath0
radio.1.mode=master
radio.1.rate.max=54M
radio.1.rate.auto=enabled
radio.1.ieee_mode=ast
radio.1.txpower=13
radio.1.acktimeout=21
radio.1.ack.auto=enabled
radio.1.rx_antenna_diversity=disabled
radio.1.tx_antenna_diversity=disabled
radio.1.rx_antenna=1
radio.1.tx_antenna=1
wireless.status=enabled
wireless.1.status=enabled
wireless.1.devname=ath0
wireless.1.security=wep64
wireless.1.ssid=PogoZone
wireless.1.hide_ssid=disabled
bridge.status=enabled
bridge.1.devname=br0
bridge.1.fd=1
bridge.1.port.1.devname=eth0
bridge.1.port.2.devname=ath0
netconf.status=enabled
netconf.1.status=enabled
netconf.1.devname=eth0
netconf.1.ip=0.0.0.0
netconf.1.netmask=255.255.255.0
netconf.1.up=enabled
netconf.1.promisc=enabled
netconf.2.status=enabled
netconf.2.devname=ath0
netconf.2.ip=0.0.0.0
netconf.2.netmask=255.255.255.0
netconf.2.up=enabled
netconf.3.status=enabled
netconf.3.devname=br0
netconf.3.ip=10.125.7.254
netconf.3.netmask=255.255.0.0
netconf.3.up=enabled
dhcpd.status=disabled
dhcpd.1.status=disabled
dhcpc.status=disabled
dhcpc.1.status=disabled
dhcpc.1.devname=br0
route.status=enabled
route.1.status=enabled
route.1.devname=br0
route.1.gateway=10.125.0.1
route.1.ip=0.0.0.0
route.1.netmask=0
ebtables.status=enabled
ebtables.1.status=disabled
ebtables.1.cmd=-t nat -A PREROUTING --in-interface ath0 -j arpnat --arpnat-target ACCEPT
ebtables.2.status=disabled
ebtables.2.cmd=-t nat -A POSTROUTING --out-interface ath0 -j arpnat --arpnat-target ACCEPT
ebtables.3.status=enabled
ebtables.3.cmd=-t broute -A BROUTING --protocol 0x888e --in-interface ath0 -j DROP
wireless.1.wds=disabled
radio.1.channel=42
wireless.1.macclone=disabled
radio.1.countrycode=840
radio.1.mcastrate=6M
radio.1.clksel=0
radio.1.chanshift=5
wireless.1.authmode=2
wireless.1.ap=
wireless.1.mac_acl.status=disabled
wireless.1.mac_acl.policy=allow
wireless.1.mac_acl.1.status=disabled
wireless.1.mac_acl.1.mac=
wireless.1.mac_acl.2.status=disabled
wireless.1.mac_acl.2.mac=
wireless.1.mac_acl.3.status=disabled
wireless.1.mac_acl.3.mac=
wireless.1.mac_acl.4.status=disabled
wireless.1.mac_acl.4.mac=
wireless.1.mac_acl.5.status=disabled
wireless.1.mac_acl.5.mac=
wireless.1.mac_acl.6.status=disabled
wireless.1.mac_acl.6.mac=
wireless.1.mac_acl.7.status=disabled
wireless.1.mac_acl.7.mac=
wireless.1.mac_acl.8.status=disabled
wireless.1.mac_acl.8.mac=
wireless.1.mac_acl.9.status=disabled
wireless.1.mac_acl.9.mac=
wireless.1.mac_acl.10.status=disabled
wireless.1.mac_acl.10.mac=
wireless.1.mac_acl.11.status=disabled
wireless.1.mac_acl.11.mac=
wireless.1.mac_acl.12.status=disabled
wireless.1.mac_acl.12.mac=
wireless.1.mac_acl.13.status=disabled
wireless.1.mac_acl.13.mac=
wireless.1.mac_acl.14.status=disabled
wireless.1.mac_acl.14.mac=
wireless.1.mac_acl.15.status=disabled
wireless.1.mac_acl.15.mac=
wireless.1.mac_acl.16.status=disabled
wireless.1.mac_acl.16.mac=
aaa.status=disabled
aaa.1.status=disabled
wpasupplicant.status=disabled
wpasupplicant.device.1.status=disabled
wireless.1.wmm=disabled
wireless.1.wmmlevel=-1
radio.1.rts=off
radio.1.frag=off
wireless.1.fastframes=enabled
wireless.1.frameburst=enabled
wireless.1.compression=enabled
wireless.1.signal_led1=94
wireless.1.signal_led2=80
wireless.1.signal_led3=73
wireless.1.signal_led4=65
tshaper.status=disabled
wireless.1.sper=disabled
netconf.2.allmulti=disabled
wireless.1.l2_isolation=disabled
radio.1.thresh62a=40
radio.1.thresh62b=40
radio.1.thresh62g=40
aaa.1.wpa=1
aaa.1.wpa.1.pairwise=TKIP CCMP
aaa.1.driver=madwifi
aaa.1.devname=ath0
aaa.1.br.devname=br0
aaa.1.ssid=PogoZone
aaa.1.wpa.psk=p0rkf0rk
aaa.1.wpa.key.1.mgmt=WPA-PSK
wpasupplicant.device.1.profile=
wpasupplicant.profile.1.name=
wpasupplicant.profile.1.network.1.key_mgmt.1.name=
wpasupplicant.profile.1.network.1.psk=p0rkf0rk
wireless.1.security.default_key=1
wireless.1.security.1.key=1abcddcba1
snmp.status=enabled
snmp.community=pU8l1c
snmp.contact=support@pogozone.net

