[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: sorry to push , but need to know



On Mon, 7 Jun 1999, Janssen, S.A. wrote:

> I realise this group is not responsible for answering my questions and it is
> possible that questions remain unanswered.

that's my experience... questions in HTML are more likely to remain
unread, I think, so I'll have a stab at yours since it's plaintext.


> I'd like to make a last request
> to the ns-developers about the following:
> 
> - How would you implement a protocol stack with a double IP-layer (plus
> possible double TCP or UDP layers?

I'm not sure; in general, either you write your own agents to support
disjoint TCP connections or IP tunnelling, or you look at tweaking the
existing Mobile-IP wireless work. MobileNode/MIPBS do encapsulation
and decapsulation - on wireless nodes only at present - and may
provide hints as to how a more general encapsulation/decapsulation
could be done.

This is in 15.2.2 of the May notes and documentation on page 106 (the
contents claim it's on 101, but that's because the i, ii, iii separate
numbering of the contents has been broken for a couple of months. Can
someone please fix?)

It really depends on what you're interested in simulating. You might
just add a counter to the packet indicating number of times
enc/decapsulated in the simplest case.

Double TCP or UDP layers? Disjoint communications, fine, but I've no
idea what you're talking about here. Doesn't sound like source
routing...

> I believe this must have come into the minds of the developers and otherwise
> it might be something to think about. It is also not an unrealistic protocol
> stack since GPRS, the GSM successor for data, has this double IP layer.

GPRS does? This is news to me.

AFAIK GPRS doesn't define anything involving IP at all, and you're
really thinking of mobile-ip tunnels running across GPRS...

cheers,

L.

<[email protected]>PGP<http://www.ee.surrey.ac.uk/Personal/L.Wood/>