Ask Slashdot: Updating a Difficult Campground Wi-Fi Design?

MahlonS writes "I am a retired network hack wintering in my RV in a campground in southern GA. 3 years ago I reconfigured the Wi-Fi system to a marginal working ability; It's now ready for a serious upgrade, prompted by a new cable net connection replacing a weak DSL. 5 dual-radio HP Curve access points connect to a 6th via single or double radio hops (effectively a Wireless Distribution System) in heavily wooded space. Unidirectional antennas at the APs (the APs are in water resistant enclosures) are placed on poles above the RVs, about 15 feet above ground. Primary hops are about 300 feet to 3 of the APs, secondary hops about the same. Signal measurements indicate that there is adequate RF between the access points. In 2008, average user count averaged about 30 users; newer devices (smart phones, etc) will likely increase that number (winter population total is about 80 RVs). While the old design worked OK when lightly loaded, I suspect that the single DSL line generated so many packet resends that the APs were flooded. This is a quasi-State Park, so money is always an issue, but there is enough squawk from the user community that a modest budget might be approved. The main AP connects to an old Cisco router. Burying wire is frowned upon, due to shallow utilities, and campfire rings that float around the campsites — sometimes melting TV cables. Since I'm not up on current Wi-Fi tech, are there solutions out there that would make this system work much better?"

Openmesh

[Anonymous Coward]

I'd look into some of the fairly inexpensive openmesh routers...they're great for extending networks (or running jasager).

http://www.open-mesh.com/ [open-mesh.com]

Too high

[BitZtream]

Lower your transmitters a little. Signals propagate horizontally (perpendicular from the antenna), this is why you need to have an AP on each floor in a house to get good signal. Not because you're on different floors so much as the signals just aren't going in the right directions.

I know you're trying to broadcast over the RVs, but going over them also means no signal is getting to them in this case.

our setup

[Anonymous Coward]

we've got 5 outdoor ruckus ap's spread across our park. (fairly cheap too)
http://www.ruckuswireless.com

they'll mesh with indoor wifi ap's if you don't want to run ethernet to each one individually.

the "smart antenna" design is actually pretty good. it supports dynamic beamforming, multiple signal paths etc. basically it just takes the path of least resistance, which helps a lot when dealing with a lot of walls/trees etc.

you can give them a call w/ any questions you might have.

Layers

[swalve]

I would run two networks, a backbone at say channel 6, and alternate APs at 1 and 11. Get highly directional antennas for the backbone, and either corner-directional antennas or omnidirectional antennas for the access points. Run the backbones up high, and the APs 12 feet or so.

Try to eliminate any double hops via short cable runs and/or smarter backbone placement.

Re:Fiber

[vlm]

Why fiber and not cooper?

Lightning. Za Pow!

Never run a piece of copper from one building to another if you can at all avoid it.

Engenius APs

[Mr. Lwanga]

Take a look at Engenius' access points, I think their multihop repeater solution might fit your needs http://www.engeniustech.com/index.php/networking/solutionsdatacom/ [engeniustech.com]

uBiquity

[smpoole7]

I don't think one vendor will supply everything that you need, but you definitely need to take a look at uBiquity. We've used their NanoBridges in studio-to-transmitter links several times and have been pleasantly surprised. The stuff is ridiculously cheap -- so cheap that we honestly wondered what could be wrong with it until we tried it. (Less than $160 for a pair of NanoBridges!)

Ubiquity's Website [ubnt.com]

Re:Wireless N would help

[Adriax]

http://www.open-mesh.com/ [open-mesh.com]
The single band series is .11G mesh, $60 for a router and another $20 for the outdoor enclosure.
The dual band does N, $100 for a router and $40 for the enclosure.

Either way you get mesh networking that's really damn simple to configure and has a public and a private network. Public can be open or encrypted, supports individual bandwidth limits, and has a splash page feature for logins or selling airtime. Private network is encrypted and unrestricted.