[MaximumISP]

have you tried disabling compression and then testing again

[UBNT-keba]

Hi,

My suggestions are:

1. Setup fixed ACK Timeout. Just check what is ACK Timeout is on Status page and set it in advanced page. Also try to make it bigger or smaller (not much with in rage 3-4)

2. Use different rate algorith. I'd suggest Optimistic to push for the Max throughput.

3. Use fixed rates like 36Mbps or 48Mbps. It is some times better than 54Mbps fixed because if you have retries/errors on high rates throughput will never be better that with lower rate without retries and errors.

4. Do all these adjustment on both sides.

5. According to RSSI levels it seems that units' anetnnas are not 100% adjusted. The RSSI should be +-5dBm the same on both sides.

BTW: what is the firmware version? Is AP also PS5?

[timd]

Firmware version: we are at the 3.0 version
The AP and the Station are both PS5. Is this alright or would you recommend something else for the AP?

I'll do the suggestions 1 thru 4 and get post back soon and then do some alignment adjustments on the units soon to see if I can get the units more inline as far as RSSI values.

[WHT]

WHT: it is real 25Mbps TCP/IP traffic. It is not 25Mbps wireless rate which is usualy = ~12Mbps TCP-IP traffic.
I'm a little unclear the way you phrased your answer. So are you saying that if I had a 6 Mbps DSL connection and set two Nanostations fifty feet apart for the 6 Mbps data rate, I should see speeds approaching 6 Mbps at the far end or a mere 3 Mbps?

[timd]

I tried the items you suggested (ie ack timeout, different rate algorithms, and using fixed rates) and this is what I noticed:
1) I cannot set the ACK timeout elow 21 - I did set it higher and the speeds got worse as I set it higher.
2) Tried different rate algorithms and Optimistic seemed to be the worst as far as actual throughput using the built in speed test utility on the radios (it was over a half of meg worse then using EWMA).
3) Setting it to 36 megs did improve the rate by about 0.2 Mbps.
4) I also lowered the power on the radios which helped a tad bit.

Overall I am still only seeing about 14.5 Mbps for both Rx and Tx speeds using the built in speed test utility on the radio (doing the speed test from the AP side). Has anyone been able to achieve faster then these speeds at a 100 ft distance?

Any other ideas?

[UBNT-keba]

Hi,

Can you try to use any speed test utility like iperf, netpert, netpim or chariot for PC to PC not internal speed test utility on PS5 itself. PS5 is 180MHz CPU only and all LS/NS/PS devices are optimized to do traffic pass through the device not to device. If you get ~18Mbps with internal speed test so I'm pretty sure you will get ~23Mbps through the devices.

Also 100 feet is very short distance for devices with high power 26dBm and 22dBm antennas inside. Try to lower down the power as much as possible (to 8dBm).

[WHT]

PS5 is 180MHz CPU only and all LS/NS/PS devices are optimized to do traffic pass through the device not to device. If you get ~18Mbps with internal speed test so I'm pretty sure you will get ~23Mbps through the devices.

That sounds a little too vague to answer my basic question that was reduced to the most basic terminology to avoid nebulous, convoluted and vague answers.

