Re: [manet] Experiment with 2000 nodes with LOAD

Thanks Thierry for your comments, this is useful.

That being said, let's bear in mind that one deployment is not sufficient to draw performance conclusion. This is especially true with reactive protocol
where the control traffic is driven by the user traffic, by contrast with reactive routing where the topology is built regardless of the user traffic profile.
But let's discuss about it once we have the new revisions of the drafts, see what the options are, …

ROLL co-chair hat on, if Load-ng deals with LLNs, all I am asking is to make sure that ROLL gets involved considering the expertise of the WG on these
environments (as you rightfully pointed out, AMI PLC are clearly a very constrained LLNs).

Best Regards.

JP.

On Oct 17, 2012, at 6:00 PM, Thierry LYS wrote:

Hi Cédric Chauvenet,

As mentionned in Lavenu's previous mail we have rolled-out a 2000 G3 PLC  (an internationally standardized PLC technology) node trial and obtained excellent performances :

  *   daily collection success rate of 98 %
  *   dicovery rate of 99,7 % (the remaining meters is facing in majority HW problems)

It is more than a year and a half that this experiment runs !

...and believe me, PLC is a very harsh media and routing is playing a critical role here. We see along the day a very stable communication rate even in duty hours (around 8 AM and 6-9 PM)
This experiment proves that LOAD is a good protocol for large PLC networks characterized by low bandwidth, unstable links and link asymmetry.
We use our experience to improve it and that's how LOADng has come up !
Finally, PLC media is probably not so far from mobile networks since attenuation and noise is changing with time !

Best regards,

Thierry Lys
ERDF
<Mail Attachment.gif>    Thierry Lys
Direction Comptage / Metering Division
Building Crysalis - 345 Avenue Georges Clémenceau - 92000 Nanterre - France
Tel : +33 (0)1 81 97 67 77

Hi Cédric, (Hope people will follow which cedric is talking !)

Le 16 oct. 2012 à 18:58, Cedric-2 LAVENU a écrit :

Dear Cédric,

I'm not sure all this discussions really make sense :

> An LLN network is definitely a type of MANET

As JP pointed out, if LLNs challenges can be addressed  in MANET, why would have ROLL being created ?
I think that a protocol intend to LLNs, or if LLNs are included in the scope should be reviewed by the ROLL working group , as it is the place dedicated by the IETF.

Does that makes sense ?

according to what Adrian said (he has been quoted several times in the past mails). One of the fileds LOADng is intended to be used are AMI PLC networks with low bandwidth (few kbps in the harshest environments), but can be extensible to other types of MANETs.

And regarding your comment about experience with LOADng :

> LOADng is a protocol for which several running implementations exist and interoperate as shown in draft : http://tools.ietf.org/html/draft-lavenu-lln-loadng-interoperability-report-02

yes it has been discussed, and explained that the goal of these test were focused on validating the protocol behavior, not the performance.

> In addition, LOAD (the previous version) was successfully run in a 2000 PLC node trial.

Very nice !
Would you mind to share some details on this ?

Cédric.

I think that all the facts are on the table to say that LOAD would be suitable for MANETs (LLNs being a subset of MANETs). In addition field and lab experience does exist and demonstrated that LOADng can be very efficient.

Regards,
Cédric _______________________________________________
manet mailing list
manet@ietf.org<mailto:manet@ietf.org>
https://www.ietf.org/mailman/listinfo/manet