[Tinyos-devel] packet timestamping and packet-level timesynchronization

[Adler, Robert P]

We previously used 64-bit time values for both the timestamps and the
GlobalTime interface because our timebase is a 13MHz clock.  At 13MHz, a
32-bit value overflows every 5.5 minutes.  Since our desire was to be
able to take advantage of the extra processing capabilities of the
imote2, it made a lot of sense for us to keep the large values and to
use such large datatypes.  It's also worth noting that the particular
application that made use of timesync required microsecond timesync, so
using a slower clock wasn't an option.

-Robbie

-----Original Message-----
From: Janos Sallai [mailto:sallai at isis.vanderbilt.edu] 
Sent: Wednesday, May 14, 2008 1:17 PM
To: Janos Sallai; Adler, Robert P
Cc: tinyos-devel at millennium.berkeley.edu; TinyOS Core WG
Subject: RE: [Tinyos-devel] packet timestamping and packet-level
timesynchronization

>> it would be important for the timestamps and
>> for the timesync protocol to be able to support parameterizable width
>> and precision. (PacketTimeStamp interface)

It appears that we do not have a consensus on this.

Reasons for parameterizing with precision _and width_:
- It's more flexible and probably future proof.
- Would allow for 64-bit time values. Is this _really_ required for the
imote2?
- Would allow for 16-bit time values, to save some space in the message
payload.

Reasons against it:
- Sources would get more verbose.
- LocalTime is not parameterized by width, either. To access non-32-bit
local time, a custom interface would be required anyway.

Janos