[ns] Performance change due to change data order in packets header: Could anybody explain???
ghadaayed at yahoo.com
Mon Feb 25 02:24:36 PST 2008
I have modified AODV protocol to support other routing metric than hop-count, at the final stages of performance testing I have realized the following:
I have added new fields to RREQ and RREP packets, lets say I have added three quantities of type double (x, y, z) to the RREQ to allow intermediate nodes to calculate the new route metric. What I have realized that the protocol performance will be different for the following two cases:
-- if I have added these values as an array to the RREQ header (i.e.double new_metric),
-- if I add them as a three elements (i.e. double x, double y, double z).
Also if only one element of the array is filled (i.e. new_metric, for example) the performance will be different from having all array elements are filled.
I am measuring the total network throughput, the results are different for different RREQ format elements arrangements.
Can any body explain this to me? what is happening here?
Looking for last minute shopping deals? Find them fast with Yahoo! Search.
More information about the Ns-users