On 10/22/2013 08:59 AM, Ramon Hofer wrote:
>
> Hi all
>
> I'm writing my master thesis and would like to use NS to test a
> protocol which uses SCTP. It is about routing various hybrid subnets
> including Ethernet, Fiber, WiFi, ZigBee, PLC, GPRS and sending data
> with SCTP.
>
> According to your wiki, SCTP is not yet implemented in NS-3 [1]. But I
> found in the archive of this mailing list a thread from last year where
> somebody was interested in implementing SCTP for NS-3 [2].
>
> According to this I should use ns-2, right?
>
> I will write a communication framework in C++, which I think would be
> better to use in ns-3?
>
> But I could probably find more ready to use modules for ns-2?
>
>

Ramon, I think arguments could be made for either ns-2 or ns-3, in the 
case of SCTP.

The group at the University of Delaware put a lot of effort into 
developing the ns-2 model, over the course of several years, so if the 
emphasis is really on studying the finer points of SCTP, perhaps you 
will find that code useful.

For ns-3, there is no code available yet, but it appears that a group 
may post some code for SCTP in ns-3 shortly; please review this thread:
https://groups.google.com/forum/#!topic/ns-3-users/SfJ8cLXoXhA

There is also the ns-3 DCE kernel code framework; since SCTP is in the 
Linux kernel, one might be able to reuse this code (although I'm not 
sure whether in your use case it would be suitable).  I checked with 
Hajime Tazaki about whether SCTP was supported yet for Linux kernel DCE, 
and he replied "not yet" and that it would probably be a few weeks of 
work to support CONFIG_IP_SCTP.

In general, all things being somewhat equal, you will probably find more 
help on the lists when using ns-3, since it is actively maintained.

You mentioned needs for a number of other models:  Ethernet, Fiber, 
WiFi, ZigBee, PLC, GPRS.  I don't think either ns-2 or ns-3 has all of 
these.

Given all of this, probably it makes sense to try both out (existing 
ns-2, and ns-3 once the SCTP code becomes available).

- Tom


Reply via email to