I thought that I should share with you the following episode. It may save 
someone else a lot of grief.

I'm developing an application based on TinyOS 1 and (moteiv) Multihop. When 
testing the network in the most convenient open space (my back garden) I had 
lots of wireless problems as shown by the 'Bad Message' errors in 
SerialForwarder.
After doing a bit of googling I came across an article on Collection protocol 
which mentioned, in passing, that MultiHop was somewhat unreliable. Because of 
that I did a test running the TinyOS 2 demo, Octopus. In 2 hours of execution 
no wireless problems were seen. My immediate conclusion was that Collection is 
more reliable then MultiHop and that I should start thinking about migrating to 
TinyOS 2.

However a colleague brought along  a spectrum analyzer and we saw that my back 
garden is a mess of electronic interference. The default TinyOS 1 channel, 11, 
being right in the worst part. The default TinyOS 2 channel, 26, was in a much 
quieter zone.
So, in fact, Collection performed better than MultiHop because of the default 
channel selection.

Spectrum analysis showed that channel 22 was the quietest channel, so I 
recompiled my MultiHop application for channel 22.
Bingo! Huge improvement, no 'Bad Message' errors after 15 hours and 150,000 
transmissions.

Conclusion? Don't jump to the first conclusion.
2nd Conclusion. Always do a spectrum analysis of your test site.
_______________________________________________
Tinyos-help mailing list
[email protected]
https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help

Reply via email to