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

Re: [ns] Wireless + Exponential/Pareto Traffic Generator doubt



On Monday 02 July 2001 11:15, you wrote:
> Hi Brian,
> Thanks for the help and information. In case of wired
> scenario all the traffic generators seem to work fine.
> [ ref. tg.tcl in the tcl/ex directory]. So as you have
> said this is the problem with the wireless model only
> right ?

I think the problem is with tracing.  The generators are generating 
data; however the traffic generators are changing the packet types.

Your transport agent is an UDP agent.  The agent should set (at some 
level) the packet type to be PT_UDP.  The traffic generator seems to 
be overriding this setting.  The trace functions don't know what to 
do with packets of type PT_EXP, so they give up - specifically 
CMUTrace::format() doesn't know what to do.

The code below, from CMUTrace::format seems to be the cause.

default:
  format_ip(p, offset);
  offset = strlen(pt_->buffer());
  switch(ch->ptype()) {
    case PT_AODV:
      format_aodv(p, offset);
      break;
    case PT_TORA:
      format_tora(p, offset);
      break;
    case PT_IMEP:
      format_imep(p, offset);
      break;
    case PT_DSR:
      format_dsr(p, offset);
      break;
    case PT_MESSAGE:
    case PT_UDP:
      format_msg(p, offset);
      break;
    case PT_TCP:
    case PT_ACK:
      format_tcp(p, offset);
      break;
    case PT_CBR:
      format_rtp(p, offset);
      break;
    case PT_DIFF:
      break;
    case PT_GAF:
      break;
    default:
      fprintf(stderr, "%s - invalid packet type (%s).\n",
                      __PRETTY_FUNCTION__,
                      packet_info.name(ch->ptype()));
      exit(1);

> I could not understand the meaning of changing source
> in my code that trace would understand... Is it that I
> need to use for eg. CBR instead of Expo/Pareto as
> thats supported in trace...

When I talk about changing source code, I'm talking about changing 
the C/C++ source code and rebuilding ns.  The script is probably 
fine.  Does the script work correctly if you turn off tracing?

> Is it true that in case of wireless model Expo and
> Pareto were no tested before ?

I don't know.  I've been working with TCP Vegas and TCP Reno in wired 
settings exclusively.

-- 
Brian Lee Bowers	|	RADIANT Team (Summer Intern)
[email protected]	|	Los Alamos National Laboratory