Love mesh networking. No cables, network topography isn't set in stone, you just toss another router into the mix wherever needed and you can cover wherever you want.

Check out Ubiquiti...

[Above]

Ubiquiti [ubnt.com] has some very cool products and customer support, you might want to look into their gear.

If you can get line of site from the remote sites back to the central site you should use 5Ghz for the backhaul, and 2.4Ghz for the client side radio. This will reduce your interference. Also, the backhaul should use _very_ directional antennas since the two endpoints are known. This will also prevent interference. It doesn't sound like any of your distances are enough to require a multi-wireless hop, although your sight lines may require it. Avoiding a double hop will increase performance.

You'll also want some intelligent QoS on both the WiFi and cable modem side. You don't want one user to be able to make the experience really bad for all the other users. For instance, if you had a 20Mbps cable modem you might want to limit any one IP/MAC to 5Mbps, or so. WRED or similar can also be your friend. Make sure there is a good local DNS server, as well

Re:Fiber

[JWSmythe]

    Fiber can be purchased rather cheaply. It's really worth it for outside runs. As someone else said, lightning strikes.. Even the extra equipment required (transceivers, fiber ready switches, etc) can be purchased fairly cheaply on eBay.

    I did it to replace a mess of copper and wireless between offices in a complex once. If I remember right, it was something like 600' of fiber for about $200. I did it in segments, so if someone were to damage one segment, it could be easily replaced. For their end points, I picked up a lot of 6 Cisco Catalyst 2924's with 4-port 100baseFX cards. I think the total price on switches was $300, and that let me replace all kinds of consumer-grade crap switches.

    His problem with fire pits and the like can be reduced by laying the fiber along the edge of the roads, and burying at a sufficient depth. Hell, they run power and water to each campsite already. Parallel runs to existing infrastructure would be fine. Fiber doesn't have that nasty tendency to pick up inductive signals.

Re:Layers

[Anonymous Coward]

I agree with the separate networks, i would go further and use different freq band to go to the netework.

If you can get get line of sight to the aps you could use a 5.8 signal to go from point a to b. ubiquity seems to have some cheap direction radio+antenna combinations.

Ubiquiti Networks - hands-down, IMO

[GSloop]

Look at ubiquiti's stuff. M5 Wireless bridges out to to the AP's and UniFi [normal or long-range] for the clients.
www.ubnt.com
Nanostation M5 [5Ghz]: http://ubnt.com/nanostationm [ubnt.com]
UniFi: http://ubnt.com/unifi [ubnt.com]

Not as slick as Ruckus or some other stuff, but incredibly cheap. [Bridges are about $200 for a pair - and super solid, massive through-put. UniFi is about $70 per AP.]

You also get the ability to help pay for the system via UniFi. [Paypal subs, no admin reqd. Vouchers for "free" use etc.] That's all included for "free" in their system.
Plus you can use Pico's for outdoor use. Already weather-proof.

[I've not run the Pico's - so check it out in the forum: http://www.ubnt.com/forum/forumdisplay.php?f=48 [ubnt.com] - you should be able to get your answers there.]

It's really some of the best bang-for-the-buck for non super-high-density WiFi use around, IMO>

-Greg

Re:Wireless N would help

[icebike]

http://www.open-mesh.com/ [open-mesh.com]
The single band series is .11G mesh, $60 for a router and another $20 for the outdoor enclosure.
 

Mesh is great in urban environments where property rights restrict you from crossing roads and other people's land.
But mesh really makes very little sense in this environment. Its a fairly obscure technology, and getting it fixed and keeping it running may be problematic when the campsite geek's RV pulls out for the season.

Look, they have power to all of these router anyway. Why not STRING the cable or use WIFI over Powerline to feed the routers?

Putting in ground-burial cat5e [cat-5-cable-company.com] is not that hard and poses no risk to the existing utilities. Its about $300 for a thousand foot roll [amazon.com]. Two guys and a rented Mini Trencher [eztrench.com] can probably install all the cable runs in one day.

