Hi,

Last year I installed an outdoor wireless network consisting of 5 WRT54GL APs built into outdoor enclosures. I equipped them all with a single external 9dbi omnidirectional antenna, and connected them using WDS. As a gateway I use Chillispot and Freeradius running against a MySQL server. This setup was functioning well.

However, over time I have seen that the radio signals has been weaker and weaker, and I am now having serious problems with the APs seeing each other. I have tried all OpenWRT Whiterussian releases from RC5 and up, with the same bad result. I have also tried all antenna diversity/power settings, as well as changing channels etc.

My suspicion is now that there is something wrong with the hardware, but it seems very unlikely that all APs suddenly are showing the same low signal level, so I am a bit puzzled here to.

What really confuses me is that this setup was working great when I installed it, but now has degraded over time.

Any suggestions, or similar experiences?

BR

Arne-J.

Check for humidity in the antenna and coax cables (including the pigtails), or just try swapping them out. Even if the connector is not directly exposed to rain, humidity will creep in unless you have a perfect seal, and that destroys the cable, and possibly the connector.

Good tip.

Would I see this visually, or is there any other way to measure if this has happened?

BR

Arne-J.

Have the trees grown more leaves in the area in which you live as the signal has degraded?  Might as well start with the obvious.

"However, over time I have seen that the radio signals has been weaker and weaker"

Please post actual data; it is important to distinguish between loss of antenna signal (degradation of the antenna and cable hardware or new obstacles) and increased levels of interference. The 2.4 GHz ISM band gets ever more crowded, more microwave ovens, more AV links, more wireless lan systems.

For diagnosis:
wl scan ; sleep 1 ; wl scanresults
wl rate

from clients: wl rssi, wl noise
from access points: wl rssi <mac of client>

and stability of ping delay times, pinging the next hop towards internet and towards the end user.

I only get an error when running this sequence:

wl scan ; sleep 1 ; wl scanresults

eth1: Operation not supported

The other commands return as follows:

wl rate
rate is 1 Mbps

wl rssi <eth>
rssi is 0

wl noise
noise is -12

(Last edited by ajauberg on 7 Jun 2007, 13:00)

There are no visual signs of moisture inside a cable.  The white insulator in most coax cables will actually soak up moisture like a sponge. 

Interference will not make your signals any weaker only harder to understand.  If your rssi has dropped look for moisture damage.  Change out the pigtails and antenna on an affected unit and see if you have any improvements. 

Make sure your wrap all your connectors with something like coax-seal and cover that with a good electrical tape to shield it from sun light.  I also like to put silicone dielectric grease inside the connectors and on all Ethernet and power connectors on outdoor radios.   This will keep the connections clean, dry and corrosion free.    The coax-seal can be found on ebay and from several wisp hardware venders.  You can find silicone dielectric grease at an auto parts store.

@ajauberg
> the wl scan command can only be done while not in access point mode; so temporarily issue 'wl ap 0" and then do the scan.

wl rate
rate is 1 Mbps

> It cannot get worse; something definitely wrong. Reasonable to good 802.11g links should operate at 36|48|54 Mbps

wl rssi <eth>
rssi is 0

> From an access point you have to give the mac address of the other side 'wl rssi <mac address>'; you can see the mac addresses with 'wl assoclist'

wl noise
noise is -12

> on an wrt54gl this value should be in the -90 <> -96 dBm range. This command should also be issued while NOT in ap mode

(Last edited by doddel on 7 Jun 2007, 14:16)

well, now it does not complain anymore, but I get no output:

wl ap 0
wl scan; sleep 1; wl scanresults

wl rssi 00:02:2D:21:E7:9D
rssi is -75

wl noise
noise is -92

with a S/N ratio ~ 17 dB you should be able to get much better result than 1 Mbps datarate.

Has this datarate been better in the past ? (am asking because if it never was good the distance may be too big and you need to set the wl0_distance parameter).

If the modulation data rate was much higher in the past I can only explain the problem from interference on the channel, from other signals than wlan.
FM transmitters, as used in AV links, are very capable of destroying wlan datatransport while not being properly detected by the wlan hardware. Can only suggest to choose a radically different channel, e.g. 14 for testing purpose, and see if it makes a difference.

a typical example of a 'good' wrt54gl client (the one this message goes through):
# wl rssi ; wl noise ; wl rate
rssi is -66
noise is -96
rate is 54 Mbps

a weak brother in our network (also a wrt54gl) but in a noise free rural environment over more than 1 km distance using a directional antenna on the client side and a 7 element collinear array (omni-directional antenna with vertical polarization) at the access point
# wl rssi ; wl noise ; wl rate
rssi is -88
noise is -96
rate is 24 Mbps

The last result illustrates that the problem in your case probably is not the signal strength but some sort of interference, reflections from nearby objects, or timing issues.

(Last edited by doddel on 7 Jun 2007, 22:01)

OK, I have finally solved this.

It appears that my antenna cable was the problem. I appearantly have not sealed the cable properly, and have now replaced it with a properly sealed cable. The results I get now are:

wl rate: rate is 48 Mbps
wl noise: noise is -12

Also, my antenna and enclosure connectors are both pointing downwards, so luckily I did not have to replace the antenna and pigtail.

The lesson learned from this is: Always properly seal your antenna connectors on outdoor solutions!!

Arne-J.