If I had a 6 Mbps DSL connection and set two Nanostations a few hundred feet apart for the 6 Mbps data rate, I should see speeds approaching 6 Mbps at the far end or a mere 3 Mbps?

[UBNT-keba]

If I had a 6 Mbps DSL connection and set two Nanostations a few hundred feet apart for the 6 Mbps data rate, I should see speeds approaching 6 Mbps at the far end or a mere 3 Mbps?

Answer is ~3Mbps because:
TCP-IP Data Rate ~= Wireless Data Rate / 2.
I'm talking about 802.11a/b/g, 20Mhz channel, no Turbo, no 11N, no extra features like Fast frames, Bursting and so on.

Set Wireless Data rate to 12Mbps and you will see ~6Mbps.

[WHT]

Set Wireless Data rate to 12Mbps and you will see ~6Mbps.
I just wanted to double check what Ubiquiti was saying. I've always mapped out my coverage and end-point user speeds by taking the manufacturer's published speeds and divide it by two.

For the customers that we sell a 10 Mbps connection, I map out the coverage using published receiver specs for 24 Mbps.

[ddunkin]

I work with timd. I specifically asked about what to get, given what we were trying to do, before we ordered the equipment. UBNT-Mike said we could get 35-40Mbps with the 5GHz equipment.
http://ubnt.com/forum/viewtopic.php?t=1686

So is that not really possible?

[UBNT-keba]

WHT you are absolutely right.
PHY rate * efficiency = Throughput.
In 11a/b/g efficiency is 55% in best case. All other is protocol overhead.

So assume there is 1538 byte packet, PHY rate 54Mbps, efficiency 55% then throughput is 29Mbps in ideal case.

[WHT]

Hmmm...my response didn't show up, so I'll try again. Coincidently I posted it, or thought I did, a mere few minutes before Kebe above posted his response.

Depends how you ask the question.

If your data speed, throughput speed, data rate, wireless speed, or WHATEVER phase you want to use is 35 to 40 Mbps, then your client, CPE, subscriber, end user, end point, or WHATEVER you want to call it will enjoy a 17 to 20 Mbps experience.

Thats why I explicitly asked the question like I did:


If I had a 6 Mbps DSL connection and set two Nanostations a few hundred feet apart for the 6 Mbps data rate, I should see speeds approaching 6 Mbps at the far end or a mere 3 Mbps?

Answer is ~3Mbps because:
TCP-IP Data Rate ~= Wireless Data Rate / 2.
I'm talking about 802.11a/b/g, 20Mhz channel, no Turbo, no 11N, no extra features like Fast frames, Bursting and so on.

Set Wireless Data rate to 12Mbps and you will see ~6Mbps.

[UBNT-keba]

To ddunkin.

UBNT-Mike was saying "...5Ghz band, using 40Mhz turbo, and you could achieve around 35-40Mbps..."
In his case channel width is 2 times larger so in theory (ideal case) you can get ~54Mbps. But reality is ~35Mbps right now.

[WHT]

For the customers that we sell a 10 Mbps connection, I map out the coverage using published receiver specs for 24 Mbps.
THEN...I take my -86 dBm receiver sensitivity for 24 Mbps data rate and subtract 20 dB for fade margin and plot my signal contour for -66 dBm.

So in real life, using a NS5 with an external 12 dB onmi antenna (max gain allowed for U.S. EIRP limit) will RELIABLY deliver a 10 Mbps experience to another NS5 with the internal 14 dB antenna about a 8/10 mile away.

Now 20 dB fade margin might be a little overly cautious, but considering the greatest reason for subscriber churn in the industry is loosing the signal, I want to be on the safe side.

[ddunkin]

UBNT-Mike was saying "...5Ghz band, using 40Mhz turbo, and you could achieve around 35-40Mbps..."
In his case channel width is 2 times larger so in theory (ideal case) you can get ~54Mbps. But reality is ~35Mbps right now.

We've observed no performance difference between 20 and 40MHz channels. Has anyone actually achieved anything like 35Mbps, even in a test environment?

[ewc]

UBNT-Mike was saying "...5Ghz band, using 40Mhz turbo, and you could achieve around 35-40Mbps..."
In his case channel width is 2 times larger so in theory (ideal case) you can get ~54Mbps. But reality is ~35Mbps right now.

We've observed no performance difference between 20 and 40MHz channels. Has anyone actually achieved anything like 35Mbps, even in a test environment?

