Running TCP protocols in Wireless mesh networks in Ns-3

905 views
Skip to first unread message

mark_ga...@hotmail.com

unread,
Sep 6, 2011, 9:02:00 AM9/6/11
to ns-3-users
Hello everyone,

I'm new to Ns-3 and I've just finished going through the tutorials and
NS-3 manuals and I'm about to begin to attempt to build a wireless
mesh network. What i wish to do is have a group of static nodes with
one gateway to the Internet and one or two mobile client nodes.
Ultimately my goal is to look at the throughput of TCP NewReno and
other transport layer protocol implementations in-order to see how
well they differentiate between packet losses caused by congestion and
packet loss caused by contention, mobility etc in order to see if
other transport protocols work better in such a way has to not
incorrectly invoke their congestion mechanisms when congestion is not
a factor so as to maintain a high a level of throughput as possible.

What i wish to know is if anyone could give me some tips on how i
might begin to put such a thing together (what steps i would need to
do). I will begin to post the code as i work through this but some
help at the beginning in order to stop me heading down blind alley
ways or to avoid common errors and issues would be very much
appreciated.

Thank you
mark

Lalith Suresh

unread,
Sep 7, 2011, 9:01:46 AM9/7/11
to ns-3-...@googlegroups.com
Hello,

Have a look at the src/mesh/examples/mesh.cc example script. You can then begin modifying it bit by bit to reflect your simulation scenario (like replacing UDP pings with a TCP application, topology, mobility models etc.). The flow monitor is the best way to monitor end-to-end statistics like throughput.
 
Thank you
mark

--
You received this message because you are subscribed to the Google Groups "ns-3-users" group.
To post to this group, send email to ns-3-...@googlegroups.com.
To unsubscribe from this group, send email to ns-3-users+...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/ns-3-users?hl=en.




--
Lalith Suresh

mark_ga...@hotmail.com

unread,
Sep 7, 2011, 1:27:58 PM9/7/11
to ns-3-users
Thanks for your advice, I'm working to understand the mesh example now
and I have a question, in the diagram the nodes that are the ping sink
and ping source node(2) and node(6)? I am looking to attach a wired
gateway and client node and send all traffic from the client node to
the wired gateway. Hopefully I should have something to show in the
next few days if I can get understand it all and add what i need.

* m_xSize * step
* |<--------->|
* step
* |<--->|
* * --- * --- * <---Ping sink _
* | \ | / | ^
* | \ | / | |
* * --- * --- * m_ySize * step |
* | / | \ | |
* | / | \ | |
* * --- * --- * =- _
* ^ Ping source



Mark

On Sep 7, 2:01 pm, Lalith Suresh <suresh.lal...@gmail.com> wrote:
> Hello,
>
> On Tue, Sep 6, 2011 at 3:02 PM, mark_gallagh...@hotmail.com <

Raymond Zhang

unread,
Sep 7, 2011, 1:44:32 PM9/7/11
to ns-3-...@googlegroups.com
Hi, Mark
I am also working on the mesh example, but I have a different
question. Maybe you can help me. My task is to let two nodes on two
physical machine can talk to each other.(like send a ping packet and
response). The two machines are connected via wired cable(ethernet). A
mesh protocol(routing protocol) should be applied to both two nodes so
that the two nodes can discover each other by mesh protocol then they
can send packets to each other.

My current hurdle is that, I found the mesh-related components in NS3
are closely related to wifi-related stuffs. Is there a method that I
can port the mesh device to a ethernet interface(eth0) so that it can
talk to real remote device and still, the upper level mesh routing
protocol is still kept?

Thank you

--
Yours sincerely
Raymond(Lei Zhang)
Ph.D Student in Computer Science,
Georgia State University
Email: wonde...@gmail.com,
          lzha...@student.gsu.edu

mark_ga...@hotmail.com

unread,
Sep 7, 2011, 1:59:52 PM9/7/11
to ns-3-users
I do know a fair amount about networking but I have only just started
attempting to implement the code in NS-3 and my programming skills are
very rusty so i cant give you much help yet. From what I've I would
think that it would be very difficult to separate the two since it
appears that the point of the mesh routing protocol (I'm guessing you
mean HWMP) is to work over a wireless mesh, neither of which you seem
to be creating. Just out of interest why are you wishing to use a mesh
routing protocol if you have 2 wired links?

On Sep 7, 6:44 pm, Raymond Zhang <wonders...@gmail.com> wrote:
> Hi, Mark
>     I am also working on the mesh example, but I have a different
> question. Maybe you can help me.  My task is to let two nodes on two
> physical machine can talk to each other.(like send a ping packet and
> response). The two machines are connected via wired cable(ethernet). A
> mesh protocol(routing protocol) should be applied to both two nodes so
> that the two nodes can discover each other by mesh protocol then they
> can send packets to each other.
>
> My current hurdle is that, I found the mesh-related components in NS3
> are closely related to wifi-related stuffs. Is there a method that I
> can port the mesh device to a ethernet interface(eth0) so that it can
> talk to real remote device and still, the upper level mesh routing
> protocol is still kept?
>
> Thank you
>
> On Wed, Sep 7, 2011 at 1:27 PM, mark_gallagh...@hotmail.com
> > For more options, visit this group athttp://groups.google.com/group/ns-3-users?hl=en.
>
> --
> Yours sincerely
> Raymond(Lei Zhang)
> Ph.D Student in Computer Science,
> Georgia State University
> Email: wonders...@gmail.com,
>           lzhan...@student.gsu.edu

Anand Hegde

unread,
Oct 29, 2012, 9:12:45 AM10/29/12
to ns-3-...@googlegroups.com
Hi,

I am a CS student from NITK, India. I am doing research in wireless data centers. I was wondering if you could share your ns-3 simulation code of your wireless mesh simulation. I have just started learning ns-3 and I am at the same point you were a year ago, just starting out. I would like to simulate channel assignment in wireless mesh networks.

Regards,
Anand Hegde.
Reply all
Reply to author
Forward
0 new messages