Re: [6lowpan] Fundamental concerns about 6lowpan ND

["Colin O'Flynn" <coflynn at newae.com>]

Hello,

> But if you do that then routers may need to answer on  
> behalf of hosts as those hosts may be sleeping asynchronously to  
> eachother if the router is e.g. powered.

So those arguing for 4861, is doing a proxy NS answer acceptable?

Regards,

  -Colin

-----Original Message-----
From: Zach Shelby [mailto:zach at sensinode.com] 
Sent: October 14, 2009 7:57 AM
To: Colin O'Flynn
Cc: 'Jonathan Hui'; '6lowpan'
Subject: Re: [6lowpan] Fundamental concerns about 6lowpan ND


On Oct 14, 2009, at 1:27 , Colin O'Flynn wrote:

> Hi Jonathan,
>
> I agree with you there - these nodes would *not* be participating in  
> real IP
> or mesh forwarding.

Right, not all nodes are routers. We do have the concept of a host,  
and it is a useful one also in LoWPANs.

>
> However the ND process needs to function with these nodes. If we  
> rely on
> 4861, these nodes which are going to miss packets must have a way to  
> defend
> their address!

Jonathan is assuming the link provides a service guaranteeing nodes  
within radio range can communicate within some bounded latency using  
duty-cycling or other techniques. This is an assumption you may be  
able to make. But if you do that then routers may need to answer on  
behalf of hosts as those hosts may be sleeping asynchronously to  
eachother if the router is e.g. powered.

It is safe to assume that there is synchronization on the host-router  
and router-router interfaces - OK. It is not safe to assume  
synchronization between peer hosts attached to the same router. Nor  
between different sections of a LoWPAN. No, we are not trying to solve  
DTN.