Slow speeds in internal wlan

[reddy]

Hello,

I've just bought RT-N66U hoping to increase transfer speeds between machines in my home network. I simply installed it leaving the majority of config options to default. Initially I used the same SSID for 2.4GHz and 5GHz bands. SSID is not hidden, wireless mode set to auto, channel bandwidth default 20/40 MHz, Authentication method is WPA2-Personal with quite a long key (34 chars). QoS is disabled by default.

Then I tested the connection between different machines using iperf (Linux and Win 7 versions). All machines within 3-5 meters from the router, not much of obstacles between them. I have one machine with Intel N 5300 card (5GHz capable) one with Intel N 6300 (5GHz capable) and two with Atheros AR9285 (2.4GHz only).

Even though it seems the cards get to high bit rate:

 Mode:Managed  Frequency:5.22 GHz  Access Point: 
          Bit Rate=405 Mb/s   Tx-Power=15 dBm
          Retry  long limit:7   RTS thr:off   Fragment thr:off
          Encryption key:off
          Power Management:off
          Link Quality=56/70  Signal level=-54 dBm
          Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
          Tx excessive retries:43772  Invalid misc:628   Missed beacon:0

iperf hardly ever reaches 28Mbps usually stays ~24Mbps...

# iperf -c 192.168.1.224
------------------------------------------------------------
Client connecting to 192.168.1.224, TCP port 5001
TCP window size: 21.6 KByte (default)
------------------------------------------------------------
[  3] local 192.168.1.50 port 43481 connected with 192.168.1.224 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0-10.0 sec  28.9 MBytes  24.2 Mbits/sec

When I set separate SSIDs for 2.4GHz and 5GHz to make sure cards connect to the correct bands I can get those ~24-28Mbps. Transfers within the same band seems to be even slower.

Initially I had firmware 3.0.0.4.270 so I updated to 3.0.0.4.374_257 with no significant change.

What's the most interesting, that I can get close to 40Mbps when I test external connections with speedtest.net over those wlan links, which is the maximum connection speed for my provider... So, wlan to WAN can get almost twice as fast (or faster, at the speed of the link to the Internet is the limit here) then internally wlan to wlan.

What's wrong? What shall I change in the settings to get higher speeds?

 

[reddy]

I've just tried Merlin's firmware 3.0.0.4_372.31_0, resetting the router to factory settings right after the upgrade and then setting back the same SSIDs and WPA2 keys (so I wouldn't have to reconfigure all the machines in the network). All other settings left to default.
Unfortunately, I see no improvement in the speed. I checked with inSSIDer and 2.4GHz link gets score between 41 and 48 with 3 nets overlapping and 2 sharing the same channel. Tried other channels, too (1, 6 and 11) - not much of a difference.

Any hints?

 

[joegreat]

I...then setting back the same SSIDs and WPA2 keys (so I wouldn't have to reconfigure all the machines in the network).

Hi,

You should try also to delete the client WLAN config and start fresh there. It can make a difference...

With kind regards
Joe

 

[reddy]

You should try also to delete the client WLAN config and start fresh there. It can make a difference...

Thanks for reply!
Do you mean remove all network config from Windows clients? I've just tried that and after reconnecting I still get very slow transfers

I haven't reconfigured Linux machines (it wouldn't change anything, I think. Linux configs are pure txt files). And there's slow connection on Linux clients as well.

I split 2.4GHz and 5GHz bands into separate SSIDs, disabled everything what makes sense (I followed http://blog.ittoby.com/2013/01/setup-and-tweak-your-new-asus-rt-ac66u.html) and still no progress.

BTW, I noticed very unstable pings, too.
For example, within 5GHz band:

# ping 192.168.1.7
PING 192.168.1.7 (192.168.1.7) 56(84) bytes of data.
64 bytes from 192.168.1.7: icmp_req=1 ttl=128 time=199 ms
64 bytes from 192.168.1.7: icmp_req=2 ttl=128 time=3.14 ms
64 bytes from 192.168.1.7: icmp_req=3 ttl=128 time=3.12 ms
64 bytes from 192.168.1.7: icmp_req=4 ttl=128 time=3.15 ms
64 bytes from 192.168.1.7: icmp_req=5 ttl=128 time=140 ms
64 bytes from 192.168.1.7: icmp_req=6 ttl=128 time=24.9 ms
64 bytes from 192.168.1.7: icmp_req=7 ttl=128 time=26.3 ms
64 bytes from 192.168.1.7: icmp_req=8 ttl=128 time=3.13 ms
64 bytes from 192.168.1.7: icmp_req=9 ttl=128 time=3.13 ms
64 bytes from 192.168.1.7: icmp_req=10 ttl=128 time=82.0 ms
64 bytes from 192.168.1.7: icmp_req=11 ttl=128 time=4.10 ms
64 bytes from 192.168.1.7: icmp_req=12 ttl=128 time=196 ms
^C
--- 192.168.1.7 ping statistics ---
12 packets transmitted, 12 received, 0% packet loss, time 11015ms
rtt min/avg/max/mdev = 3.126/57.503/199.983/74.572 ms

