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

RE: [manet] Post processing simulation data



To being the discussion on probably a touchy subject, I believe simulation
is far too overused.  Real experimentation is the only way to attain true
values for the variables being studied.  Simulation is good for working out
the initial bugs in the design, but final testing and design verification is
too often relegated to simulation as well.  Remember, all that really
matters is what works in the real world anyway.

Simulation also has a tendency to report the same results over and over
again.  I'd venture to say that 80% of the output we collect is garbage,
i.e.. regurgitated simulation results that either have little bearing on
reality or at least duplicated and/or foreseen results that have little
bearing on the research at hand.  I ask, what are we really learning if
there are 1000 identical AODV simulations running on ns-2 on 1000 different
workstations?  Answer, nothing.  What I'd really like to see are 1000 actual
AODV clusters in day to day use.  Or at the very least a 1000 DIFFERENT
simulations coded by different individuals on 1000 different workstations.
Such techniques would improve our colected results, but ultimately only real
world experimentation would give us the all the answers.  

Only once we move out of the world of endless simulation and back into the
real world of implementation, would we really begin to learn the limitations
of our protocols and our simulations and consequently improve the work that
we do.






-----Original Message-----
From: manet-admin@ietf.org [mailto:manet-admin@ietf.org]On Behalf Of D
Kothris, Electrical & Electronic Engineering
Sent: Monday, January 26, 2004 9:03 AM
To: manet@ietf.org
Subject: [manet] Post processing simulation data


Hi,

I am sure that to all of those involved in MANET research, post processing 
of simulation data is becoming increasingly complex, tedious and error 
prone. This is mainly due to the shear amount of data generated and the 
multiplicity of data types involved.

I would like to probe the community in order to share information about 
possible file/data management and manipulation tools  (scripting languages, 
specialized tools) that can make our lives easier, and furthermore to 
stimulate thoughts and discussion on how to deal with this enlarging 
problem at both ends > [simulation tools and post processing tools]

Cheers

Dimos



_______________________________________________________________________
***********************************************************************
Dimosthenis Kothris	      Centre for Communications Research,
			      University of Bristol,
			      Network & Protocols group

Address:		      Contact numbers: 	
_______________________________________________________________________
Network & Protocols
Center for Communcations research
Merchant Venturers Building         Tel   :            +44 117 903 1104
Woodland Road                       Fax   :            +44 117 954 5206
Bristol BS8 1UB                     Mobile:           +44 7779 40 48 42
United Kingdom                      Email :     d.kothris@bristol.ac.uk
***********************************************************************

_______________________________________________
manet mailing list
manet@ietf.org
https://www1.ietf.org/mailman/listinfo/manet

_______________________________________________
manet mailing list
manet@ietf.org
https://www1.ietf.org/mailman/listinfo/manet