You don' have to trench it in more than 4 inches deep, and you only need to go that deep to keep people from tripping on it. You can hand trench when you get near your pipes and power runs.

Re:Too high

[Lyttek]

I'll have to disagree with you on this, having actually setup a wifi system at an RV park. It is different than setting up a wifi hotspot in a building. The park owner asked for my recommendations about installing wifi to cover the park, and I gave them. The cost figures that I presented were higher than what he wanted and asked if we could do it another, less expensive way. What he wanted to try was a couple of telephone poles with an omnidirectional antenna for the access point, directional antenna for the backhaul link. If you were within a certain range (not far) it worked ok, but more than one or two rows of RV units away, it was no good. Keep in mind that RV units are essentially big metal cans... not the most conducive to getting a wifi signal into from the outside. What I wanted to do, and we eventually ended up doing, was installing a 70' tower with directional antennas pointed at an angle down. By using the technical specs of the antenna, we could figure the angles to get a pretty fair amount of coverage over the park, with almost line-of-sight from the antenna to each RV. This last bit was the key. By having the antennas too low, they would HAVE to penetrate multiple tin cans to get to the farthest units... and that just doesn't work, even with a 1-watt transmitter. A second park pulled me in for some consulting on the same type of thing, and they had antennas located about 10-12 feet in the air... I can guarantee you that unless you want to install an access point at each campsite, go higher. Lower does NOT work in this type of situation. We did keep one of the omni-directional antennas, because it worked so well. While most antennas are either horizontally or vertically polarized, this one was constructed to basically take the signal in any polarization that reached it. This park has trees as well, but how it would compare to yours is hard to say. Trees will absorb the signal quite a bit. One of the things I used to get the owner to put up the cash for the tower and etc was a bunch of signal-strength charts generated by netstumbler.

Re:Fiber

[Technician]

All wire has resistance and inductance. A high current nearby lightning strike will induce voltage and current in nearby conductors. This is why you never stand near a metal fence in an electrical storm. The fence may be grounded at both ends, but in relation to the nearby ground the fence can be lethal. One of the biggest strikes I had to clean up was a radio station transmitter. The antenna was properly grounded. The local utilities were properly grounded. A nearby lightning strike blew out diodes in the power suppy and there was obvious arc marks between the utility ground and the utility neutral. On the other end of the wire at the AC panel, the neutral is bonded to ground and connected to the building ground. The final in the transmitter was fine. The power suppy took the hit with the high voltage differential between ground and ground due to the high current. On the wall, there was arc marks between the coax to the antenna and the upper ground ring in the room. There were several points of arcing between ground and ground. Two panels on the wall showed explosive discharge between the frame of the panel and the conduit between them, even with the ground wire in the conduit in the panels tying them together. Transformer action into the conduit created high current in the conduit. Conduit joints and box to conduit joints showed arc marks. A semiconductor anything in that area would have taken the hit. Just tying it to ground doesn't work for high energy pulse discharges.

Re:Fiber

[vlm]

Having pulled cable for a number of years I can tell you that you can pull on both fiber and copper until your eyes bulge from your skull, you aren't going to hurt it. I've seen 24-strand SMFO pulled taught by a truck driving at 10MPH with no damage to the cable (OTDR verified).

I know outside plant is infinitely tougher than inside. 20 years ago I shattered some escon by exceeding the min bend radius under a raised floor, got a talking to from the bosses boss, and this was literally dropping it in place and lightly one hand tugging for slack control. The OTDR showed the shatter right at the under floor turn so it was all me... Supposedly even just whipping escon could shatter it, donno about that.

Pull straight on single mode and you are correct you could probably hang from it quite easily. At least in the olden days, min bend radius was like 2 feet so take the same fiber and wrap it around your hand and try to hang from it and it'll shatter instantly.