You're right,
Paul,sim_time() returns an uint32_t measure.I see this number in
the program :
*DEBUG (0): Booted at time 1024
DEBUG (1): Booted at time 1320
DEBUG (2): Booted at time 6000
.
.
DEBUG (1): Timer fired at 1175691828
DEBUG (1): Timer fired at -1943585060
DEBUG (1): Timer fired at -767894652
DEBUG (1): Timer fired at 407795756
.*
.
.
What is this negative numbers ? Is it because of overflow ? my
boot times are :
*t.getNode(0).bootAtTime(1024);
t.getNode(1).bootAtTime(1320);
t.getNode(2).bootAtTime(6000);*
and I use this statement to print the time in the code :
*dbg("Base","Timer fired at %d\n",sim_time());*
And timer period is 1000.But when is use sim_time_string() it
returns the correct form of time as I said before.
Thanks,
On Tue, Nov 17, 2009 at 8:51 PM, Paul Johnson <[email protected]
<mailto:[email protected]>> wrote:
Yes, you can do that, but it won't accurately model a real
clock because of clock skew. In general, clocks do not run
at the exact same rate, their frequency tends to drift. This
drift depends on a lot of factors such as temperature,
humidity, etc. In TOSSIM there is no support for clock
skew. Another drawback of simulation is that it doesn't
simulate processing time. This means that you will never
have any jitter (the non-deterministic timing differences
between receiving a packet and processing it, or preparing a
packet, and sending it).
These two factors are what makes time synchronization a
non-trivial problem. Using TOSSIM you could check your
algorithm to see if it sends messages correctly, parses
received messages correctly, corrects the random offset
between nodes, etc, but you will not be able to see how
accurate the synchronization is using pure simulation.
-Paul
mojtaba raznahan wrote:
I think i should solve this problem by specifying random
numbers as clock offset between motes,and sim_time() as
global time.
On Mon, Nov 16, 2009 at 10:31 PM, mojtaba raznahan
<[email protected]
<mailto:[email protected]>> wrote:
Paul,
Thanks again for your clear answers.
Yes, I think like you.So due to this issue I can't test
time synchronization algorithms in TOSSIM,am I right ?Or
maybe there is some interface for doing this job .. ?
sincerley,
Mojtaba
On Mon, Nov 16, 2009 at 9:58 PM, Paul Johnson
<[email protected] <mailto:[email protected]>> wrote:
Mojtaba,
Actually, i believe the bootAtTime is in 100's of
pico seconds (10^-10) seconds. So 1024 * 10^-10 =
0.0000001024, and apparently sim_time_string()
returns time in seconds (up to nano seconds
(10^-9)). There are actually 1024 milli "ticks" per
second, so this is why each timer fired is adding
not quite 1.0 to the time.
As for your other issue, this is actually by
design. If a node transmits a broadcast packet, all
other nodes in the transmission range will receive
it at the same time. The boot times only change
when the node powers up, but doesn't skew it's
clock. In fact, i can't say for 100%, but I think
that sim_time_string() grabs the global simulation
time, and there is no local clock simulation is TOSSIM.
-Paul
mojtaba raznahan wrote:
Hi Janos,
thanks for reply.
What's the meaning of this numbers ? I set the
bootAtTime event as these :
*t.getNode(0).bootAtTime(1024);
t.getNode(1).bootAtTime(1320);
t.getNode(2).bootAtTime(6000)*;
And i get the time by *"sim_time_string()*" method
and it print the boot times as this :
0:0:0.000000102 and 0:0:0.000000132 and
0:0:0.000000600 . As i can guess 102 means "1024"
MilliSec and 132 means "1320" Milil.
And I set the startPeriodic at 1000 ,some thing is
ambiguous.The last statement prints the time equal
to 0:0:0.000000142 the next statment is
*Timer.startPeriodic(1000)*,but in the output when
i print the time in *Timer.fired()* event, it
prints the time equal to *0:0:0.976562642* and the
later *0:0:1.953125142* ,* 0:0:2.929687642*,
*0:0:3.906250142* what does it mean ?
And also when i send a packet, the 2 other
recievers get the packet at the same time!!(I've
set the bootTimes differently!).
Thanks,
Mojtaba
On Mon, Nov 16, 2009 at 7:34 PM, Janos Sallai
<[email protected]
<mailto:[email protected]>> wrote:
In tossim, the node object in python has a
bootAtTime method which
lets you set when the mote boots. You can use
this to make the local
clock's offsets different from each other. By
the way, TOSSIM does not
simulate clock skew. This might limit the
validity of the simulation
results.
Janos
On Mon, Nov 16, 2009 at 7:55 AM, mojtaba raznahan
<[email protected]
<mailto:[email protected]>> wrote:
> Hi all,
>
> I want to know how can I set the local time
of each mote programmatically ?
> I want to set different local time for each
node then test my
> synchronization algorithm on them.
> I used the LocalTime interface to get the
local time but i can't set
> different initial local time for each
node.And LocalTime interface has 2
> type "TMilli" and "TMicro" but when i use
TMicro then in configuration
> section there isn't any component provider
for it! .I'm using Tinyos 2.1.0
> and TOSSIM.
>
> Regards,
> --
> Mojtaba Raznahan
> BS of Computer engineering
> TMU university
> www.raznahan.com <http://www.raznahan.com>
>
> _______________________________________________
> Tinyos-help mailing list
> [email protected]
<mailto:[email protected]>
>
https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
>
--
Mojtaba Raznahan
BS of Computer engineering
TMU university
www.raznahan.com <http://www.raznahan.com>
------------------------------------------------------------------------
_______________________________________________
Tinyos-help mailing list
[email protected]
<mailto:[email protected]>
https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
--
Mojtaba Raznahan
BS of Computer engineering
TMU university
www.raznahan.com <http://www.raznahan.com>
--
Mojtaba Raznahan
BS of Computer engineering
TMU university
www.raznahan.com <http://www.raznahan.com>
--
Mojtaba Raznahan
BS of Computer engineering
TMU university
www.raznahan.com <http://www.raznahan.com>