Between 2.4GHz and 5GHz looks a bit more stable but also some fluctuations visible:

# ping 192.168.1.116
PING 192.168.1.116 (192.168.1.116) 56(84) bytes of data.
64 bytes from 192.168.1.116: icmp_req=1 ttl=128 time=3.48 ms
64 bytes from 192.168.1.116: icmp_req=2 ttl=128 time=3.27 ms
64 bytes from 192.168.1.116: icmp_req=3 ttl=128 time=13.0 ms
64 bytes from 192.168.1.116: icmp_req=4 ttl=128 time=6.44 ms
64 bytes from 192.168.1.116: icmp_req=5 ttl=128 time=9.33 ms
64 bytes from 192.168.1.116: icmp_req=6 ttl=128 time=8.53 ms
64 bytes from 192.168.1.116: icmp_req=7 ttl=128 time=16.3 ms
64 bytes from 192.168.1.116: icmp_req=8 ttl=128 time=27.1 ms
64 bytes from 192.168.1.116: icmp_req=9 ttl=128 time=3.13 ms
64 bytes from 192.168.1.116: icmp_req=10 ttl=128 time=3.13 ms
64 bytes from 192.168.1.116: icmp_req=11 ttl=128 time=6.70 ms
64 bytes from 192.168.1.116: icmp_req=12 ttl=128 time=13.0 ms
64 bytes from 192.168.1.116: icmp_req=13 ttl=128 time=3.56 ms
64 bytes from 192.168.1.116: icmp_req=14 ttl=128 time=13.9 ms
^C
--- 192.168.1.116 ping statistics ---
14 packets transmitted, 14 received, 0% packet loss, time 13020ms
rtt min/avg/max/mdev = 3.133/9.372/27.199/6.624 ms

I simply have no further ideas what else I can check...

 

[thiggins]

You will always get half speed going wireless <-> wireless than wired <-> wireless.

Radio has to use double the airtime to move the packet.

 

[reddy]

You will always get half speed going wireless <-> wireless than wired <-> wireless.

Radio has to use double the airtime to move the packet.

I wouldn't mind if I get at least that
Even if I'd take single channel cards capable of (theoretically) 150Mbps, then I should get transfers like 75Mbps (or, let's say, ~50Mbps, as it would never reach the max). However, I reach ~28Mbps max in real life, while usually it's much less...

And why ping times vary that much?

 

[Deleted member 22229]

You will always get half speed going wireless <-> wireless than wired <-> wireless.

Radio has to use double the airtime to move the packet.

That's not true I get the same speeds wired or wireless all devices 60 down 11 up. In fact my wireless works so good that all my devices are now wireless the only time I use a wired connection is for firmware upgrades and trouble shooting if I need to.

 

[reddy]

I started to think that there's something wrong in the apartment I live now as I remember every time I tried to bore some hole in the wall I encounter something metallic inside. So I checked with the detector and indeed it seems I have something like metal plates in the walls... It's not a complete Faraday cage as I can see several networks from my neighbours but I wonder if that may cause so much reflections that affects my signal so heavily...

But, on the other hand, I managed to get speeds close to 80Mbps late evening from one machine in my 5Ghz network to the Internet (checked with speedtest.net) and even now I can get ~60Mbps. So the wlan is capable of delivering that speed outside, but not within LAN...

I run one more test - iperf between the same machine and another one connected with Ethernet cable (so, it should be similar to speedtest.net scenario, but within LAN) - and I got 28Mbps... I simply don't understand what can cause it.

 

[RogerSC]

That's not true I get the same speeds wired or wireless all devices 60 down 11 up. In fact my wireless works so good that all my devices are now wireless the only time I use a wired connection is for firmware upgrades and trouble shooting if I need to.

I believe that the question being addressed was data transfer between two wireless clients versus data transfer from a wired to a wireless client. That has nothing to do with internet download speed, just transfer speed between clients. So, yes, it is true *smile*.

 

[Deleted member 22229]

After reading the post again I believe your right thanks for the correction.

