[Nfd-dev] NDN Link Layer (2.5) Retransmissions

Klaus Schneider klaus.schneider at uni-bamberg.de
Thu Feb 19 12:55:50 PST 2015


Thank you. I read the NDNLP techreport and it answers my question.

I still wonder if it wouldn't be better to do retransmissions on the 
strategy layer to exploit multiple paths.

The link layer has basically two options when faced with packet loss 
(correct me if I'm wrong about anything of the following):

1. Request the packet again
2. Announce the packet loss to upper layers (after a timeout or a 
specified number of retries)

With multiple paths the strategy layer has more options:
1. Send the Interest on the same path again
2. Send the Interest on a different path
3. Send the Interest simultaneously on multiple paths (a sort of forward 
error correction)
4. Announce the packet loss to upper layers

A good example would be burst errors. Retransmissions on the link layer 
would probably create a longer delay than switching the path on the 
strategy layer. One might find other examples where using the 
performance information of the strategy layer might be beneficial.

I would like to hear your thoughts on this.

Best regards,
Klaus



Am 19.02.2015 um 15:13 schrieb Beichuan Zhang:
>
> On Feb 19, 2015, at 6:23 AM, Klaus Schneider <klaus.schneider at uni-bamberg.de> wrote:
>
>> Hi everyone,
>>
>> while reading a redmine discussion
>> (http://redmine.named-data.net/issues/1999) I found the idea to
>> introduce a NDN link layer (2.5) to perform retransmissions below the
>> strategy layer.
>>
>>> Updated by Alex Afanasyev about 1 month ago
>>>
>>> As I remember, the way to address link problems was to introduce a
>>> limited recovery inside the link (2.5) layer. This layer will do
>>> retransmissions much more efficiently and quicker than usually coarse
>>> RTO estimation (the initial value is 1 second).
>>
>> I find this concept very interesting. However, I didn't find much about this on the web.
>>
>> Can you tell me the current state of these ideas?
>
> There is a design and implementation in NFD for the 2.5 layer, but it currently doesn’t have the retransmission part.
> http://redmine.named-data.net/projects/nfd/wiki/NDNLP-TLV. We just started the redesign of this 2.5 layer including limited loss recovery. This may take a while to get fully designed and implemented in NFD.
>
> Beichuan
>
>>
>> Thanks a lot,
>> Klaus
>>
>>
>> --
>> Klaus Schneider
>>
>> Mail: klaus.schneider at uni-bamberg.de
>> LinkedIn: https://www.linkedin.com/in/schneiderklaus
>> _______________________________________________
>> Nfd-dev mailing list
>> Nfd-dev at lists.cs.ucla.edu
>> http://www.lists.cs.ucla.edu/mailman/listinfo/nfd-dev

-- 
Klaus Schneider

Mail: klaus.schneider at uni-bamberg.de
LinkedIn: https://www.linkedin.com/in/schneiderklaus



More information about the Nfd-dev mailing list