 a horrible drawing I made |
I have further refined my idea for an implementation of a electronically steerable directional switched antenna array.
My old idea to switch it during the preamble decoding was fundamentally flawed and would require a re-write of the Atheros driver and hardware that I don't think exists yet.
To do this fast enough such a task would require a CPU that is at least 3ghz running a real time operating system and in the real world when you get it all loaded up and running a linux or bsd it would probably require closer to a 5-6GHZ cpu. Relays can also not switch that fast and I can't find something else that is appropriate for this function. Even in long mode the preamble just happens way too fast.
Speed is far less critical for the switching with my new idea of how to do it.
Have each antenna go to an 802.11 radio that is receive only just looking for station mac address's and recording the RSSI.
It will pass the RSSI info to the site transaction co-ordinator to determine the the direction of arrival. the site transaction co-coordinator controls the relays via GPIO. Information from receive radios would be sent to the site transaction co-coordinator over TCP.
The site co-coordinator would need its own ATH9K radio and a working TDMA implementation. It would use the right antenna for the right client's time slot.
When the center conductor gets high and the radio at the base is transmitting have relays that disconnect the receive only radio from the transmitting antenna in the array to not fry the receive only radios.
The things that look kinda like bullets are the receive only radio's with relays on the end connected to the center conductor. They would need power for switching like the relay boxes for switching the antennas from the AP. The power connections were not included in the diagram.
Someone please make a 2.4 GHz product that can do this, I don't know if I can get a working TDMA system developed and added to ATH9k on my own but im going to try.
If any other wisps here have any good programmers we should all work together to develop this system and open source it.
If a finished open source TDMA system was developed for us wisp's the vendor lockin ubiquiti has over us would be gone. We would be free to use anyone's devices that we choose and a new golden age of the wisp could be ushered in. Where all hardware vendors could be used and multiple vendors could be free to implement a free TDMA solution.
This would be as revolutionary as to the WISP industry as the introduction of the MS-DOS and the Compaq PC was to the computer industry.
This system would allow for a previously unprecedented amount of spectrum re-use since your TX duty cycle is switched among 6 different directions.
Since they terminated the SDK, sold me miles of bad wire and discontinued the nanobridge M5 to replace it with the nanobeam M5. I have really lost much of my faith in Ubiquiti as a company and I really don't trust them to get a fully competent product to market in a timely manor.
BelAir Networks and Nortel have already developed a similar system in the past but for mesh networks. As far as I know nobody has really done a switched antenna array before with a TDMA fixed wireless system.