Problems with DWL-7700 bridges

[mboyaci]

Hi,

I have two D-Lİnk DWL-7700AP units set up in a point to point bridge configuration in 802.11a mode.

I have two dish antennas on either side. RSSI indication on both units is 100% (lowest so far is 96%). So the signal seems to be coming through. There is a distance of just under two miles between the bridges with clear line of sight.

My problem is that whatever I have done I cannot go over 45K/s transfer speed (measured from ftp server attached to one bridge to ftp client attached to the second bridge).

When I look at the statistics I see a transmit success rate of under 10% (sometimes as low as 2%). Nearly 50% receive duplicate rate.

I have tried everything I can think of including reducing the transmit power to as low as 1/8th and the transfer speed remains the same.

Can anyone please help?

Thanks in advance.

Mehmet

[jagajutt]

Hi Mehmet,

RSSI shows signal strength ... do these bridges show the SIGNAL QUALITY. There is a big difference between the two. I will post a nice paragraph succinctly describing the two if I can find where I saved it!

I guess from the results we can assume that the Quality is poor. Try changing the channel in case there is an interfernce source i.e. other wireless network, microwave frequencies in the area.

Also by 45 k/s do you mean 45kb/s == 360 kilo bits per second.

I am not sure how 802.11a mode decides on the transfer rate - I know if the link is not clear Ap's etc usually decrease the transmit speed i.e from 54 Mbps to 11 to 2 Mbps etc in a attempt to reduce packet loss. Do the bridges used have logs ???

Post back

Jag

[mboyaci]

Jag,

Thanks for the reply.

I know about signal strength and signal quality. We have a network of Cisco 350 bridges. The Cisco units display a lot more statistics including signal quality and they have logs you can download. Unfortunately the only measurement the D-Link units display is the RSSI indication.

I have tried every available channel with the same result. The reason why I chose 802.11a was because 802.11b/g is becoming very popular and is everywhere. I was hoping there would be less interference with 802.11a.

I let the unit decide the transfer speed automatically. I always get around 45KB/s (about 400 kilo bits/sec).

I am using 90cm dish antennas with a clear line of sight. I was hoping with the focused beam the quality would be much better.

Also I don't know if this means anything but I can ping one bridge from the other in 1 ms with an occasional 2-3 ms lapse. When I increase the packet size to 1000 bytes the duration goes up to 3 ms (occasionally 4-7 ms). When I increase the packet size to 10000 bytes the time goes up to 360-380 ms. At 65000 bytes the time goes up to 2700 ms.

Mehmet

[jagajutt]

Mehmet,

I think you might be on the right track by changing the TCP MTU (maximum transfer unit). I understand the ping times went up once the MTU was set to 10000 bytes - this makes sense because each packet is much larger and would take longer to travel hence RTT ping should increase - but did you try transfering a file afterwards what was the data throughput??

I think searching for software that monitors TCP datagrams may be an option ..... or papers on the topic of TCP over wireless.

But basically increasing the MTU size reduces TCP/IP header overhead which should improve data throughput!

Will post back soon.

Jag

[mboyaci]

Jag,

Thanks again for the reply.

Transmission speed did not change after my tests.

The problem as far as I can see seems to be signal strength versus signal quality as you suggested. I have very good signal strength but apparently very low signal quality.

I tried searching the net for tools to test signal quality but couldn't find anything. Can you recommend anything that I can use to test signal quality. The device itself does not measure signal quality.

Regards,

Mehmet

[jagajutt]

don't know of any tools - i hope someone will help with that.

Also try putting the bridges on G mode (2.4GHz) just to see if it helps the overall transfer rates. Try a few different channels. Unfortunately, with no logs or quality meter, you will have to use the transfer rates you achieve and assume the quality is better or worse or the same.

I would assume 2.4 GHz is more congested but because it is lower frequency is should travel further with less attenuation.

As you are probably aware TCP has a back-off alogrithm, if packets are having to be retransmitted TCP assumes the link/medium is congested so backs off --> where with wireless it is probable that the link quality is poor and this will never improve. So the FTP transfer in unecessarily holding off.

Do the bridges tell you what the speed the link is i.e. 54 Mbps, 22 etc?

Jag

[mboyaci]

I can set the speed manually but it does not matter whether I set it to 1 Mb/s or 54 Mb/s the transfer rate remain the same.

I used to have a Cisco Aironet 350 bridge (802.11b) but the performance with that was even worse. That is why I switched to 802.11a.

I will keep trying and look for software that might be useful to check signal quality.

Regards,

Mehmet

[Elena]

hi everyone,

i have a problem with my D-link DWL-7700AP access point. we bought it down from the roof and seems the two antennas are damaged. tried pinging to the server and still no rpely. I am about to order the two antennas as i think this could be the problem. is there another way i could confirm this before i order the two antennas.

thanks

[Alan87i]

Can you login to the device?
If the antennas are damaged there could be a chance water has gotten inside that could ruin the device.