Re: [manet] review of draft-dearlove-manet-olsrv2-rmpr-optimization-00

[Jiazi Yi <[e-mail]>]

And one nit:

In the reference, we have

    [RFC6966-to-be]
               Clausen, T., Dearlove, C., and P. Jacquet, "Rationale for
               the Use of Link Metrics in the Optimized Link State
               Routing Protocol Version 2 (OLSRv2)", RFC 6966, TBD 2013.

I suppose the name of the draft is

   Link Metrics for the Mobile Ad Hoc Network (MANET) Routing Protocol 
OLSRv2 - Rationale

http://tools.ietf.org/html/draft-ietf-manet-olsrv2-metrics-rationale-04 ?

Jiazi

> Jiazi Yi <mailto:[e-mail]>
> 11 Nov 2013 12:12
> Hi,
>
> I had a review of draft-dearlove-manet-olsrv2-rmpr-optimization-00. 
> Here are some comments (or some of them, I would rather call questions :)
>
> 1. In Section 4: Routing MPR Selection
> ...
> then x_1 to x_n may be removed from the set of routing MPRs, if selected.
> ...
>
> The first note: I think we should use RFC2119 term "MAY" here?
> In fact, this is the part that confuses me most. As the only "action" 
> described in the draft, the use of "may" without further explanation 
> when this action should be taken or not is too vague. In the end of 
> the section, another "RECOMMENDED" is used. Therefore, if I put them 
> together, it is something like "It is RECOMMENDED that one MAY do foo, 
> bar..."
>
> 2.  If x_1, x_2 ... x_n are the largest set with corresponding y_1, 
> y_2, ... y_n. Should we check all the subsets {x_i} of x_1, ... x_n 
> with the criteria
>
> d1(x_0) + d2(x_0,y_1) + ... + d2(x_m-1,y_m) < d1(x_m) for m belong to 
> subset of x_1, ... x_n?
>
> If so, we probably will have a lot combinations to check in a dense 
> network. If not, the criteria seems to be too harsh to be actually 
> "triggered"?
>
>
> 3. Should this be applied to all types of metrics? A potential affects 
> of this draft is that, we can probably have more "hops" around the 
> neighborhood of the original router, i.e., there is higher possibility 
> of interference. In some of metrics, the interference is not much 
> considered (because it is hard to measure). Will it be a problem if we 
> introduce more transmission in some scenarios?
>
> 4. We have a "Willingness" value to "represents the router's 
> willingness to be selected as an MPR." Should this need to be 
> considered also? What about if x_0 has low willingness, while the 
> others x_1, ... x_n have higher willingness?
>
> 5. In some scenarios, we probably want to have certain redundancy in 
> MPR in some lossy scenarios?
>
> sincerely
>
> Jiazi
>
>
>