Hi,

I appreciate your help. And I still have some difficulty understanding the
inage field, please see my comment in line.

On Wed, Jun 3, 2009 at 5:19 AM, Rémi Villé <[email protected]> wrote:

> I have a few questions regarding link estimation of tinyos 2.x:
>
>
> I'm working on it too since a few month, so I can help you, but don't
> hesitate to correct me.
>
>>
>> 1 Are they identical, the link estimator under $TOSDIR/lib/net/4bitle and
>> the one under $TOSDIR/lib/net/le? If yes, why bother keeping both? If no,
>> what is the difference between these two?
>
>
> net/le is the simplest estimator, 4bitle add the concept describe in this
> paper : http://sing.stanford.edu/pubs/sing-07-00.pdf, it uses 4 bits too
> have a metric which efficiently mix the qualities of the physical, link and
> network layers, it uses the CompareBit component with the event
> shouldInsert(...).
> For now I have only used and modified net/le.
>
>>
>> 2 What is the inage field in each entry of neighbor table? Why is it
>> initialized to MAX_AGE, rather than 0?
>
>
> I didn't  look closely to  this parameter but it allow to estimate the
> number of round (a round correspond to have received BLQ_PKT_WINDOW beacon
> packets) we haven't be able to update the in-bound quality link of a
> neighbor (we haven't received beacon from this neighbor).
> It's initialized to MAX_AGE and decremented to 0 every BLQ_PKT_WINDOW
> beacon packets received.
> When you look in the updateNeighborTableEst(...) function we can see that
> if ((ne->inage == 0) && (ne->outage == 0)) then ne is set to be a invalid
> entry (like if the neihbor was evicted).
>

I'm still not quite sure why we need this inage field. It is set to
MAX_AGE every
time a beacon is received (in function updateNeighborEntryIdx()) and
decreased by 1 every round.


>
>> 3 Why is linkest_footer_t defined this way? Does it imply only one
>> neighbor entry can be accommodated in the footer?
>> typedef nx_struct linkest_footer {
>>   neighbor_stat_entry_t neighborList[1];
>> } linkest_footer_t;
>>
>
> I'm not yet comfortable with nesC concepts but neighborList[1] doesn't
> meens that theres only one neighbor entry in the footer, on the contrary if
> you look in the sources or the LEEP frame describes in the Link Estimator
> Exchange Protocol TEP, there's as many entry in this list that there are
> neihgbors in the neihbor table.
> This information is broadcast and the neighbors can update their out-bound
> quality link.
>
> Note that in 4bitle the out-bound quality link is not used for a reason I
> don't know, but the LEEP frame stay identical as the net/le estimator to
> respect the protocol.
>

-Xiaohui
_______________________________________________
Tinyos-help mailing list
[email protected]
https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help

Reply via email to