On Feb 20, 2010, at 11:04 PM, Xiaohui Liu wrote:
> Hi,
>
> Thanks. Still one thing not sure. CIL.
>
> On Sat, Feb 20, 2010 at 6:28 PM, Omprakash Gnawali <[email protected]>
> wrote:
> On Fri, Feb 19, 2010 at 10:15 PM, <[email protected]> wrote:
> > Hi,
> >
> > Please help me understand the following uncertainties regarding 4 bit link
> > estimation (4bitle):
> > 1) 4bitle is using both data-driven and beacon approach to estimate link
> > qualities in terms of ETX. Data-driven is estimating bidirectional link
> > quality since it leverages ACK, whilst beacon is only measuring inbound link
> > quality. Is it problematic to combine these two estimated ETX into a single
> > ETX? What does this single ETX mean? Inbound link quality, or bidirectional?
>
> Bi-directional. Beacons are only used for bootstrapping.
> But beacons can be sent throughout the running lifetime of the protocol, such
> as in CTP. In the long run, isn't this going to interfere w/ the accuracy of
> the bi-directional ETX measured via data-driven approach by mingling it w/
> inbound-only ETX?
They are just estimates: there is no "true" ETX. Generally, if a link is being
used, there are so few beacon estimates in comparison to datapath estimates
that they don't really matter. Don't forget that the retransmission policy is
such that if a node thinks a link is good based on inbound ETX yet the link
isn't, there will be a lot of retransmissions and therefore datapath estimates.
This will cause the link estimator to quickly realize the link is bad and stop
using it.
Phil
_______________________________________________
Tinyos-help mailing list
[email protected]
https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