I have.

But it is unachievable with the current nano and powerstations due to their lack of CPU power, but for me this is fine, because they are low cost CPEs if you wish to pass that much data it is more than achievable with MT, especially their new 400 series routerboards. Not that I should be mentioning brands, but as UBNT is currently not compeating in this area of the game I guess it's not too bad, please advise if otherwise.
I have tested a nanostation-----nanostation link and I can advise there was little to no benefit in running it in 40Mhz mode.

Regards

Aidan
:)

[ddunkin]

I understand it's a low cost unit, it just isn't cool to say something will do 40Mbps if it can only do 25. Thanks for corroborating that 40Mhz channels don't help.

[ewc]

I understand it's a low cost unit, it just isn't cool to say something will do 40Mbps if it can only do 25. Thanks for corroborating that 40Mhz channels don't help.

Ahhh yes, If I have a dollar for every time I had heard that!

Imagine this though. A MikroTik RIC/522C is next to a Nanostation, the MikroTik will be advertising its ability to bridge 108Mb/s while the Nanostaton has sign saying it can bridge 25Mb/s TCP, which would you buy? (sorry you probably have a better idea, i mean which would the new WISP owner buy)
Realistically both wont do over 25Mb/s if that. Its an unfortunate reality of the wireless world and one that gets you nice and confused.

I know it sucks, it really does, but its all part of the game :)

Aidan

[WHT]

EWC..Reminds me of a cartoon back in the days before cable TV. Salesman was telling a couple the TV picture antenna would of course be better with an outdoor antenna at their house and you could see he had a 500 element antenna on the store roof.

[ewc]

@ WHT

lol i bet, this is pretty typical of the industry though, with AMD and their 2400 CPU's that only run at 2GHz or whatever, just because people saw the Intel P4's being 2.4GHz and didn't realise a 2GHz athlon equalled a 2.4GHz P4.

@ whomever is listening

What we have to remember is that when we are transferring 27Mb/s TCP we are actually transferring close to 54Mb/s over all, so it isn't really incorrect per se :twisted:

Aidan
:)

[DrNutbush]

Does this pic interest anyone? I have done somewhat extensive testing now for weeks and in this pic are the best speeds I get from NS5's. It is with DD-WRT software, which is where I notice the difference between 20MHz and 40MHz the most. At 20MHz I don't notice much difference between the AirOS and the DD-WRT...
http://picasaweb.google.com/DrNutbush/AllgaeuPCStuff#5246568557805737570

This test was indoors, though I am getting similiar results outdoors up to 500 meters. Am about to begin 2 Kilometer testing...

Just my 2 cents

[Wifi442]

Does this pic interest anyone? I have done somewhat extensive testing now for weeks and in this pic are the best speeds I get from NS5's. It is with DD-WRT software, which is where I notice the difference between 20MHz and 40MHz the most. At 20MHz I don't notice much difference between the AirOS and the DD-WRT...
http://picasaweb.google.com/DrNutbush/AllgaeuPCStuff#5246568557805737570

This test was indoors, though I am getting similiar results outdoors up to 500 meters. Am about to begin 2 Kilometer testing...

Just my 2 cents

Any update on this? Has anyone else been able to reproduce these results? Looks interesting!

[rconaway]

We have confirmed several things with the PS5

1) 20MHz channel - throughput 25-26Mbps.
2) 40MHz channel - 34-35Mbps processor limited.

A lot of it depends on your settings. Just do an FTP download locally to test that.

[Wifi442]

We have confirmed several things with the PS5

1) 20MHz channel - throughput 25-26Mbps.
2) 40MHz channel - 34-35Mbps processor limited.

A lot of it depends on your settings. Just do an FTP download locally to test that.

That is what I figured, DrNutbush's post was showing closer to 45Mbps speeds with DD-WRT (see his screen shot). I thought maybe DD-WRT might have less overhead or something. Last night I installed the DD-WRT trial on 2 of my bullet5's and the speed did not improve. There are some cool features but for now I'm sticking with simple and effective Air OS :)

[rconaway]

We are proposing large projects based on AirOS. I'm hoping that a mesh product comes out soon but I am a firm believer in the KISS principle. AirOS gives a 90% solution for 10% of the cost and I'm committing it to commercial and government installations for that reason. Now if it came out with multiple SSID's and better WPA/WDS compatibility, I would really be happy.