On Monday 18 September 2006 01:11 pm, Gurkan Cepni wrote:
> Actually, I am not planning to sample the signals continously.  That's ok
> for me to sample for just 2 secs. (using 30ksps sampling rate: I want to
> sample it fast so that the environment will not change much)

I wrote a test program on a Telos-like platform (MSP430, 4MHz) that sampled 3 
channels at 26kSa/sec (aggregate rate of 78kSa/sec), performing a simple 
sum-of-squares calculation on each group of three samples and saving the 
largest vector.  The program did nothing else and was definitely maxing out 
the processing power of the MSP430 (it was using the coprocessor to calculate 
the squares).  This code later went into a sampling application where it was 
throttled back to 10kSa/sec (30kSa/sec aggregate), sampling very infrequently 
and only for 500ms at a time.

IOW, you can probably hit 30kSa/sec for 2 channels if you can figure out how 
to handle the amount of data being generated.

> Based on your analysis, 2 secs. of data is 960kbps*2sec=1920 kb. This can
> be easily stored on the memory and be transferred to base station (it is
> not critical to how fast the mote transfers data to base station.).

Actually, 1920kb is nearly 2 megabits.  It would fit in flash, but I'm not 
sure if the TelosB is fast enough as a whole to drive the flash fast enough.  
If storing all the raw data, 2 second sample may still be too big to reliably 
handle.

>
> In short, I will push a button on the remote mote, the mote will sample 2
> external signals (with 30ksps) for 2 secs, store data on the memory, send
> it to base-station, then laptop, then save it on the computer. The next
> time I push the button for the second recording can be 20 sec.s or  more
> later, no problem.

Can the mote do some processing of the sample data before sending it to the 
computer?  If you're looking for a certain kind of event, perhaps all or part 
of the processing required to characterize that event could run on the mote 
and reduce the size of the data set considerably.
_______________________________________________
Tinyos-help mailing list
Tinyos-help@Millennium.Berkeley.EDU
https://mail.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help

Reply via email to