> ISTR The way the 1-wire things do it is by doing a progressive
> bit-by-bit polling scheme on th e
> address,  i.e. : "are there any devices with serial number bit 0
> set ?" (none, one, or many devices
> may respond), and then narrowing down the search until each
> device is found - the only requirement
> is that it can reliably distinguish between no devices and (1 or
> more) devices - with the 1-wire
> open drain system this is easy, and I suspect something similar
> is possible with RFID.
>
> Another way to look at it is like a binary search : imagine the
> 'space' of all possible numbers.
> You initially look for anything in the first half of that space -
> if there are no devices, discard
> it, if there are devices, split it in two and check each half
> again, and so on, discarding all the
> spaces that contain no devices.
> As the vast majority of spaces will contain no devices once you
> get a few steps down the "tree", the
> search will discard many spaces early on, reducing the number to
> be searched for further devices to
> a manageable number.

        Interesting, however, considering the low bandwidth involved with RFID I
don't see this technique being feasible. There will be possibly trillions of
"serial numbers" (maybe even more), doing this type of search will probably
still take to long, especially when you have to add error detection,
collision resolution and all the other protocol "features" that are needed.
TTYL

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.