Re: [5gangip] 6G White Paper

[Alexandre Petrescu <[e-mail]>]

Le 03/07/2020 à 17:14, Behcet Sarikaya a écrit :
> Hi all, Please check this out:
> 
> http://jultika.oulu.fi/files/isbn9789526226842.pdf

Thanks for the pointer.  It does refer to many concepts from IETF, which
is excellent.  I too will refer to it when people ask me what is 6G.  So 
there are ideas to think about.

But I would also like to comment on it.

I think 6G Flagship needs input from experts when publishing this text.
  It might be that their statements are too much away from reality, and
there might be a need to be more humble when making such statements,
especially if these statements are public in nature, as the above paper
seems to be.

It is a question of Ethics.

The paper says:
> For 6G services, the fundamentals of the IP layer may need to be 
> entirely redesigned.

I wonder whether the authors have conscience of what it means to
re-design the IP layer, entirely.  It sounds pompous, if I write that
correctly.

Because there might be some difficulty in proposing IP re-designs
through the 6MAN WG which maintains the current and future version of
IP.  I know this from personal experience of trying to propose even
slight changes such as relaxing a 64bit bound rule in 6MAN WG.

I think 6G could start by analyzing why simple IPv6 is still absent from
5G (GTP carries IPv6 flows in UDP/IPv4, 64share is a non-std RFC),
despite the talk of IPv6 since the 3rd generation.  That analysis might
lead to some sensible measures, if performed following certain requirements.

The requirements of analysis of why IPv6 is absent from 5G would be
somewhere among following:
- ask the implementers.
- get feedback from the experts.
- make public some example packet dumps from deployments.
- ask the end users.
- ask the operators.
- ask the association of end users.
- ask the people who oppose 5G for health reasons.

These requirements will lead to many solutions, among which the use of
IP and IPv6.

> As it is known, the current Internet Protocol (IP) is designed based 
> on two basic concepts: statistical multiplexing and best-effort 
> forwarding.

I am 100% sure IP does not do statistical multiplexing.  Otherwise I
would have seen an implementation of some statistics function like a
Poisson distribution somewhere in the core IP stack.

Maybe instead of multiplexing the authors wanted to say bidirectional
transmission of data, as opposed to single direction.  But multiplexing?

Maybe they meant not just simplex, not just duplex, but a universal
3D-like multiplexing.  But multiplexing probably means nothing for IP(?)

If one wants to mention a basic concept of IP that is in many IP stacks,
especially in the core IP routers, it is a longest prefix match
algorithm.  It has no statistics but works more like Dirac functions: 0
or 1.

> Inter-networking is evolving away from the current single public 
> Internet due to a combination of factors such as data-concentration 
> in the multi-access edges, centralization of data in global public 
> clouds, and the onset of global broadband satellite networks.

There is a draft about this, I think.  It is about limited domains.

But the limited domains concept does not challenge the Inter-networking
per se.  The Inter-networking is still there, and Internet will still be
there.

Alex

> 
> 
> Regards,
> 
> Behcet
> 
> 
> 
> _______________________________________________ 5gangip mailing list
>  [e-mail] https://www.ietf.org/mailman/listinfo/5gangip
>