[ndnSIM] RTTEstimator

Alex Afanasyev alexander.afanasyev at ucla.edu
Tue Apr 16 16:58:22 PDT 2013


Thanks Cheng!

I completely agree that the way existing ndn::Consumer* apps behave is not necessary the ideal or the right way at all.  We attempted to emulate very simple TCP behavior, which is very well known not to work "too well" in wireless scenarios.

The "problem" can be solved in a couple of different ways:

1) The most straightforward way, you can write your custom app (based on the existing one) that doesn't doubles RTO as much as it does currently.  

2) Another way is to set maximum value of RTO to some small value, acceptable for you simulations (by default, max RTO is 200 seconds):
Config::SetDefault ("ns3::ndn::RttEstimator::MaxRTO", TimeValue (Seconds (4.0)));

3) Write a completely different (simple) app with fixed RTO, if it is acceptable (not sure) or different logic for RTO calculation.

---
Alex

On Apr 16, 2013, at 7:54 AM, Cheng Yi <yic at EMAIL.ARIZONA.EDU> wrote:

> Hi Gabriel,
> 
> IncreaseMultiplier is called whenever there is a NACK or an Interest timeout, which implies potential congestion in the network. The implementation of RttEstimator is based on the Karn's algorithm which is used in TCP RTT estimation. Whether it fits NDN well is debatable and still under investigation. Hope this helps.
> 
> Cheers,
> Cheng
> 
> On 4/16/2013 7:32 AM, Gabriel Marçal de Brito wrote:
>> Hi Alex.
>> 
>> I am studying the impact of named-data networking over wireless MANETs and within this work I am performing some simulations with ndnSIM.
>> 
>> I noticed an increasing average delay using high chunk request rates and digging deeper, found that RttEstimator::IncreaseMultiplier() is called whenever there is an Interest, doubling the RTT estimate.
>> 
>> My guess is that since my scenarios are pretty dynamic, some consumer nodes would be able to retrieve the desired content if they re-request it sooner than waiting for the very long ever-doubled RTT (in case of discontinuities, for exemaple).
>> 
>> Is there a reason for doubling this estimate?
>> 
>> Thanks in advance,
>> Gabriel
>> 
>> 
>> 
>> 
>> _______________________________________________
>> ndnSIM mailing list
>> ndnSIM at lists.cs.ucla.edu
>> http://www.lists.cs.ucla.edu/mailman/listinfo/ndnsim
> 
> _______________________________________________
> ndnSIM mailing list
> ndnSIM at lists.cs.ucla.edu
> http://www.lists.cs.ucla.edu/mailman/listinfo/ndnsim

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.lists.cs.ucla.edu/pipermail/ndnsim/attachments/20130416/793841fb/attachment.html>


More information about the ndnSIM mailing list