Re: LZ: Teleconferece


[Prev][Next][Index][Thread]

Re: LZ: Teleconferece



(please note: I reserve the right to be wrong as hell)


> Now, we believe that if several TIs are set on the same freq that they
> should be able to comunicate together at once.  


Uhm, no.  I haven't looked at the RT link plans (I'll go do that, after I
get some sleep), but I am going to make a WILD assumption that at best it
is set up to send on one freq, and receive on another (much like amature
radio).  What you will end up with by running 4 calcs (better get them
"tuned") running on the same freq's (or freq) would be like trying to run
modems on conference phone calls.


I cannot think of any two way communication method hosting
multiple concurrent users that operates on just a couple of frequencies.
(without a server of some sort) 


(further desc of how I think RT-Link works and why your plan won't):


Some how RT link signals what gets translated to the binary for the calc.
so, imagine little 1's and 0's flying in the air... well... the 1's and
0's are just that 1's and 0', no further identification.  So ANY 1's and
0's from any RT link that frequencey will be "accepted" (I'd hope there is
some form of error correction on the software)... but if you get 3 calcs
sending 1's and 0's none of the calcs is going to know which is sending
what... (and you thought line noise was bad).


You could, possibly, manage to set up a "ring" for transfering data
between multiple calcs, but if the ring got broken, you'd make a mess. 
Data would about have to be transfered in packets (to at the very least ID
where it came from initially) (example)


(all numbers are arbitrary)
3 calcs: A, B and C   A receives on 80Mhz, transmits on 85MHz. B on 85 and
90, and C on 90 and 80.  A marks it's data packets with an A, B marks
it's packets with a B (and so on). (note those ID are only for packets
originating from that calc initially)  A would send all data pakcets it
recieves from C to B, unless it had an A mark (ID) on it.  (meaning it
came from A originally. (a calc would not replace the ID on any packets
that it was just resending (A sends evertything that doest have an A on it
that it recieves to B... and on)


> Each TI would need to assign itself some random name.  Then whenever another
> TI sends out any string, it would be affixed with the name of the TI so that
> the other TIs would understand what was going on.  This type of design might
> allow for leaving and joining in.


Names are yucky... (many bytes).  Numbers (a byte or two) would be a bit
more effiecient.  A name system could be devised, but it should just
translated to a number (byte or two)... just like som big networks.  (the
Name could be identified in a later packet with a special ID or header).


I am too tired to talk about protocols..


Jacob






<pre>
-- 
  ab976@seorf.ohiou.edu                  
  Jacob.Hopkins@family.mveca.ohio.gov      
  513 393 0582  up to 28.8Kbps  No Luck BBS  PCB 15.3beta /5 
</pre>


References: