[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [ns] AODV in a wired-cum-wireless environment




Hi,

   I got the same problem with the combination of Mobile Ip and AODV .

   I think one possible reason is that AODV doesn't handle broadcast pkt
right now. I find only AODV its own broadcast pkts are handled in
aodv.cc, but for broadcast pkts it recvs, AODV just drops them.
Since the beacon sent by HA or FA  is relayed to the binded routing
agent, it will get dropped. So in the wired-cum-wireless example, using
AODV won't have the similar result as DSDV, since the MH can't hear the 
advertisement from HA or FA.

  That's my opinion after some simulation and modification of AODV code.
don't know if it's the real reason. 

  Looking forward to more discussion of AODV here, cause i'm also doing
some research on it now. 

  Thanks :)

Yuan

Joe Macker wrote:
> 
> As one possibility of what could be going wrong, you could check the data
> type of your test traffic.  As written, I noticed, the "on-demand" AODV
> code allows packets to be dropped for packets not defined as DATA PACKETS
> in packet.h. In my case, this did not include at present some traffic
> agents that I had created (e.g., new UDP transport agents) and other
> preexistent agents I was using (e.g., REAL_AUDIO). If your traffic agent(s)
> are not defined as such you could get the behavior you described and you
> may need to modify packet.h to define traffic of interest as DATA_PACKET types.
> 
> -Joe
> 
> At 03:14 PM 7/5/2001 +0200, G�ran Nylen wrote:
> >Hi!
> >
> >I'm currently doing my master's thesis and along with that I'm updating the
> >AODV implementation to draft version 8 using the latest NS-2.1b8a using Linux.
> >
> >I am interested in using AODV in a wired-cum-wireless environment. An example
> >of this resides in the directory "ns-root/tcl/ex/wired-cum-wireless-sim.tcl".
> >This simulation features the DSDV protocol in a wired-cum-wireless
> >environment in a very illustrative way.
> >
> >Now, chaning
> >   set opt(adhocRouting)   DSDV
> >to
> >   set opt(adhocRouting)   AODV
> >
> >would accoringly to all documentation and logics, enable the AODV protocol
> >instead of the DSDV protocol, and this is certainly the case.
> >
> >However, when AODV is used the simulation fails horribly. The nodes doesn't
> >seem to connect to each other and the tracefile barely contains any
> >traceinformation at all (Using DSDV -> tracefile size=11 Mb, using AODV ->
> >tracefile size=3.7k)!
> >
> >Can someone *please* help me out here? Is there something special about the
> >DSDV implementation which isn't covered in the AODV implementation?
> >(I have tried both the AODV implementation which comes bundled with the
> >NS-2.1b8a package, and the Felix Ko's port of Mahesh Marina's AODV
> >implementation, both ending up with the same result).
> >
> >ANY hints are very welcome!
> >
> >Best Regards,
> >G�ran Nylen
> >www.ipunplugged.com