[ns] MobiCom'98 paper results

[Wang Qiang <engp0442@nus.edu.sg>]

Hello ns users,

I have a question regarding the simulation results presented in CMU's
MobiCom'98 paper "A performance comparison of multi-hop wireless ad hoc
network routing protocols".  I'm trying run the simulation for DSDV-SQ, DSR,
AODV-LL and TORA as mentioned in the paper and see if the results could be
reproduced.
I'm using ns-allinone-2.1b7 package installed on RedHat Linux 7.0.  So far
the results for DSDV-SQ and DSR are close to those presented in the paper.
However, I CAN'T get the same output for AODV-LL protocol!  I set the
following constants as listed in the paper:

#define AODV_LINK_LAYER_DETECTION

#define MY_ROUTE_TIMEOUT  	600
#define ACTIVE_ROUTE_TIMEOUT 	300
#define REV_ROUTE_LIFE		3
#define BCAST_ID_SAVE		3

#define NETWORK_DIAMETER	50

#define NODE_TRAVERSAL_TIME	0.04

#define HELLO_INTERVAL		1
#define ALLOWED_HELLO_LOSS	3
#define BAD_LINK_LIFETIME	3

#define RREP_WAIT_TIME		6.0
#define LOCAL_REPAIR_WAIT_TIME 0.15
#define MAX_RREQ_TIMEOUT	10.0

#define TTL_START			1
#define TTL_THRESHOLD		7
#define TTL_INCREMENT		2

#define MaxHelloInterval 	( 1.25 * HELLO_INTERVAL )
#define MinHelloInterval 	( 0.75 * HELLO_INTERVAL )

But after a few runs using the same scenarios files and movement pattern
files as used for DSDV-SQ and DSR, at 20m/s max. speed, 0s pause time and 20
CBR sources, the AODV_LL packet delivery ratio value is around 97.2%, BUT
the routing overhead value insists to be around 55489.5, which is an order
of magnitude less than the value of over 120,000 presented in the paper!!!
Even after I changed MY_ROUTE_TIMEOUT to 60 and ACTIVE_ROUTE_TIMEOUT to 50,
the outcome remain the same!  Could someone please give me some help?  How
should I set those parameters to produce the same settings as in the
MobiCom'98 paper?
Your help is sincerely appreciated!

Wang Qiang