 

[reddy]

I'm starting to think I might have a defective unit. I simply don't have better explanation for this behaviour.
One more thing makes me wonder - high values of "Tx excessive retries" which quickly reaches hundreds of thousands.

Shall I try e.g. to downgrade the firmware to some old, stable version (I keep seeing in various threads that 4.270 is quite ok, where to get it? But, on the other hand, Merlin build I have installed now is using the same driver versions as that one...)
What other things I still can try to get decent speeds?

 

[reddy]

I couldn't understand why I kept getting better throughput on speedtest.net then to wired LAN, so I made another test, this time with iperf server running on Linux connected with Ethernet to the AP. I set both bands to N only to make sure none of legacy devices can slow down the network and double-checked all Bluetooth devices are off.

And this time iperf clients connecting from 5GHz wlan were able to get constantly 150-170Mbps. Nice. But clients from 2.4GHz can get 30Mbps max, usually stay close to 18Mbps... So, the problem is confirmed. In general:

5GHz   (Intel N 5300)   -> Ethernet                150-170Mbps
2.4GHz (Atheros AR9285) -> Ethernet                27-28Mbps 
2.4GHz (Atheros AR9285) -> 5GHz (Intel N 5300)     13-18Mbps
2.4GHz (Atheros AR9285) -> 2.4GHz (Atheros AR9285) 9.5Mbps

Have I missed something or my unit is defective?

 

[reddy]

I've done even more tests.
First, I reset the router to factory settings (from the GUI and then once again with reset button, just to be sure) and set up WLAN on completely new SSIDs so none of my devices could directly connect to it (so, I made sure none of the legacy devices are connected). Connected only test machines, run iperf tests - the same results... 5GHz network client to wired iperf server gives 150-170Mbps, 2.4GHz network client connects to the same server with max 18Mpbs...

As I thought my unit is defective, I turned on another access point (Inteno DG201) which is 2.4GHz only. Similar network setup (the same keys, AES encryption, WPA2) just different SSID - still can't get faster than 18Mbps...

So, either both routers are defective in the same way (very unlikely) or there's something weird in my environment which prevents 2.4GHz network from running with decent speed. InSSIDer still shows I have a very good link (for example, now link score is 70, co-channel nets: 6, overlapping 3) but the throughput is crappy...

I run out of ides. Any help, please?

 

[CaptainSTX]

Try all the channels starting at one and work your way up. It is possible that someone is heavily using channel six or even one of the overlapping channels which is impacting your through put.

One of my APs works on channel nine much better than any other channel. No way to know until you test.

 

[RMerlin]

Also keep in mind that InSSIDer can only show you a very small subset of potential interference sources. In addition to wifi, a lot of other applications use the 2.4 GHz band, and can interfere with wifi performance: cordless phones, microwave ovens, Bluetooth devices, baby monitors, wireless keyboard/mouse, and so on.

 

[CaptainSTX]

Dect 6.0

Just curious if anyone is still using cordless phones on a land line that isn't DECT 6.0? When did they stop making cordless phones that aren't DECT 6.0?

The frequency that DECT 6.0 phones use should not interfere with 2.4 Ghz WiFi which was the whole purpose of going to that technology/ frequency.

Microwave ovens on the other hand can really cause problems with the lower WiFi channels.

 

[RMerlin]

Just curious if anyone is still using cordless phones on a land line that isn't DECT 6.0? When did they stop making cordless phones that aren't DECT 6.0?

The frequency that DECT 6.0 phones use should not interfere with 2.4 Ghz WiFi which was the whole purpose of going to that technology/ frequency.

Microwave ovens on the other hand can really cause problems with the lower WiFi channels.

Large retailers still sell cheap junk that isn't DECT 6.0. And the vast majority of customers have no idea what DECT 6.0 means, so they won't pay attention to that detail when shopping for a new cordless phone.

 

[CaptainSTX]

I live in the Caribbean and we get a lot of third world junk foisted off on us. The primary retailer on island is Kmart and I haven't seen a non Dect 6 phone in their stores in the last couple of years.

Still interested in knowing how many people on this board have a non Dect 6 phones.

 

[dread111]

Non Dect 6.0 phones

I still have a couple of 900 mhz and a 2.4 ghz phone The rest are 5.8ghz. Never seem to have any interference problems because the 2.4 doesn't get used much and isn't near any wifi devices.

 

[RMerlin]

Looks like things have changed recently. When I last shopped for a new phone a few years ago, Futureshop still had separate categories for 900 MHz, DECT 6.0 and I forgot which other radio band,

That only leaves people who already have a cordless phone that's a few years old then.