[ns] Question on Error Model

[Chan Kwang Mien <Kwang-Mien.Chan.GP@icn.siemens.de>]

	hi, 
	 
	     I have some queries on using Error Model in NS.

	a)      When using the Uniform distribution for the Error Model in
NS, can we specify the unit as byte ?
	        i.e.

				    	set myEM  [new ErrorModel/Uniform
0.2 byte]
	 
		  When i looked through the ns-errmodel.tcl, there is only
definition for the unit packet and no definition for the unit byte.
	   
	    	  If the unit is in pkt, what does 0.2 means ? does it mean
that 20% of the packets sent by the sender would be dropped ?

	 
	 b.   Does the Uniform ErrorModel suppose to distribute the error
evenly ?

		 From the tracefile that i obtained, it does not seem to
reflect the right amount of errors that i expected. 

	       For e.g.  I use this :

				    	set myEM  [new ErrorModel/Uniform
0.2 pkt]
	 
	      however, from the tracefile, i do not seem to see that 20% of
the packets are dropped. The tracefile shows that about 13% of packets 
	are dropped.  Is that correct ?


	 c.    Does anyone know what kind of error model i should use to
simulate packet corruption due to bit error over a GSM link ?


	    Thank you.


	rgds,
	Kwang Mien