[ndnSIM] Number of requests / interests expressed by ConsumerZipfMandelbrot
jpbaugh at umich.edu
Sat Jun 17 20:04:08 PDT 2017
1. Does anyone have suggestions regarding enabling a finite number of
content, but a more realistic and randomized request pattern as described
in my previous e-mail?
2. If this type of behavior is not currently implemented, are there any
suggestions about implementation of a custom subclass of Consumer or its
subclasses? E.g., does "NS_OBJECT_ENSURE_REGISTERED" ensure that when
using the Consumer Helper, that the class is recognized and can be
installed on the nodes?
Thanks a ton,
On Sat, Jun 17, 2017 at 4:43 PM, John Baugh <jpbaugh at umich.edu> wrote:
> Greetings again,
> I think maybe I need to rephrase my issue/question(s):
> - It seems that no matter which of the built-in Consumer classes that
> I use, the first content object requested gets the most frequent requests,
> then the second, then the third, etc.
> - Are there any Consumer classes and/or setting combinations that will
> lend to a more random set of requests from various consumers?
> - E.g., Consumer1 might request content object %FE%04 more
> frequently than 03 or 01
> - E.g., Consumer2 might request content object %FE%05 most
> frequently, then 03, then 01, then 02, etc.
> I just want something a little more realistic.
> Do I need to create a custom Consumer subclass for this behavior or are
> there settings that will allow for this. I found the
> SetAttribute("Random", ....) doesn't really make anything truly random, so
> I suspect maybe there's another setting I'm missing.
> On Sat, Jun 17, 2017 at 4:26 AM, John Baugh <jpbaugh at umich.edu> wrote:
>> Greetings ndnSIM friends,
>> I was under the impression that ZipfMandelbrot was going to be fairly
>> random as to *how many* of each content object a consumer requests. I set
>> the number of contents to 5 just to get a basic picture of what is going on.
>> I have noticed that content %FE%01 is always requested many more times
>> than %FE%02, which is requested more than %FE%03 and so on...
>> Aren't there supposed to be some consumers arguably that should be
>> requesting these at random? So sometimes 01 is most popular, sometimes 03,
>> And yes, I've tried it with the Random attribute (using SetAttribute) set
>> to none, exponential, and uniform. There isn't a significant difference as
>> far as *which* is requested most frequently.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the ndnSIM