[Tinyos-devel] Re: [Tinyos Core WG] time sync interfaces

[Janos Sallai]

>> What time accuracy do you call a useful timesync on our current
>> platforms?
>
> Well I was asking that question too, actually ;-) I'm not an
> application developper/expert, so I can't really answer.
>
> David
> PS: I suspect with reasonable evidence that millisecond level timesync
> is practical.

We have a couple of projects where we had to timestamp acoustic events.
Sound travels roughly a foot in a millisecond, so a millisecond resolution
might be too low. For us, a 32768Hz clock was sufficient; this way sound
travels about 1cm in a clock tick. I don't see any reason why someone
would want higher precision than that with acoustics. (Maybe for
ultrasonic ranging with subcentimeter accuracy?)

Another possible application of timesync is coordinated action, e.g., when
we want to make the motes do something at the same time. In our
radio-interferometric ranging/tracking application, where the motes need
to start sampling the ADC at the same time (with 17kHz sampling rate), the
32768Hz clock is sufficient.

However, a TDMA-like layer in the radio stack (that relies on
timestamping) would need higher precision than that. We have an ongoing
project where we measure sound propagation in steel, which will probably
require higher precision, as well.

So the question is: Do we really want to specify a "network time
precision", or let it be hardware/implementation specific (where the
frequency of the clock used for timestamping can be queried through an
interface).

Janos
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3096 bytes
Desc: not available
Url : https://www.millennium.berkeley.edu/pipermail/tinyos-devel/attachments/20071218/c48962ac/smime.bin