Wireless across half-mile?
-
You need directional antennas.
This might interrest you:
http://www.hubersuhner.ch/co-ch/de/mozilla/products/hs-p-rf/hs-rf-wireless-data-links.htm
Although the SL60 series is quite expensive. (multiple 10'000 $).
But you get a guaranteed full-duplex 100Mbit/s link.WLAN is a lot cheaper.
Usually i work in the 5GHz band for links due to less distortions, and you can send with 30 dBm instead of 20 dBm in the 2.5 GHz band.
I've had some links up to 8km at minimum 16 Mbit/s.
At 4 km. we had the full 36Mbit/s 802.11a provides.If you run 802.11g with one of the turbo modes, you can get about 70 Mbit/s.
I've had good experiences with this card:
http://www.compex.com.sg/fullDescription.aspx?pID=24
For 802.11g mode only this one
http://www.compex.com.sg/fullDescription.aspx?pID=29
would be even better since you have 28 dBm TX power availlable.For the antennas: I've made good experiences with "Huber and Suhner" Antennas.
I have a few of them together with outdoor-cases at home. Can't look up the exact part number right now, but some similar antennas:http://www.hubersuhner.ch/co-ch/de/mozilla/products/hs-p-search/hs-rf-product-search/hs-sup-search-prod-rf-ant-2g.htm?applied=1&selected=1&search=wiz&display=50&unit=si&step=1&col=&sort=&vt=16081&list=d&listed=2&prm=B77840D275C6A610BB988D9C8E2F87C0
http://www.hubersuhner.ch/co-ch/de/mozilla/products/hs-p-search/hs-rf-product-search/hs-sup-search-prod-rf-ant-2g.htm?applied=1&selected=1&search=wiz&display=50&unit=si&step=1&col=&sort=&vt=16081&list=d&listed=2&prm=F9EAD0C820D9A756445FEC5D88B323A6
http://www.hubersuhner.ch/co-ch/de/mozilla/products/hs-p-search/hs-rf-product-search/hs-sup-search-prod-rf-ant-2g.htm?applied=1&selected=1&search=wiz&display=50&unit=si&step=1&col=&sort=&vt=16081&list=d&listed=2&prm=1D4C39592919EBEFAE31F9E877DE604C
The usually have a nominal gain of around ~ 20dBi.
-
Well, like I said, price isn't really a huge concern because a half-mile run of fiber would likely be more than any wireless system I could deploy.
GruensFroeschli, those are some excellent recommendations and I appreciate that you told me what your results were with that hardware. I'm going to give those a read, thanks.
-
I have used Cisco AIR-BR1310G-A-K9 with integrated antenna. New they are over $1,000 each. Used they are about $200 each and you can still get smartnet on them for unlimited replacement and unlimited phone support for about $100 each per year. I have 4 deployed and get about 26 mbit throughput with them. Personally I would recommend using them since the price is just hard to beat compared to a homebrew wireless bridge solution. We tried many semi-pro systems Like netgear's business line and Dlink's business line but the minute we switched to the Cisco bridges we have never had them go down besides maintenance upgrades in 4 years. Don't get me wrong…I love using opensource as much as possible but sometimes the cost on opensource for something like this can end up costing you more to maintain and support.
Forgot to say my deployment is as such:
4 Bridges
A to B - switch - C to D (A has no ability to get internet and there is no direct line of site from A-D but B and C can see both locations.
A-B and C-D have a distance of about 3/4 of a mile.
About 24-26 throughput. And that was with mounting them and trying to point them at each other with no technical knowledge of degrees. :-)
-
You can get some very nice and inexpensive integrated modules from Ubiquiti. Basically they (NanoStation/PowerStation) are a weatherproof box with an integrated directional antenna and wireless radio/router powered by PoE. The Bullet devices are interesting as well, basically screw them to any antenna and they provide a radio & router in a tiny weatherproof enclosure. Get yourself some outdoor-rated Cat5e, a PoE injector and find a place to mount them and use the RSSI LEDs to align them (you can fine tune in software later). These products are great, the price is low for what you get, the performance is high, and there's not really any 'integration' concerns; everything comes in the box already set up for you, just log in and configure it.
I'd suggest you stick with the 802.11a (5GHz) gear, it's a lot more reliable, and generally performs better IME. You should be able to do 1/2mi easily with a pair of NanoStation5 devices and decent LoS. You should be able to get 40mbit or so I would estimate, but that might require some tuning. 802.11g 'Turbo' performance might be slightly better, but any wireless network tech will tell you that it's a lot less reliable than 802.11a. Certainly it's possible to get a fairly high-speed, high-quality link over that distance (it's pretty short for serious wireless gear).
If you need better than that you can either look to 802.11n (which has little 'enterprise grade' equipment available, you'll be homebrewing, and it's still at 2.4GHz where reliability issues are common) or some of the telecom industry unlicensed high capacity backhaul gear from Alvarion, Proxim, Redline and the like ($$$$$$$$$). It might be possible to aggregate serveral radios into a single faster link using e.g. Linux bonding driver, but I've never really investigated this.
