2 mile wireless link. What card and settings should I use?

[DingbatCA]

Having fun with long range wireless.
What wireless card (MiniPCI) and setting should I use to get the best link. Best link, as in uptime, not speed. Trying to keep this link up when it rains or snows.

Both sides are using the same hardware. The system is running 802.11B. I have 250mw bi-directional amps. The antenna's are 19bd dishes and I am using Intel 2200BG wireless cards.
Is there a better card? More power will not help as the amp takes the card up to 250mw.
What setting should I use on the wireless card to get a better link? like `iwconfig eth1 rate 1M`

[alex.blackbit]

you can use a cantenna for long range links where the antennas are never moved and where you have a free line-of-sight.

[DingbatCA]

Using a cantenna would be a very large step backwards. Here is what I am currently using.
http://www.hyperlinktech.com/web/hg2419g.php

[John R. Graham]

Typo was confusing: "19bd". 19dB is impressive. What issues are you having, specifically?

- John
_________________
I can confirm that I have received between 0 and 499 National Security Letters.

[DingbatCA]

Link works about 80% of the time. I am trying to make up that last 20%. I can upgrade the cards, or change the settings. I can not change out any other hardware.
I found that if I drop the cards to 1M (iwconfig eth1 rate 1M) I can get better number. I have also been playing with the sens option. These help and I can gain back 5% more, but still have 15% down time. Thats in a clear weather environment.

When it rains the link goes down. 2.4GHz does not like water :-). The boss is not happy when the link is down. With 24dB dishes, and 1watt amps, I could get it. But not with the current hardware. So, is there a better card out there, or are there any setting I can use to help with the problem?

[John R. Graham]

Well, a couple of things. Given that rain kills the link, seems like your signal strength (or antenna gain) might be marginal. What is the reported signal strength?

Are all hardware manipulations out of the question? If not, I'd try the following:

• Make sure the antennas are aligned. This requires swinging them horizontally and vertically with access to a signal strength meter. I've done this with microwave antennas and it's a pain.
• At those frequencies, rain water is a good conductor. Consider putting in suitable awnings to keep water off the antennas.

It's true that the lower data rates use more total energy per bit, so the signal to noise ratio will be better at lower data rates. Nevertheless, I don't think network card manipulations are going to solve your problems.

- John
_________________
I can confirm that I have received between 0 and 499 National Security Letters.

[DingbatCA]

Antennas are aligned, spent a few hours on each side with my laptop doing so.
Antennas are under an overhang an protected from the weather.

Is the Intel 2200BG card the best choice, or is there something better for this?

[Cyker]

I recall a team doing this at an older DEFCON, where they got some ridiculous range on standard un-boosted cards...

... but they were using satellite comms-grade dishes, and that's some serious gain...

[John R. Graham]

DingbatCA, what's your signal strength in fair weather conditions? If it's good, then your "15-20%" issue is probably unrelated to the "dropout during the rain" issue. I recall some older 802.11b equipment that would periodically disconnect under unpredictable conditions. Might be worthwhile trying any different NIC hardware, just ensuring that it isn't based on the same chipset.

Regarding the rain issue, are you able to shorten the path length as an experiment? (And then wish for rain, hee hee.) If you halved the path length, it will result in roughly a 6dB gain increase, which is equivalent to the gain you'd see switching to 1w amplifiers from 250mw.

Working on the real problem is always a plus.

- John
_________________
I can confirm that I have received between 0 and 499 National Security Letters.

[DingbatCA]

I have a few extra cards, prism, ralink 25XX, and a broadcom... I will give those a try. I got permission to take the system down over Christmas and tinker. I will also be moving one of the dish's closer, about 1000ft, not much, but it should help some.

Link quality on a clear day is averaging 30~35/100, so poor. I would love to boost that to 50%

[John R. Graham]

1,000ft closer from 2 miles (roughly 10,000ft) would be 10*log((10000/9000)^2) dB, or just under 1dB, better. However, if you see a measurable improvement in your reliability, you have your answer: more gain (amps and/or antennas).

- John
_________________
I can confirm that I have received between 0 and 499 National Security Letters.

[barophobia]

All cards should use the same techniques to overcome noise and signal loss, short of reading detailed specs of each chip or trying every card you will probably be stabbing the the dark. The only difference is the antenna, how much good signal you can collect how much bad noise you do not collect. After that you can try running some filters, kalman filter pops in mind, to clean up your signal but I am no signals expert. I would try to collect some sample background noise data and build a filter from there. I hope you stayed awake in your signals and systems class I know I didn't. Anyways have fun!

Just remember the less you have to filter the better.

You know on a second thought maybe the network card does all that filtering for you already but who knows.
_________________
An apple is an apple unless you say it is not an apple!

[John R. Graham]

DingbatCA, one thing that I forgot to mention is that, at those frequencies, losses in the coax can be significant. If the equipment can be relocated to reduce the coax length substantially, I think you'll see significant improvement. There are also lower-loss varieties of cable available.

Good luck over the Holidays.

- John
_________________
I can confirm that I have received between 0 and 499 National Security Letters.