I'm sorry but you've lost me there, Captain. So what you're saying is that the signal to noise ratio on a 5.8GHz WUG will decrease as its coverage increases? And then you talk about fresnel lenses and LEDs. That sounds like some kind of OPTICAL technology to me?
Please would you repeat what you said in layman's terms?
Lets presume 30 people connect to 10 houses via Ronja. These 10 houses in turn aim ONE 5.8ghz parabolic dish to the JAWUG highsite. Thus 30 people share one Wi-Fi transmitter, instead of each of these 30 people aiming 30 individual 5.8ghz dishes to a JAWUG highsite sector antenna using up limited bandwidth.
The houses adjacent share between them a DSLAM network. Each house has 3 RONJA
http://scratchpad.wikia.com/wiki/FreeSpaceOptics to connect distant 1.4km users. The DSLAM network in turn parcels out the bandwidth to each of the
40 users from the single 5.8ghz antenna. Each RONJA provides 10meg, thus the local LAN created has more than enough bandwidth, since each RONJA would first go through the local DSLAM network before connecting to the other RONJA. The bottleneck is with the limited RF bandwidth on 5.8ghz. Wi-Fi signal to noise ratio S/N is not that of Wimax , thus it isn't optimizing the spectrum as much as wimax does. Wimax and LTE has major multi-path dispersion issues because the signals bounce of the buildings. Establish RONJA hotspots which in turn link up with a parabolic to a Wimax or LTE tower.
http://scratchpad.wikia.com/wiki/TransimpedenceAmp describes how to build an FSO receiver. The code and schematics are from Chicago University in an optic application they built for the CERN collider. The LED transmission side RONJA handles very well , it is the receiver side that I can't fathom what they are trying to achieve since of the shelf ICs are available(but expensive), you don't need to hack a design with BJTs. Building a pure analogue system as RONJA means that you won't be able to go above 10Meg , only ICs will allow 100Meg the switching limit of a LED.
The input stages are Femionics PIN input diode coupled to a Transimpedance amplifier and then Post amplifier.
ST-M13A306 optical input -> noise filter -> Phillips TZA3033 Transimpedance -> noise filter -> TZA3034 Post Amplifier ->LV-PECL to LVDS
Study the datasheets on the PHillips Pre and post amplifiers of optic signals and it is clear that building a 155Mbit receiver system isn't complicated. The Chicago design uses the same ICs as in a fiber optic SONET network.
== Switching led now at 400Meg ==
http://www.firecomms.com/tech-RCLED.html seems to have made a breakthrough, claiming 400Meg. Flash the led through a Fresnel lens to a Transimpedence amp on the other side and you should get a 155Meg link. Catch is it seems they won't sell individual RCLEDs. Only solution is then to buy
http://www.molex.com/cmc_upload/0/000/-14/401/smipof_ds.pdf Transceiver units and remove the RCLED from the package. Available in SA from
http://dataweek.co.za/news.aspx?pklNewsId=26040&pklCategoryID=48