Hi!
On Wed, 9 Sep 2009, Alban Hessler wrote:
> Hi,
>
> In the LPL stack, in the PowerCycleP file:
>
> I don't understand why it is needed to probe 400 times
> (MAX_LPL_CCA_CHECKS) the signal strength to detect if there is
> currently a transmission. The sender on the other side sends
> continuously, there is just a bit of radio silence between two
> duplicates.
>
> 400 times is like 50ms (according to another post), which is a lot.
The check is actually much smaller, around 16.5ms on TelosB. Figure 7 and
Table 1 from here shows this:
http://cs.jhu.edu/~razvanm/ipsn2008koala.pdf
> Finally, if you put a node to sleep 100 ms, then it's actually with
> the radio on for half of the time (!)
>
> Then, if I decrease this MAX_LPL_CCA_CHECKS to 80, then I get about
> 50-60 percent packet losses. (LPL with receive sleep between 100ms and
> 300ms). Whereas with the value set to 400, I have a residual 1 %
> packet loss.
>
> Can somebody explain me the reason behind this 400 value? Why is it so
> large? Why is for example 80 not enough?
As far as I know the value of 400 for the number o checks was picked
because, as you also noticed, it is giving better performance than a small
one. :-) It would be nice to investigate more the behavior of the CCA pin
when a stream of packets are received.
A good paper that talks about the LPL from TinyOS is this one:
http://sing.stanford.edu/pubs/sing-08-00.pdf
--
Razvan ME
> The nodes are close to each other and send at 0 dbm, so I don't expect
> the CCA threshold to be an issue. The CC2420 spec says that the RSSI
> prob is +-6db, but I expect the RSSI to be anyway 12db more than the CCA
> threshold.
>
> That question was already asked in the following post, but remained
> unanswered:
>
> http://mail.millennium.berkeley.edu/pipermail/tinyos-help/2009-February/038272.html
>
> Thanks for any help,
>
> Best regards,
>
> Alban
> _______________________________________________
> Tinyos-help mailing list
> [email protected]
> https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
>
_______________________________________________
Tinyos-help mailing list
[email protected]
https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