With that said, if you have the land right of way, it might still make sense to just run fibre. It's not that far, and it'll be quite a bit more costly, but you'll get much better performance out of it, and if you lay down several strands it gives you tons and tons of room for expansion.
-
Hmmmm..is there a possibility of bonding 2 wireless bridges? 2 on each end? this could definitely increase the bandwidth without laying fiber. Depending on where it is getting permits to do something like that could be very very expensive as I know telco companies generally do not allow 3rd party businesses to run aerial cable on their poles.
-
Hmmmm..is there a possibility of bonding 2 wireless bridges? 2 on each end? this could definitely increase the bandwidth without laying fiber. Depending on where it is getting permits to do something like that could be very very expensive as I know telco companies generally do not allow 3rd party businesses to run aerial cable on their poles.
I think it should work, but there are enough potential complications that it may not. I'd be more comfortable using a pair of wireless NICs in the same machine, but I'm not 100% sure that will work either. Worst case you could always set up the multiple wireless links in separate IP subnets and then use some type of Ethernet over IP to get 3 virtual Ethernet interfaces and then aggregate those. Rather clumsy, but it would work.
Of course any of these techniques have the same limitations as any other link aggregation (in a nutshell: single session performance won't improve, and you'll never get 100% utilization).
-
Ok…maybe wireless bridge with Cisco WASS or some other kind of data compression technology. You could possibly get 100 mbit under $10,000 using a single bridge on each end. Depending on the current infrastructure at each end the physical network might already be limited to 100 mbit on the wire.
-
Was part of a project where this was implemented and there was a very big difference in the speed on a point to point T1 link.
-
How does Cisco WASS fit in? I'm definitely not a Cisco guy, but from a Google and a browse of their site it doesn't seem related at all. I don't think any of the 'enterprise grade' solutions will come in under $10,000 for a faster-than-802.11a link. Motorola looks to be about double that for 100mbit.
Another option, if routing is acceptable, might be to set up the two bridges on separate network interfaces on a 'router' (pfSense might even be able to do this…) and then use OSPF or even static routing to load balance between them.
Nothing beats a single high-speed link for versatility though, and that stuff really costs. I think for the distance you're running anything more than 802.11-based gear is going to price itself out of contention, assuming you control the land to run fiber. 802.11n on 5Ghz is a promising solution, but if you want to go that route you can't do it as 'cookie cutter' as using the Ubiquiti gear mounted on a pole. They do have gear to do it though, the SR71 NICs are 802.11n 5Ghz compatible, and they've got some nice embedded boards to run them. Add an enclosure/couple antennas and you'd be in business. I hadn't realized this gear was released, I'll have to try it out. Should be able to do a full link at pretty good speed for under $1000.
Edit: A good way to do this seems to be to buy a dual-polarization antenna (or if you can't find one, two regular antennas would work) and just do 2x MIMO on each side. You should be able to achieve approximately 80mbps half-duplex on such a link, and I think this is fully doable for $1000. Double this setup and you can quite easily set up the routing to route traffic going one way down one link and the other way down the other link for 80mbps full duplex (approx). Doing better than this would mean link agg.
-
I have a 1km 802.11a (Turbo) link that has around 50Mbit throughput using 18dBi directional panel antennas and CM9 cards, using pfSense on both ends.
Also have a few 802.11g links with distances from 0,5 to 4km, and their throughput floats from 15 to 30Mbit using m0n0wall/pfSense boxes with atheros cards. -
hm, it would be very interesting if it is possible to make wireless trunk link aggregation or however people like to call them with PFS.
Two or more NICs on both side for link, that way it would be possible to make good 100mbps link.
-
I have a 1km 802.11a (Turbo) link that has around 50Mbit throughput using 18dBi directional panel antennas and CM9 cards, using pfSense on both ends.
Also have a few 802.11g links with distances from 0,5 to 4km, and their throughput floats from 15 to 30Mbit using m0n0wall/pfSense boxes with atheros cards.Perfect. 802.11a it is (until Ubiquiti releases some N Bullets). I've already been playing with the Bullet 2HP and the range is phenomenal, even with a small 7dBi antenna. I've not been able to do more than 2.5MB/s (20Mbit/s), but it's rock solid at that rate, and I'm guessing that 802.11a will bump me a little higher (I've never had good luck with 802.11b/g).
-
The problem with Ubiquiti (bullet/nanostation) and similar equipment is they are normally based in atheros SoC that generally clock at 180Mhz. While this is ok with 802.11a/g non-turbo 20-25Mbps, when you go with Turbo Mode the CPU simply isn't capable of achieving much more than that. The testing i've done in the last years using x86 hardware I concluded that to fully saturate a radio in 802.11a Turbo (40Mhz channels) you need more than 300Mhz CPU power, but you can get 50Mbps with a single radio :)
