2.4GHz Bi-directional Booster

I was just looking around DSE web site and look what i fund not bad for the price :P
Cat No. XH6825

500mW, now we're cookin' 8)
Could be a tad slow due to being half duplex.
Dosen't specify what standards it supports.
Yea I know "it just repeating the signal" but you never know.

Or if you are good with pcb construction
http://www.qsl.net/n9zia/wireless/appendixD.html

http://www.rflinx.com/2400_G_Series.htm
200mW 802.11g
us$250

Thanks for the heads up.
Didn't consider going down this path until I saw the price.

Then there the fact clients can hear the AP but can the AP hear the clients?

the radios in the APs are only half duplex, I believe the switching time on these nessesitates the use of a long preamble on the radio which may reduce thruput somewhat.

Depends what you use it for really tho. Im getting one to go on my omni when they come in. Worth it for a play.

500mW into an antenna gives me 9dB of gain before im past the legal limit. I have an 8dB here. Might not be worth it because the line of site to my antenna would most likly be about the same either way, but it may help get the signal to some places closer with trees in the way since these dlink accesspoints are so under powered..

I've been consdering designing a full-duplex radio link myself, but would mean I would use two wi-fi channels (AP's at each end) and a router at each end. Understandably as Wi-Fi equipment uses the same carier (channel frequency) for receive and transmit, so is only half-duplex as mentioned. I thought of routing traffic in this sense between both segements.


router 1| 10.0.0.1 ---->--ch11---->----- 10.0.0.2 |router 2
router 1| 10.0.0.4 ----<--ch10----<----- 10.0.0.3 |router 2

The router would be effectively a cheap PC (or hardware router), which has a two network cards, and windows setup to forward IP packets (something that can be done via the windows register to make it a router). Traffic on router 1 would send (metric priorty 1) non-local traffic outbound via 10.0.0.1 to router2 awaiting at 10.0.0.2 which it would forward over its network. Then router2's table would send non-local traffic back via 10.0.0.3 to router1 destined for its network etc... Router1 could have a route table entry (metric priority 2) should ch11 die, to try to send data next in reverse flow, then the same at the other end so it could kill one link leaving the other used in half-duplex. I'm not sure how to setup a dynamic table in this sense that can open and close each line but I think I would need real stand-alone routers with RIP etc... so as not to always have a delay in sensing if the main flow line is working before trying the second.


Cheers,
Gavin.
Radio Engineer (getting in to pc networking, sick of mobile internet costs).

You'd be far better off using some sort of propriatary ethernet bonding protocol. I think Linux will do this for you. This could potentially double the bandwidth because it can use both links to send not just one.

The real problem that half-duplex has is the latency because the airtime has to be chopped up. You could to completely redesign the protocols to fix that.

Good point, never thought about the latency being far less aswell. That's a good bonus.

I'm currently doing some tests at the moment, I managed to figure out the registry tweak in windows to allow it to act as a router between two network adaptors (not bridge mode that the user interface allows for selection). Works a treat between two lan's leaving the router with its own IP address on each network for other tasks.. such as a server for clients on each network, not just a router.

Next thing (which I don't have enough of yet) is to plonk two wi-fi cards in to two seperate pc's, and simply setup the route table via a command line to point to a default gateway of one device for outbound traffic say over channel 11 wi-fi adaptor, then on the other pc make a default route for comign back via its channel 10 device. The only problem I think I might run in to, is that all traffic will attempt to go back over the pipe in once it which came, rather than zap back over the next one.

I think the only way to solve this is to have each pc with 2x wi-fi cards in it, to have it's own 100Mbps LAN with a subnet all of its own, that way traffic will eventually have a destination of a non-wi/fi adaptor so as when traffic comes back it can be routed back via an alternative pipe, than going back over the one it came from.

IE: Subnet 192.168.1.X has a default route of 192.168.1.254 (LAN port on its routing PC). Then the PC looks up it's routing table and finds that out of the 2 wi-fi ports it has, it will pipe traffic to wi-fi1 10.0.0.1, over channel 11, then the receiving router sends traffic to its LAN on 192.168.2.X. Traffic coming back will then have a defaut route on the router for its wi-fi1 being 10.0.0.3, which lands back at wi-fi2 10.0.0.2 on the router of the orginating subnet of 192.168.1.X. I think this will work. Just means Two Wi-Fi cards in each PC with a 100Base-T port for its local traffic.

This would be a great way to bring lots of subnets together over a wi-fi link without having to assign an individual 10.0.0.X IP to every single client wishing to be part of the network. So instead of 128 concurrent users max on an access point, an entire subnet of other computers could still link through the wi-fi system with a total of 64 networks linked, hundreds+ LAN nodes.

Hmmm, this is just me thinking out loud.

Gavin.