I apologize if this post is in the incorrect forum; didn't see one that would be more germane. I'm looking for recommended, reasonably priced hardware on which I can install an open-source solution to transmit a 2.4 GHz signal from my neighbor's house to my house. I found one solution from a company but their radios use a proprietary OS. I was able to do a test with two modified Linksys routers, but they don't offer a good, all-weather solution for me as the radios would likely be on my TV mast near the antenna.

Any suggestions are appreciated.

Hi.

Ubiquiti NanoStation2
http://ubnt.com/products/ns2.php

I strip WRT54GL units from their shells and mount them in RooTenna housings all the time.  I use a couple of sticky patches and zip ties to adhere them to the backplate but not touching so no short danger.  The L-shaped bare board leaves plenty of room for the PoE adapter.  I keep the Linksys shells in a big box.

The RooTenna units now include RJ45-ECS units which I had to retrofit on all my old models.  It's a good IP68-compliant waterproof housing and I've been using them for years.

I recommend the full-depth Roo2 units if you are planning on putting in a standard 802.3af PoE adapter so you can run power & data on the same wire.  We have some older low-profile units also but those are only deep enough to accomodate our home-made 12V passive PoE adapters not enough depth for real PoE adapters.  What else, oh yeah I mount my RooTenna for horizontal polarization and usually downtilted to reduce picking up all the vertical interference in the area.

Check PacWireless.com for the parts.

(Last edited by vincentfox on 22 Aug 2008, 06:26)

48v to 12v DC/DC converters for 48v PoE are pretty small.
http://focus.ti.com/docs/prod/folders/p … 01120.html
That module is only slightly larger than one square inch and less than half an inch tall. Even smaller modules are available.

I've recently meshed an entire village using a few strategically placed WRTs in bridged-repeater mode. You could have one in each house and either attach powerful omnidirectional antennas, or a directional antenna at each end to bridge your wireless link and an omni plugged into the second port of each to provide local connectivity to the clients. You must kill diversity mode on the antennas for this to work properly, but that's just a couple of simple commands.

So you have two antennas which will be receiving very different signals. A google for antenna diversity yields

So there wouldn't be any redundant data since the signals are very different. What kind of effects do you see turning diversity mode on or off with your 2-antenna setup? Wouldn't you want antenna diversity mode ON when you have one directional and one omni directional antenna plugged in to the same radio?

No. Diversity mode is a solution to a phenomenon known as multipath distortion. Basically a single omnidirectional antenna emits RF frequencies in a way very similar to a pebble hitting a pond: waves travel out from a central point getting wider and weaker as they go. Each ring is called a "wave front" and if that front strikes a dense enough surface, it can reflect back (or more likely outward at an angle), thereby creating another wavefront travelling in the opposite direction which can interfere with the first, and any subsequently emitted wavefronts.

Now imagine the chaos that happens inside an average house - furniture (think sofa, bed, etc with metal springs, etc), walls, electrical items, RF interference, ceilings, mirrors with foil backs, etc each surface either reflecting, absorbing (some or all), distorting your signal around. It's a veritable nightmare with deadspots and, worse still, multiple wavefronts. You stream a file from one host to another across your network, host A is located downstairs in your lounge whilst host B is in the back bedroom. The first wavefront hits a mirror, then a wall, then the ceiling and 5 other flat sufaces before it gets to host B whilst the second packet gets hit by reflected wavefronts emitted by the first so this packet takes a totally different route and, due to the way it gets knocked about, arrives at host B from 3 different directions at the same time. Some of the data hits rebounding waves head on and get totally nullified, other weakened wavefronts get slammed from behind by a stronger one which strengthens it.

I'm sure you are getting the picture, basically it's messy and [i[diversity mode[/i] is an attempt to improve matters. With DM, 2 antennas spaced just inches apart both transmit and receive the same datastream, but each listens and the one with the best signal - if one is in a null area (deadspot), then the other will take over. DM improves local coverage, it does not improve distance.

Now let's say you wanted to span a large hop, a few miles, between 2 networks with a single device. The best option here would be to mount the AP equidistant between the 2 networks (assuming LoS) attach 2 strong directional antennas and point each in opposite directions. Since the datasteams will be heavier on one side than the other at any given time (a user is streaming a video or a download), DM will try to load balance between the 2 antennas which obviously cannot work as the other one is pointing away. DM can handle this in small doses, but if this setup is permanent it simply has an identity crisis - the algorithm it works on cannot deal with this behaviour.

A much better explanation is offered here:

http://www.cisco.com/en/US/tech/tk722/t … f646.shtml

Now for a local omni / omni setup, DM is fine, but for a directional / omni setup, it is useless as load balancing here is just not possible.

You can kill DM with the following command:

nvram set wl0_antdiv=[antenna]

-1=auto, 0=main (nearest the power jack), 1=aux (the other one), 3=diversity)

but check which is the main antenna for your specific version of router because somewhere around Version 2 they reversed the order.

(Last edited by bappy on 27 Aug 2008, 23:19)

great reply, thanks.

I would I have thought when diversity mode was turned off, the AP would just select one antenna and send that one signal to the radio. 

But you say the signals, say from an omni + directional, or directional + directional, are combined and the radio receives the two very different signals from both antennas?

Does an RTS/CTS mechanism have to be used to prevent the FAR away radios from transmitting at the same time, thereby creating collisions at the multi-antenna node?

Can the multi-antenna radio transmits on both antennas at the same time with diversity mode turned off?  Wouldn't that halve the Tx power?

Thanks for your help, just trying to sort this stuff out in my head
C