[Tinyos-help] how to improve reliability of the multihop sample?

Ákos Maróy darkeye at tyrell.hu
Mon Oct 8 06:15:34 PDT 2007


I've just deployed 40 T-Mote Sky motes, with a slightly modified version
of the multihope oscilloscope sample. basically the modification is that
there are 4 sensor inputs sent instead of one (internal voltage and 3
ADC inputs)

as the motes were deployed unto a park and the rooftop of a neighouring
building, well, the results are not too good. it seems that the motes
which are close to the root mote (and thus a lot of traffic would seem
to go through them) hang on a regular basis. this basically cuts of a
lot of motes at once. (I have two such crucial motes)

an alternative topology could form, bbut somehow it doesn't.

on the motes that hang, some of them have the red light (led0) lit,
which signals that the report_problem() call was called from the mote.
some simply hang (no leds alight, do nothing). some have the blue led
lit (led2), but this is seldom.

sampling time is still per second. I figures this shouldn't be an issue
even if all 40 motes have to send their data through 1, that would mean
40 messages x 28 bytes / sec, that sounds like 1120 bytes / sec - I
guess the infrastructure should be able to handle that...

are there generic ways to improve the reliablity of the motes?

I have an exhibition starting on Friday, and we're counting on these
babies :)


Akos

PS: I can provide the source code to our motes, naturally, if needed for
analysis


More information about the Tinyos-help mailing list