Thank for the answer, now it's clear.
One last thing:
"The cell managed by MSF are all TX=1 RX=1 and SHARED=1 type of cell
but associated to a dedicated neighbor"
This means then nodes run backoffs in all dedicated cells? (i.e. to
avoid collisions between parent and child)
Kind regards,Esteban
On Mon, 2017-11-13 at 21:29 +0100, Tengfei Chang wrote:
> Hi Esteban,
> I replied inline :-)
> 
> On Mon, Nov 13, 2017 at 8:19 PM, Esteban Municio <esteban.municio@uan
> twerpen.be> wrote:
> > Hi,
> > 
> > After reading the draft I have some small comments.
> > 
> > When requesting the first 6P ADD, the CellsOptions are
> > TX=1,RX=1,SHARED=1 which means that the cell is bidirectional.
> > Is there any way to prevent that future EBs and DIOs are sent in
> > this shared cell? Or the implementation ensures that they will only
> > be sent in ts=0,ch=0?
> 
> In this case the implementation will ensure that but through the
> neighbor associated to the cell. For the reserved cell, it associated
> to its parent/child address. Only a packet send to/received from
> parent/child are allow to send on the cell. 
> > When allocating/deallocating cells due to parent switching, is this
> > initial bidirectional cell allocated with the same CellsOptions in
> > the new parent as well?
> 
> The 6P transaction for allocation to new parent will happen on the
> minimal cell. After new cells are reserved to new parent, the 6P
> transaction for deallocation to the old parent will happen on the
> previous installed cell to the old parent. After this process, there
> will be no cell to it's old parent. 
> > Then if I understand well the node should always have at least one
> > cell with TX=1,RX=1,SHARED=1 with its preferred parent 
> 
> Correct! 
> > and has to "re-allocate" it parallely
> > in a different 6P message than the used for the dedicated cells
> > with TX=1,RX=0,SHARED=0. Am I right?
> 
> The cell managed by MSF are all TX=1 RX=1 and SHARED=1 type of cell
> but associated to a dedicated neighbor. (It's a dedicated cell since
> it only send/received packet from a dedicated address) 
> > When the nodes start sending DAOs? If it is at Step 6, it could be
> > interesting mention it there.
> > Are the DAOs allowed to be sent in the minimal cell?
> 
> No, the DAO is send through the scheduled dedicated cell.  Hence even
> a node get a RANK through the DIOs it received, it can't send any
> packet except 6P until it has a dedicated cell installed.The DAOs and
> all other packets should be sent after a node reach to the End State.
> > And maybe a bit off-topic,
> > Is it possible for a node to have TX cells to neighbours that are
> > not currently its preferent parent?
> 
> For MSF, the node only schedule the TX|RX|SHARED cell to preferent
> parent. The TX cell or schedule non-preferent parent are out of scope
> of MSF.It could be another SF though. 
> > Kind regards,
> > Esteban 
> 
> Let me know if you have any future issue or suggestion on MSF. Thank
> you for asking! They are good questions. I hope I answered clearly!
> Tengfei
> > -- 
> > Esteban Municio
> > IDLab - Dept. Mathematics and Computer Science 
> > University of Antwerp - imec
> > Middelheimlaan 1 , 2020 Antwerp, Belgium 
> > Office M.G.322 
> > 
> > 
> > _______________________________________________
> > 
> > 6tisch mailing list
> > 
> > [email protected]
> > 
> > https://www.ietf.org/mailman/listinfo/6tisch
> > 
> > 
> 
> 
> -- 
> Chang Tengfei,
> Pre-Postdoctoral Research Engineer, Inria
> 
-- 

Esteban Municio

IDLab - Dept. Mathematics and Computer Science 

University of Antwerp - imec

Middelheimlaan 1 , 2020 Antwerp, Belgium 

Office M.G.322 


Attachment: signature.asc
Description: This is a digitally signed message part

_______________________________________________
6tisch mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6tisch

Reply via email to